and evolutionary ecology of movement behaviour in the...

Habitat fragmentation and evolutionary ecology of movement behaviour in the speckled wood butterfl y (Pararge aegeria L.)

Thomas MerckxPromotor: Prof. Dr. Erik Matthysen (UA) // Co-promotor: Prof. Dr. Hans Van Dyck (UA / UCL)

PROEFSCHRIF T VOORGELEGD TOT HE T BEHALEN VAN DE GR A AD VAN DOC TOR IN DE WE TENSCHAPPEN A AN DE UNIVERSITEIT ANT WERPEN TE VERDEDIGEN DOOR

Antwerpen, 2005

and evolutionary ecology of movement behaviour and evolutionary ecology of movement behaviour and evolutionary ecology

in the speckled wood butterfl y (

Thomas Merckx

UNIVERSITEIT ANTWERPEN

FACULTEIT WETENSCHAPPEN

DEPARTEMENT BIOLOGIE

Habitat fragmentation and evolutionary ecology of movement behaviour in the speckled wood butterfly (Pararge aegeria L.)

Habitat fragmentation and evolutionary ecology of movement behaviour in the speckled wood butterfly (Pararge aegeria L.)

Thomas MerckxPromotor: Prof. Dr. Erik Matthysen (UA) Co-promotor: Prof. Dr. Hans Van Dyck (UA / UCL)

PROEFSCHRIF T VOORGELEGD TOT HE T BEHALEN VAN DE GR A AD VAN DOC TOR IN DE WE TENSCHAPPEN A AN DE UNIVERSITEIT ANT WERPEN TE VERDEDIGEN DOOR

Antwerpen, 2005

UNIVERSITEIT ANTWERPEN

FACULTEIT WETENSCHAPPEN

DEPARTEMENT BIOLOGIE

Table of contents

{ 5 }

1. Introduction 3 1.1 Habitat fragmentation 4 1.1.1 A mini review 4 1.1.2 Evolutionary consequences 4 1.2 Micro-evolutionary changes in phenotype 4 1.2.1 Morphology 4 1.2.2 Behavioural phenotypes 4 1.2.3 Plasticity versus divergence 4 1.3 Butterfly movement biology 4 1.3.1 Multi-functionality of flight 4 1.3.2 Thermoregulation 4 1.3.3 Sexual differences 4 1.4 Model species: Pararge aegeria 5 1.5 Study areas 4 1.6 Aims & outline 4

Table of contents

{ 6 }

2. Landscape structure and phenotypic plasticity in flight morphology 5 in the butterfly Pararge aegeria Merckx T & Van Dyck H [submitted to Oecologia]

3. Habitat fragmentation and thermal flight budget in a woodland butterfly: an experimental assessment 6 Merckx T, Van Dongen S, Matthysen E & Van Dyck H [submitted to Functional Ecology]

4. Sex- and landscape-related differences in thermal under-limit for flight ability in a woodland butterfly 7 Merckx T, Karlsson B & Van Dyck H [submitted to Biology Letters]

5. Mate location behaviour of the butterfly Pararge aegeria in woodland and fragmented landscapes 9 Merckx T & Van Dyck H (2005) [accepted in Animal Behaviour]

6. The evolution of movements and behaviour at boundaries in different landscapes: a common arena experiment with butterflies 10 Merckx T, Van Dyck H, Karlsson B & Leimar O (2003) [Proceedings of the Royal Society of London Series B, 270, 1815-1821]

7. Landscape-related differences in the ability to find habitat between populations of a woodland butterfly 11 Merckx T & Van Dyck H [submitted to Journal of Animal Ecology]

8. General Discussion 12 8.1 Habitat fragmentation, movement behaviour and thermoregulation 12 8.2 Sexual differences 12 8.3 Landscape, conservation and variation in functional connectivity 12 I. References 13II. Summary 3III. Samenvatting 3IV. Dankwoord 3

Chapter 1 { Introduction }

{ 7 }

1}

Introduction

1.1 Habitat fragmentation

Chapter

1.1.1 a mini reviewHumans have become an evolutionary force of extraordinary influence (Palumbi 2001). The high selection pressures that result from their activities cause evolutionary chang-es that often occur on contemporary timescales, often even within decades (Kinnison & Hendry 2001, Reznick & Ghalambor 2001). For instance, human activities select for rapid changes in pathogen biology, and hence contribute to the emergence and spread of new diseases (Altizer et al. 2003). Another example of the evolutionary impact of human activity is the influence of fisheries on life history characteristics (e.g. size, maturity) of intensively exploited fish populations (Jennings et al. 1998). But, one of the most obvious effects of human activity is habitat fragmentation.

Habitat fragmentation can be viewed of as a landscape-scale process involving two co-occurring components: habitat loss and the breaking apart of habitat – or fragmenta-tion sensu stricto (Lord & Norton 1990, Fahrig 2003). The process results in a transforma-tion of a large habitat into a number of smaller patches isolated from each other by a landscape matrix unlike the original habitat (Wilcove et al. 1986). Since ancient times, man has modified landscapes and consequently caused the fragmentation of habitats. During the last century the rate of change ever has accelerated due to dramatic changes in land use (e.g. the intensification of agriculture). Nowadays, habitat fragmentation is almost ubiquitous (Wiens 1997, Hanski 1999). Habitat loss is generally considered to be one of the most important bottlenecks for biodiversity (Meffe & Carroll 1997): it is regarded as the main cause of the current mega-extinction crisis world-wide (Pimm & Raven 2000). However, other major players are global climate change and biological invasions (Thomas et al. 2001, 2004, Clark et al. 2003, Clobert et al. 2004). The dramatic, negative effects of habitat loss on biodiversity (a/o Fahrig 2002) mask in most cases the

{ 8 }


{ 9 }

}1.2 Micro-evolutionary changes in phenotype

mate change, for instance, habitat fragmentation may function as a constraint. Though global warming forces organisms to move or adapt (IPCC 2001, Noss 2001), a fragmented landscape is generally expected to limit movements and to slow down the spread of adaptive genes across the landscape, whilst small population sizes might simultaneous-ly compromise the ability of these organisms to respond adaptively to environmental change (Stockwell et al. 2003). For example, butterflies in Britain were expected to re-spond positively to climate warming, since all of them are heliothermic organisms. Yet, three-quarters of them – the habitat specialists – declined because they were envisaging highly fragmented environments (Warren et al. 2001). Nowadays, the British butterfly community is impoverished and dominated by mobile and widespread habitat general-ists (Warren et al. 2001). Moreover, this example shows that movement rate, relative to the availability of habitat, is a trait which has a particularly strong effect on survival probabilities in fragmented landscapes (Thomas 2000) (see also 1.2.1.). Of course, life history traits other than dispersal may also change as an evolutionary consequence of trade-offs due to increased habitat fragmentation. The oögenesis-flight syndrome, for in-stance, occurs as a trade-off between reproduction and mobility, because of limitations in the total resource available to be allocated to both functions. For example, allocation to the abdomen (containing reproductive organs) decreased with declining habitat area in the butterfly Plebejus argus (Thomas et al. 1998). In this context, it is important to realise that responses are highly species-specific and can be opposite for two species liv-ing in the same fragmented landscape but using a different set of resources (Van Dyck & Matthysen 1999). Nevertheless, in the next paragraph some general consequences of selection related to movement (and movement-related morphology) will be discussed.

1.2.1 morphologyMovement is a central concept in population-, evolutionary-, and conservation biol-ogy and ecology (Clobert et al. 2001, Hanski 2001). The ‘rescue-effect’ (Brown & Kodric-Brown 1977), which is associated with a relatively high mobility, states for instance that colonising individuals may rescue local populations in a fragmented landscape, which would otherwise face higher extinction risk associated with small population sizes (e.g. small effective population size, higher stochasticity for dramatic events). In the same vein, a higher movement rate may lead to recolonisations of local populations that went extinct (Thomas 1994). Mobility is a central issue for a species’ life history, and it takes different forms. Some organisms barely move, while individuals in other species move considerable distances (Dingle 1996). Dispersal is mostly viewed of as a process that consists of one or several movements. However, not every movement is important to dispersal, which is commonly considered to determine interpopulation movements and, hence, gene flow. Consequently, one often focuses on movements between the natal area

weaker effects of habitat fragmentation sensu stricto. The latter can be either positive or negative (Fahrig 2003). Moreover, theoretical studies suggest that the effects of frag-mentation sensu stricto should only become apparent below a critical threshold of habitat area in a landscape (e.g. 20-30% threshold: Andrén 1994, 1999, With & Christ 1995, Fah-rig 1997, 1998, Flather & Bevers 2002).

According to the classic theory of island biogeography (MacArthur & Wilson 1967), species richness increases with patch size and decreases with patch isolation (a/o Tho-mas & Harrison 1992, Hanski et al. 1994, Steffan-Dewenter & Tscharntke 1999, 2000, Rick-etts et al. 2001). Possible reasons for these species-area-distance relationships are coloni-sation-extinction dynamics and increased habitat diversity in large compared to small areas (MacArthur & Wilson 1967, Connor & McCoy 1979). Depending on the geometry of the fragment, edge effects may further affect habitat quality, and hence, species diver-sity and community structure (Didham et al. 1998, Thomas et al. 2001). However, due to the recent focus on spatial population dynamics (Bjornstad et al. 1999), metapopulation biology (Hanski & Gilpin 1997) and landscape ecology (Wiens et al. 1993), it has become apparent that the size and isolation of individual habitat patches alone is often insuf-ficient for predicting population persistence. A landscape integrated view may be more appropriate (Hanski & Ovaskainen 2000, Weibull et al. 2000, Fahrig 2001, Söderström et al. 2001). More specifically, the spatial arrangement of patches (Collinge & Forman 1998) and the spatial and temporal heterogeneity of the surrounding landscape are im-portant determinants of population dynamics and viability in fragmented communities (a/o Marino & Landis 1996, Gustafson 1998, Thies & Tscharntke 1999, Steffan-Dewenter et al. 2002, Schtickzelle & Baguette 2004). Landscape connectivity is an important is-sue in this respect. This species- and population specific parameter comprehends both structural and functional connectivity of habitat fragments. Structural – or geographic

– connectivity is determined by variation in the availability of physical corridors and barriers (With et al. 1997, Beier & Noss 1998, Collinge 2000, Tischendorf & Fahrig 2000, Steffan-Dewenter & Tscharntke 2002). Nonetheless, what ultimately matters is func-tional connectivity. This is achieved by dispersal, and hence depends on the scale at which organisms interact with spatial structure (Andrén 1994, Jonsen & Fahrig 1997, With et al. 1999, Tischendorf & Fahrig 2000, Weibull et al. 2000, Steffan-Dewenter et al. 2001, 2002). So, in addition to a black-and-white depiction of the landscape (struc-tural connectivity), the characteristics (e.g. hostility, resources) of the matrix as well are important for landscape connectivity (a/o Debinski & Holt 2000, Norton et al. 2000, Zschokke et al. 2000, Fahrig 2001, Ricketts 2001, D’Eon et al. 2002, Goodwin 2003).

1.1.2 evolutionary consequencesGiven the nature of habitat fragmentation (see 1.1.1.), it is straightforward to see that habitat fragmentation can cause new selection regimes. Moreover, it can both provoke and impede adaptive responses to other selection pressures. Within the context of cli-

{ 10 }


{ 11 }

als (Kareiva 1985, Hill et al. 1996, Sutcliffe et al. 1997, Thomas & Hanski 1997). On the other hand, when resources have a more scattered distribution in the landscape, and are no longer concentrated in single fragments, but neither too isolated from one each other, selection for higher mobility is expected to occur (a/o Olivieri & Gouyon 1997; Hill et al. 1999a, Van Dyck & Matthysen 1999).

Summarising, individuals may respond adaptively to habitat fragmentation by means of changing movement rates. Depending on the scale of isolation, the success of dispersers can be reduced, and hence selection against mobility is expected. However, one can also expect fragmented landscapes to favour higher mobility when necessary resources are no longer concentrated but rather spread over different fragments.

1.2.2 behavioural phenotypesPersistence of wildlife populations in the face of landscape change depends not only on adaptive morphological variation but also on behavioural responses. Several behavioural consequences of habitat fragmentation have been described. For example, activity and hunting behaviour of tawny owls (Strix aluco) is greatly affected by fragmentation: they have longer inter-perch distances and perch times in fragmented woodlands, which is explained by an altered availability of small mammals (Redpath 1995). Another example is the trade-off between foraging and anti-predatory behaviour of black-capped chicka-dees (Poecile atricapilla). In more deforested landscapes, chickadees ventured farther into the open. This altered behaviour can be explained by a constraint (increased energy stress), which forces the birds to take more risks. Therefore, it may increase mortality through greater exposure to predators (Turcotte & Desrochers 2003).

Although these examples only make up a very small selection of numerous behav-ioural responses toward fragmentation, we now want to know whether movement be-haviour, in particular, is affected by fragmentation. There exist several observations on intra-specific variation in movements among populations in different landscapes (a/o Roland et al. 2000, Tischendorf & Fahrig 2000, Ricketts 2001, Brotons et al. 2003, Mennechez et al. 2003). Such altered movement patterns may compensate for a reduced structural connectivity of the landscape: more mobile individuals may still be able to lo-cate resources, in spite of the larger distances between them. However, compensation at the population-level may be hampered when fragmentation is too severe or happens too fast, leading to spatial isolation of patches. Then, fragmentation may result in reduced movement among patches (e.g. beetles: Goodwin & Fahrig 2002 ; butterflies: Baguette et al. 2003, Schtickzelle & Baguette 2003; voles: Wolff et al. 1997) (see also 1.2.1.).

In order to know the mechanism behind the observed patterns, several authors have recently pleaded for mechanistic studies, since different patterns can be explained by different mechanistic scenarios. Dissections of the observed behaviour, using experi-mental set-ups, allow to better understand and predict how animals react to landscape elements and how they move through the landscape. Dispersal is basically a three-stage

and the area where reproduction first takes place (i.e. natal dispersal), and on move-ments between two successive breeding areas (i.e. breeding dispersal). However, this is a view typical for territorial and/or nestbound species. The distinction between these and other types of movement is not clear-cut (Clobert et al. 2001, Wiens 2001), and as a consequence some scientists contrastingly consider any movement to be dispersal, or at least relevant for dispersal (a/o Ide 2002). Anyway, dispersal can be realised in two different ways: as a by-product of routine, explorative movements (e.g. foraging, mate location), or, as special, directed movements designed for displacement (see Johnson 1969, Shreeve 1992a, Van Dyck & Baguette 2005).

Changed mobility patterns may be associated with morphological adaptations since a number of morphological traits is found to be functionally related to flight perform-ance and dispersal abilities (e.g. bees: Steffan-Dewenter & Tscharntke 1999; beetles: Niemela & Spence 1991; bugs: Palmer & Dingle 1989; butterflies: Kuussaari et al. 1996, Steffan-Dewenter & Tscharntke 1997, Berwaerts et al. 2002, Van Dyck 2003; crickets: Fairbairn & Roff 1990). Evidence for morphological change in association with frag-mentation has accumulated. The striking wing reduction in birds on oceanic islands (McNab 1994, Williamson 1981), for instance, is a clear example that habitat isolation may have evolutionary consequences related to mobility. However, the fragmentation of terrestrial habitats can also affect the morphological ‘design’ in more subtle ways. Empirical research on insects has provided interesting suggestions for different direc-tional evolutionary change in flight morphology with landscape structure. Time-series of Papilio machaon and Maculinea arion butterflies suggested that flight muscle mass de-creased with isolation (Dempster et al. 1976, Dempster 1991), whereas Calopteryx maculata damselflies and Plebejus argus, Hesperia comma and Melitaea cinxia butterflies were found to have a more ‘mobile morphology’ in fragmented landscapes (Taylor & Merriam 1995, Thomas et al. 1998, Hill et al. 1999a, Norberg & Leimar 2002). The reasoning behind these evolutionary changes in flight morphology is that costs and benefits of dispersal alter with landscape structure (a/o Olivieri & Gouyon 1997). If dispersers are not a random genetic sub-sample of populations, but if they are characterised by a more ‘costly’ mo-bile and heritable morphology, isolated fragments might evolve toward lower mobility (Thomas et al. 1998, Hill et al. 1999b, Van Dyck & Baguette 2005). Besides the cost of re-source allocation toward flight muscles instead of toward fecundity, other costs associ-ated with dispersal are the costs of time and energy which cannot be used for other activities, and the higher risk among mobile emigrants to get lost into the matrix, or to get predated. These individuals are not likely to be replaced by mobile immigrants (i.e. recolonisation) since individuals traversing the hostile matrix experience higher mor-tality (Schtickzelle & Baguette 2003). Hence, the resulting mobility pattern is achieved as a balance between selection at the local patch and selection at the metapopulation level (i.e. ‘metapopulation’ effect). Selection for a reduced emigration rate is likely to be strongest in small remnants due to relatively high proportions of emigrating individu-

{ 12 }


{ 13 }

mental variation (i.e. the reaction norm) can be under genetic control (West-Eberhard 2003). Natural selection on traits can only occur when traits are variable and heritable (i.e. genetic basis). When considering phenotypic plasticity as a trait, the variability lies within the slopes of the reaction norms. If different genotypes possess different reaction norms one defines this variability as genotype-by-environment interactions. In that way, the pattern of sensitivity to external conditions may be altered adaptively (Blanckenhorn 1991, Price et al. 2003, West-Eberhard 2003).

Both pathways can be followed at the same time. For instance, we know that many speciation events result from colonisation of new environments. Such events go to-gether with strong selection pressures but in the mean time they also directly induce changes in behaviour, morphology and physiology (i.e. phenotypic plasticity) (Price et al. 2003). Thus, phenotypic plasticity may permit recently colonised populations to survive in new environments during the first generations. In other words, phenotypic plasticity may bridge the time-gap during which individuals are not yet sufficiently adapted to the novel environment (Price et al. 2003, Yeh & Price 2004). Hence, phenotypic plastic-ity can be viewed of as a trait that facilitates adaptation. In the absence of phenotypic plasticity, the trophic radiation and subsequent speciation as observed in Old World haplochromine cichlids, for example, most probably would have been less extensive and explosive (Stauffer & Gray 2004).

Butterflies have been popular study organisms for a long time: they were and are used to study questions relating to physiology, development, behaviour, ecology, evolution, and systematics (Boggs et al. 2003). Butterfly movements have attracted an increasing interest: Ehrlich (1961) was among the firsts to perform mark-release-recapture stud-ies. Later, butterflies played a major role in the development of metapopulation theory (Hanski & Gilpin 1997). Nowadays, research on butterfly movements gradually focuses on the behavioural underpinning of movements in a landscape ecological context (Burel & Baudry 2003, Conradt et al. 2003, Schtickzelle & Baguette 2003).

1.3.1 multi-functionality of flightIn many insects, including butterflies, flight is multifunctional (Shreeve 1992a, Dud-ley 2000, Watt 2003). For instance, it is used for different activities like mate location, predator-avoidance, foraging, and finding oviposition sites. Thus, flight performance may have an important impact on realised fitness (Berwaerts & Van Dyck 2004). In order to locate receptive females, male butterflies typically ‘wait or seek’. Patrollers actively search, while perchers sit-and-wait at particular spots (Scott 1974, Wickman & Wiklund 1983, Shreeve 1987). A study on several temperate butterfly species demon-strated that flight design was related to mating system. Males of perching species had

}1.3. Butterfly movement biology

process, consisting of emigration, transience and immigration. A more thorough and mechanistic knowledge of the complete process will increase the ability to manipulate it in favour of conservation purposes (Lima & Zollner 1996, Morales & Ellner 2002, Wil-son & Thomas 2002, Schooley & Wiens 2003, Schtickzelle & Baguette 2003). Recently, behavioural aspects of dispersal have received increasing attention (a/o Jonsen & Taylor 2000a, b, Conradt et al. 2001, Bowman et al. 2002, Hanski et al. 2002). Several studies focus on behaviour at patch boundaries and effects of corridors on dispersal (a/o Haddad 1999, Ries & Debinski 2001, Schultz & Crone 2001, Berggren et al. 2002, Harris & Reed 2002), or on the effect of patch area on emigration decisions (Sutcliffe et al. 1997, Baguette et al. 2000, Schtickzelle & Baguette 2003). Studies on perceptual ranges are also of high relevance to understand how animals moving through the landscape respond to their environment (a/o Yeomans 1995, Zollner & Lima 1997, Zollner & Lima 1999, Zollner 2000, Schooley & Wiens 2003, Bonte et al. 2004). Yet, for most groups, information on dispersal behaviour is scarce. This makes it difficult to assess how isolated or hostile a specific environment is for a specific population (Tscharntke et al. 2002). Nevertheless, such information is critical for predicting effects of fragmentation on populations as shown by simulation models (Fahrig 2001).

1.2.3 plasticity versus divergenceCoping with changing environmental conditions, and hence with variable local selec-tion pressures, is a fundamental aspect of organic evolution (Pigliucci 1996). The neo-Darwinian view of adaptation maintains that populations tend to evolve characteristics that make them well suited to the particular conditions experienced over a long period of time and a small spatial scale. However, organisms experience a variety of environ-ments during their life cycle, and it is common that a given genotype can produce dif-ferent phenotypes when exposed to different environments. So, in order to cope with varying or changing environments, populations can follow two general pathways. The first one is genetic adaptation and divergence: natural selection of present genetic vari-ation results in local adaptation, at least when the amount of gene flow is not too high (e.g. Postma & van Noordwijk 2005). The second pathway is phenotypic plasticity. This option is characterised by single genotypes producing different phenotypes based on environmental cues (Agrawal 2001). Phenotypic plasticity is a measure of how different the phenotypes produced in distinct environments are from each other. In other words, it is a measure of sensitivity to external conditions. Reaction norms are functions de-scribing these phenotypes over a certain range of environmental conditions. The study of phenotypic plasticity and reaction norms has attracted much attention in modern evolutionary ecology (Schlichting 1986, Sultan 1987, West-Eberhard 1989, Scheiner 1993, West-Eberhard 2003). Most of the time, plastic responses are highly specific and are co-ordinated by an array of regulatory genes acting at different hierarchical levels (Pigliucci 1996). The degree to which a genotype responds in a plastic way to environ-

{ 14 }


{ 15 }

Butterflies have been used extensively as ‘model’ animals for the study of ther-moregulation and the co-evolution of morphology and thermoregulatory behaviour (a/o Digby 1955, Clench 1966, Wasserthal 1975, Kingsolver 1985a, b, Heinrich 1986, Dennis 1993, Rutowski et al. 1994, Srygley 1994, Kemp & Krockenberger 2002, Watt 2003). In order to fly efficiently, butterflies need to elevate the temperature of their thoracic musculature considerably above air temperature (Watt 1968, Heinrich 1993): 34-37°C in several temperate butterflies (Vielmetter 1958, Douwes 1976, Rutowski et al. 1994, Dreisig 1995). Since internal heat production by shivering is rare in butterflies, which are typi-cally heliothermic, most species select strategic microhabitats and apply specific bask-ing body orientations and postures (Casey 1988, Dennis 1993, Rutowski et al. 1994). Heat transfer is further facilitated by morphological adaptations (Dennis & Shreeve 1989, Schmitz 1994, Berwaerts et al. 2001, Berwaerts & Van Dyck 2004), such as colour pattern and amount of body fur. Fur functions as an insulator. It counteracts the convective cooling effect of cool air that moves over the body (Kingsolver & Moffat 1982, Kingsolver 1983, Watanabe & Imoto 2003). In this respect, size is known to be a crucial factor: smaller butterflies are more constrained to be on the wings during marginal weather conditions (Heinrich 1986). Melanisation of the basal wing parts increases solar absorp-tion during basking (Van Dyck & Matthysen 1998, Berwaerts et al. 2001). Trapped heat is then conducted to the nearby thorax and flight muscles. This mechanism allows dark-coloured individuals to raise their body temperature to the thermal optimum for activity faster than pale individuals (Kingsolver 1987, Van Dyck & Matthysen 1998). Conversely, pale morphs can be at advantage in environments in which overheating is likely (Ottenheim et al. 1999). As a result, for instance, basal melanism increases with altitude and latitude (Guppy 1986, Kingsolver 1983, Kingsolver & Wiernasz 1987, Ellers & Boggs 2004), since even small increases in pre-flight warm-up rates owing to mela-nisation appear to have a strong selective pressure (Watt 1968). Wing surfaces play an important role in the process of heat transfer, either by conductance of absorbed heat from the basal wing areas to the thorax (Wasserthal 1975, Kingsolver 1987), reflectance of solar radiation onto body tissues (Kingsolver 1985a) or via shielding body tissues from solar radiation (Rawlins 1980). Basically, three species-specific basking mecha-nisms are proposed for butterflies. Dorsal and lateral basking are most widespread and involve orientation of the body so that either dorsal or ventral wing surfaces are per-pendicular to the plane of incident solar radiation (Shreeve & Dennis 1992, Rutowski et al. 1994). The thorax is warmed by direct exposure to the sun, and from heat conducted from basal wing regions. A third posture involves angling the wings downward so that the distal margins are appressed to the substrate. This appression posture could be viewed of as an extension of dorsal basking most often adopted under relatively cool conditions (Kemp & Krockenberger 2002). These mechanisms imply that large areas of wing surfaces bear limited thermal consequences. In the absence of genetic correla-tions between the colour of basal versus more distal wing parts (but see Kingsolver &

larger thorax/body mass ratio, higher wing loading and more pointed wings than pa-trolling species (Wickman 1992). A similar relationship was found at the intra-specific level in the speckled wood butterfly Pararge aegeria (Van Dyck et al. 1997a). Just as is the case for mate location strategy, the need to evade predatory birds by flight affects flight patterns and flight morphology. For example, neotropical butterflies that are palatable

– and hence fly more erratically – have more of their mass allocated to flight muscles and tend to have shorter wings than unpalatable ones, that fly more slowly and straightly (Chai & Srygley 1990, Srygley & Chai 1990a, b, Marden & Chai 1991). Yet, whether and to what extent movements used for these different functions take part in dispersal between habitat patches is a matter of discussion (see also 1.2.1.). In case they are not relevant for inter-patch movements, it would be best to refer to them as ‘trivial move-ments’, or as ‘station keeping’ (sensu Southwood 1962 and Dingle 1996, respectively). But in some cases these movements could have a significant impact on inter-patch move-ments and, hence, they could be of significance for gene flow and dispersal. Moreover, even truly trivial movements could confound interpretations with regard to disper-sal. For instance, when predation pressure or mating system change in parallel with a change in habitat structure, it is difficult to disentangle the effects of habitat structure on dispersal capacity versus the effects on ability to escape predators or to locate mates. Merely looking at flight morphology is not sufficient, due to the multi-functionality of flight (Van Dyck & Matthysen 1999). Anyway, flight performance in ectothermic insects directly depends on thoracic (i.e. flight muscle) temperature, which is the outcome of both heat loss due to convective cooling and heat gained by means of thermoregulation (Watanabe & Imoto 2003).

2.2.2 thermoregulationFlight requires considerable quantities of energy. In ectotherms, flight performance and thermoregulation are inextricably linked (Kingsolver & Huey 1998, Chown & Nicolson 2004). In order to deliver the required amount of energy, body temperature typically needs to be situated within critical thermal limits (Dudley 2000, Chown & Nicolson 2004). Any factor that affects body temperature may have a strong impact on meta-bolic activity and flight performance and, hence, on individual fitness (Kingsolver 1987, Gilchrist 1995, Forsman 1999). Convective cooling is such a factor. It results from the impact of wind, but is also caused by flying itself. The endogenous heat produced when flying cannot compensate for the heat dissipated through convective cooling (Shelly & Ludwig 1985, Heinrich 1993), except for some larger insects (Heinrich 1975, Heinrich & Mommsen 1985, Dudley 2000). Therefore, individuals need to stop frequently in order to behaviourally warm-up toward an optimal temperature by basking. Thermoregulatory adaptations are thought to account for a significant proportion of inter- and intra-spe-cific phenotypic diversity in animals, including diversity due to phenotypic plasticity (Kingsolver 1987, Kingsolver & Wiernasz 1987, Kingsolver 1995).

{ 16 }


{ 17 }

been suggested to be the more dispersive sex in butterflies (Baker 1984, Hill et al. 1999a, Merckx et al. 2003, Hanski et al. 2004). While females spend most of their active time searching for host plants, male behaviour and distribution is completely dependent on female behaviour and distribution (Wiklund 2003). As a result of the ample sexual dif-ferences, and hence of different reactions and sensitivities, it is important to look at both sexes when it comes to study effects of habitat fragmentation.

We chose P. aegeria as a model to study effects of habitat fragmentation on the evolution-ary ecology of movement behaviour. P. aegeria is a temperate-zone satyrine butterfly (Lepidoptera: Nymphalidae) that primarily occurs in woodland with dappled sunlight (Tolman & Lewington 1997). The butterfly also occurs in more fragmented agricultural landscape and other habitat types with some kind of woodland aspect, including parks, gardens and along hedgerows in the more central and southern parts of its distribu-tion (Dover & Sparks 2000). Throughout Europe it is a non-threatened, widespread spe-cies and it has recently expanded in several countries, meanwhile extending its range northwards (Maes & Van Dyck 1999, Asher et al. 2001). These changes in distribution are associated with two factors: climate and habitat change. In many parts of Europe, wood-land habitats have substantially changed: the decline of coppicing during the twentieth century affected butterflies of shady woodland in a less detrimental way than it af-fected early successional species, such as fritillaries (a/o Gorissen et al. 2004). The spread of P. aegeria also suggests the strong influence of recent climate changes. However, the ability of P. aegeria to colonise areas as they become climatically suitable is likely to be constrained by the availability of potential habitat (Hill et al. 1999c, 2001).

Eggs are laid singly on leaves of various grasses. In spring and autumn, females tend to lay eggs in warm, sheltered spots on the edges of woodland, but in summer they choose more shady positions within woodland (Wiklund & Persson 1983, Shreeve 1986a, b). The cryptic larvae rest under the grass blades, and feed on the edges of the leaves. The pupae are formed on the host plant or on other vegetation nearby. The life style is unusually complex: this species hibernates at either the larval or pupal stage and three developmental pathways are possible (a/o Lees & Tilley 1980, Wiklund et al. 1983, Van Dyck & Wiklund 2002). These pathways are characterised by complex reaction norms cued by photoperiod and temperature (Nylin et al. 1989, 1993, 1995, Gotthard et al. 1994, Sibly et al. 1997, Nylin & Gotthard 1998, Van Dyck & Wiklund 2002). As a result of the different pathways, there are usually three generations each year in Belgium. However, the flight periods of each generation tend to overlap. Consequently, this species can be seen at almost any time from March to October.

Both P. aegeria males and females rely on sunlit patches at the forest floor to ther-moregulate behaviourally by dorsal basking (sensu Clench 1966). When the butterfly

}1.4 Model species: Pararge aegeria

Wiernasz 1991), the outer wing parts are isolated from selection for thermoregulatory performance (Wasserthal 1975). Then, they may freely function adaptively in other con-texts, such as aposematic signalling and mate recognition (Vane-Wright & Boppré 1993, Jiggins et al. 2001, Kapan 2001, Ellers & Boggs 2003).

Since both thermoregulation and convective cooling are strongly influenced by mi-croclimatic parameters (e.g. Shipp et al. 1987, Peng et al. 1994), flight performance and mobility of insects are dependent on the physical structure of the local environment, and, hence, it is dependent on microclimate (Vielmetter 1958, Kingsolver & Watt 1983, 1984). Habitat fragmentation goes along with altered landscape structures. Such changed structures cause alterations in microclimatic conditions, such as ambient temperature, ground temperature, wind speed and the amount of solar radiation that penetrates the habitat. Altered microclimatic conditions may select for adaptations in optimal thoracic temperature for flight in ectothermic insects. In the same vein, fragmentation may influence thermoregulation behaviour and/or thermoregulation-related morphology in order to counteract effects on thoracic thermal budget (Meyer & Sisk 2001).

1.3.3 sexual differencesEffects of habitat fragmentation, that cause microclimates to change also, may be dif-ferent for males versus females, since both sexes differ in many respects (e.g. mobility, flight-morphology, thermoregulation-related morphology). Consequently, trade-offs in one sex may be differently affected by fragmentation in the other sex. In particular, females are limited by the number of zygotes they can produce and, therefore, will max-imise offspring quality through resource acquisition or mate choice. Males are limited by the number of receptive mates (Andersson 1994). In most butterfly species, females mate soon after eclosion and often mate only once (Wedell et al. 2002). Hence, from a male’s perspective, females are rare resources. As a result, and as predicted from sex-ual selection theory, males will favour early emergence (i.e. protandry) over large size under most circumstances (Wiklund & Fagerström 1977, Fischer & Fiedler 2000, 2001). Male and female behaviour during courtship and mating agrees well with sexual se-lection theory (Rutowski 1984, Silberglied 1984, Andersson 1994). Males are aggressive to rivals and persistent in locating and courting females. Females are more ‘reserved’ and effective at rejecting males (Wiklund & Forsberg 1991). Such decisions seem to be dominated by olfactory cues. Although butterfly sexes differ in coloration pattern, and though male coloration is in general more brilliant (Darwin 1859), female choice is not governed by stimuli from the coloration pattern of males. Females use visual cues pri-marily in the UV (Rutowski 1981). Hence, the major selective agent responsible for bril-liant male coloration is intra-sexual communication between males (Silberglied 1984). Butterfly sexes differ considerably in flight behaviour relating to their different ecologi-cal functions (Shreeve 1992a, Van Dyck 2003). Males spend most of their active time to mate location (Shreeve 1992a), but this mainly covers local movements. Females have

{ 18 }


{ 19 }

orchards, tiny woodlots and several sunken roads with hedgerows (‘Hoegaarden’ and ‘Rillaar’). The ‘Hoegaarden’ study area (555 ha) consisted of 33 sunken roads with hedge-rows that made up a distance of 8.9 km in total. Only 3% of the study area was covered by six small (0.4 to 11.2 ha) and mainly poplar woodlots. The ‘Rillaar’ study area (401 ha) consisted of 15 sunken roads with hedgerows (3.9 km) and six small oak woodlots (3% of the study area; 0.3 ha to 6.3 ha). The depth of the sunken roads – eroded agricultural tracks bordered by steep grassy and woody banks – varied from 0.5 m to several meters; their width from 5 to 25 m and their lengths from 80 to 1370 m (but sometimes with gaps of some meters). Distances between study landscapes range from 7 to 60 km and are situ-ated within regions with similar topography, soil and macroclimate (Figure 1.1).

Figure 1.1: Distribution of the five study sites on a map covering parts of the provinces Vlaams-Brabant and Antwerpen (Flanders, Belgium) and that distinguishes between optimal and suboptimal habitat for P. aegeria. Optimal habitat is oak woodland; Suboptimal habitat is all other habitat (e.g. deciduous woodland, structure rich conifer woodland, orchards, hedgerows, sunken roads) (Based upon a biological validation map of Flanders: version 1.0-2.0-2.1, Institute of Nature Conservation, Brussels).

starts to fly voluntarily after a period of basking, body temperature is usually within a range of 30-34°C (Shreeve 1984, Van Dyck & Matthysen 1998). During flight, body tem-perature decreases by convective cooling, which forces the butterfly to bask again (Van Dyck & Matthysen 1998). Hence, the duration of a single flight bout increases with am-bient temperature (Shreeve 1984). Both sexes mainly feed on honeydew and are rarely seen feeding on flowers, except early and late in the year when aphid activity is low (Asher et al. 2001). The behaviour of males makes them more conspicuous than females, which spend more time basking or feeding (Shreeve 1992a). Males adopt one of two strategies for locating mates: perching in a territory (i.e. a small spot of sunlight, from where they rise rapidly to intercept any intruder) or patrolling through the habitat. Van Dyck et al. (1997a) have shown that males with paler wing colour are predominantly territorial and darker males are predominantly patrollers. This is thought to be because darker males warm up more quickly when basking, and, therefore, are able to spend longer periods in flight (Van Dyck & Matthysen 1998). So, thermal requirements differ between the behavioural strategies, and behavioural differences between phenotypes relate to differences in thermal ecology (Van Dyck et al. 1997a, Van Dyck & Matthysen 1998). Perchers also have a relatively large thorax, a difference which is explained to be adaptive. It facilitates individuals to accelerate at a higher level, which could be of advantage in escalated fights (Van Dyck 2003 and references therein).

We chose the speckled wood (Pararge aegeria L.) as a model to investigate several aspects of evolutionary ecology of movement behaviour in the context of habitat fragmentation. A preliminary analysis revealed differences in morphological design among differently fragmented landscapes (Berwaerts et al. 1998). Moreover, several morphological traits in males and females have been found to possess a genetic basis (Van Dyck et al. 1998, Ber-waerts 2004), and morphological design has been interpreted in terms of performance (Berwaerts et al. 2002). Furthermore, P. aegeria is one of the most studied butterfly species; extensive background knowledge made it easier to test the research questions.

The research described in the following chapters is based upon observations during field studies as well as upon data collected by means of experiments. Experimental designs included reciprocal transplant experiments and observations in common garden set-ups. Observed individuals originated from five study sites. Those were situated in central Belgium in three contrasting types of landscape that reflect different degrees of frag-mentation: (i) large continuous oak woodlands [‘Meerdaalwoud’ (1255 ha) and ‘Bos ter Rijst’ (280 ha)], (ii) a set of small (1.3 to 19.0 ha) oak woodland fragments, covering 8% of the study area (‘Boshoek’), with relatively small inter-patch distances (85 to 865 m), and (iii) highly fragmented, agricultural landscapes composed of intensively used fields and pastures (the large majority of the area) and to a lesser extent houses and farms,

}1.5 Study areas

{ 20 }


{ 21 }

heating rate) among these landscapes, and we estimated the impact of microsite selection (i.e. vegetation structures, flight height) on thermal budgets, using P. ae-geria dummies as models {Chapter 3}

- We tested for differences in thermoregulation-related morphology among popula-tions from these contrasting types of landscape {Chapter 3}

- We looked for differences in flight ability between types of landscape and sex, by means of testing laboratory-reared offspring from both types of landscape in flight cabinets at several low temperatures {Chapter 4}

Movements:- We observed and compared mate location behaviour in the two contrasting types

of landscape {Chapter 5}- We studied movements and habitat boundary crossing of laboratory-reared indi-

viduals that originated from these landscapes, using an experimental landscape as a common environment {Chapter 6}

- We performed a release-experiment, with a reciprocal transplant set-up, to test for differences in habitat-detection capacity between individuals from these contrast-ing types of landscape {Chapter 7}

This dissertation is presented as a compilation of published, accepted and submitted articles. It presents the results in different chapters. At the moment of writing, chapter 6 is published in the Proceedings of the Royal Society of London series B, chapter 5 is accepted for publication in Animal Behaviour, chapter 2 is submitted to Oecologia, chapter 3 to Func-tional Ecology, chapter 4 to Biology Letters and chapter 7 to Journal of Animal Ecology. Chapter 8 concerns a general discussion in which we discuss the results in a larger framework and in which we formulate future directions.

Within the context of habitat fragmentation we address evolutionary ecological ques-tions related to movement behaviour in the butterfly P. aegeria. Habitat fragmentation is considered to be the major cause of the current mega-extinction crisis (Pimm & Raven 2000). However, fragmentation may also affect species in more subtle ways. Our study species copes with highly fragmented, agricultural landscapes. Nevertheless, P. aegeria is a common and widespread butterfly species (Maes & Van Dyck 1999, Asher et al. 2001). In order to examine how P. aegeria deals with habitat fragmentation, we observed and experimentally compared populations from differently fragmented landscapes. Since phenotypic plasticity is a manner to deal with varying environments (West-Eberhard 2003), we first examined whether butterflies show landscape-related plasticity in flight morphology. Altered landscape structures imply altered distributions of resources, that may select for changed abilities to cross distances among resources. Hence, it may be adaptive to respond with changes in flight morphology (a/o Dempster et al. 1976, Van Dyck & Matthysen 1999). In the case of ectotherms, microclimate is another factor that directly affects the cost of flying (Kingsolver & Huey 1998, Dudley 2000, Chown & Nicol-son 2004). Therefore, we first examined to what extent woodland landscapes differed from highly fragmented agricultural landscapes in terms of microclimate. Next, we tested whether these contrasting microclimates had an impact on thermal budgets for flight. Finally, we addressed how P. aegeria could compensate behaviourally and morpho-logically for microclimatic differences. Given these morphological and micro-climati-cal differences among both contrasting types of landscape, we experimentally tested whether individuals originating from both types of landscape differ in their ability to fly at suboptimal to even critically low temperatures, using large climate chambers. Ad-ditionally, we experimentally tested whether flight ability was differently affected with changing temperatures in both sexes. Since differences in thermal properties and in the distribution and frequency of resources among landscapes may directly affect the cost of flying, the level of movements (e.g. mate location behaviour, dispersal propen-sity, habitat-detection capacity) may differ among differently fragmented landscapes.

Phenotypic plasticity in flight morphological traits:- We tested for landscape-related plasticity in flight morphology using a recipro-

cal transplant experiment with offspring of females from differently fragmented landscapes {Chapter 2}

- We examined whether this plasticity is able to respond to natural selection {Chap-ter 2}

Thermoregulatory needs for flight:- We compared microclimate among contrasting types of landscape {Chapter 3}- We investigated differences in thermal budgets for flight (i.e. cooling distance,

}1.6 Aims & outline

{foto: Jeroen Mentens}

Chapter 2 { Landscape-related phenotypic plasticity }

{ 23 }

2}

Landscape-related

Summary

Landscape structure and phenotypic plasticity in flight morphology in the butterfly Pararge aegeria

{ Thomas Merckx & Hans Van Dyck }

phenotypic plasticityChapter

In evolutionary time, varying environments may lead to different morphs as a result of genetic adaptation and divergence or of phenotypic plasticity. Landscapes that differ in the extent of habitat fragmentation may provide different selection regimes for disper-sal, but also for other ecological functions. Several studies on flying insects have shown differences in flight morphology between landscapes, but whether such differences result from plastic responses have rarely been tested. We did a reciprocal transplant experiment with offspring of speckled wood butterfly females (Pararge aegeria) from three types of landscape differing in fragmentation: woodland landscape, landscape with woodland fragments and agricultural landscape with only hedgerows. Young cat-erpillars were allowed to grow individually on potted host grasses in small enclosures under the three landscape conditions (split-brood design). Mortality in caterpillars was much higher in agricultural landscape compared to the other landscapes. Additive to the effect of landscape of development, landscape of origin also affected mortality rate in a similar way. Flight morphology of the adults resulting from the experiment dif-fered significantly with landscape. Independent of the landscape of origin, males and females that developed in agricultural landscape were the heaviest and had the great-est wing loadings. Females that developed in agricultural landscape had heavier tho-races (i.e. greater flight muscle allocation) in line with adaptive predictions on altered dispersal behaviour with type of landscape. In males, relative thorax mass did not respond significantly relative to landscape of development, but males originating from landscape with woodland fragments allocated more into their thorax compared to males from the other types. We found significant G x E interactions for total dry mass and wing loading. Our results suggest the existence of landscape-related phenotypic plasticity in butterfly flight morphology.

{ 24 }


{ 25 }

et al. 2003). There is considerable knowledge on the functional flight morphology in this species (Berwaerts et al. 2002, Van Dyck 2003 and references therein), and earlier work has suggested morphological differences between landscapes (Berwaerts et al. 1998). Here, we performed a reciprocal transplant experiment with offspring of females from three landscapes differing in the degree of fragmentation: continuous woodland land-scape, agricultural landscape with several nearby woodland fragments, and intensively used agricultural landscape with only hedgerows and a few woodlots (hereafter re-ferred to as woodland landscape, landscape with woodland fragments, and agricultural landscape, respectively). Following a split-brood approach, offspring of each individual female was distributed evenly among these three landscapes. Young caterpillars were allowed to grow individually on potted host plants in small enclosures under local field conditions. Hence, it is also possible to monitor mortality during development and to explore survival rate relative to landscape of origin and of development. After eclosion, adults were collected to analyse their morphology.

We test for adaptive differences in flight morphology assuming that dispersal rate has changed with landscape structure. Flight morphology includes relative thorax mass (i.e. allocation to flight muscles), wing loading and forewing aspect ratio (i.e. wing shape). For a detailed discussion of the functional significance see Dudley (2000), Ber-waerts et al. (2002) and Van Dyck (2003). However, butterfly flight is multifunctional (Shreeve 1992b, Watt 2003) and is not exclusively related to dispersal ability, which may confound interpretations (Van Dyck & Matthysen 1999). P. aegeria males and females differ considerably in their flight behaviour relating to their different ecological roles (Van Dyck 2003). Males spend most of their active time in mate location. They either adopt an aggressive territorial sit-and-wait strategy on a sunlit patch (‘perching’) or a search strategy in which a male explores a wider area looking for receptive females (‘patrolling’) (Wickman & Wiklund 1983, Shreeve 1987). In either case, mate location mainly covers local movements. On the other hand, there is evidence that females are the more dispersive sex. At considerable distance outside woodland habitat, Baker (1984) only observed P. aegeria females traversing the landscape. This is anecdotal evidence, but stronger support comes from comparative work on flight morphology between recently colonised sites (at the edge of the distribution) versus permanently populated sites (at the core area of the distribution). Hill et al. (1999b) found significant differences in P. aegeria females only; females from newly established populations were larger, had larger thoraces and had lower aspect ratios. In our earlier work testing variation in dispersal propensity in experimental cages, we found only clear responses in P. aegeria females (Merckx et al. 2003). So, it makes sense to restrict a test on dispersal-related variation in flight morphology to females only. As a result of altered costs and benefits of dispersal, fragmented landscapes are expected to select for higher mobility since resources are no longer concentrated but scattered over different fragments (e.g. Olivieri & Gouyon 1997, Van Dyck & Matthysen 1999). Based on the literature of functional flight-morphology in

}IntroductionIn evolutionary terms, organisms have in principle two alternatives to deal with varying environments. They can either follow the pathway of genetic adaptation and divergence, or phenotypic plasticity. For instance, seasonal morphological variation can result from differential success of different genetic forms (e.g. Cepaea nemoralis – Cain & Sheppard 1954), or, of plastic responses of the genotype based on environmental cues producing dif-ferent phenotypic forms (e.g. seasonal polyphenism – Shapiro 1976). The reaction norm or the degree to which a genotype responds in a plastic way to environmental variation can be under genetic control, and hence, be subject to selection (Price et al. 2003).

Habitat fragmentation is a widely recognised phenomenon that alters the environ-ment in multiple ways for several organisms (Meffe & Carroll 1997, Hanski 1999, Fahrig 2003) providing different selection regimes. In this vein, several studies using flying in-sects as study models have shown differences in flight morphology between landscapes (e.g. Dempster 1991, Taylor & Merriam 1995, Thomas et al. 1998, Hill et al. 1999a, Norberg & Leimar 2002). Morphological patterns have typically been interpreted in terms of as-sumed changes in dispersal rates with habitat fragmentation, such as selection against mobility with increased isolation in line with Dempster et al. (1976).

Several of those flight-morphological traits (including total body mass, relative tho-rax mass, wing loading and forewing aspect ratio) have a heritable basis (Hill et al. 1999b, Berwaerts 2004), which is a prerequisite for evolutionary change. In seasonal environ-ments, however, there is evidence for adaptive phenotypic plasticity in the same mor-phological and life history traits in multivoltine butterflies (Fric & Konvicka 2002, Van Dyck & Wiklund 2002). Reaction norms for morphological traits (and life history traits) may differ – ultimately in adaptive terms – between geographic regions (Nylin et al. 1995) or even between populations at close proximity (Sibly et al. 1997). We are not aware of studies testing adaptive morphological plasticity in relation to landscape structure in a context of habitat fragmentation. Moreover, plasticity studies have only rarely been done under field conditions (Lorenzon et al. 2001). In the laboratory, environmental regimes with changed means without changes in variance appear to be the rule (but see Miner & Vonesh 2004). Besides practical constraints, the obvious reason to avoid field experiments relates to the fact that field conditions are typically too complex and vari-able. However, to test associations between phenotypic plasticity and landscape struc-ture, reciprocal transplant experiments provide an interesting scope. Such experiments are of course limited because they do not allow insight in the environmental cues and causal factors that are operating.

In this paper, we evaluate to what extent a particular genotype produces different phenotypes in different landscapes using the speckled wood butterfly (Pararge aegeria L.) as a model. This butterfly is primarily a woodland species (Tolman & Lewington 1997), but it also occurs in fragmented, agricultural landscape with only small and scattered pieces of ‘woodland’ habitat like woodlots and hedgerows (Dover & Sparks 2000, Merckx

{ 26 }


{ 27 }

reciprocal transplant experimentIn summer 2001, we caught females in three different landscapes in central Belgium: (i) a landscape dominated by deciduous oak woodland (‘Meerdaalwoud’, 1255 ha), (ii) a set of small oak woodland fragments (‘Boshoek’, 1.3 to 19.0 ha) with relatively small interpatch distances (85 to 865 m), and (iii) a highly fragmented, agricultural landscape composed of intensively used fields and pastures (the large majority of the area) and to a lesser extent houses and farms, orchards, tiny woodlots and several sunken roads with hedge-rows (‘Hoegaarden’). This hedgerow landscape consisted of 33 sunken roads with hedge-rows that made up a total distance of 8.9 km. Only three percent of the study area was covered by six small (0.4 to 11.2 ha) and mainly poplar woodlots (Table 2.1). The caught females were transported to the laboratory and allowed to lay eggs on potted tufts of grass (Poa trivialis) in individual cages. Offspring of two females of each landscape was used in the experimental set-up: thirty second instar larvae of each female were placed in individual enclosures (11 cm wide x 11 cm long x 35 cm high) containing a fresh tuft of potted Poa trivialis. Grasses had been reared in the laboratory under standardised condi-tions for light, ambient temperature, nutrition, water and soil substrate (Wiklund et al. 1983). 180 enclosures were equally spread among the three types of landscape where fe-males were collected. They were placed at sites in the landscape where larvae of P. aegeria also naturally occur (TM, pers. obs.). Cages were inspected frequently and we recorded when larvae or pupae were dead or when adults were not fully eclosed. Freshly eclosed adults were collected, killed and stored by freezing (-20°C) for morphological measure-ments. The experiment covered a period from May 16 to July 26. Next, we repeated the experiment from July 24 to September 25 (n=180). Given summertime conditions, all individuals followed a direct development to the adult stage (see Wiklund et al. 1983).

morphological measurementsMorphological measurements were according to the methods of Van Dyck & Wiklund (2002). Prior to the measurement, specimens were dried in an incubator until constant mass during 24 h at 60°C. Total dry mass was measured with a microbalance (Mettler

landscape type study area optimal habitat suboptimal habitat

woodland 138 ha ca. 85% ca. 15%

woodland fragments 757 ha 9% 2%

agricultural 555 ha 0% 3% + 8.9 km sunken roads

Table 2.1. Overview of percentages optimal and suboptimal habitat surface area for given study areas with-in three landscapes: (i) woodland landscape, (ii) landscape with woodland fragments, and (iii) agricultural landscape. Optimal habitat is oak woodland; Suboptimal habitat is all other habitat (deciduous woodland, structure rich conifer woodland, orchards, hedgerows, sunken rural roads). Based upon a biological valida-tion map of Flanders (version 1.0-2.0-2.1, Institute of Nature Conservation, Brussels).

butterflies (Dudley 2000, Van Dyck 2003) and as applied by Hill et al. (1999b), we predict females to have a larger size (larger total body mass), to have higher wing loading, to al-locate more mass to flight muscles (higher relative thorax mass), and to possess lower as-pect ratio with increasing degree of fragmentation. Elsewhere we have shown that mate location is more frequently of an intermediate type between perching and patrolling in agricultural landscape compared to woodland landscape (Merckx & Van Dyck 2005), but this may have little impact when average values of flight morphology are tested among landscapes. So, we predict no differences in male flight morphology.

We can also test a prediction on size that follows from a constraint hypothesis and contrasts with the earlier predictions. We know that fragmented, agricultural land-scape is on average warmer than woodland landscape (Merckx, Van Dongen, Matthysen & Van Dyck, unpublished). Higher ambient temperature makes larvae growing faster which leads to smaller adults (Sibly et al. 1997, Nylin & Gotthard 1998). Hence, from this constraint hypothesis, we predict smaller individuals in more fragmented landscape in males and females, which is in contrast with our earlier predictions.

Finally, we also use the experimental data on flight morphology to test for signifi-cant genotype x environment interactions as families (or genotypes) may differ in their plastic responses. Such variability is a prerequisite for phenotypic plasticity to respond to natural selection and hence to be adaptive (West-Eberhard 2003).

study speciesThe speckled wood (Pararge aegeria L.) is primarily a woodland butterfly (Tolman & Lew-ington 1997), but it also breeds in other habitat types with some kind of woodland as-pect, including hedgerows (Dover & Sparks 2000). This butterfly can use three develop-mental pathways corresponding to different seasonal cohorts: (i) development with a pupal winter diapause resulting in early-spring adults, (ii) development with a larval winter diapause resulting in late-spring adults and (iii) direct development resulting in summer generation adults (Wiklund et al. 1983, Nylin et al. 1989). Honeydew is the main adult feeding resource and several grasses can be used as larval host plant (Shreeve 1986a). Eggs are deposited on isolated grass plants that grow under shaded, rather wet conditions surrounded by some bare ground (Wiklund & Persson 1983). Our recent work has pointed to several differences between speckled woods living in woodland land-scape versus fragmented agricultural landscape, including dispersal propensity (Merckx et al. 2003), temperature-related female fecundity (Karlsson & Van Dyck 2005), mate location (Merckx & Van Dyck 2005) and habitat finding ability (Merckx & Van Dyck, unpublished).

}Methods

{ 28 }


{ 29 }

the agricultural landscape were heavier and had higher wing loadings than individuals that developed in woodland landscape and landscape with woodland fragments (Figure 2.1). Females that developed in the agricultural landscape had heavier thoraces than females that developed in the other two types of landscape. In males, relative thorax mass did not respond significantly in relation to landscape of development, but a land-scape of origin effect was present. Males originating from the landscape with woodland fragments allocated more mass to the thorax than individuals from woodland land-scape and agricultural landscape. Forewing aspect ratio did not differ with landscape of development, or with landscape of origin (Table 2.2). Hence, we observed landscape-related phenotypic plasticity for total dry mass, wing loading and relative thorax mass. For the latter trait, however, plasticity was only present in females.

In both sexes, individuals from different families had different responses to the same environmental conditions for total dry mass and wing loading. These differences in the observed phenotypic plasticity among genotypes imply that G x E interactions were present for these traits. With regard to relative thorax mass, a G x E interaction was only present in males. No G x E interaction could be detected for aspect ratio.

��

��

��

��

��

��

��

��

��

��

��

��

��

��

variable effect F P

males

total dry mass landscape of development F2, 18.6

11 0.001

relative thorax mass landscape of origin F2, 10.1

7.9 0.009

total dry mass F1, 105

772 <0.0001

wing loading landscape of development F2, 21.8

6.2 0.008

females

total dry mass landscape of development F2, 22.3

6.6 0.006

relative thorax mass landscape of development F2, 96.6

11 <0.0001

landscape of origin F2, 7.66

3.8 0.07

total dry mass F1, 93.1

580 <0.0001

wing loading landscape of development F2, 17.3

4.2 0.03

Figure 2.1: Total dry mass of female P. aegeria offspring from different landscape origin that developed under woodland landscape, landscape with woodland fragments and agricultural landscape conditions in a reciprocal transplant enclosures experiment. We used means ± SE

Table 2.2. Summary of the analyses of different flight-morphological traits of P. aegeria offspring from a reciprocal transplant enclosure experiment in relation to landscape of origin (woodland landscape, land-scape with woodland fragments or agricultural landscape) and landscape of development (idem) (see Meth-ods). Mixed procedures (SAS) were run for both sexes separately.

Toledo 5). Next, specimens were dissected by separating head, thorax, abdomen, legs and wings. Thorax mass was also weighed. Forewings were photographed (Olympus Ca-media C-3030) to measure forewing length and forewing area using an image analyser system (Optimas 1999). Morphological characteristics were grouped into three meas-ures: (1) size: total dry mass, (2) wing shape: aspect ratio (4*forewing length²/forew-ing area) and (3) relative allocation: relative thorax mass and wing loading (total dry mass/forewing area). Both forewing length and forewing area was correlated with total dry mass (Pearson correlation coefficients: P<0.0001). Repeated measurements for mass and wing traits on 20 individuals showed high repeatabilities (between 0.99 and 1.00). Individuals with substantial wing wear were excluded from the analyses.

analyses and statisticsCounts of dead larvae and pupae and counts of not fully eclosed adults were summed and contrasted with counts of full-grown individuals. Differential survival with ‘landscape of development’ and ‘landscape of origin’ was tested by means of χ²-tests. Flight-mor-phological traits were analysed in function of ‘landscape of origin’ and ‘landscape of development’. Main effects were tested using mixed regression models (Proc Mixed, SAS 2001). In line with the predictions, models were run separately for males and females. Models also included ‘family’ (i.e. ‘genotype’) (nested within ‘landscape of origin’) and ‘session’ as random effects. For each trait it was tested by log-likelihood ratio whether the interaction of the ‘family x landscape of origin’ random parameter with ‘landscape of development’ made a significant contribution to the model. If so, it meant that a sig-nificant G x E interaction was present. After removing outliers of two males, original data were normally distributed. Final statistical models were obtained by backward elimination of non-significant factors (highest P-values). Residuals of the final models were normally distributed.

The proportion of individuals that survived to the adult stage was significantly smaller in agricultural landscape (58%) compared to woodland landscape (74%) and landscape with woodland fragments (76%) (n=346, χ²

2=10.59, P=0.005). Independent of landscape

of development, there was also a significant effect of landscape of origin on survival rate (n=346, χ²

2=13.42, P=0.001): Fewer individuals that originated from the agricul-

tural landscape survived to the adult stage (57%) than did individuals that originated from landscape with woodland fragments (72%) (n=229, χ²=5.44, P=0.02). Individuals that originated from woodland landscape had the highest survival (79%), although this proportion was not significantly different from landscape with woodland fragments.

Independent of their landscape of origin, males and females that developed as larvae in

}Results

{ 30 }


{ 31 }

an environmental parameter with different means. They can also respond morphologi-cally to the amount of environmental variability (Brakefield & Mazzotta 1995, Miner & Vonesh 2004). Hence, both average temperature as well as temperature variability could be cues within directly developing butterflies in order to predict future landscape structure conditions. Seasonal plasticity, on the other hand, operates among pathways. Van Dyck & Wiklund (2002) compared the three seasonal cohorts for the same flight morphological traits, and found evidence that photoperiod functions as the prime cue for this kind of plasticity in P. aegeria, although temperature can play a role as well (Stevens 2004).

It may not be a coincidence that we find landscape-related plasticity in a species that also shows seasonal plasticity. Once a species has developed plastic responses to seasonal environmental conditions, it may have the advantage of using the same mechanisms to respond to other (micro) climate-related variation like variation between landscapes. This leads to the hypothesis that species with high levels of plasticity in morphology and life history have a greater capacity to deal with changing and highly variable land-scapes than other species. Contrary to several other butterfly species that show clear to dramatic declines (Maes & Van Dyck 2001), P. aegeria is one of the few species that increased highly significant in distribution and abundance, at least in NW-Europe (e.g. UK: Asher et al. 2001, The Netherlands: van Swaay & Groenendijk 2004).

Although our experimental set-up principally aimed to evaluate plastic responses in flight morphology, it also revealed significant differences in juvenile mortality among types of landscape. A much higher mortality in agricultural landscape for caterpillars of this woodland butterfly points to a novel aspect of habitat fragmentation that requires further research. Besides the direct effect of higher mortality under agricultural land-scape conditions, the additive significant effect of landscape of origin suggests some divergence among the populations and types of landscape with respect to the capacity of larvae to deal with (or tolerate) different microclimatic conditions. It shows that evolu-tionary ecological studies in a context of habitat fragmentation should not be limited to adult ecology alone. Moreover, it needs to be evaluated to what extent increased mortal-ity biases the variation in phenotypic expression in adults between the landscapes.

Thanks to S. Van Dongen for valuable comments on statistics and to K. Borghgraef for practical assistance.

The University of Antwerp (GOA 15R/3942) and the Belgian Federal Office of Scientific, Technical and Cul-

tural Affairs (OSTC-PADD II EV/06/16B) funded the research. TM was research aspirant with the Fund of

Scientific Research Flanders-Belgium (F.W.O.). The experiments comply with the current laws of Belgium.

}Acknowledgements

Our reciprocal transplant experiment in different landscapes, following a split-brood approach, showed that adult offspring of P. aegeria females differed significantly in adult flight morphology. This suggests the existence of landscape-related phenotypic plastic-ity as the same genotype produced a different morphology in, for instance, a woodland landscape versus a highly fragmented agricultural landscape. Interestingly, the observed phenotypic plasticity for body size contrasts with the constraint hypothesis, which pre-dicts smaller individuals with fragmentation in both males and females due to higher growth rates with higher ambient temperatures (Sibly et al. 1997, Nylin & Gotthard 1998). Our results demonstrated the opposite. Hence, they point to adaptive differences. In line with the assumption of changed dispersal rates with landscape structure, we showed that individuals that developed in the agricultural landscape were heavier and had higher wing loading than those that developed in woodland landscape and landscape with woodland fragments, irrespective of their landscape of origin. Although these ef-fects were similar among males and females, phenotypic plasticity for relative thorax mass was only present in the female sex. This is in line with the prediction that an adaptive response of variation in flight morphology is easier to detect in females. As predicted according to the adaptive hypothesis, females that developed in the agricul-tural landscape allocated more mass to flight muscles than did females that developed in woodland landscape and landscape with woodland fragments. However, effects were not consistent for all flight-morphological traits: in both sexes no significant differ-ence among landscapes was present for aspect ratio. Relative thorax mass was the only trait for which landscape of origin played a role, but only in males. Males originating from the landscape with woodland fragments allocated more mass to the thorax than individuals from the woodland and agricultural landscape. Clearly this needs further investigation as it cannot be explained in terms of an adaptive response to changes in dispersal capacity with landscape structure as in females. Finally, our experimental data indicated that individuals from different families responded in a different way to the same landscape structure. In other words, we showed that different genotypes possessed different reaction norms. Significant G x E interactions were present for total mass, wing loading, and male relative thorax mass. Such interactions indicate that the observed landscape-related phenotypic plasticity is adaptive (West-Eberhard 2003) and contributes to evolutionary changes in dispersal – and probably other flight related be-haviours that need further research – with changed landscape structure.

Which environmental cue(s) induce plastic responses in flight morphology among different landscapes? Our field experiment does not provide direct answers. Ambient temperature is a candidate because we know that sites in agricultural landscape where P. aegeria thrives are on average warmer, and show more variability in temperature than corresponding sites in woodland (Merckx, Van Dongen, Matthysen & Van Dyck, unpub-lished). Organisms are not only able to induce different morphologies as a response to

}Discussion

{foto: Rollin Verlinde}

Chapter 3 { Habitat fragm

entation and thermoregulation }

{ 33 }

3

}

Habitat fragmentation

Summary

Habitat fragmentation and thermal flight budget in a woodland butterfly: an experimental assessment

{ Thomas Merckx, Stefan Van Dongen, Erik Matthysen & Hans Van Dyck }

and thermoregulationChapter

1. We assessed the impact of vegetation structure and type of landscape on microcli-matic conditions for flight activity, using the heliothermic butterfly Pararge aegeria as a model.

2. Dummies that initially were warmed up to similar thorax temperatures were able to cover larger distances, and they heated up faster, with higher ambient temperature and with lower wind speed.

3. Convective cooling decreased with increasing degree of shelter, and with lower flight height. The smaller impact is reflected in lower values of SE’s. Flight bouts bridged 140 ± 23 m under unsheltered, 81 ± 13 m under partly sheltered and 29 ± 6 m under sheltered conditions.

4. Ambient temperature and wind force were higher in highly fragmented, agricultural landscapes than in continuous woodland landscapes.

5. Only males had a higher amount of fur in agricultural landscape. This sexual differ-ence is interpreted in terms of differences in life history among the sexes: from an adaptive viewpoint, selection pressure to extend flight activity during periods with less favourable microclimatic conditions, as is more often the case in agricultural landscape, should be higher in males, since males are predicted to maximise the time spent to mate location, while females should allocate energy toward egg-laying rather than flight-activity.

{ 34 }



{ 35 }

been shown to be useful instruments to study – at least some – aspects of their thermal ecology (e.g. Berwaerts et al. 2001, Kemp & Krockenberger 2002). Dummies have largely similar thermal properties as living butterflies (Heinrich 1986). Obviously, living but-terflies influence or regulate their body temperature by adjusting their posture and by microclimate selection. So, dummies are no full equivalent to living butterflies, but

– as it is generally the case with models – tools that simplify complexity in order to dis-sect sources of variation, in this case operational thermal differences among different landscapes.

First, we compared two thermoregulation-related traits of adult butterfly morphol-ogy between samples of woodland and agricultural landscapes: degree of dorsal melani-sation of basal wing parts and amount of fur on thorax and basal wings. Higher levels of dorsal melanisation of basal wing parts result in higher heating rate during basking (Van Dyck & Matthysen 1998, Berwaerts et al. 2001). Body fur has been shown to be func-tionally significant for thermal isolation by reducing convective cooling (e.g. Kingsolver 1983). So, we predict agricultural landscape butterflies to have darker dorsal wings and more fur than woodland butterflies, in order to counteract associated higher levels of convective cooling. Berwaerts et al. (1998) have shown preliminary evidence for such an effect, at least for wing melanisation.

Next, we did two experiments with butterfly dummies. In a first experiment we esti-mated the distance that can be covered by a dummy until thorax temperature is cooled down from optimal temperature for flight activity (33°C) to a suboptimal temperature (25°C). This is what we call a thermal budget for flight. It allows testing for differen-tial effects of general vegetation structure and type of landscape on convective cooling, and hence on the value of a thermal budget in terms of flight distance. In a second experiment we examined heating rate of the thorax from suboptimal (25°C) to optimal temperature for flight (33°C) under the same environmental conditions as in the first experiment. Measurements of the two experiments were done close to the substrate (0.1 m) and at 1.0 m height because other studies have suggested the importance of shelter in this respect (Willmer 1986, Dover & Sparks 2000). To test differential effects of veg-etation structure we distinguished between three structural categories: (i) unsheltered: open field (only in agricultural landscape), (ii) partly sheltered: along hedgerow (agri-cultural landscape) or along woodland fringe (woodland landscape), and (iii) sheltered: in sunken roads (agricultural landscape) or inside woodland (woodland landscape).

Thermal budget for flight is expected to be larger under warm weather conditions but to be smaller under conditions of strong wind (convective cooling). So, we predict larger variation in cooling distances and heating rates in agricultural landscape than in wood-land landscape. Similarly, we predict more variation when shelter providing vegetation is lacking compared to (partly) sheltered conditions in the two types of landscape (Dover 1996, Noss & Csuti 1997, Morecroft et al. 1998), and we predict lower impact of convective cooling (i.e. less variation) close to the substrate (Willmer 1986, Dover & Sparks 2000).

}IntroductionIn ectotherms, the relationship between body temperature and performance typically shows an optimum bounded by critical thermal limits (Dudley 2000, Chown & Nicolson 2004). The thermal sensitivity of performance, especially locomotor performance, has become an important research focus in evolutionary physiology and ecology, and flight performance and thermoregulation of insects are inextricably linked (Kingsolver & Huey 1998, Chown & Nicolson 2004). In many flying insects (including butterflies), flight is a multifunctional trait relating to different behaviours with direct fitness consequences like predator escape, mate location, foraging, and oviposition (Dudley 2000, Watt 2003). Their flight performance directly depends on body temperature of the thorax where all flight muscles are located (Dudley 2000). Thoracic temperature is the net result of heat gained by behavioural thermoregulation and by physiological processes associated with muscle contraction and of heat lost by convective cooling (Clench 1966). Convective cool-ing is proportional to the difference between body temperature and air temperature and it is proportional to the velocity of the airflow over the insect body (Kingsolver 1983, Watanabe & Imoto 2003). Since both thermoregulation and convective cooling are influenced by variation in microclimate (e.g. Shipp et al. 1987, Peng et al. 1994), insect flight performance may depend on the physical structure of the local environment (Kingsolver & Watt 1984). Habitat fragmentation results in altered landscape structure and composition providing different microclimatic conditions (e.g. ambient tempera-ture, ground temperature, wind speed, amount of solar radiation that penetrates the habitat; Matlack 1993, Malcolm 1998, Honnay et al. 2001). In ectothermic insects, altered thermal environmental profiles may lead to evolutionary change in thermal properties of local populations. So, habitat fragmentation may influence thermoregulation behav-iour and thermoregulation-related morphology in order to counteract effects of altered microclimatic conditions that would take flight muscle temperature out of the optimal range for flight activity. However, so far only few studies have addressed the potential link between changing landscape structures and microclimate for heliothermic insects like butterflies (Meyer & Sisk 2001). We are not aware of studies evaluating variation in thermal flight budgets among landscapes differing in degree of fragmentation.

In this article, we use adults of the butterfly Pararge aegeria (L.) to test for differences in thermoregulation-related morphology and to assess the relative value of a thermal budget in terms of flight distance in relation to landscape structure in two contrasting types of landscape: woodland landscape and highly fragmented, agricultural landscape. When fully active, thoracic temperature in P. aegeria is within the range of 30-34°C (Shreeve 1984, Van Dyck & Matthysen 1998). As in other flying heliotherms, this but-terfly interrupts flight bouts with periods of basking since flight typically imply con-vective cooling towards suboptimal body temperatures (Van Dyck & Matthysen 1998). Basking is done by spreading the wings dorsally (Clench 1966). Butterfly dummies (i.e. dead specimens, with spread wings, on a hypodermic needle with a thermocouple) have

{ 36 }



{ 37 }

basal part to allow assessment of the amount of fur. The amount of fur on these basal wing parts was scored into three classes: (1) few short hairs, (2) medium amount of fur; the wing itself is clearly visible, and (3) lot of fur covering most of the basal wing surface. Repeated measurements for basal wing colour on a sample of 20 individuals showed a high repeatability (0.94). Only one out of 20 individuals was scored differ-ently for fur in an independent assessment. For 35 females we checked the relationship between amount of fur on the basal forewings and on the thorax (three classes as well). We could only show a weak relationship: 15 individuals were scored to the same class, 16 individuals to an adjacent class and four individuals were scored to an opposite class. Individuals with substantial wing wear were excluded from the analyses.

ambient temperatureDuring almost three months (May, 3 2003, 05.00 hrs. - July, 29 2003, 16.00 hrs.), 12 ther-moprobes connected to dataloggers (Hobo Pro Series) measured ambient air temperature each five minutes at a height of 1 m above the substrate. In each of four study sites we placed three thermoprobes. In total, we distinguished between six vegetation/land-scape classes: (1) inside woodland (‘core’), (2) at the edge of clearings within woodland (‘clearing’), (3) at the outer fringe of woodland (‘fringe’), (4) between both woody sides of sunken roads (‘inner sunken road’), (5) outer edges of woody sunken roads (‘outer sunken road’), and (6) salient, exposed locations within agricultural landscape (‘open’). BoxCar Pro 4 software was used to calculate average ambient temperatures for each hour and maximal ambient temperature of each day for different thermoprobes.

cooling distance experiment By placing a butterfly dummy in a sunlit patch, it was heated to a thoracic tempera-ture of 33°C. At this temperature, the observer started to walk with the dummy at a constant speed of c. 1 m/s. This is a relevant velocity in free flying P. aegeria (T. Merckx, unpublished data, see also Bunker 1993). The head of the dummy was always positioned forward and the spread wings were held horizontal. We recorded time and associated distance until dummy thoracic temperature dropped to 25°C (with a maximum of 300 m or c. 5 min). For each trial, we also recorded ambient temperature and wind speed (BF). Experiments were performed under three categories of vegetation structure: (i) unsheltered: open field, at least within a 50 m radius (only in agricultural landscape), (ii) partly sheltered: along tall hedgerow (± 4 m) (agricultural landscape) or along wood-land fringe (woodland landscape), within 1 and 2 m sideways, and (iii) sheltered: in sunken roads with tall hedgerows along both sides (agricultural landscape) or inside woodland, at least 50 m from a clearing or the woodland fringe (woodland landscape). For each specific location (n=98), the experiment was repeated three times, both at 1 m and 0.1 m height above the substrate. We used 11 specimens of an independent popula-tion of intermediate landscape structure (CDE-campus, University of Antwerp, which

study speciesP. aegeria is a satyrine butterfly primarily of woodland throughout Europe, but it also occurs in open, fragmented landscapes with some woodland aspect, like agricultural landscapes with hedgerows and small patches of woodland (Dover & Sparks 2000, Mer-ckx et al. 2003). Males either defend a sunlit spot at the forest floor as a territory or they patrol between several sunlit spots to find mates (Van Dyck 2003 and references therein). This bivoltine species is characterised by adaptive seasonal plasticity for flight- and thermoregulation-related morphology (Van Dyck & Wiklund 2002).

study landscapesFor morphological measurements, we sampled spring and summer individuals in four Belgian areas representing two types of landscape: (a) two landscapes dominated by deciduous oak woodland: Meerdaalwoud (1255 ha) and Bos ter Rijst (280 ha), and (b) two highly fragmented, agricultural landscapes composed of intensively used fields and pas-tures (the large majority of the area) and to a lesser extent houses and farms, commer-cial orchards, tiny patches of woodland and several sunken roads with hedgerows: Ril-laar and Hoegaarden (see Merckx et al. 2003). In both woodland landscapes, butterflies were collected within study plots of c. 135 ha of 100% woodland habitat. The Hoegaarden study area (555 ha) consists of 33 sunken roads with hedgerows (8.9 km) and six small patches of woodland. The Rillaar study area (401 ha) consists of 15 sunken roads with hedgerows (3.9 km) and six small patches of woodland. In both areas, small patches of woodland (0.3 to 11.2 ha) cover 3% of the area. Study landscapes are characterised by similar topography, soil type and macroclimate. Distances ranging from 7 to 60 km separate these four landscapes from one each other.

Experimental dummy measurements were done in Hoegaarden (agricultural land-scape) and Meerdaalwoud (woodland landscape). They were performed simultaneously in both areas during periods when climatic conditions allow butterfly activity: sunny weather and wind speed maximally 3 Beaufort (BF). We performed experiments during three periods (1st: April, 14-24; 2nd: May, 28 - July, 8; 3rd: July, 11 - August, 2). The heat-ing rate experiment was only performed during the first and second period.

morphological measurementsIn spring and summer of 2000 and 2001, P. aegeria individuals (n=650) were captured at the four study landscapes and stored at -20°C. Prior to measurement, specimens were dried in an incubator during 24 h at 60°C. Next, wings were carefully separated from the thorax. Forewings were photographed (Olympus Camedia C-3030) in order to measure basal darkness (grey values) using an image analyser (Optimas 6.5). From all captured females (n=133), and from a separate set of males (n=283), we additionally photographed a close-up (binocular microscope Leica MZ125: magnification = c. 46) of the forewings’

}Methods

{ 38 }



{ 39 }

temperature, walking velocity (after log transformation and for the cooling distance experiment only) and period were added as fixed effects. The specific location (98 and 110 levels in the cooling distance and heating rate experiment, respectively), dummy (11 and 6 levels in the cooling distance and heating rate experiment, respectively), and person (3 levels in both experiments) were added as random effects. Cooling distanc-es were log transformed and, as indicated above, if distances were larger than 300 m (=2.477 after log transformation) the experiment was stopped. Heating times were arc-sin transformed and the experiment was terminated when the temperature of 33°C was not reached after 180 s. Therefore, in both experiments data were right-censored. This was incorporated in the analysis by treating these censored observations as missing datapoints originating from a truncated normal distribution. We now explain in detail how this incorporation was done. The linear mixed model is traditionally formulated according to Laird and Ware (1982) as:

where Xß reflects the fixed effects part and Zb the random effects part of the model, and ε the residual error assumed to be normally distributed with zero mean and variance σ2. If we set the fixed effects and random effects equal to a linear predictor θ = Xß + Zb, then this model can be easily reformulated as:

These two formulations are generally and interchangeably used in linear mixed models (Verbeke & Molenberghs 2001). In order to avoid having to replace the censored observa-tions by an arbitrary score, we adapted the ‘traditional’ mixed model developed above in order to allow explicit estimation of these missing censored datapoints. Censored data were set as missing and assumed to result from a distribution without any mass below the upper limit of the data. Formally, in order to incorporate the censoring, this model was slightly adapted as:

The above model allows making inferences about the means through the linear predic-tor θ. In addition, we were interested in the variability in cooling distances and heating rates in relation to landscape, shelter, and height. We therefore fitted heterogeneous variance models to allow separating the residual variance (σ2) components to be esti-mated for the different levels of these three factors. This was achieved by a multiplica-tive model at the level of σ2:

represents a half-open park type of landscape). This selection of specimens showed little inter-individual variation in basal wing colour and fur and the use of specific dummy individuals was randomised over the different measurements.

heating rate experimentA dummy was brought to a thoracic temperature of 25°C by placing it in the shade. At 25°C, the dummy was placed with its wings parallel to the ground in a sunspot and time was recorded. The experiment was stopped when thoracic temperature of the dummy reached 33°C. For each trial, we also recorded ambient temperature and wind speed. Experiments were performed under three conditions of shelter (see above). We repeated the experiment three times for each specific location (n=109), and both at 1 m and at 0.1 m above the substrate. In total, six dummies were used in a random order. Again, they originated from the same independent population. Dummy specimens did not show sub-stantial inter-individual variation in basal wing colour and amount of fur.

statistical analysesBasal wing colour was compared between seasons and types of landscape. Main effects and their interaction were tested via mixed regression models (Proc Mixed, SAS 2001), which were run for both sexes separately. In order to control for possible effects of year and study area, we added year (2 levels), study area (nested within type of landscape) (2 levels) and the interaction of study area and season as random effects. To obtain final models, we applied backward elimination of non-significant factors (highest P-values). Final model residuals had normal distribution. The frequency of different fur classes was analysed using Fisher’s Exact tests (Proc Freq, SAS 2001).

Ambient temperature profiles (average temperature for each hour and maximal day temperature) were compared between vegetation/landscape classes (6 levels), and test-ed via mixed regression models (Proc Mixed, SAS 2001). In order to control for possible effects of study area, we added study area (4 levels) as a random effect in both models. Model residuals had normal distributions. Log-likelihood tests were performed to check for interclass differences in variance. In addition, and also via mixed regression models, average and maximal temperatures were compared between types of landscape (agri-cultural versus woodland landscape) with vegetation/landscape class as a random effect. Model residuals were normally distributed.

We performed survival analyses to handle the right-censored cooling distance and heating rate data. However, since we were confronted with several fixed and random effects, we opted for an adapted linear mixed model approach in order to estimate cen-sored data points. Cooling distances and heating times were compared between land-scapes, vegetation structure classes (shelter) and height classes. All main effects and interactions were tested. In order to control for possible confounding factors, wind speed [4 classes: 1=0BF (no wind), 2=1BF, 3=2BF, 4=3BF (relative strong wind)], ambient

{ 40 }



{ 41 }

vegetation/landscape classes (average temperature: F5,5.04

=15.95, P=0.0042; maximal temperature: F

5,12=7.76, P=0.0018) (Table 3.1). There was no significant difference in

variance among the classes. Regardless of interclass variation, average and maximal temperatures were higher in agricultural landscape than in woodland landscape (aver-age temperature: F

1,4.04=10.07, P=0.033 – agricultural: 17.0 ± 0.20°C versus woodland: 16.1

± 0.20°C; maximal temperature: F1,3.53

=12.23, P=0.031 – agricultural: 23.8 ± 0.51°C versus woodland: 21.3 ± 0.50°C).

cooling distance experimentLarger distances could be covered by moving butterfly dummies when ambient tempera-ture was higher (β

temp=3.50, 95% CI: 1.36 – 5.66), and smaller distances with increasing

wind speed [differences in cooling distances after log-transformation using wind class 4 (relatively strong wind) as reference: β

1-4=31.66, 95% CI: 16.8 – 44.0; β

2-4=18.0, 95%

CI: 4.0 – 28.9; β3-4

=11.9, 95% CI: -1.4 – 22.3]. There was a two-way interaction between vegetation structure (shelter) and flight height: cooling distances were shorter at 1 m than at 0.1 m height, but this effect was smaller under more sheltered conditions (dif-ference between high and low height: unsheltered: β

high–low=-36.5, 95% CI: -53.3 – -19.4;

partly sheltered: βhigh–low

=-12.8, 95% CI: -19.2 – -6.54; sheltered: βhigh–low

=-4.4, 95% CI: -7.69 – -1.17) (Figure 3.1a, b). From the viewpoint of differences in cooling distances between the three classes of vegetation structure this interaction can be interpreted as show-ing a stronger effect of shelter (longer cooling distances in sheltered versus unsheltered conditions) at lower height (0.1 m) (lower height: β

partly sheltered–unsheltered=-41.4, 95% CI: -82.4

– -3.94; βσheltered– unsheltered

=-95.2, 95% CI: -137 – -57.2; β sheltered–partly sheltered

=-53.8, 95% CI: -78.1 – -31.8; higher height: β

partly sheltered–unsheltered=-17.7, 95% CI: -57.5 – 18.4; β

sheltered–unsheltered=-63.1,

95% CI: -104 – -26.3; βsheltered–partly sheltered

=-45.4, 95% CI: -69.3 – -23.6) (Figure 3.1a, b). A two-way interaction between flight height and type of landscape was found as the dif-ference between high and low height (i.e. shorter cooling distances at high than at low height) was observed in woodland (β

high–low=-4.42, 95% CI: -7.69 – -1.17), but not in the

agricultural landscape (βhigh–low

=1.81, 95% CI: -2.39 – 6.04) (Figure 3.1a, b). There was no

landscape vegetation/landscape class average temperature (°C ± SE) max temperature (°C ± SE)

core 15.9 ± 0.23 21.0 ± 0.43

woodland clearing 15.9 ± 0.23 21.3 ± 0.43

fringe 16.5 ± 0.23 21.5 ± 0.43

inner sunken road 16.7 ± 0.23 22.8 ± 0.43

agricultural outer sunken road 17.1 ± 0.23 23.4 ± 0.43

open 17.4 ± 0.25 25.2 ± 0.57

Table 3.1: Summary of least squares means of average and maximal ambient temperature during adult flight activity in suitable P. aegeria habitat for six vegetation/landscape classes in two types of landscape (woodland versus agricultural landscape).

A ratio (R) of two residual variance components equal to 1 hereby indicates that the vari-ances did not differ significantly. This type of complex models can be easily fitted in a Bayesian framework using Monte Carlo Markov Chain (MCMC) simulation techniques. We performed all analyses in the computer package WINBUGS (version 1.4) which is freely available at http://www.mrc-bsu.cam.ac.uk/bugs. Bayesian analyses are based on the likelihood principle but also involve making prior assumptions about the parameters in the model in the form of a prior distribution. As we had no prior knowledge about the parameters, we used weak prior distributions. For all fixed effects parameters, a normal prior distribution with zero mean and a variance of 1000 was applied. For variance com-ponents, a uniform prior on the standard deviation was used with mass between zero and 1000. The prior distributions of the ratios R were truncated normal distributions with mean equal to 1 and a variance of 1000, and no mass for negative values. In each analysis we ran five independent chains of 10000 iterations after discarding the first 4000 iterations to avoid dependence on the initial values. Convergence was checked by visual inspection of the MCMC’s and by the Gelman and Rubin shrink factor. Posterior distributions were summarised by their mean and 95% credibility interval (CI). For more details we refer to Gelman et al. (1995) and Carlin & Louis (2000). Model selection was performed in two steps. First, heterogeneity in residual variance was investigated in a model containing all fixed and random effects. Next, a backward elimination proc-ess was performed on the factors landscape, shelter and height and their interactions. All other covariates were kept in the model.

thermoregulation-related morphologyMales had more fur than females (χ²

2=14.40; P=0.0007; n=416), and they had more fur

in agricultural than in woodland landscapes (Fisher’s exact, P<0.0001); within both types of landscape there was no difference between study sites. Females showed a trend to have more fur during summer than during spring (Fisher’s exact, P=0.066), but showed no difference between the two types of landscape. Basal wing colour did not differ between agricultural and woodland landscapes. Males were darker in spring than in summer (F

1,2.17=22.07, P=0.036), but there was no such a pattern in females.

ambient temperatureWe observed differences in average and maximal ambient temperature among the six

}Results

{ 42 }



{ 43 }

landscape (difference between high and low height: unsheltered: no data available; partly sheltered: β

high–low=0.019, 95% CI: 0.014 – 0.025; sheltered: β

high–low=0.028, 95% CI:

0.017 – 0.040) (Figure 3.2a, b). Looking at the main effect of vegetation structure, heat-ing rates did not differ between partly sheltered and unsheltered conditions, but heat-ing rates were higher in sheltered relative to unsheltered conditions (β

shelter sheltered–unshel-

tered=0.027, 95% CI: 0.011 – 0.044) (Figure 3.2a, b). The period in which observations were

done, made no significant contribution to the model. Residual variances did not differ between the two classes of flight height (R

height=0.85, 95% CI: 0.56 – 1.21) nor between

the two types of landscape (Rlandscape

=0.85, 95% CI: 0.63 – 1.20). Residual variance was smaller in partly sheltered relative to unsheltered conditions (R

shelter2=0.52, 95% CI: 0.34

– 0.83) but not so when comparing sheltered and unsheltered conditions (Rshelter3

=1.04, 95% CI: 0.66 – 1.72) (Figure 3.2a, b).

Fragmented landscapes may provide different thermal environments for flying heli-otherms like butterflies. Our parallel measurements of ambient temperature at sites where the speckled wood butterfly Pararge aegeria occurs in continuous woodland versus corresponding sites in fragmented agricultural landscape indicated different thermal profiles. Ambient temperature in and along hedgerows and small patches of woodland in agricultural landscape was on average higher than in woodland landscape. Woodland landscape provides much shadier, cooler conditions that also buffer ambient tempera-ture. Typical sunlit patches that P. aegeria uses to bask and as rendezvous sites (Shreeve 1984), are better sheltered against convective cooling. In agricultural landscape, the tiny patches of woodland and narrow hedgerows do proportionally have larger amounts

�

��

��

��

��

��

��

��

��

�

��

��

��

��

��

��

��

��

Figure 3.2: Heating times of P. aegeria dummies in relation to vegetation structure and type of landscape (open circles: highly fragmented, agricultural landscapes; filled circles: woodland landscapes). We used means ± SE’s. Figure 3.2a: high flight height (1 m). Figure 3.2b: low flight height (0.1 m).

}Discussion

three-way interaction between type of landscape, flight height and vegetation structure, nor a two-way interaction between type of landscape and vegetation structure. The period in which observations were done, did not contribute significantly to the model. Residual variances did not differ between the two flight height classes (R

height=1.27, 95%

CI: 0.90 – 1.74) nor between the two types of landscape (Rlandscape

=0.95, 95% CI: 0.74 – 1.27). Residual variance was smaller in partly sheltered conditions relative to unsheltered conditions (R

shelter2=0.39, 95% CI: 0.24 – 0.65) and even stronger when comparing shel-

tered and unsheltered conditions (Rshelter3

=0.15, 95% CI: 0.10 – 0.24). Thus, variability in cooling distances was highest under unsheltered conditions and decreased (up to a fac-tor 6) when vegetation structure provided more shelter (Figure 3.1a, b).

heating rate experimentThere was an increased heating rate between 25 and 33°C when ambient temperature was higher (β

temp=-0.0085, 95% CI: -0.010 – -0.0068). Effects of wind speed on heating

were complex and non-linear: albeit there was no significant difference in heating rate between relatively high wind speed and windless conditions (β

1-4=0.0024, 95% CI:

-0.005 – 0.010), heating rate was slower with increasing wind speed (β2-4

=0.020, 95% CI: 0.011 – 0.028; β

3-4=0.039, 95% CI: 0.025 – 0.052). There was a three-way interaction

between type of landscape, vegetation structure and flight height. From the viewpoint of the effect of flight height (smaller heating rates at high versus low height), this inter-action can be interpreted as being a consequence of the fact that the difference in heat-ing rates between high and low height increased with degree of shelter in the agricul-tural landscape (difference between high and low height: unsheltered: β

high–low=0.025,

95% CI: 0.015 – 0.035; partly sheltered: βhigh–low

=0.017, 95% CI: -0.002 – 0.037; sheltered: β

high–low=0.009, 95% CI: 0.001 – 0.017) whereas the opposite was true for the woodland

�

��

��

��

��

��

��

��

��

��

��

�

��

��

��

��

��

��

��

��

��

��

Figure 3.1: Cooling distances of P. aegeria dummies in relation to vegetation structure and type of land-scape (open circles: highly fragmented, agricultural landscapes; filled circles: woodland landscapes). We used means ± SE’s. Figure 3.1a: high flight height (1 m). Figure 3.1b: low flight height (0.1 m).

{ 44 }



{ 45 }

those of agricultural landscape. The combined results suggest physiological adaptations among differently fragmented landscapes that provide different thermal environments.

Our results also showed evidence for morphological adaptation in line with predict-ed thermal variation. Males of agricultural landscapes had more fur on thorax and basal wings than did woodland males. More fur is known to promote thermal isola-tion like has been observed in lowland versus highland populations of Colias butterflies (Kingsolver 1983). So, in a next step comparative experiments with dummies should take this variable into account to estimate the benefit in terms of gained flight distance or duration with and without fur. Another test would be to release woodland males into agricultural landscape and quantify to what extent they are more restricted in their flight time budget or more restricted to sheltered micro-sites in the agricultural land-scape than local males. Experiments in which fur is removed from the thorax provide additional tools to estimate the benefit of this trait relative to landscape structure.

Variation in the degree of wing melanisation also affects thermoregulation and, hence, flight activity (Kingsolver & Watt 1984, Guppy 1986, Watt 2003, Ellers & Boggs 2004). Earlier work on males of P. aegeria has suggested darker individuals in agricul-tural landscape compared to woodland (Berwaerts et al. 1998); darker wings allow faster heating (Van Dyck & Matthysen 1998) which would be in particular an advantage in environments with high risk of convective cooling. Elsewhere we have shown using a field experiment that P. aegeria males which developed in agricultural landscape had indeed darker basal wings than males that developed in woodland (Merckx & Van Dyck, unpublished). However, there was no significant relationship with basal melanisation in the wild caught specimens of our current study. But it has recently been shown that the degree of melanisation is affected by environmental stress (like desiccation of host grasses) (Talloen et al. 2004). Such factors may to some extent confound the melanisa-tion patterns between landscapes, at least in some years.

Effects on the thermoregulation-related morphological variation appear to be lim-ited to males. This makes sense relative to the different behavioural repertoires of both sexes in butterflies (Pivnick & McNeil 1986). In males, extended flight during periods with less favourable microclimatic conditions increase the probability of locating vir-gin females; virgin females may emerge even under relatively poor weather. Knowing that there is a trade-off in females between allocations for reproduction versus flight (e.g. Hughes et al. 2003), females need to be more economical in terms of flight activity. So, butterfly females typically spend less time flying than males (Shreeve 1984). Hence, thermal isolation to extend flight bouts under relatively poor conditions (high convec-tive cooling) is most frequently important to males. This may result in differential se-lection regimes for such traits between the sexes.

At the landscape level, our results attract the attention to the importance of veg-etation structures as potential corridors for movement across fragmented landscapes. Moreover, the importance or significance of such structural characteristics will be

of irradiated surface, but more sites are exposed to the cooling effect of higher wind speed in the surrounding open land. So, an agricultural landscape appears to be more variable in operational temperature profile for butterflies.

We experimentally tested the potential impact of the different types of landscape for P. aegeria flight activity and performance using butterfly dummies (i.e. dead butterfly on a thermoprobe). In butterflies, dummies have largely similar thermal properties as living butterflies (Heinrich 1986). The distance or time that we could walk with a dum-my before it was cooled from optimal (i.e. 33°C) to the threshold of basking (i.e. 25°C) un-der woodland conditions at 1 m height (27 ± 4 s; n=102) corresponded well with values on voluntarily flight bouts in a woodland population of P. aegeria studied by Shreeve (1984) (26 ± 4 s in males and 18 ± 6 s in females). This suggests that the technique provides rel-evant estimates (see also Berwaerts et al. 2001, Kemp & Krockenberger 2002). Dummies were found to cover larger distances – and they also heated up faster - with higher am-bient temperature and lower wind speed. In the same vein, we observed that variation in cooling distance and in heating rate was higher in agricultural landscape compared to woodland landscape. Heating rates were significantly higher in sheltered relative to unsheltered conditions of vegetation structure. Although we could cover on average longer distances before dummies were cooled down under unsheltered conditions, vari-ation in flight distance was highest under such conditions (i.e. open agricultural land) and was reduced by a factor up to six when dummies were moved under more sheltered conditions. Earlier, Merckx & Van Dyck (2002) have shown for two grassland butterflies that occur in agricultural land that they also were more confined to the vicinity of hedgerows when risk of convective cooling was high (i.e. relatively high wind speed). Additionally, moving close to the ground or substrate, particularly at places with shel-tered conditions, had also a significant effect to lower the impact of convective cooling on flight distance or duration.

In the same vein, Rutowski et al. (1994) have shown that male butterflies adjusted perch height in relation to ambient temperature. We have found based on independent behavioural observation data that P. aegeria males flew more frequently at low height in agricultural landscape than in woodland landscape (Merckx & Van Dyck, 2005).

Although butterflies have some opportunities to avoid the problem of convective cool-ing by seeking sheltered vegetation structures in open agricultural landscape, the physi-ological cost of daily life may still be different in highly fragmented, agricultural land-scapes versus sheltered woodland. From an evolutionary viewpoint, this may result in different life styles among different landscapes. In another experiment where real but-terflies were induced to fly under standardised conditions at low suboptimal body tem-peratures, individuals of woodland landscape origin flew for longer than did agricultural landscape butterflies (Merckx, Karlsson & Van Dyck, unpublished). Within the range of temperatures of voluntary flight, Merckx et al. (2003) have shown in a common arena test in outdoor cages that P. aegeria of woodland landscape origin were more at flight than

{ 46 }

weather or climate dependent. If butterflies are only able to make relative short flights, a finer grained ecological network of corridors would be required. Moreover, for ecto-therms like butterflies hedgerows are likely to be rather stepping stones for thermal rea-sons than really continuous conduits which they may be for endotherms. Alternatively, it may be that the types of ‘daily’ or ‘routine’ butterfly movements we have focused on are not necessarily representative for real dispersal movements (Van Dyck & Baguette, 2005).

Finally, our study has focused on thermal differences for adults only. Of course, thermal profiles of the landscapes may have profound effects on the larval stages as well. In this context, variation in oviposition site choice among landscapes provides, for instance, an interesting perspective of further research as female behaviour and morphological design may be affected by selection on oviposition site choice.

We are grateful to K. Borghgraef, J. Cortens, M. Jacobs, and several students who assisted with experiments

and/or taking images of wings and thoraces. TM is research aspirant with the Fund of Scientific Research

Flanders-Belgium (F.W.O.). This research is funded by the University of Antwerp (GOA 15R/3942) and the

Belgian Federal Office of Scientific, Technical and Cultural Affairs (OSTC-PADD II EV/06/16B).

}Acknowledgements

Chapter 4 { Sex, landscape and butterfl

y flight ability }

{ 47 }

4

}

Sex, landscape

Summary

Sex- and landscape-related differences in thermal under-limit for flight ability in a woodland butterfly

{ Thomas Merckx, Bengt Karlsson & Hans Van Dyck }

and Chapter

Active time budgets of flying ectotherms depend on body temperature. Thermal knowl-edge of flight ability and performance is currently, however, heavily biased towards the range of voluntary flight activity. However, flight under suboptimal conditions also may be important in terms of fitness. Here, we studied flight ability and duration of laboratory-reared Pararge aegeria (L.) butterflies that originated from a woodland and an agricultural landscape, under low temperatures (10-21°C) in a common-garden set-up. We predict that males are able to fly at lower temperatures than females as males have lower wing loading. Since woodland is on average cooler than agricultural landscape we also predict that flight ability at low temperature is better developed in woodland individuals. Individuals showed flight activity at all tested temperatures and flights were longer with increasing temperature. Males flew for longer than did females. How-ever, there was no difference with sex at the lowest temperature, but an increasing difference with increasing temperature. We showed that woodland individuals flew for longer and had higher wing loading than agricultural landscape butterflies. Our results, that shed new light on the thermal ecology of flight at suboptimal temperatures, are discussed from both proximate and ultimate points of view.

butterfly flight ability

{ 48 }


y flight ability }

{ 49 }

}IntroductionThe idea of economic analysis of costs and benefits is central to the evolutionary ecologi-cal understanding of behaviour (Krebs & Davies 1997). Active flight has, for instance, clear benefits to track resources and to escape from predators, but it is energetically costly (Dudley 2000). Flying heliotherms (including butterflies) use solar radiation by basking behaviour as an external heat source, which is energetically less costly than physiological heat production (Dennis 1993). But the reverse side to this cheap option is that their active time budget can be severely constrained by poor weather.

In butterflies, voluntary flight activity is typically within a range of body tempera-tures of 28-38°C (e.g. Dennis 1993). In temperate-zone environments this is usually well above ambient temperature. Therefore, they thermoregulate behaviourally by basking and microsite selection to keep body temperature optimal (Shreeve 1992a). Thermal knowledge of flight ability and performance is currently biased towards the range of voluntary flight activity, but is far more incomplete for flight under suboptimal condi-tions, including the under-limit for flight activity. This has lead to the view that tem-perature is ‘dictating’ a butterfly’s behaviour. But recent evidence has contested this view. Although Pararge aegeria butterflies accelerated faster during take-off under opti-mal than under suboptimal thermal conditions, some individuals were able to perform at top level under both conditions (Berwaerts & Van Dyck 2004). Experiments by Kemp & Wiklund (2004) have shown that P. aegeria males won territorial conflicts independent of thermal asymmetries among the contestants. This contrasts with earlier views that disputes were settled on the base of differential deviance from optimal body tempera-ture (Hardy 1998).

Even if they are short, flights at low temperature outside the range of voluntary flight activity can be significant in terms of fitness (e.g. predator escape from an at-tack during resting). Here, we test flight ability and duration in P. aegeria under low ambient temperature in an experimental set-up. Males and females may show different behavioural responses at low temperature, in line with their different behavioural rep-ertoires and ecological functions. Females typically mate only once soon after eclosion and, hence, they are a rare resource for males. Since male reproductive success is lim-ited by the number of receptive females (Andersson 1994), we can predict selection on early emergence (protandry – Wiklund et al.. 1992) and early activity or activity under rather dull conditions in males (Pivnick & McNeil 1986) to maximise time spent to mate location (Shreeve 1992a). But females are limited by the number of produced zygotes (Parker 1978). Assuming that the cost of flight can be reduced by flying under optimal thermal conditions (see trade-off between flying and egg-laying), females’ flight activ-ity can be expected to be much more limited than males’ (Pivnick & McNeil 1986).

At the proximate level, males and females also differ in flight morphology (e.g. wing loading). Flight morphology interacts with ambient temperature to realise a particular level of flight performance (Berwaerts & Van Dyck 2004). Lower wing loading in males

(Berwaerts et al. 2002) allows them to fly with lower wingbeat frequency to generate the same amount of lift (Casey 1981). Since wingbeat frequency depends on temperature (Stevenson & Josephson 1990), we predict that males are able to fly at lower temperature than females.

Besides this sexual aspect, we are particularly interested in testing evolutionary dif-ferences in temperature-related flight ability between landscapes that differ in degree of habitat fragmentation by testing butterflies originating from a continuous woodland landscape versus a fragmented agricultural landscape (see Merckx et al. 2003). Wood-lands are typically cooler and better buffered environments than more open land (Noss & Csuti 1997). So, from an adaptive viewpoint we predict that flight ability at low tem-perature is better developed in woodland populations.

study speciesThroughout Europe, P. aegeria is primarily a woodland butterfly, but also occurs in frag-mented agricultural landscape with patches of woodland and hedgerows. Voluntary flights after basking are associated with thorax temperatures of 30-34°C (Shreeve 1984, Van Dyck & Matthysen 1998). The average duration of spontaneous flight bouts in the field increases with ambient temperature from 10 s at 17°C to 90 s at 25°C (Shreeve 1984).

sampled landscapes and breeding procedureIn May 2004, females were captured at several sites in a woodland landscape (Meerdaal-woud) and in an agricultural landscape (Rillaar) in Belgium. Intersite distance of cap-tures within each type was 0.5-5.0 km. Meerdaalwoud is a closed landscape dominated by deciduous oak woodland. Rillaar is composed of intensively used fields and pastures (80% of the area), houses and farms (10%), commercial orchards (5%), small patches of woodland (2.5%) and several sunken rural roads with hedgerows (2.5%). Females were allowed to oviposit on tufts of the grass Poa annua in small cages. Larvae were further reared under controlled climatic conditions that allow direct development (22°C, LD 22 : 2 h). After emergence, adults from five agricultural and three woodland landscape families were used for the experiment. Prior to observations, individuals were stored at 8°C and LD 12 : 12 h and they were not fed or mated. Finally, butterflies were killed and stored at -20°C. Next, they were dried during 24 h at 60°C and weighed (Sauter micro-balance AR1014). Forewing area was measured from digital pictures (Olympus Camedia C-3030) using an image analyser system (Optimas 1999).

testing flight abilityIndividual flight ability and duration was measured for males and females originat-ing from the two types of landscape in experimental flight cabinets (height x length x

}Methods

{ 50 }


y flight ability }

{ 51 }

width: 2.4 m x 4 m x 2 m). We used five identical flight cabinets differing in ambient temperature: 10, 13, 15, 17 and 21°C. Prior to the experiment, there was an acclimatiza-tion period of 30 min. After this period, thorax temperature was at the prevailing am-bient temperature in the cabinet (thorax thermoprobe measurement in 10 individuals; see Van Dyck & Matthysen 1998). Flight ability was tested for each individual under each temperature, but the order was randomised. The test was similar to Pivnick & McNeil (1986): a butterfly sitting with closed wings was picked up from its plastic box with a pair of tweezers and was released from a standard height of 2 m. The behaviour was observed and timed using a stopwatch. Observation lasted until the butterfly was alighted. An individual was tested three times for each of the five treatments. Between trials individuals rested in plastic boxes while others were tested. In total, we tested 88 individuals: 22 males and 22 females of woodland landscape origin and 28 males and 16 females of agricultural landscape origin. Since P. aegeria is protandrous (Nylin et al. 1993), tested males were significantly older than females (X ± SE=5.1 ± 0.11 versus 4.2 ± 0.31 days since eclosion; Student T test: t

45.8=-2.75, P=0.009). There was no age difference between

individuals from the two types of landscape.

statistical analysisWe used mixed regression models (Proc Mixed, SAS 2001) to analyse flight duration in relation to the factors of interest: ambient temperature (5 classes), landscape of origin (2 classes), age (i.e. days since eclosion), wing loading (total dry mass/forewing area) and sex. Flight duration was log-transformed and wing loading was transformed by taking the arc sine of the square root to obtain normally distributed model residuals. Because wing loading covaried strongly with sex, the analysis was done separately for each sex. Model selection starting from a full model (all factors of interest and interaction terms) was done by stepwise backward elimination of nonsignificant factors. Factors were not eliminated as long as they were incorporated in interaction terms. As individuals were tested repeatedly, the model took into account ‘individual’ and the interaction ‘indi-vidual’ x ‘trial’ as random effects.

We observed some to considerable flight activity at all tested temperatures (range of flight duration 1.1-600.8 s). At 10°C flights lasted very short but on average at least three times as long as the falling of a dead specimen and butterflies always alighted in a con-trolled manner on the legs. Butterflies flew for longer when ambient temperature was higher and males flew for longer than females (F

1,131=19.63, P<0.0001). However, there

was no difference with sex at the lowest temperature, but the increasing difference with temperature explains the highly significant temperature x sex interaction (F

4,1012=17.39,

P<0.0001; Figure 4.1). At 21°C males flew on average five times as long as females.

Table 4.1 shows the results of the mixed regression model for flight duration in males and females. Butterflies originating from the woodland landscape flew for longer than did agricultural landscape butterflies. In males this was a strong main effect, but the landscape effect also interacted with temperature on the one hand and with age on the other hand. The difference of landscape of origin increased gradually with temperature, but was smallest at 21°C. Younger males flew for longer, but especially at higher tem-peratures. The effect of age was stronger in males of woodland landscape origin. High wing loading was associated with longer flights at all temperatures except for the treat-ment of 17°C (10°C: slope (β)=2.4; 13°C: β=4.0; 15°C: β=3.6; 17°C: β=-10.2; 21°C: β=14.0).

In females, there was a tendency for a similar effect of landscape of ori-gin as in males. Age was significant only in inter-action with temperature: younger individuals flew for longer than did older ones, especially at higher temperatures. Females

}Results

effect F P

(i) males temperature F4,525

11.74 < 0.0001age F

1,12815.84 0.0001

wing loading F1,42.2

2.39 0.13landscape F

1,1144.55 0.035

age x temperature F4,536

4.48 0.001wing loading x temperature F

4,52411.20 < 0.0001

temperature x landscape F4,533

3.94 0.004age x landscape F

1,1205.69 0.019

(ii) females temperature F4,195

3.38 0.011age F

1,21.50.44 0.51

wing loading F1,18.8

0.02 0.89landscape F

1,16.73.94 0.064

age x temperature F4,196

4.84 0.001wing loading x temperature F4,196 4.86 0.001

Table 4.1. Summary of the analyses of flight endurance of (i) P. aegeria males and (ii) P. aegeria females in relation to the wing loading of the butterfly, air temperature of the flight cabinet, age and landscape of origin (see Material & Methods) using Mixed procedure (SAS).

��

��

��

��

��

��

��

��

��

Figure 4.1. Temperature x sex interaction effect on flight duration

{ 52 }


y flight ability }

{ 53 }

flew for longer when they had lower wing loading but this relationship was much steeper at low ambient condition than at the relatively higher temperatures (10°C: β: -2.9; 13°C: β=-3.9; 15°C: β=-0.5; 17°C: β=-0.5). At 21°C only, we observed a reversed effect (β=7.1).

Our experiment showed that the butterfly P. aegeria is able to show flight activity at much lower temperatures (10-21°C) than the voluntary flight range (30-34°C – Shreeve 1984). So, the threshold for flight activity must be even below 10°C. This is much lower than in several other butterfly species (e.g. Pivnick & McNeil 1986, Masters et al. 1988, Norberg 2003). P. aegeria is able to fly at low ambient temperature, but this is generally avoided (i.e. short flights after experimental induction only) suggesting a considerable physi-ological cost. In the same vein, older butterflies that have already used much of their reserves (Karlsson 1994) had shorter flights. In line with our key predictions, males and females responded differently and the same was true for butterflies of woodland versus agricultural landscape origin.

Since males were significantly older than females, the result of males flying for longer than females should be considered a conservative result. Ultimately, the ob-served sexual difference of a lower willingness to fly at suboptimal temperature in females relates to the fundamental differences in their life history relative to males. For males extended periods of flight increase the probability of locating virgin females. Even under relatively poor weather virgin females may emerge. Knowing that there is a trade-off in females between allocations for reproduction versus flight (e.g. Hughes et al.. 2003), females need to be more economical in terms of flight activity. So, butterfly females typically spend less time flying than males (Shreeve 1984), which most likely contributes to the explanation of male-biased samples in mark-release-recapture studies (e.g. Wahlberg et al. 2002).

At the proximate level, the difference in flight duration at low temperature can be attributed to lower wing loading in males (Gilchrist 1990). It allows flight at lower wing-beat frequency (Casey 1981) and wingbeat frequency is temperature-dependent (Steven-son & Josephson 1990). Minimum temperature for flight in Thymelicus lineola increased with an experimental increase in wing loading (Pivnick & McNeil 1986). Sexual differ-ence in wing loading in P. aegeria is mainly caused by differences in body mass rather than forewing area (Berwaerts et al. 2002). So, the smaller male size can be seen as an adaptation to extend flight activity.

P. aegeria of woodland landscape origin flew for longer at low ambient temperatures than did agricultural landscape butterflies. Within the range of temperatures of vol-untary flight, Merckx et al.. (2003) have already shown that P. aegeria of woodland land-scape origin are more at flight than agricultural landscape butterflies. Karlsson & Van Dyck (2005) have experimentally shown that females of woodland landscape origin laid

more eggs at low ambient temperature than did agricultural landscape females, but the pattern reversed at high ambient temperature. Hence, all these results point at tempera-ture-related adaptation in the cooler environment of a continuous woodland landscape versus a warmer, but thermally more variable fragmented agricultural landscape. At the proximate level, the observed differences in wing loading among landscape types (higher wing loading in a woodland than in an agricultural landscape) are likely to con-tribute to the effect, since high wing loading was associated with longer flights at 21°C in females and at all but one temperature tested in males. Conversely, in the tempera-ture range of voluntary flight a lower wing loading allows flying with lower wing-beat frequency (Casey 1981) and, hence, it allows longer flights.

Our study provides a novel perspective to the examination of flight ability and per-formance below optimal body temperature and to what extent it contributes differently in males and females to landscape-related patterns of adaptation. It is timely now to quantify experimentally the physiological cost of flight under suboptimal thermal con-ditions in terms of fitness in both sexes.

Thanks are due to K. Lauwers and M. Jacobs for assistance and to S. Jakobsson for provided facilities at Tove-

torp research station. This research was funded by the University of Antwerp (GOA 15R/3942), the Belgian

Federal Office of Scientific, Technical and Cultural Affaires (OSTC-PADD II EV/06/16B) and a grant to B.K.

from the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning. T.M. had

a PhD-grant of Scientific Research Flanders-Belgium (FWO-research aspirant).

}Discussion

}Acknowledgements

Chapter 5 { Variation in mate location }

{ 55 }

5

}

Variation

Summary

Mate location behaviour of the butterfly Pararge aegeria in woodland and fragmented landscapes


Chapter

Visually cued mate location behaviour in insects such as butterflies is typically clas-sified by ‘wait or seek’ dichotomies. Perching males adopt a sit-and-wait strategy at a particular spot rising to intercept passing females (which is often done in an aggressive territorial way), whereas patrolling males are permanently on the wing searching for females. The potential influence of changes in landscape caused by habitat fragmenta-tion on mate location behaviour has only rarely been addressed. We investigated this behaviour among populations of the speckled wood butterfly, Pararge aegeria, living in continuous woodland versus highly fragmented agricultural landscape with hedgerows and small patches of woodland. In the latter landscape males showed higher levels of aggressive fast take-offs (an indicator of territorial perching), but also higher levels of displacement (an indicator of patrolling). In an independent census, behaviour inter-mediate between perching and patrolling was much more frequent in the agricultural landscape than in the woodland landscape. Our results suggest that the dichotomy of perching versus patrolling as typically observed in woodland fades away in highly frag-mented agricultural landscape. We discuss our results in relation to differences in den-sities and thermal properties of both types of landscape.

in mate location

{ 56 }


{ 57 }

}IntroductionVariation in visually cued mate location behaviour in insects has widely been classified using ‘wait or seek’ dichotomies (e.g. Odonata: Heinrich & Casey 1978; Thysanoptera: Crespi 1988; Hymenoptera: Alcock & Houston 1996). In butterflies, Scott (1974) defined the variants as perching and patrolling, respectively. Perching males adopt a sit-and-wait strategy at a particular spot rising to intercept passing females, whereas patrolling males are permanently on the wing searching for females. As male butterflies spend most of their active time locating mates (Shreeve 1992a), different mate location strate-gies may have consequences for their local distribution and daily movements. Some but-terfly species are exclusive patrollers and others predominantly perchers, whereas in some other species both behavioural strategies occur (Dennis & Williams 1987, Wickman 1988, 1992, Shreeve 1992a, Van Dyck 2003). Dennis & Shreeve (1988) argued that patrol-ling is the ancestral behaviour; the evolution of perching depends on habitat structure, because the presence of distinct landmarks (e.g. hills, host plant clumps and sunlit patches) is essential for a perching strategy. Male butterflies can also show aggressive, territorial behaviour. Although territoriality is most often connected to perching, they are not synonymous. Some species show perching without any aggression towards con-specific males (Shreeve 1992a). Assuming that territorial behaviour is costly, the territo-ry economics hypothesis predicts that monopolizing a perch with aggressive behaviour is viable only under low to moderate male density (Parker 1978, Thornhill & Alcock 1983, Rutowski 1991). In other words, high population density would favour nonterritorial patrolling (Scott 1974, Willmer 1991).

Although alternative mate location strategies have attracted much attention (Thorn-hill & Alcock 1983, Rutowski 1991, Wiklund 2003 and references therein), the poten-tial influence of changing landscape structures on mate location behaviour has only rarely been addressed. Habitat fragmentation provides a typical framework for studies on changes in dispersal behaviour (e.g. Hill et al. 1996, Thomas et al. 1998), but other be-haviours, including mate location, may also be changed (Van Dyck & Matthysen 1999). Habitat fragmentation may simply affect densities and, hence, frequencies of territorial perching and nonterritorial patrolling males in systems where both strategies occur. However, things can be more complicated in fragmented landscapes when conditions favouring one of the strategies change as well, but not necessarily in parallel with changes in density. Hence, habitat fragmentation may alter cost - benefit balances of alternative mate location strategies.

We tested for differences in mate location behaviour among populations of the speckled wood butterfly, Pararge aegeria L., living in continuous woodland versus highly fragmented agricultural landscape with some small patches of woodland and hedgerows. Males either perch in an aggressive territorial way on a sunlit patch on the forest floor, or they patrol by flying from one sunlit patch to the other through more shady conditions (Wickman & Wiklund 1983, Shreeve 1987, Van Dyck 2003 and references therein). Under

relatively cool ambient conditions, as in spring or in closed shaded coniferous woodland, the value of defending a sunlit patch is high and territorial perching is then observed as the dominant strategy (Wickman & Wiklund 1983, Shreeve 1984, 1987). Furthermore, in these heliotherms patrolling flights are constrained by local climatic conditions, which are the result of both the weather and the habitat structure (Dennis & Shreeve 1988, Wickman 1988). Flight duration increases in relation to air temperature, convective body cooling being slower at high temperatures when the difference between optimal body temperature and air temperature is less (Shreeve 1984, Van Dyck & Matthysen 1998).

From a speckled wood’s perspective, highly fragmented landscapes are likely to pro-vide a different template for mate location behaviour than continuous woodland, for three reasons that we investigated in this study. First, densities per unit of habitat are likely to be higher in fragmented landscapes. Woodland represents a continuous area in terms of resources in general and for mate location sites (e.g. sunlit patches) in particular. In highly fragmented landscapes, however, resources are restricted to ‘lines’ (e.g. hedgerows) and ‘bits’ (e.g. patches of woodland) which may lead to a con-centration of individuals (Chardon et al. 2003). In other words, densities per unit of habitat are expected to be higher in fragmented landscapes. Hence, in line with the territory economics hypothesis, we predicted less perching but more patrolling for frag-mented landscapes. Second, thermal conditions differ considerably between woodland and hedgerows (or patches of woodland) in highly fragmented agricultural landscapes. Measurements of ambient temperature at suitable P. aegeria habitat during flight activity showed that average and maximal daily temperatures are higher in fragmented agri-cultural landscapes than in woodland (Merckx, Van Dongen, Matthysen & Van Dyck, unpublished). So, from a thermal constraint hypothesis, we predicted a shift towards higher levels of local mobility with fragmentation. To reduce the impact of convective cooling in open fragmented landscapes, butterflies should seek shelter and fly closer to the vegetation than in woodland. Finally, the occurrence of landmarks for mate loca-tion may vary with landscape structure. Sunlit patches are landmarks where territorial perching males settle and patrolling males fly from one sunlit patch to the other. There are no oviposition or feeding sources in such a patch: it is only a spot clearly delineated by solar radiation penetrating through holes in the canopy. In fragmented agricultural landscapes, a considerable proportion of P. aegeria habitat occurs as relatively small lines such as hedgerows and sunken roads. Clearly delineated sunspots are rare and a large part of the hedgerow is irradiated. Under such conditions males are predicted to make longer flights than typical territorial perchers do in woodland. In woodland, the major-ity of males show either a clear perching or patrolling behaviour, although a minority may have an intermediate behaviour (Van Dyck et al. 1997b). We tested these predictions using data from behavioural observations and mark - release - recapture programmes in woodland and a highly fragmented agricultural landscape in central Belgium.

{ 58 }


{ 59 }

study speciesThe speckled wood is a temperate-zone satyrine butterfly (Lepidoptera: Nymphalidae). Throughout Europe, it is primarily a woodland species (Tolman & Lewington 1997). In the northern parts of the European distribution it is confined to woodland, whereas more to the south it also occurs in more fragmented landscapes with some woodland aspects such as agricultural landscapes with hedgerows and small patches of woodland (Dover & Sparks 2000, Merckx et al. 2003). Honeydew is the main adult feeding resource and several grasses are used as larval host plants (Shreeve 1986a). In woodlands, females lay eggs singly on isolated grass plants that grow under shaded, rather wet conditions (Wiklund & Persson 1983). In Belgium, P. aegeria is a common and widespread species (Maes & Van Dyck 1999).

study areasWe studied male mate location behaviour of P. aegeria in central Belgium at four differ-ent sites representing the two contrasting landscapes: woodland and a highly fragment-ed agricultural landscape. Meerdaalwoud (1255 ha) and Bos ter Rijst (280 ha) represent closed landscapes dominated by oak woodland. Hoegaarden (603 ha) and Rillaar (361 ha) represent open, agricultural landscapes. The latter type of landscape was composed mainly of intensively used fields and pastures (Hoegaarden: 85%; Rillaar: 67% of the to-tal area/landscape), urban areas (4% and 20%, respectively), tiny patches of woodland (3%) and sunken roads, which are eroded agricultural tracks, bordered by steep grassy and woody banks (7% and 3%, respectively). The sunken roads are<0.5 m to several m deep, 5 - 25 m wide (Gulinck et al. 1991) and from 80 (but sometimes with gaps of some m) to 1370 m long.

behavioural observationsIn spring and summer 2002, we observed the behaviour of 62 P. aegeria males under fine weather conditions that allowed butterfly activity: 26 in the woodland landscape (Meerdaalwoud: 17; Bos ter Rijst: 9) and 36 in the agricultural landscape (Hoegaarden: 22; Rillaar: 14). During an observation session, an individual male was followed for 5 min. Different activities (flying, basking and resting) were recorded with a minidisk (Sony MZ-N710). When the male took flight, we alos recorded take-off speed (slow ver-sus fast) and flight altitude (± 0.5 m). Labelled sticks were placed at positions where individuals rested or changed flight direction. After the observation session, we meas-ured the positions of all sticks and distances between sticks, with a laser instrument (Leica DISTO classic3, Leica Geosystems AG, Heerbrugg St. Gallen, Switzerland). Meas-urements were stored on schematic drawings of the flight tracks. We calculated the fol-lowing measures for each observation session: proportion of time spent flying, number of displacements, (proportional) number of fast take-offs, mean altitude of flight and

mean and total distance moved. For each observation session, we also recorded ambient temperature (± 0.1°C) and relative wind velocity (Beaufort scale). After the observation session, individuals were captured by hand net and killed and stored at -18°C. Individu-als that did not adopt mate location behaviour (but exclusively foraged or rested) were excluded from analyses. As a result, analyses were run on 50 individuals. During be-havioural observations, ambient temperature was lower in the agricultural than in the woodland landscape (19.6 ± 0.3°C versus 23.1 ± 0.7°C; Student T test: t

25.3=4.74, P<0.0001),

and mean wind force was higher in the agricultural than in the woodland landscape (Fisher’s exact test: P=0.006).

mark - release - recapture dataWe collected mark - release - recapture (MRR) data during six surveys (8 - 13 August 2000) at one agricultural landscape area (Rillaar) and during five surveys (23 - 29 August 2000) at one woodland landscape area (Meerdaalwoud). Surveys were conducted only during periods with fine weather conditions. To obtain comparative sampled areas, we did MRR experiments in a part of each study landscape: 81 ha/4.9 km and 138 ha/7.4 km (area/perimeter ratios of Rillaar and Meerdaalwoud 165.3 and 186.5, respectively). The largest recordable movements were 1550 m and 2200 m, respectively. During each survey, we checked the whole area for males and females, and altered the trajectory regularly. We captured butterflies by hand net and marked them individually at first capture with unique numbers on the ventral side of the left hindwing with fine, non-toxic, permanent markers (Staedtler Lumocolor Permanent marker 313, Staedtler, Nürn-berg, Germany). On first capture, butterflies were released at the site of capture. For second captures, butterflies were collected and stored in the laboratory (-18°C). For each capture we recorded data, time, exact position and mark number. Before each capture, we observed the mate location behaviour for several minutes, to classify a male as (1) percher, (2) patroller or (3) showing a behaviour intermediate between perching and patrolling (see method applied by Van Dyck et al. 1997b). Distances of movements be-tween capture and first recapture were measured on detailed maps of the study sites. Distances were log

transformed prior to analyses.

statistical analysesWe used mixed regression models (Proc Mixed, SAS 2001) to analyse the behavioural data. All time, length and number variables were expressed as proportions of total observation time. Since we were mainly interested in the effects of landscape (open versus closed) and of ambient temperature, we used these variables and their interaction as fixed effects in the models. As two study sites represent each landscape structure, ‘site’ nested within ‘landscape’ was used as a random effect in each model. As data were collected during two periods (period 1: end of April - start of June; period 2: mid August), we also included pe-riod as a random effect. Backward selection of nonsignificant factors was done to obtain

}Methods

{ 60 }


{ 61 }

a final model. All final models had normally distributed residuals. Fisher’s exact tests (Proc Freq, SAS 2001) were used to compare frequency distributions between woodland and agricultural landscapes for mean flying height (< 1.5 versus>1.5 m).

For the analyses of the MRR data we used mixed regression models (Glimmix macro, SAS 2001). Since tracked distances did not follow a normal distribution, we reduced this information to a binomial variable: movements<50 m or>50 m. Landscape (woodland versus agricultural landscape) was used as a fixed effect. Next, mixed regression models (Proc Mixed, SAS 2001) were run on the subsample of males that covered distances of more than 50 m, with tracked distance as the dependent variable. Again, landscape (woodland versus agricultural landscape) was used as a fixed effect. Encounter rate of butterflies was used as a proxy for butterfly density among sites. We calculated this rate by dividing the number of observed individuals by the total observation time for each MRR survey day. Next, we compared mean values between the study sites or landscapes (Proc Ttest, SAS 2001). Frequencies of three types of mate location behaviour (perch-ing, patrolling and intermediate behaviour) within the MRR survey were compared between woodland and agricultural landscapes (χ², Proc Freq, SAS 2001).

Densities of both sexes, expressed as encounter rate within suitable habitat, were sig-nificantly higher in the agricultural landscape than in the woodland landscape (0.19 versus 0.13 individuals/min; t

8.89=-3.26, P=0.010). In the agricultural landscape, males

performed significantly more fast take-offs (X ± SE=2.9 ± 0.4, range 0 - 8 versus 1.5 ± 0.3, range 0 - 5 per 5 min) but also more displacements (5.2 ± 0.6, range 1 - 16 versus 3.2 ± 0.5,

range 1 - 7 per 5 min) than did males in the woodland landscape (Figure 5.1a, b, Table 5.1). The frequency of fast take-offs also tended to increase with increasing ambient temperature. Males from the agricultural landscape covered larger total distances dur-ing the behavioural observation session than conspecifics from the woodland landscape (X ± SE=48.2 ± 6.9 m, range 3.5 - 173.5 m versus 32.4 ± 5.1 m, range 3.3 - 103.4 m). However, this difference switched at relatively high ambient temperatures (Figure 5.1c, Table 5.1). Similarly, the mean displacement or step length was larger for males in the agricultural landscape than for woodland males at low ambient temperatures, but the relation re-versed at relatively high ambient temperatures (Figure 5.1d, Table 5.1). Landscape had no effect on the proportion of time spent flying among males, but males flew more frequently at low heights in the agricultural landscape than in the woodland landscape (Fisher’s exact test, P=0.008).

}Results

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

Figure 5.1. (a) Number of fast take-offs (X ± SE), (b) number of displacements (X ± SE), (c) total displace-ment length and (d) mean displacement length of male P. aegeria butterflies in a woodland landscape (n=20) and in a fragmented agricultural landscape (n=30) under natural conditions, proportional to observation time.

variable effect df F Pnumber of fast take-offs landscape

temperature 1, 461, 46.6

10.223.72

0.0030.060

number of displacements landscapetemperature

1, 461, 46.2

5.760.54

0.0210.47

total displacement length landscapetemperaturelandscape x temperature

1, 461, 461, 46

7.750.586.97

0.0080.450.011

mean displacement length landscapetemperaturelandscape x temperature

1, 45.81, 45.91, 45.8

8.151.518.57

0.0060.230.005

Table 5.1. Results of mixed regression models (Mixed procedure, SAS) for the proportional number of fast take-offs and displacements and total and mean displacement length in relation to the landscape of origin of male P. aegeria butterflies (woodland or fragmented agricultural landscape) and ambient temperature at the start of the observation

(a) (b)

(c) (d)

{ 62 }


{ 63 }

At the agricultural landscape site, we captured 158 males and 47 females; 30.4% of the males were recaptured, but no females were. At the woodland landscape site, we captured 105 males and five females; 37.1% of the males and one female were recaptured. Females therefore had lower recapture frequencies than males, especially in the agri-cultural landscape site (χ²

1=18.64, P<0.0001). Overall recapture percentages (i.e. for both

sexes) and the proportion of mobile (≥50 m) to sedentary (< 50 m) males did not differ with landscape structure. Within the subsample of mobile males, tracked distances did also not differ between the landscape types.

The frequency of males showing pure territorial perching behaviour was significant-ly lower in the agricultural landscape than in the woodland landscape (based on first capture data; 42 versus 80%; χ²

1=37.05, P<0.0001). Mate location behaviour intermediate

between perching and patrolling was relatively rare in the woodland landscape, but much more common in the agricultural landscape (10 versus 48%; χ²

1=40.82, P<0.0001).

Our study showed significant differences in the mate location behaviour of male P. ae-geria butterflies living in a woodland landscape versus a highly fragmented agricultural landscape. In the latter landscape, males more frequently performed fast take-offs from a resting posture than in the woodland landscape. Such take-offs are typically associ-ated with territorial perching (see Rutowski 1991 and references therein). Since densi-ties per unit of habitat were higher in the agricultural landscape, this observation is in contrast to the territory economics hypothesis and our first prediction of more patrol-ling rather than perching in the fragmented landscape. On the other hand, agricultural landscape males made more and wider displacements, associated with patrolling, than woodland males. This result is in line with the thermal constraint hypothesis and the second prediction of higher mobility in the fragmented landscape; during observations, however, ambient temperature was lower in the agricultural landscape, so that the as-sumed higher average ambient temperature is not likely to explain the difference. In the independent census, we found in the same vein a much higher frequency of behav-iour intermediate between perching and patrolling in the agricultural landscape than in the woodland. This suggests that the clear dichotomy of perching versus patrolling as typically observed in woodland (Van Dyck 2003 and references therein) fades away to some extent in a highly fragmented, agricultural landscape.

In earlier studies on P. aegeria the vast majority of males could be classified as either percher or patroller (Davies 1978, Wickman & Wiklund 1983, Shreeve 1987, Van Dyck et al. 1997a, b). Males are consistent in adopting one of the two strategies, although switches may occur (Van Dyck et al. 1997b). In addition to males switching, we observed that some males also combine traits of both strategies into intermediate mate location behaviour. Fischer & Fiedler (2001) showed in another butterfly that aggressive, territorial males also used a flexible combination of perching and patrolling. Our study shows that the

degree to which P. aegeria males do so depends on the type of landscape. Although we found clear differences in mate location behaviour with landscape, the

underlying mechanism requires further analysis. Mate location strategies have been de-scribed in relation to demography but in many ectotherms (including butterflies) mate location behaviour is strongly affected by thermal conditions (Willmer 1991). Wickman (1985, 1988) concluded for two other butterflies that population density is of small, often negligible, importance for determination of mate location behaviour.

Fast take-offs from a resting posture are facilitated at higher temperatures (Ber-waerts & Van Dyck 2004). We found a trend towards increased frequency of fast take-offs with increasing ambient temperature. In an open agricultural landscape, ambient temperature is typically higher than in deciduous woodland, where there is a cooler, more buffered microclimate (Merckx, Van Dongen, Matthysen & Van Dyck, unpub-lished). However, the influence of wind by convective cooling can be much more severe along a hedgerow in open landscape than on a sunlit patch in woodland (Merckx, Van Dongen, Matthysen & Van Dyck, unpublished). In line with this, we observed that males more frequently flew low down close to the vegetation, which helps to avoid convective cooling (Merckx, Van Dongen, Matthysen & Van Dyck, unpublished). In two other but-terflies, individuals compensated for higher levels of wind speed, and hence convective cooling, by flying closer to structures that provided shelter (Merckx & Van Dyck 2002).

The higher frequency of intermediate mate location behaviour in the agricultural versus woodland landscape can be explained from an adaptive point of view. The com-bination of aggressive territoriality and movements could permit males in more linear, relatively narrow landscape structures to adopt a kind of ‘customs strategy’ towards conspecific males and females. Males would thus be able to keep an entire section of a hedgerow under surveillance. The underlying assumption is that females in an agri-cultural landscape fly especially along or in hedgerows and sunken roads rather than through the open space. Circumstantial evidence of this comes from censuses where we collected females in this landscape (TM, pers. obs.). Furthermore we have experimen-tally shown that females originating from an agricultural landscape have lower pro-pensities to cross woodland - open space boundaries than do females from a woodland landscape (Merckx et al. 2003).

On the other hand, the higher frequency of intermediate mate location behaviour in the agricultural than in the woodland landscape could be nonadaptive and related to physical, optical differences in the occurrence of sunlit patches with landscape struc-ture. These landmarks for mate location are more clearly delineated in woodland, where solar radiation penetrates through holes in the canopy; in a fragmented agricultural landscape such clearly delineated sunspots are scarcer and a considerable proportion of the linear habitat structures are completely irradiated. Under such conditions, males cannot use the contrast of sunshine and shade as a physical border of a territory. In woodland, males follow sunlit patches when patches move during the day, indicating

}Discussion

{ 64 }

that only the simple light - shade contrast is used rather than particular vegetation structures per se. This nonadaptive, constraint explanation is obviously not mutually exclusive with the former explanation.

In conclusion, our study illustrates that behavioural consequences of habitat frag-mentation are not limited to, for instance, dispersal, but behaviours related to reproduc-tion such as mate location may also change. There are no reasons to believe that such changes are unique to butterflies.

Thanks are due to E. Janssens, K. Borghgraef and J. Cortens who assisted with fieldwork. The University of

Antwerp (GOA 15R/3942) and the Belgian Federal Office of Scientific, Technical and Cultural Affaires (OSTC-

PADD II EV/06/16B) funded this research. T.M. is research aspirant with the Fund of Scientific Research

Flanders-Belgium (F.W.O.).

}Acknowledgements


entation and butterfly m

ovements }

{ 65 }

6

}As landscapes change, mobility patterns of species may alter. Different mechanistic sce-narios may, however, lead to particular patterns. Here, we tested conflicting predictions from two hypotheses on butterfly movements in relation to habitat fragmentation. Ac-cording to the resource distribution hypothesis, butterflies in more fragmented land-scapes would have higher levels of mobility as resources are more scattered. On the oth-er hand, these butterflies could have lower levels of mobility as they experience more frequently ‘hard’ habitat boundaries (i.e. higher crossing costs) compared to butterflies in landscapes with continuous habitat; i.e. the behaviour-at-boundaries hypothesis. We studied movements, habitat boundary crossing and habitat preference of laboratory reared individuals of Pararge aegeria that originated from woodland and agricultural landscapes, using an experimental landscape as a common environment (outdoor cages) to test the predictions, taking into account sexual differences and weather. Woodland butterflies covered longer distances, were more prone to cross open-shade boundaries, travelled more frequently between woodland parts of the cages and were more at flight than agricultural butterflies. Our results support the behaviour-at-boundaries hypoth-esis, with ‘softer’ boundaries for woodland landscapes. Since the butterflies were reared in a common environment, the observed behavioural differences rely on heritable vari-ation between populations from woodland and agricultural landscapes.

Habitat fragmentation

Summary

The evolution of movements and behaviour at boundaries in different landscapes: a common arena experiment with butterflies

{ Thomas Merckx, Hans Van Dyck, Bengt Karlsson & Olof Leimar }

butterfly movementsChapter

and

{ 66 }



ovements }

{ 67 }

As habitats become more fragmented, changes in habitat patch geometry and quality may affect the costs and benefits of dispersal, resulting in changed dispersal rates among differently fragmented landscapes (Thomas 2000). If a continuous distribution of resources has broken up to a scattered distribution, responses towards higher or lower mobility can be expected depending on the spatial scale of fragmentation, or the spread of resources, relative to the maximum range of exploration and dispersal of a species (Van Dyck & Matthysen 1999). Both theoretical models (e.g. Leimar & Norberg 1997, Ol-ivieri & Gouyon 1997) and indirect empirical evidence from changes in flight morphol-ogy (e.g. Dempster 1991, Thomas et al. 1998) have suggested evolutionary responses as a result of selection on mobility in relation to landscape structure.

More recently, behavioural aspects of dispersal have received more attention (e.g. Jonsen & Taylor 2000a, Conradt et al. 2001, Bowman et al. 2002, Hanski et al. 2002) and several studies suggest a direct relationship between variation in behavioural responses at habitat boundaries and variation in dispersal (e.g. Haddad 1999, Schultz & Crone 2001, Berggren et al. 2002). Ries & Debinski (2001) showed that a less mobile habitat specialist butterfly returned more frequently at habitat boundaries than did a more mobile habitat generalist. Moreover, the specialist responded strongly to different types of boundaries, while the generalist responded only to boundaries of high contrast. At the intraspecific level, returning behaviour at boundaries occurred more frequently in a highly fragmented habitat network compared to a more aggregated network in the localized butterfly Proclossiana eunomia (Schtickzelle & Baguette 2003).

In this paper, we compare movements and behaviour of speckled wood butterflies (Pararge aegeria L.) from differently fragmented landscapes – woodland versus agricul-tural landscape – in an experimental common environment consisting of large outdoor cages. Our main purpose is to test predictions on adult butterfly mobility that follow from a resource distribution hypothesis versus a behaviour-at-boundaries hypothesis. For butterflies, the presence of suitable egg-laying sites is an important habitat charac-teristic. For P. aegeria in agricultural landscapes, habitat occurs in the form of hedgerows and tiny woodlots, and will be much more fragmented than the corresponding habitat in woodlands. One might then predict higher mobility in agricultural landscapes, for instance a higher incidence of long flights, as an adaptation to the scattered resource distribution (Taylor & Merriam 1995). On the other hand, one can make the opposite prediction when considering differential behavioural responses at habitat boundaries with type of landscape. For butterflies from a woodland landscape, most boundaries are soft boundaries in the sense that crossing them will bring a butterfly to another piece of woodland habitat after a few to tens of meters of open, but mostly sheltered, areas such as rides or clearings. In contrast, butterflies from agricultural landscape would rather use boundaries, like hedgerows, as their habitat per se, and leaving them can be inter-preted as crossing hard boundaries in the sense of encountering open, unsheltered ar-

eas for tens to hundreds of meters (e.g. fields and pastures). Hence, as boundaries mean on average different things in the two landscapes, one may predict higher incidence of boundary crossing, or higher mobility at this local scale, in woodland butterflies com-pared to agricultural landscape butterflies.

Comparative field studies between landscapes may point at relevant behavioural dif-ferences, but are less powerful to test for selection on mobility as they cannot rule out behavioural flexibility. We are particularly interested in selection on behavioural re-sponses and movements relative to landscape structure, which has often been assumed – for instance to interpret morphological variation among landscapes (e.g. Hill et al. 1999a)

– but has far less been directly tested. Therefore, we adopted an experimental approach by releasing butterflies from woodland and agricultural landscapes simultaneously in a common environment. We used naive butterflies that were reared under controlled conditions in the laboratory from females collected in both types of landscape. Hence, any behavioural differences would be a consequence of heritable variation.

Besides the effect of landscape of origin, we also examined sexual differences in mobility and how movements and behaviour were affected by weather conditions. In several butterflies (including P. aegeria), females have been considered the more disper-sive sex (Baker 1984). However, males are likely to be more frequently observed flying as they spend most of their time locating females. Two strategies of mate location behav-iour co-occur in P. aegeria: territorial perching (i.e. aggressive sit-and-wait behaviour on a sunlit patch) and patrolling (i.e. searching behaviour in more or less permanent flight in a wider area) (Davies 1978, Wickman & Wiklund 1983). Being flying heliotherms, weather conditions may have a severe impact on several aspects of butterfly biology, in particular on activity and movements (Dennis 1993). As ambient temperature increases, both males and females of P. aegeria fly for longer and they change microhabitat selec-tion within woodland habitats (Shreeve 1984). So, in this paper we test the conflicting predictions in relation to landscape structure, as well as sexual and weather related differences and their interaction effects.

study speciesThroughout Europe, the speckled wood (Pararge aegeria L.) is primarily a woodland but-terfly (Tolman & Lewington 1997), but may also occur in other habitat types with some kind of woodland aspect, including parks, gardens and along hedgerows. Adults are rarely seen on flowers, but mainly feed on other resources like honeydew. Several grass-es can be used as larval host plant (Shreeve 1986a). Eggs are deposited on isolated grass plants, surrounded by some bare ground, that grow under shaded, rather wet conditions (Wiklund & Persson 1983).

}Introduction

}Methods

{ 68 }



ovements }

{ 69 }

sampled landscapes and breeding procedure In Augusts 2002, females of P. aegeria were captured at several sites in two different types of landscape in Belgium. The first was a closed landscape dominated by decidu-ous woodland with differently sized clearings, pastures and rides (Meerdaalwoud). The other was an open, agricultural landscape composed of intensively used fields and pas-tures (c. 80% of the area), houses and farms (c. 10%), commercial orchards (c. 5%), tiny woodlots (c. 2.5%) and several sunken roads with hedgerows (c. 2.5%) (Rillaar). Hence, from a speckled wood’s perspective, Meerdaalwoud is a rather continuous landscape as far as egg laying sites and sheltered sunlit patches for mate location are concerned, while in Rillaar such sites can only be found along or in hedgerows and tiny woodlots that are scattered in the landscape. Collected females were allowed to oviposit on tufts of grass (Poa annua) in small cages in the laboratory. Next, eggs were transported to Stockholm where the individuals were reared on Poa annua in climate rooms under con-trolled conditions (22 °C, LD 22 : 2 h). After emergence, adults – from 6 woodland and 12 agricultural landscapes families – were sexed and individually marked by writing a unique number on the ventral side of both hindwings with a fine permanent, non-toxic and waterproof pen with silver ink. Until release in the outdoor cages, adults were stored in climate chambers (8 °C, LD 21 : 3 h).

observations in outdoor cagesBehavioural observations were done by one person between September 11-18 2002 in a manipulated environment in two identical, oblong outdoor cages with half-circle shaped cross sections (Manufacturer: Serres de France S.A., w x l x h: 8 x 30 x 4 m – Norberg et al. 2002). Both cages were located side by side with their longitudinal axes positioned in EW orientation in a sheltered pasture near Tovetorp field station (c. 100 km SW of Stock-holm, Sweden). The first cage was somewhat more exposed than the second cage that was positioned between the first cage and a third cage that was not used. The cage roof and walls consisted of fine-meshed netting. The experimental cage set-up was similar to Leimar et al. (2003): both ends – 8 m each – had a closed, rather shady habitat condition (i.e. mimic for woodland structure) and the central zone – 14 m – was an open meadow. The closed habitat, providing shade with scattered sunlit patches, was created by the use of camouflage netting and non-transparant tarpaulin in which holes of varying size were cut. In addition, several sheets of camouflage netting were hanging down from the roof and two artificial christmas trees were placed in each end part. In each cage we provided six adult feeding stations (i.e. sugar solution on sponge).

In each of the two cages five individuals of each sex and origin (woodland or agri-cultural landscape) – hence 20 individuals per cage in total – were placed at random positions. Densities were kept constant over time; if an individual was not observed during an entire day, it was replaced by another one of corresponding sex and origin. After five days of observations, all individuals were collected and replaced by a new

sample of individuals according to the same experimental design. In total, 92 individu-als were released.

A first group of individuals was observed during a three day period that was sig-nificantly warmer than the period of the second group of individuals (mean ambient temperature at noon: 20 ± 1 °C versus 14 ± 1 °C, U=0, n=3 and 5, P=0.036). Hence, weather condition was taken into account for the analyses as a categorical variable (warm versus cold weather).

During one-hour-long observation sessions, the observer walked slowly through the cage and binoculars were used to read individual marks on the wings from a distance without disturbing the butterflies. The location of each butterfly (using a 2 x 2 m grid system) and its behaviour (three categories: flying, resting and basking) were recorded. A subsample of flying individuals (n=75) was also followed until they were alighted; individual tracks and behaviours were recorded. From the observations (i.e. locations and behaviours) we inferred the following variables: (a) traversal rate (i.e. the number of movements from one shady end of the cage to the other divided by the time between the first and last observation, including only those parts of the day when weather con-ditions would allow butterfly activity; available time for active flight was assessed for each day separately), (b) moved distance rate within an observation session (i.e. the summed moved distance divided by the summed time between the first and last ob-servations within the session), (c) tracked distance (i.e. the summed moved distance during the times the individual was followed divided by the number of these tracking events), (d) crossing tendency (i.e. the number of open-shade boundary crossings per tracking event), (e) proportion of observations in the open (the middle third, not in-cluding the 2 m wide boundary sections of the cage), and (f) proportions of observations an individual was flying, basking or resting.

statistical analysesPreliminary tests (including variable transformation choice, tests for variance homo-geneity and detecting outliers) were done using R software version 1.6 (Ihaka & Gentle-man 1996). The individual butterfly was used as statistical unit. Data were analysed in relation to landscape of origin (woodland or agricultural landscape), weather condition (warm or cold), and sex. We also included cage as a factor, as well as interaction terms, but we dropped such terms from the final models if they neither improved the fit of a model nor gave rise to P-values smaller than 0.1.

To analyse the data on traversals, we used SAS version 8 (SAS 2001; GenMod proce-dure). Assuming that an individual has a given probability per time unit to traverse the open middle part of a cage, the number of traversals over a period of time will follow a Poisson distribution. For this reason, we used generalized linear models with Poisson response (and log link function). Effects of variation in the time available for traversal were taken into account by using the logarithm of this time as so-called offset variable

{ 70 }



ovements }

{ 71 }

in the models (this is analogous to a covariate). We judged the fit of a model by noting how close the scaled deviance per degree of freedom was to one.

For the other analyses we applied general linear models using the same SAS software (GLM procedure). We used the so-called Akaike Information Criterion (AIC) to judge the fit of a model. In the analysis of moved distance rate within an observation session, we excluded individuals with less than 15 min summed within-session time, since the random variation in the rate becomes very high for short times. For the measures based on proportions of observations (measures (e) and (f) above) we excluded individuals with five or less observations in total, to reduce random variation in the proportions. By suitable choice of transformations for the dependent variables, and in some analyses removal of outliers, we achieved homogeneous variances in the different tests (we also checked that any statistical significances did not depend on the removal of outliers). Means are given ± SE.

movements between shaded partsMales seemed to be somewhat more variable in their traversal rate between the shaded ends of the cage than females. This could reflect variation in mate location strategies, in which case the model assumption of Poisson distribution of the number of traver-sal might be violated. The best fitting model that included both sexes showed a higher traversal rate in woodland butterflies compared to agricultural butterflies, and some non-significant tendencies for the other factors and interaction terms (Table 6.1a). If sexes were analysed separately, we found a very good fit for the females (scaled devi-ance per df of 1.006). Woodland females moved more frequently from one shaded part to the other than did agricultural females and traversal rate was higher under warmer

weather conditions (Table 6.1b, Figure 6.1). On the other hand, the analysis for males showed a less good fit (scaled deviance per df of 1.124) and no significant effect of either landscape of origin or weather conditions (Table 6.1c, Figure 6.1).

movements and boundary crossing within observation sessionsFlight distances per time unit within observation sessions were longer in woodland butterflies than in agricultural butterflies and were longer under cooler weather con-ditions (Table 6.2a, Figure 6.2a). There was a non-significant tendency for a difference in those distances between the two cages as distance moved tended to be longer in the second cage. Within the subsample of followed individuals, there was no significant ef-fect on distance covered per tracking event (Table 6.2b). However, woodland butterflies were found to cross open-shade boundaries more often than did agricultural butterflies (Table 6.2c, Figure 6.2b).

habitat preferenceOnly 20% of all butterfly observations were in the open middle part of the cages. The open part was, however, more used under cooler conditions than under warm condi-tions, with no significant differences between sexes or landscape of origin (Table 6.2d, Figure 6.2c). Considering only the boundary stripes, butterflies were observed predomi-nantly along the open irradiated site of the boundary rather than in the shaded site (t

=6.73, d.f.=74, P<0.0001).

variable effect χ2 P

a) traversal rate (movements/h) both sexes

landscape sexweather landscape x sexlandscape x weathersex x weather

9.190.102.992.023.272.97

0.0020.750.0840.160.0700.085

b) traversal rate (movements/h) females landscape weather

17.3310.60

<0.00010.001

c) traversal rate (movements/h) maleslandscapeweather

1.770.06

0.180.81

Table 6.1. Results of a generalized linear model for the traversal rate between shaded ‘woodland’ parts in the outdoor cage in relation to the landscape of origin of the speckled wood butterflies (woodland or agri-cultural landscape) and weather condition (see Methods) when taking into account both sexes (a), females (b) and males only (c). The GenMod procedure was applied (SAS) using type III sums of squares.

��

��

��

��

��

��

��

��

��

��

��

}Results

Figure 6.1. Traversal rate of female (N = 43) and male (N = 45) speckled woods from woodland landscape origin and agricultural landscape origin under warm (white bars) and cooler weather conditions (black bars) in large outdoor cages. Traversal rate is expressed as the number of movements from one shady end of the cage to the other plotted against time between first and last observation, including only those parts of the day when weather conditions allow butterfly activity. We used weighted means (with corresponding time present as weight) ± SE (by bootstrapping).

{ 72 }



ovements }

{ 73 }

flight activityWoodland butterflies were more often observed flying than agricultural butterflies, but this effect of landscape was larger for females than for males. Nonetheless, males were more observed at flight compared to females. The proportion of observations of flying butterflies differed between the cages as butterflies were more often at flight in the sec-ond cage. There was no significant effect of weather condition (Table 6.2e, Figure 6.2d). As there were no significant differences for basking, the results for resting largely rep-resent the inverse of those for flying: woodland butterflies rested less than agricultural butterflies, and females more than males. There was an interaction effect of landscape of origin with weather: under relative cooler conditions butterflies were observed less frequently resting, and this effect was larger for woodland than for agricultural but-terflies (Table 6.2f).

Our experimental approach to compare behavioural responses and movements among speckled wood butterflies in a large cage set-up revealed differences with landscape of origin. Woodland butterflies covered longer distances, were more prone to cross the open-shade boundaries, travelled more frequently between the shaded cage ends and were more frequently at flight than agricultural butterflies. Hence, the results support the prediction of the behaviour-at-boundaries hypothesis rather than the contrasting

variable effect F P

a) moved distance rate within session (m/h)

F1, 63

landscapesexweathercage

4.620.566.143.36

0.0360.460.0160.071

b) tracked distance (m)

F1, 71

landscapesexweather

2.341.002.70

0.130.320.10

c) crossing tendency (crossings/tracking event)

F1, 71

landscapesexweather

6.653.130.06

0.0120.0810.80

d) proportion of observations in open part

F1, 70

landscapesexweathersex x weather

2.100.2417.332.85

0.150.63<0.00010.096

e) proportion of observations flying

F1, 67

landscapesexweathercagelandscape x sex

14.2111.401.0711.134.68

0.0010.0010.300.0010.034

f) proportion of observations resting

F1, 66

landscapesexweathercagelandscape x weathersex x weather

9.4011.405.983.804.443.33

0.0030.0010.0170.0550.0390.072

Table 6.2. Summary of the analyses of different movement and habitat preference variables in relation to the landscape of origin of the speckled wood butterflies (woodland or agricultural landscape), sex, weather condition and cage (see Methods) using general linear models (GLM procedure - SAS). Type III sums of squares were used.

}Discussion

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

Figure 6.2. Movement, habitat preference and behavioural variables of speckled woods in relation to land-scape of origin (woodland versus agricultural landscape), sex and weather (white bars = warm; black bars = cold – see Methods) as derived from observations in large outdoor cages: (a) distance within session per time unit, using weighted means (with corresponding time present as weight) ± SE (by bootstrapping), (b) boundary crossing tendency, (c) proportion of observations in the middle third of the cages for individuals with six or more observations (line represents random expectation), and (d) proportion of observations of flying butterflies.

(a) (b)

(c) (d)

{ 74 }



ovements }

{ 75 }

prediction of the resource distribution hypothesis. Particularly for the traversal rate between the shaded ends, the differences between landscapes were clearly present in females. In a similar cage set-up, Leimar et al. (2003) found higher traversal rates in speckled woods compared to another woodland satyrine Lopinga achine (both from Swe-den) which agrees with their differences in habitat exploration and dispersive nature. Within P. aegeria, a Swedish woodland population (N-European subspecies tircis) was found to be more mobile than a Madeiran population (S-European subspecies aegeria) living in more open habitat (Leimar et al. 2003). Our results suggest similar behavioural differences at a much smaller geographic scale within Belgian populations occurring in woodland versus agricultural landscapes.

The naive butterflies used in our experiment were the F1-generation (reared under common garden conditions in the laboratory) of wild-caught females from both land-scape types. Hence, this study provides evidence for heritable behavioural differences

– a prerequisite for selection – and suggests that differences among landscapes do not follow from behavioural flexibility alone. So, in addition to environmental differences among the landscape types that may directly mediate differences in behaviour, there are also intrinsic differences in responses.

Although there are no empirical studies yet available on long-distance dispersal movements and capacity in speckled woods (Hill et al. 1999a), one may argue whether the observed differences have any relevance for dispersal. Observed movements in the outdoor cage mainly reflect explorative or ‘daily’ movements (Leimar et al. 2003, Nor-berg et al. 2002). In a study by Van Dyck et al. (1997a) recapture distance of males within a complex of woodland fragments was on average less than 40 m within woodland veg-etation, but some males moved between fragments covering up to 450 m. The spatial dimensions of the used cages do not allow a direct comparison of dispersal capacity per se. However, several recent studies have pointed out the significance of behaviour at boundaries to emigration decisions (Haddad 1999, Ries & Debinski 2001, Schultz & Crone 2001, Morales 2002, Schtickzelle & Baguette 2003). It is important to realise that behaviour at boundaries relates to dispersal propensity and not necessarily to dispersal capacity (in terms of covered distance), although both could be interrelated as well. Variation in the behaviour at boundaries can be studied at relatively small scales, like in our cages. Therefore, our results are at least relevant in terms of habitat exploration and dispersal propensity. Lower traversal rates and boundary crossing for speckled woods in agricultural landscapes do make sense as resources are much scarcer and travelling in open, unsheltered land is likely to be more costly for this primarily woodland butter-fly. So, our study suggests that behavioural responses that affect dispersal in speckled woods have been changed by selection as woodland landscapes change to agricultural landscapes (i.e. increasing degree of habitat fragmentation).

Only little is known on landscape perception distances in butterflies (Conradt et al. 2000). However, the two separated ‘woodland parts’ in the experiment are within the

perceptive – most likely visual – detection range of speckled woods (Merckx & Van Dyck, unpublished). Therefore, it would be highly relevant to test their movements in a much longer cage (e.g. length of 100 m or so). The potential role of visual stimuli (like another piece of woodland within a few meters of open area) relative to boundary crossing deci-sions requires further research. We hypothesize that this may contribute significantly to traversal rates (particularly in woodland species), and that its role varies among popula-tions from woodland or agricultural landscapes. Hence, we predict that, in a much long-er cage, woodland butterflies would move less between the woodland parts compared to agricultural butterflies, and that the overall mobility will be lower than in the current experiment, assuming all other factors (e.g. weather conditions, densities) being equal.

Speckled wood males were more frequently observed flying than females. This is in agreement with basic behavioural differences between the sexes in butterflies as males spend most of their active time budget to mate location, while females rest and bask during considerable time periods between oviposition, foraging and dispersal flights (Shreeve 1992a). This sexual difference in time budget use probably caused that the effect of landscape on flight activity – woodland butterflies were more often observed flying than agricultural butterflies – was larger in females than in males (i.e. signifi-cant landscape x sex interaction). Higher flight incidence is not always related to higher levels of net displacement (Merckx & Van Dyck 2002). The two distinct types of mate location behaviour that co-occur in speckled woods and also in our cages – territorial perching versus patrolling – provide an explanation for the less good fit to Poisson dis-tribution of traversals among males compared to females. Territorial perchers are more likely to stay in one part compared to patrollers who will move on.

In contrast to movement rates and crossing tendency, we did not find differences in habitat preference among the woodland and agricultural butterflies; all prefered the shady, ‘woodland’ parts. However, there was a significant effect of weather condition. Under cooler weather, more butterflies were observed in the open, more irradiated part of the cage. This is in agreement with thermal microhabitat selection under natural con-ditions; in a British woodland population Shreeve (1984) observed that speckled woods were more frequently present on rides and clearings (i.e. higher levels of irradiance and ambient temperature) than on small sunlit patches in the woodland when ambient temperature was low, but the use of the latter type of microhabitat increased with tem-perature. Weather condition also affected female traversal rate in the predicted way; flight duration increased with ambient temperature (Shreeve 1984), which resulted in higher traversal rates under warmer weather conditions. There was no effect of weather on flight incidence, but as habitat preference changed towards warmer, irradiated sites under cooler conditions (see above), the butterflies were capable to compensate and keep the time budget for flight unchanged (at least within the weather conditions of the ex-periment). Proportions of butterflies seen at flight or at rest are obviously not entirely independent (see Results), but besides the effect of sex, there were highly significant

{ 76 }

differences with landscape of origin for both variables. Moreover, in case of the analysis of proportion of resting, it is intruiging to notice that the effect of weather was larger in woodland than in agricultural butterflies (i.e. significant weather x landscape inter-action). Being flying heliotherms, both landscape types have quite different thermal properties for butterflies and the degree of variation is likely to interact with weather conditions. In our common arena approach, both populations were faced with the same environmental set-up, and so the differences point at intrinsic differences how wood-land and agricultural speckled woods deal with the environment. Such behavioural differences are likely to be underpinned by thermal physiological differences.

So far, the majority of research on dispersal has in particular focused on movement patterns (e.g. frequency distributions of distances, maximum distance, etc.). However, mechanistic research on movements is, as a complementary field, also required for a better understanding of dispersal and how this is affected in fragmented landscapes. Hill et al. (2001) observed a c. 45% slower expansion rate in speckled woods in a land-scape that had 24% less woodland. However, different mechanistic scenarios could lead to such a pattern, and remediating conservation strategies in fragmented landscapes would benefit considerably from a better understanding of the mechanisms. Hedgerows have often been assumed to be corridors for woodland species, but for butterflies (among several other taxa) there is hardly empirical evidence for this (Dover & Sparks 2000). From our experiments and preliminary work on woodland perception from varying distances (Merckx & Van Dyck, unpublished), we hypothesize that woodland butterflies may be encouraged to cross a woodland boundary if there is a landscape feature that is perceived as woodland (e.g. a hedgerow). Hence, networks of hedgerows would rather function according to a stepping stone principle than to corridors sensu stricto. This re-quires careful testing, and we, more generally, believe that further experiments on be-haviour at habitat boundaries and on landscape perception will contribute significantly to conservation strategies at the landscape level.

Thanks are due to K. Borghgraef and M. Miettinen who assisted with collecting and rearing of the butterflies.

TM is research aspirant and HVD postdoctoral fellow with the Fund of Scientific Research Flanders-Belgium

(F.W.O.). This research is funded by the University of Antwerp (GOA 15R/3942), by the Belgian Federal Of-

fice of Scientific, Technical and Cultural Affaires (OSTC-PADD II EV/06/16B), by a grant from the Swedish

Research Council to OL and by a grant from the Swedish Research Council for Environment, Agricultural

Sciences and Spatial Planning to BK.

}Acknowledgements

Chapter 7 { Landscape and habitat-finding ability }

{ 77 }

7

}

Landscape

Summary

Landscape-related differences in the ability to find habitat between populations of a woodland butterfly


Chapter

1. The ability of an organism to find new habitat is likely to contribute to dispersal suc-cess in different landscapes.

2. We compared the ability of speckled wood butterflies (Pararge aegeria L.) from two types of landscape to orient towards forested habitat as a behavioural assay to the ability to perceive forested habitat at a distance (25-200 m).

3. This field release-experiment showed that butterflies of fragmented agricultural landscape origin were better able to orient towards habitat from a wider distance than butterflies of woodland landscape origin. They followed more linear tracks, dis-played flight angles that deviated less from orientation towards habitat, and showed orientations that were significantly directed towards habitat from wider distances than butterflies of woodland origin. If target habitat was more apparent, attraction was facilitated.

4. Our behavioural experiment provides novel insights into landscape connectivity and points at inter-population differences in the ability to find habitat associated with landscape type.

habitat-findingand

ability

{ 78 }


{ 79 }

}IntroductionAlthough dispersal is widely recognised to be a process with important ecological and evolutionary implications, there is yet much to learn about behavioural mechanisms, like search strategies, that dispersing animals employ to detect new habitat (Clobert et al. 2004). There is an increased interest to dissect the behavioural variation behind dispersal patterns, but so far most studies have particularly focused on dispersal pro-pensity when comparing the tendencies to cross habitat boundaries between species or populations (e.g. Berggren et al. 2002, Ries & Debinski 2001, Schtickzelle & Baguette 2003, Schultz & Crone 2001). However, dispersal is a three-stage process including emigration (crossing of habitat boundary), traversing through a landscape matrix and settlement (immigration or colonisation) (Ims & Yoccoz 1997). Dispersal propensity is only one part of the chain to understand the process of dispersal. The ability of animals to perceive habitat at a distance once they move through the landscape is also a relevant behavioural mechanism that may contribute to dispersal success in fragmented landscapes (Lima & Zollner 1996, Zollner 2000). However, empirical information on the perceptual abilities of animals in real landscapes is still limited (Desrochers & Hannon 1997, Gillis & Nams 1998, Zollner 2000, Zollner & Lima 1999, Bonte et al. 2004, Schooley & Wiens 2003).

In this article, we report on orientation behaviour in a woodland butterfly – the speckled wood Pararge aegeria (L.) – to find woodland habitat and we test the hypoth-esis of differential abilities among populations from highly fragmented agricultural landscape versus woodland landscape. In the former landscape the habitat of P. aegeria is confined to scattered bits and pieces like hedgerows and small woodlots. From an adap-tive point of view the time spent searching for forested habitat in a hostile agricultural matrix could be reduced by an increased ability to find habitat (e.g. increased percep-tual range) compared to populations living in a woodland landscape with continuous habitat. Hence, we predict that butterflies from agricultural landscape have better per-ceptual abilities than woodland butterflies. We test the responses of the butterflies at two sites to evaluate additionally the role of the dimension of the target habitat using a pasture with a high woodland fringe and a pasture with a hedgerow. A high woodland fringe is expected to facilitate habitat attraction more than does a hedgerow.

We used, in a manner similar to Zollner (2000), the ability of P. aegeria from both types of landscape to orient towards forested habitat from a distance as a behavioural assay to the ability to perceive forested habitat at a distance. These abilities were as-sessed by capturing P. aegeria at distant woodland and agricultural landscape and re-leasing them at an unfamiliar pasture at different distances from forested habitat in order to follow and quantify their flight trajectories. Critical to this approach is the assumption that movement oriented towards the forested habitat is equivalent to the perception of habitat. In an earlier experiment in outdoor cages, Merckx et al. (2003) have shown evidence that P. aegeria originating from both types of landscape preferred shady, sheltered woodland conditions to open conditions like pasture. Therefore, it is

reasonable to assume that the released butterflies are motivated to search immediately for habitat. Moreover, open, relatively short-turf pasture is a hostile environment for P. aegeria which is essentially a woodland butterfly (Chardon et al. 2003).

P. aegeria is a satyrine butterfly (Lepidoptera: Nymphalidae). Throughout Europe the prime habitat is woodland, but the species also occurs in fragmented landscapes with some kind of woodland aspect like along hedgerows and in woodlots in agricultural landscape (Merckx et al. 2003). Adults mainly feed on honeydew and to a lesser extent on nectar.

Equal numbers of P. aegeria males were captured at two different types of landscape in central Belgium (n

total=98). The first landscape of origin was closed landscape domi-

nated by deciduous woodland with differently sized clearings and rides (Meerdaalwoud). The second was open, agricultural landscape composed of intensively used fields and pastures (85% of the area), houses and farms (4%), tiny woodlots (3%) and several hedge-rows or woody sunken rural roads (Hoegaarden). Hence, the landscapes represent a con-trast between rather continuous habitat for P. aegeria versus highly fragmented landscape (habitat <5% of the area, and scattered across the landscape). Butterflies were captured from different sites in each landscape (see Merckx et al. 2003). Hence they most likely belonged to separated local populations in each landscape as the distance between indi-vidual captures was 0.5 to 5.0 km. They were captured by hand net, temporarily stored in individual plastic jars and transported in a cool box (ca. 15°C) to one of the release sites.

The release-experiment was done during nine days with fine weather conditions that allow butterfly activity between the end of July and the beginning of September 2002. Butterflies were released – one at the time – at two different sites representing open agri-cultural landscape (mainly pastures under active agricultural use) with forested habitat at one side only. None of the butterflies was familiar with the site of release. At one of the release sites, the forested habitat was a ca. 40 m high woodland fringe. At the other site, it was a tall hedgerow with a height of ca. 5 m. Release distances from the habitat edge were 25, 50, 100 and 200 m. Releases were randomised between release distances and landscapes of origin. Before release, an individual was transferred from the cool jar into a butterfly net without any direct handling by the observer. The butterfly typically crawled out the net, started to bask by spreading the wings dorsally and took off sponta-neously after a while. At that moment observations started. Two observers tracked the butterfly; one placed tagged sticks to register the flight track and the other timed – using a stopwatch – and recorded the observed behaviours. Butterflies were never followed too close (<10 m) to avoid interference from the observers. An observation session lasted until the butterfly either (i) reached the forested habitat, (ii) did not reach habitat but crossed a 250 m distance threshold and continued to fly into the open, agricultural land-scape away from the forested habitat, or (iii) interrupted its track and rested (>1 minute).

}Methods

{ 80 }


{ 81 }

The latter category (n=7) was obviously not used in the analyses. After each release, the XY-co-ordinates of the placed sticks were determined by measuring the distances to ref-erence points (i.e. triangulation) using a laser instrument (Leica DistoTM classic³).

From the flight trajectories (including positions and timing) and behavioural ob-servations we derived several parameters that typified the movement pattern. For each individual it was recorded whether or not it reached the forested habitat. Total covered distance, total flight time and mean flight velocity were measured. Flight trajectories were classified as one of four classes: (i) straight flight track, (ii) mainly straight flight track, but with one curve, (iii) flight track with several curves, without returning to the point of release and (iv) petal-like loop pattern (i.e. an at least almost complete ellipse or more ellipses from and back to the point of release). We imposed 25 m-sections from and parallel with the habitat edge on the open landscape matrix. If a flight track crossed such a 25-m line, the angle between the flight track and the straight line to the habitat (0-360°) was determined. The linearity of flights between two successive 25 m-section lines was expressed as Euclidean distance / real covered distance. Measures of linear-ity were arcsine root transformed prior to analysis to reach normality. For each release event, average wind speed and ambient temperature were measured.

Mixed regression models were used to analyse the movement parameters [movement angle (0-180°) and linearity] in relation to landscape of origin, distance to habitat and their interaction as fixed effects. Only four classes (25, 50-75, 100-150 and 175-225 m) were used for the parameter distance to habitat based on the 25-m line sections because a larger amount of classes reduced goodness of fit of the models. Since the flight track of an individual may cross several 25-m line sections, ‘individual’ was included in the model as a random effect. Due to significant three-way interaction effects (landscape of origin x distance to habitat x release site), analyses were done separately for the two release sites as landscape of origin and the interaction with distance to habitat are the factors of highest interest. Mean velocity was analysed as a function of landscape of origin, release distance, release site, and all mutual interactions as fixed effects and residual ambient temperature as a covariate (Proc Mixed, SAS 2001). The response vari-able ‘reaching habitat or not’ was analysed with a model with a binomial error structure using landscape of origin, release site and distance to habitat as fixed effects (Glimmix macro, SAS 2001). In a separate analysis for reaching habitat, the effect of type of flight track was tested. The flight track class ‘straight with one curve’ was omitted from that analysis since it could not be estimated (n=11). Wind speed was always included as a ran-dom effect. Backward selection of non-significant factors (highest P-values) was done to obtain the final statistical model. Differences of Least Squares Means were used as a posteriori tests among groups. Normality of model residuals was always checked.

Directedness towards the forested habitat was analysed using circular statistics (Oriana version 1.01, Kovach Computing Services, Anglesey, Wales). As this could not been done in a multivariate way, three separate tests were done: 1) distance to habitat

for both release sites together, 2) distance to habitat for both landscapes of origin sepa-rately, and 3) distance to habitat for both release sites separately. Here, five classes could be used for the distance to habitat (25, 50, 75, 100 and 125-225 m). Rayleigh tests were used to assess whether flight angle distribution differed significantly from randomness. Mean vector length (r) and 95% confidence intervals were calculated as measures of the variation around mean direction (Batschelet 1981).

reaching target habitat, trajectories and velocityThe probability that a released speckled wood butterfly reached the target habitat de-creased with distance from habitat (F

3, 256=16.25, P<0.0001). Success rate in reaching hab-

itat was higher at the site with a high woodland fringe as target than at the site with a hedgerow as target (F

1, 29.9=7.10, P=0.012).

A flight track with several curves, but without returning to the point of release, was shown by most butterflies (45.1%). One quarter of the observed flight tracks were straight (25.3%). A straight flight track with only one curve was followed by 13.2 % of the individuals, and a petal-like loop pattern was shown by 16.5% of the individuals.

Mean velocity increased with release distance from habitat (at 25 m: 2.0±0.2 m/s; at 50 m: 2.3±0.3 m/s; at 100 m: 2.5±0.2 m/s), but was considerably lower in the butterflies released at 200 m: 0.9±0.1 m/s (F

3, 85=4.24, P=0.0077). Velocity did not differ with land-

scape of origin, release site or ambient temperature.

linearity of flight trajectoriesAt the woodland fringe site, flight tracks of butterflies of agricultural landscape origin had a significantly higher level of linearity than flight tracks of butterflies of woodland landscape origin (0.91±0.02 versus 0.84±0.03; F

1, 48.8=6.16, P=0.017). This effect could not

be detected at the hedgerow site. Whereas there was no effect of distance present at the woodland fringe site, we observed a trend for higher linearity at 25 m from habitat than at distances from 100 m on at the hedgerow site (t

114=1.95, P=0.054).

flight angle relative to the target habitatAt the woodland fringe site, flight angles of butterflies of agricultural landscape origin deviated significantly less from straight direction to the habitat than flight angles of butterflies of woodland landscape origin (51.9°±6.3 versus 75.9°±7.1; F

1, 46.8=4.53, P=0.039).

There was an additive effect of distance from habitat as flight angles deviated more from straight direction to the habitat with distance from habitat (F

3, 117=2.81, P=0.043;

Figure 7.1). At the hedgerow site, there was no significant effect of landscape of origin, nor of release distance. However, angles deviated more from straight direction to the hab-itat at distances from 100 m on than they did at 25 m from habitat (t

118=-2.04, P=0.044).

}Results

{ 82 }


{ 83 }

flight directednessIn the overall test, the distribution of the orientation of the butterflies differed signifi-cantly from non-random until 100 m from the habitat. In case of a non-random pattern, the orientation was always towards the habitat as reflected by the mean vector direc-tions close to 0°. Vector directions became more scattered around the mean vector with increasing distance from the habitat (Table 7.1a; Figure 7.2a).

When testing for the effect of the landscape of origin, different thresholds for the distance at which flights were orientated towards the habitat were found. Butterflies of agricultural landscape origin showed significant differences from randomness until 100 m, whereas butterflies of woodland landscape origin did so up to 50 m only. So, mean vector lengths at 75 m and 100 m were larger in butterflies of agricultural landscape than of woodland landscape (Table 7.1b; Figure 7.2b). Comparing the effect of release site, butterflies at 75 m from habitat possessed orientations that differ significantly from randomness at the woodland fringe site, whereas orientations were only marginally significant directed at the hedgerow site. For the same distance, mean vector direction was close to habitat direction at the woodland fringe site only (woodland fringe: 357° versus hedgerow: 37°). The mean vector length was also larger at the woodland fringe site (woodland fringe: 0.57 versus hedgerow: 0.33 – Table 7.1c; Figure 7.2c).

Our release-experiment showed that speckled wood butterflies were less likely to orient towards forested habitat from a wider distance in open agricultural land within a range of release distances of 25-200 m. If the target habitat was more apparent (high woodland fringe versus lower hedgerow), attraction towards the forested habitat was facilitated. In line with our key prediction, butterflies of agricultural landscape populations were bet-ter able to orient towards habitat from a wider distance than butterflies of woodland landscape populations. They followed tracks with a higher degree of linearity, displayed

25 m 50 m 75 m 100 m 125-225 m

µ 8° 10° 13° 9° 219°r 0.59 0.34 0.42 0.24 0.04

95% CI 352°-24° 343°-37° 347°-39° 326°-51° 241°-196°

P 0.00 0.00 0.00 0.03 0.95

25 m

H

25 m

W

50 m

H

50 m

W

75 m

H

75 m

W

100 m

H

100 m

W

125-225m

H

125-225m

W

µ 19° 357° 8° 12° 37° 357° 354° 26° 191° 326°r 0.64 0.56 0.31 0.37 0.33 0.57 0.26 0.23 0.53 0.0795% CI 359°-39° 333°-20° 326°-51° 336°-47° 351°-83° 331°-24° 300°-48° 323°-89° 120°-262° 125°-166°

P 0.00 0.00 0.04 0.01 0.06 0.00 0.12 0.21 0.25 0.86

��

��

��

��

��

��

��

��

��

��

��

��

Figure 7.1. Flight angle (0-180°) relative to the target habitat at the woodland fringe release site in function of distance to forested habitat (4 classes: 25, 50-75, 100-150 and 175-225 m).

Table 7.1: Analyses of flight directedness towards forested habitat in function of distance to habitat (5 classes: 25, 50, 75, 100 and 125-225 m) using circular statistics.

a) Test for distance to habitat (data of both release sites and landscapes of origin combined)

25 m

A

25 m

W

50 m

A

50 m

W

75 m

A

75 m

W

100 m

A

100 m

W

125-225m

A

125-225m

W

µ 11° 6° 353° 29° 12° 14° 5° 26° 195° 302°r 0.61 0.56 0.37 0.35 0.57 0.29 0.35 0.09 0.28 0.0595% CI 349°-32° 342°-30° 317°-28° 351°-67° 345°-39° 321°-67° 328°-43° 207°-205° . 25°-220°

P 0.00 0.00 0.01 0.02 0.00 0.12 0.01 0.82 0.65 0.92

b) Test for distance to habitat separately for landscape of origin (A=agricultural landscape; W=woodland landscape)

c) Analysis for distance to habitat separately for site of release (H=hedgerow site; W=woodland fringe site)

}Discussion

{ 84 }


{ 85 }

flight angles that deviated less from orientation towards habitat, and showed orientations that were significantly directed towards habitat from wider distances than butterflies of woodland origin. Selection for directed flight in fragmented, agricultural landscape populations is in line with an adaptive response in such an environment as it reduces the time budget spent within the hostile matrix; it is likely to lower associated costs in terms of energy, predation risk and loss of active time for reproduction (Zollner & Lima 1999).

Our results add to the diverse and highly significant ecological and evolutionary ef-fects that habitat fragmentation may have on life history traits of organisms, including dispersal. Earlier, our colleagues and we have detected differences among P. aegeria pop-ulations of woodland versus fragmented agricultural landscape in dispersal propensity (Merckx et al. 2003) and in temperature-related fecundity and longevity (Karlsson & Van Dyck 2005). In both studies the butterflies that were tested were reared in a common environment after parental stock was collected in both types of landscape. Here, we directly compared captured adults of both types of landscape as was done in a similar study with small mammals (Zollner 2000). So, from a causal point of view, we cannot discriminate at this stage between intrinsic differences versus differences in experience between the butterflies of different landscape of origin. This requires further testing, but for habitat crossing behaviour we earlier showed evidence that both groups differ intrinsically, most likely relying on heritable differences (Merckx et al. 2003).

The observation of insect dispersers that need to approach potential habitat relative-ly close to be attracted has been made a few times before. Coyne et al. (1987), for instance, showed that Drosophila flies moved distances of several kilometres between oases, but they were only able to detect an oasis from <10 m distance. Euphydryas butterflies also moved up to 5 km between habitat patches, but they only oriented towards a serpentine grassland habitat at a distance of <50 m (Harrison 1989). But the differences between P. aegeria populations of different landscapes suggest that detection ability may vary considerably at the intraspecific level. Butterflies from agricultural landscape oriented towards forested habitat from twice as wide than butterflies from woodland landscape (100 versus 50 m). Besides the clear average differences with landscape of origin, there was considerable inter-individual variation. For instance, two individuals released at 200 m from habitat successfully reached habitat adopting a flight track of high degree of linearity (values of 0.98 and 0.93).

Although our study does not provide insight into the sensory mechanism of habitat perception, butterfly movements are most likely to be visually cued. However, consider-ing the relatively poor acuity of the butterfly eye, Rutowski (2003) predicts that even large objects such as trees several meters high may not be resolvable at distances of more than 20 to 30 meters. However, empirical data on landmark detection at the landscape level are – to our knowledge – lacking in butterflies (among several other taxa). The size of the compound eyes relative to total body size did at least not differ between speci-mens of both types of landscape (T. Merckx, unpublished data), but there are several

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

Figure 7.2: Angular orientation of released butterflies, defined relative to the point of release and the place at which following was stopped. The dots on each unit circle represent the angular orientations of individual butterflies. The part of the habitat closest to the point of release is defined as 0° for each release. Vectors indicate average angle (mean vectors (m), Batschelet 1981) and are surrounded by 95% CI’s. Data are given for (a) butterflies of both release sites together at five distance classes from forested habitat (25, 50, 75, 100, 125-225 m), (b) butterflies of both landscape origins separately at distances from forested habitat of 75 and 100 m, and (c) butterflies of both release sites separately at a distance from forested habitat of 75 m.

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

(a)

(c)

(b)

25 50 75

100 125 - 225

agri origin (75 m) woodland origin (75 m) agri origin (100 m) woodland origin (100 m)

release site near woodland fringe (75 m) release site near hedgerow (75 m)

{ 86 }


{ 87 }

other traits that may relate to visual detection capacity (Rutowski 2003 and references therein). We cannot exclude an additional, or even to some extent alternative, role of ol-factory cues (Baguette 2003, Fownes & Roland 2002). Hansson (1995) suggested plant-pro-duced odours to be potentially important cues for locating and evaluating foraging sites in Lepidoptera, but direct evidence has yet to be experimentally shown in butterflies.

The presence of habitat across the landscape has been shown to affect expansion rates in P. aegeria in different areas in the U.K. (Hill et al. 2001). At a smaller spatial scale, the incorporation of habitat elements into a landscape connectivity measure signifi-cantly improved modelling patterns of presence/absence for P. aegeria in woodland frag-ments compared to the use of Euclidean distance in two Belgian landscapes (Chardon et al. 2003). Merckx et al. (2003) showed that P. aegeria from woodland landscape had higher propensity to cross habitat boundaries than did conspecifics from agricultural landscape in a common arena test. In combination with our current results on differential percep-tual ability, we hypothesise that P. aegeria butterflies of agricultural landscape are better able to deal with a coarse-grained ecological infrastructure (e.g. less compact network of hedgerows or woodlot stepping stones) to move through fragmented landscapes than are butterflies of woodland landscape populations. Relatively low perceptual ability is likely to be associated with more time spent searching in the hostile matrix. This in turn may increase mortality risks that are considered important costs associated with dispersal (Waser et al. 1994, Mennechez et al. 2004). The idea of differential optimal or minimally required geometry of corridor or stepping stone networks for the same species in differ-ent types of landscape is valuable to test in several other species, particularly those of high conservation interest. The importance of strong contrasts to facilitate orientation to forested habitat is also a significant observation in this respect. The woodland fringe was eight times higher than was the hedgerow at our release sites. So, hedgerows and woodlots with much taller dimensions – particularly higher trees – are predicted to be landmarks of wider attraction across the landscape for this type of organism. Such insights can be taken into account in modelling connectivity using cost distance ap-proaches (Chardon et al. 2003). They further contribute to an improved understanding of how structural landscape connectivity is translated into functional connectivity for a particular target species (Bowman & Fahrig 2002, Schooley & Wiens 2003, Stevens et al. 2004). More generally, it illustrates that a better understanding of perceptual ranges and the effects on movements will contribute to the analyses of population dynamics in fragmented landscapes and hence to conservation strategies (Lima & Zollner 1996).

Our study also raises questions about behavioural mechanisms of movements through the landscape matrix that are interpreted as being dispersal. Conradt et al. (2000) and Conradt et al. (2001) concluded from release-experiments with two other sa-tyrine butterfly species that dispersers search systematically for habitat by flying in a succession of progressively larger ellipsoidal loops away from and back to their starting point. However, only 21.6% of the released butterflies engaged in this pattern of petal-

like searching loops (Conradt et al. 2001). One could argue that their sites of capture and release were too nearby relative to the spatial scale of movements and to the perceptual range in these species to exclude home-range like movements or real homing. It has been noticed before that homing movements differ behaviourally from dispersal move-ments in several taxa (Zollner & Lima 1999, Van Dyck & Baguette 2005). Moreover, it has been recognised that the petal-like loop movements are typical for homing (e.g. Hoff-mann 1983) and for movements to exploit local resources (e.g. Müller & Wehner 1994). In our study, all tested individuals were unfamiliar with the release sites. The propor-tion of released butterflies that engaged in the pattern of petal-like searching loops in our study was even slightly lower (16.5%) than the proportion reported in Conradt et al. (2001). Behaviourally different types of movement (like homing, and foraging as rou-tine movements on the one hand and oriented linear movements as special movements to displace on the other) should be recognised more explicitly in dispersal studies as they have different assumptions and consequences on how animals move through the landscape (Van Dyck & Baguette 2005). Animals may extend explorative movements like during foraging beyond one physical patch of resources (e.g. Fraser & Stutchbury 2004). In our experiments, one particular butterfly that was released at 100 m took off with a strong vertical component and flew at high level (30 m versus typically 1-2 m) and very fast (6.2 m/s versus average of 2.1±0.13 m/s) in a straight line to the woodland fringe. In contrast with Conradt et al. (2000, 2001), fast, directed movements have been associated with dispersal by several other authors (e.g. Johnson et al. 2002, Mate et al. 1998). Both slow, explorative movements with lots of curves in the trajectories and fast, directed movements may contribute to net displacement, although the latter are likely to represent dispersal behaviour per se (Van Dyck & Baguette 2005). The role of sexual differences may need further attention as well, particularly if the dispersive nature is more pronounced in only one of the sexes. Here we have tested males only. Returning to visual perceptual abilities, it is known in several butterflies (including P. aegeria) that females have much smaller compound eyes than do males (Rutowski 2003). Yet, we are only at the beginning of understanding the behavioural mechanisms of dispersal and particular of how organisms perceive and respond to landscape elements and target habitat once they move through the landscape matrix.

We thank E. Matthysen and F. Adriaensen for valuable comments and E. Janssens, K. Borghgraef and TM’s

mother for field assistance. TM was research aspirant with the Fund of Scientific Research Flanders-Belgium

(FWO). This study was supported by a grant from the University of Antwerp (GOA 15R/3942) and from the

Belgian Federal Office of Scientific, Technical and Cultural Affairs (OSTC-PADD II).

}Acknowledgements

Chapter 8 { General D

iscussion }

{ 89 }

8}

General Discussion

8.1 Habitat fragmentation, movement behaviour and thermoregulation

Chapter

Especially the process of habitat fragmentation brings about the current world-wide extinction crisis (Pimm & Raven 2000, Wilson et al. 2004). However, despite an intense degree of habitat fragmentation, the speckled wood butterfly (P. aegeria) succeeds to re-main a common and widespread butterfly species (Maes & Van Dyck 1999, Asher et al. 2001); its occurrence is not restricted to large patches of habitat (woodland) but also includes highly fragmented, agricultural landscapes where habitat is restricted to ‘bits’ and ‘pieces’ (e.g. hedgerows, small patches of woodland, woody sunken rural roads). Nonetheless, habitat fragmentation may have less visible evolutionary implications (reviewed in Debinski & Holt 2000, Fahrig 2003). In order to reveal this evolutionary impact, we observed and experimentally compared populations from differently frag-mented landscapes.

Landscape structures that are altered by fragmentation often imply larger distances between resources. Furthermore, fragmentation may cause microclimates to change. This was shown for sites where P. aegeria occurs: open, highly fragmented, agricultural landscapes were on average warmer, but also more windy and more variable in tem-perature than corresponding sites of the butterfly in closed, continuous woodland {Chapter 3}. This is of high relevance, especially for ectotherms like butterflies as their activity patterns are directly influenced by microclimatic conditions. This was, for example, observed in the higher traversal rate within our outdoor cage set-ups under warmer weather conditions and in the changed habitat preference for warmer, irradiat-

{ 90 }


iscussion }

{ 91 }

ed sites under cooler conditions {Chapter 6}. In the same vein, we showed that P. aegeria dummies were able to cover larger distances given a fixed thermal budget, and that they heated up faster, under conditions of higher ambient temperature and lower wind speed {Chapter 3}. The larger variation observed in cooling distance and heating rate within fragmented, agricultural landscapes, compared to woodland landscapes, is in line with the co-occurrence of higher ambient temperatures and higher wind speeds in agricultural landscapes {Chapter 3}. Therefore, when ectotherms cross a boundary of an habitat patch in an unfragmented landscape versus a similar boundary in a fragmented configuration, this may mean a different cost: (i) the probability to encounter resources is lower in fragmented landscapes due to the scarceness of and the higher distances in between resources, and (ii) the probability to experience an open, unsheltered environ-ment for a longer time is higher, and so are most likely the associated costs in terms of energy, time and mortality risk (exhaustion, predation) (Zollner & Lima 1999). These effects of altered landscape structures on boundary hardness, via combined effects on resource distribution and microclimate, may cause strong selection with regard to move-ment behaviour, such as selection against mobility with increased isolation (e.g. Demp-ster et al. 1976). The figure below illustrates the relationships and interactions between the different components involved in this study by means of a flowchart (Figure 8.1).

In the following paragraph we summarise the evidence for an ecologically-based di-vergent selection (Rundle & Nosil 2005) between P. aegeria populations from differently fragmented types of landscape:(I) In line with the assumption of higher dispersal rates with fragmented landscape

structure, due to larger inter-resource distances (Taylor & Merriam 1995), we showed by means of a reciprocal transplant experiment with split-brood design that individuals that developed in agricultural landscape were heavier and had higher wing loading than individuals that developed in woodland landscape and landscape with woodland fragments, irrespective of their landscape of origin {Chapter 2} [for a detailed discussion of the functional significance of these flight-related morphological traits see Dudley (2000), Berwaerts et al. (2002) and Van Dyck (2003)]. This landscape-related phenotypic plasticity strongly points to adaptive differences in flight-related morphology since our observations contrasted with a constraint hypothesis, which predicted smaller individuals with fragmentation due to higher growth rates with higher ambient temperatures (Sibly et al. 1997, Ny-lin & Gotthard 1998).

(II) In line with a more frequent experience of ‘hard’ (i.e. higher crossing cost) bound-aries in butterflies of fragmented landscapes than in butterflies of unfragmented landscapes, we observed in a common arena test that P. aegeria F1-individuals origi-nating from a continuous woodland landscape had a higher propensity to cross habitat boundaries than had conspecifics from highly fragmented, agricultural landscape {Chapter 6}.

(III) In line with selection for directed flight in fragmented landscape populations to reduce the time budget spent within the hostile matrix, our release-experiment is among the firsts to show evidence for intra-specific differences in the ability to find habitat relative to the landscape of origin of the tested individuals. As a key result, P. aegeria butterflies of agricultural landscape origin were able to orient towards habitat from a distance twice as wide than butterflies of woodland land-scape origin {Chapter 7}.

(IV) Besides behavioural compensations like flying low to the ground and close to shelter-providing vegetation {Chapter 3 and 5}, tests with adult offspring from both land-scape origins in flight cabinets at several low ambient temperatures pointed further to temperature-related physiological adaptation relative to type of landscape: indi-viduals of woodland landscape origin flew for longer than did agricultural landscape butterflies {Chapter 4}. Also within the range of voluntary flight, we have shown in a common arena test that P. aegeria of woodland landscape origin were more at flight than conspecifics from agricultural landscape {Chapter 6}. In the same vein, Karlsson & Van Dyck (2005) have experimentally shown that females of woodland landscape origin laid more eggs at low ambient temperature than did agricultural landscape females, but the pattern reversed at higher ambient temperatures.

��

��

��

��

��

��

��

��

� �

�

�

��

��

��

��

��

��

��

��

��

��

��

� �

Figure 8.1. Overview of the relationships, at both the level of the environment and the organism, between the major components involved within this study of the evolutionary ecology of movement behaviour of P. aegeria butterflies within a context of habitat fragmentation. Numbers on arrows relate to corresponding chapters that treat specific relationships.

{ 92 }


iscussion }

{ 93 }

(V) In line with the observed microclimatic differences between differently fragment-ed landscapes {Chapter 3 and 5}, we demonstrated that individuals of agricultural landscapes compensated for higher amounts of convective cooling as they were covered with more fur than woodland individuals {Chapter 3}. We interpret this to be a morphological adaptation since fur helps to reduce the impact of the typi-cally higher wind speeds that often leads to higher levels of convective cooling in agricultural landscape (Kingsolver 1983).

(VI) Although the underlying mechanism requires further analysis, male butterflies were observed to show a higher frequency of mate location behaviour intermedi-ate between perching and patrolling in agricultural landscapes {Chapter 5}. This illustrates that behavioural consequences of habitat fragmentation are not limited to dispersal – which has been the major issue in this field – but behaviours related to reproduction such as mate location may also change. As male butterflies spend most of their active time locating mates (Shreeve 1992a), different mate location strategies may have different consequences for rates of movements as well. Our re-sults suggest that the dichotomy of perching versus patrolling as typically observed in P. aegeria males of woodland (Van Dyck 2003) fades away in fragmented agricul-tural landscape {Chapter 5}.

The studied flight- and thermoregulation related morphological traits are known to have a heritable basis (Hill et al. 1999a, b, Berwaerts 2004), which is a prerequisite for evolu-tionary change. Furthermore, differences observed via common garden and reciprocal transplant approaches may indeed point at intrinsic, heritable differences. Hence, the differences we observed may indicate evolutionary consequences of habitat fragmenta-tion. An important, accelerating factor in such processes of micro-evolution (relatively fast local adaptation) is phenotypic accommodation (i.e. evolutionary change without genetic change), which results from phenotypic plasticity (i.e. environmental respon-siveness) (West-Eberhard 2003). To our best knowledge, we are among the firsts to show results that suggest the existence of landscape-related, adaptive phenotypic plasticity in butterfly flight morphology {Chapter 2}. Other key results are the significant genotype-by-environment (G x E) interactions that we found for some of these traits. In other words, we demonstrated that different genotypes produced a different flight-morphol-ogy in different microclimatic environments (i.e. differently fragmented landscapes) {Chapter 2}. These interactions, that may be seen as genetic variation for plasticity (Pigliucci et al. 1995), strengthen the hypothesis that the observed landscape-related phenotypic plasticity is adaptive and that it contributes to the observed evolutionary changes in thermoregulation and movement behaviour (e.g. dispersal) in differently fragmented landscapes.

Throughout our study on the evolutionary ecology of P. aegeria’s movement behaviour in a context of habitat fragmentation, we observed that evolutionary changes in ther-moregulation and movement behaviour (see 8.1.) differed relative to sex. These sexual differences can be understood in relation to the multifunctionality of butterfly flight (Shreeve 1992a, Watt 2003), which differs between males and females relating to their different behavioural repertoires and ecological roles (Van Dyck 2003). Males are the more ‘wasting’ sex. They spend most of their active time budget to mate location and, hence, mainly move locally but are more subject to convective cooling. In order to in-crease the probability of locating virgin females, males extend flight even during peri-ods with less favourable microclimatic conditions, because under relatively poor weath-er also virgin females may emerge. Females, on the other hand, are characterised by a trade-off between allocation towards reproduction versus flight (i.e. the oögenesis-flight syndrome – Hughes et al. 2003). Hence, females need to be more economical in terms of flight activity. So, butterfly females typically spend less time flying than males (Shreeve 1984) and they will preferentially fly during periods with favourable weather condi-tions: they rest and bask during considerable time periods between oviposition, forag-ing and dispersal flights (Shreeve 1992a). However, despite their lower flight incidence, females are considered to be the more dispersive sex (Baker 1984). This illustrates that general flight activity is not necessarily a good indicator of dispersal ability (Merckx & Van Dyck 2002). As a result of these different behavioural repertoires and ecological functions between the sexes, the two following selective regimes for some traits will be different between males and females. The selection pressure on thermoregulation-related morphological traits in order to maximise the time period that flight muscle temperature is within the optimal range for powerful flight, will be higher in males. On the other hand, since males spend most of their active time to mate location and, hence, mainly move locally, the selection pressure on flight-related morphology for enhanced dispersal ability will be higher in females.

In line with the predicted higher selection on some aspects of thermoregulation in males, we showed that the pattern of darker [i.e. faster heating (Van Dyck & Matthysen 1998, Berwaerts et al. 2001)] wings in springtime compared to the warmer summertime was only present in males; it was lacking in females {Chapter 3}. In the same vein, we showed that females have less fur (i.e. less thermal isolation, higher impact of convec-tive cooling – Kingsolver 1983) than males {Chapter 3} and had a lower willingness to fly at temperatures within the range of voluntary flight activity in the experimental cages {Chapter 6}. The latter observation most likely contributes to the explanation of male-biased samples in mark-release-recapture studies (e.g. Wahlberg et al. 2002). In line with the key predictions of the indoor experiment on flight ability at suboptimal temperatures, males and females responded differently: again, males were more will-ing to fly than females {Chapter 4}. Ultimately, the latter observed sexual difference

}8.2 Sexual differences

{ 94 }


iscussion }

{ 95 }

relates to the fundamental differences in life history between the sexes. From a proxi-mate viewpoint, this difference can be attributed to lower wing loading (i.e. ratio of total body mass to wing area) in males from a proximate point of view (Gilchrist 1990). It allows flight at lower wingbeat frequency (Casey 1981) and wingbeat frequency is tem-perature-dependent (Stevenson & Josephson 1990). For example, minimum temperature for flight in the butterfly Thymelicus lineola increased with an experimental increase in wing loading (Pivnick & McNeil 1986). The sexual difference in wing loading in P. aegeria is mainly caused by differences in body mass rather than forewing area (Berwaerts et al. 2002). So, the smaller male size can be seen as an adaptation to extend flight activ-ity. Furthermore, we showed that only males differed adaptively in the amount of fur between types of landscape {Chapter 3}. These sexual differences may result from dif-ferential selection regimes for thermoregulation-related morphological traits that allow maximisation of the time period that flight muscle temperature is within the optimal range for powerful flight. Comparative experiments with individuals that are exposed to different treatments to remove fur could test the hypothesis that males are con-fronted with higher selection on this aspect of thermoregulation. We predict that the negative effects of shaved thoraces on flight activity (duration of flight bouts, power…) will be more obvious in males than in females.

The predicted higher selection on the dispersal process in females is supported by comparative work on flight morphology between recently colonised sites versus perma-nently occupied sites: Hill et al. (1999b) found significant differences in P. aegeria females only; females from newly established populations were larger, had larger thoraces and had lower aspect ratios than females from old populations. In the same vein, Hanski et al. (2004) found significant effects – relating to energy metabolism – in females only in a study on variation in dispersal propensity among populations of the butterfly Meli-taea cinxia. When testing for variation in dispersal propensity and flight activity among landscapes of origin, we also found the clearest responses in P. aegeria females {Chapter 6}. Furthermore, landscape-related phenotypic plasticity for relative thorax mass was only present in females [Chapter 2}. The direction of this effect was in line with the adaptive prediction about dispersal ability. In our release experiment on differential ability to detect habitat among butterflies from differently fragmented landscapes, we have, however, tested males only for practical reasons. Thus, the role of sexual differ-ences in this context requires further testing. Nevertheless, it is known in several but-terflies (including P. aegeria) that females have much smaller compound eyes than do males (Rutowski 2003). However, the size of the compound eyes relative to total body size did not differ between the tested males of both types of landscape, despite the clear effect of landscape of origin {Chapter 7}. But there are several other traits that may relate to visual detection capacity (Rutowski 2003 and references therein). Only an experimental approach will be able to tell whether the effect of landscape of origin, as observed in males, is also present in females. Furthermore, this approach will allow

detecting whether perceptual abilities in females, the more dispersive sex, differ from males’ abilities to detect habitat. All results on differential effects of habitat fragmen-tation on thermoregulation and movement behaviour for both sexes of P. aegeria add significantly to the diverse ecological and evolutionary effects of habitat fragmentation on life history traits of organisms.

We take the view that P. aegeria individuals pay a higher cost when they leave a habitat patch within highly fragmented, agricultural landscapes, relative to the same action in a more continuous woodland landscape (see also 8.1.). In fragmented landscapes, bound-ary crossing may imply a higher chance to experience larger distances of open, unshel-tered environment. Our common garden experiment with large outdoor cages showed that P. aegeria butterflies prefer forested spots over open environments: irrespective of the landscape of origin, butterflies preferred the shaded cage ends (forested habitat) over the open middle part of the cages (open matrix) {Chapter 6}. This result supports the view that open landscape matrices are hostile environments for this butterfly spe-cies of forested habitat. The chances to get predated are probably higher in an open envi-ronment. Furthermore, resources are scarcer and inter-resource distances are larger in highly fragmented, agricultural landscapes. This may imply higher probabilities to lose ‘costly’ time and energy (Waser et al. 1994, Mennechez et al. 2004). Our experiment with common garden cage set-up revealed that P. aegeria from highly fragmented, agricultural landscape had lower propensity to cross open-shade boundaries than did conspecifics from woodland landscape {Chapter 6}. This result in combination with the higher abil-ity to detect habitat in individuals from agricultural landscape {Chapter 7} leads to the following hypothesis: P. aegeria butterflies of agricultural landscape may be better able to deal with a coarse-grained ecological infrastructure (i.e. a less compact network of corridors and stepping stones) to move through fragmented landscapes than are butter-flies of woodland landscapes. This idea of a differential optimal or minimally required geometry of corridor or stepping stone networks for the same species in different types of landscape is valuable to test in several other species, particularly those of high con-servation interest. It may be significant to make use of such knowledge when restoring landscapes with regard to the conservation of particular target species. More specifical-ly, knowledge of the optimal geometry of habitat networks may be useful when it comes to create environments that allow functional displacements of a sufficient number of individuals from particular focal species.

We hypothesised that P. aegeria from highly fragmented, agricultural landscapes may be better able to deal with a more coarse-grained ecological infrastructure than conspecifics from continuous woodland landscapes. Furthermore, we showed evidence

}8.3 Landscape, conservation and variation in

functional connectivity

{ 96 }


iscussion }

{ 97 }

for locally adapted thermoregulation- and flight-related morphology {Chapter 2 and 3}. This may further facilitate individuals from agricultural landscapes to undertake larger displacements, all the more given the on average higher ambient temperatures in agricultural landscapes {Chapter 3}. At the same time, however, our results attract the attention to the importance of vegetation structures as potential corridors for move-ment across fragmented landscapes: we observed higher heating rates and a less vari-able thermal budget for flight in the vicinity of such structures, especially when wind speed was higher {Chapter 3}. In the same vein, Merckx & Van Dyck (2002) have shown earlier in two other butterfly species that they flew more in the vicinity of hedgerows, particularly during more windy conditions. So, the importance or significance of such landscape elements may be weather or climate dependent (see also Dover 1996, Dover & Sparks 2000). If butterflies are only able to make relative short flights, a more fine-grained ecological network of corridors would be required. Furthermore, we showed by means of P. aegeria dummies that flying more closely to the substrate may be an additional way to behaviourally lower the impact of convective cooling resulting from habitat fragmentation {Chapter 3}. Accordingly, the study of Hill et al. (2001) indicated that the availability of breeding habitat is important: P. aegeria range expansion in the UK was c. 45% slower in an area that had 24% less woodland. Nevertheless, this result may largely depend on the nature and density of the network of greenveins within the intervening matrix, which have been shown to be major players in the lagging behind of species expansions to current, warmer climates (Hill et al. 1999c, 2002). So, a network of shelter providing structures in the landscape matrix may at least mitigate effects of habitat fragmentation as flying at low height in the vicinity of these structures signifi-cantly increases the thermal budget for flight {Chapter 3} and, hence, may increase landscape connectivity and extend flight distances for butterflies traversing through fragmented landscapes.

So far, the majority of research on movements has in particular focused on patterns (e.g. frequency distributions of distances, maximum distance, etc. – e.g. Clobert et al. 2001). But different mechanistic scenarios could lead to the same pattern. Therefore, mechanistic research on movements is required for a better understanding of move-ments (e.g. dispersal) and how this is affected in fragmented landscapes. As a conse-quence, remediation conservation strategies in fragmented landscapes would benefit from a better understanding of the mechanisms. For example, hedgerows have often been assumed to be corridors for woodland species, but for butterflies (among several other taxa) there is hardly empirical evidence for this (but see Dover & Sparks 2000, Dover & Fry 2001, Cant et al. 2005). From our experiments, we hypothesise that wood-land butterflies may be encouraged to cross a woodland boundary if there is a landscape feature that is perceived as woodland (e.g. a hedgerow). Hence, networks of hedgerows would rather function according to a stepping stone principle than to corridors per se. Moreover, for ectotherms such as butterflies hedgerows are likely to be rather stepping

stones for thermal reasons than really continuous conduits which they may be for en-dotherms. Of course, this requires further and careful testing. Furthermore, the im-portance of strong contrasts to facilitate orientation to forested habitat is a significant observation in this respect {Chapter 7}. From our experiment in which we compared the effect of an eight-fold difference in height of the target habitat on released butter-flies, we predict that hedgerows and patches of woodland with much taller dimensions

– particularly higher trees – may function as landmarks of wider attraction across the landscape for this type of organism. Such insights can be taken into account in model-ling connectivity using cost distance approaches (Chardon et al. 2003). They further contribute to an improved understanding of how structural landscape connectivity is translated into functional connectivity for a particular target species (Bowman & Fah-rig 2002, Schooley & Wiens 2003, Stevens et al. 2004).

More generally, we believe that further experiments on behaviour at habitat bound-aries will give us a better understanding of the decision-making process with regard to movements at patch boundaries. Furthermore, examining the scale at which indi-viduals orient and respond to landscape geometry may give a better understanding of perceptual ranges and their effects on movements. The new methodology in which also butterfly flight paths are tracked across the landscape with harmonic radar (Cant et al. 2005) promises to be of significant value for future investigations in this direction. The strategy used to orient within the landscape and the scale at which the landscape is per-ceived restrict the sustainability of populations within habitat fragments (Dover & Fry 2001). Consequently, this experimental approach will contribute significantly to under-standing the role played by landscape in population dynamics. It is this kind of knowl-edge which is vital to evidence-based conservation strategies at the landscape level.

{ References }

{ 99 }

References

Agrawal AA (2001) Ecology – Phenotypic plasticity in the interactions and evolution of species. Science 294: 321-326

Alcock J & Houston TF (1996) Mating systems and male size in Australian Hylaeine bees (Hymenoptera: Colletidae). Ethology 102: 591-610

Altizer S, Harvell D & Friedle E (2003) Rapid evolutionary dynamics and disease threats to biodiversity. Trends in Ecology and Evolution 18: 589-596

Andersson M (1994) Sexual Selection. Princeton University PressAndrén H (1994) Effects of habitat fragmentation on birds and mammals in landscapes with different

proportions of suitable habitat – a review. Oikos 71: 355-366Andrén H (1999) Habitat fragmentation, the random sample hypothesis and critical thresholds. Oikos 84: 306-308Asher J, Warren M, Fox R, Harding P, Jeffcoate G & Jeffcoate S (2001) The Millenium Atlas of Butterflies in

Britain and Ireland. Oxford University PressBaguette M (2003) Long distance dispersal and landscape occupancy in a metapopulation of the cranberry

fritillary butterfly. Ecography 26: 153-160Baguette M, Petit S & Queva F (2000) Population spatial structure and migration of three butterfly species

within the same habitat network: consequences for conservation. Journal of Applied Ecology 37: 100-108Baguette M, Mennechez G, Petit S & Schtickzelle N (2003) Effect of habitat fragmentation on dispersal in

the butterfly Proclossiana eunomia. Comptes Rendus Biologies 326: 200-209Baker RR (1984) The dilemma: when and how to go or to stay? In: Vane-Wright RI & Ackery PR (eds.) The

Biology of Butterflies, 1989 edn. Academic Press, London, pp 279-296Batschelet E (1981) Circular Statistics in Biology. Academic Press, LondonBeier P & Noss RF (1998) Do habitat corridors provide connectivity? Conservation Biology 12: 1241-1252Berggren A, Birath B & Kindvall O (2002) Effect of corridors and habitat edges on dispersal behavior,

movement rates, and movement angles in Roesel’s Bush-Cricket (Metrioptera roeseli). Conservation Biology 16: 1562-1569

Berwaerts K (2004) An Integrated Study of Flight Performance in the Butterfly Pararge aegeria L.: Performance, Morphological Design and Genetics. Thesis (PhD), University of Antwerp

Berwaerts K & Van Dyck H (2004) Take-off performance under optimal and suboptimal thermal conditions in the butterfly Pararge aegeria. Oecologia 141: 536-545

Berwaerts K, Van Dyck H, Van Dongen S & Matthysen E (1998) Morphological and genetic variation in the speckled wood butterfly (Pararge aegeria L.) among differently fragmented landscapes. Netherlands Journal of Zoology 48: 241-253

{ 100 }

{ References }

{ 101 }

Berwaerts K, Van Dyck H, Vints E & Matthysen E (2001) Effect of manipulated wing characteristics and basking posture on thermal properties of the butterfly Pararge aegeria (L.). Journal of Zoology 255: 261-267

Berwaerts K, Van Dyck H & Aerts P (2002) Does flight morphology relate to flight performance? An experimental test with the butterfly Pararge aegeria. Functional Ecology 16:484-491

Bjornstad ON, Ims RA & Lambin X (1999) Spatial population dynamics: analyzing patterns and processes of population synchrony. Trends in Ecology and Evolution 14: 427-432

Blanckenhorn WU (1991) Life-history differences in adjacent water strider populations – phenotypic plasticity or heritable responses to stream temperature. Evolution 45: 1520-1525

Boggs CL, Watt WB & Ehrlich PR (2003) Butterflies: Ecology and Evolution Taking Flight. The University of Chicago Press

Bonte D, Lens L & Maelfait J-P (2004) Lack of homeward orientation and increased mobility result in high emigration rates from low-quality fragments in a dune wolf spider. Journal of Animal Ecology 73: 643-650

Bowman J & Fahrig L (2002) Gap crossing by chipmunks: an experimental test of landscape connectivity. Canadian Journal of Zoology 80: 1556-1561

Bowman J, Cappuccino N & Fahrig L (2002) Patch size and population density: the effect of immigration behavior. Conservation Ecology 6 (1): art. 9

Brakefield PM & Mazzotta V (1995) Matching field and laboratory environments: effect of neglecting daily temperature variation on insect reaction norms. Journal of Evolutionary Biology 8: 559-573

Brotons L, Monkkonen M & Martin JL (2003) Are fragments islands? Landscape context and density-area relationships in boreal forest birds. American Naturalist 162: 343-357

Brown JH & Kodric-Brown A (1977) Turnover rates in insular biogeography: effect of immigration on extinction. Ecology 58: 445-449

Bunker SJ (1993) Form, Flight Pattern and Performance in Butterflies (Lepidoptera: Papilionoidea and Hesperioidea). Thesis (PhD), University of Exeter

Burel F & Baudry J (2003) Landscape Ecology – Concepts, Methods and Applications. Science Publishers, USACain AJ & Sheppard PM (1954) Natural selection in Cepaea. Genetics 39: 89-116Cant ET, Smith AD, Reynolds DR & Osborne JL (2005) Tracking buterfly flight paths across the landscape

with harmonic radar. Proceedings of the Royal Society of London Series B (in press)Carlin BP & Louis TA (2000) Bayes and Empirical Bayes Methods for Data Analysis. Chapman and Hall, LondonCasey TM (1981) A comparison of mechanical and energetic estimates of flight cost for hovering Sphinx

moths. Journal of Experimental Biology 91: 117-129Casey TM (1988) Thermoregulation and heat-exchange. Advances in Insect Physiology 20: 119-146Chai P & Srygley RB (1990) Predation and the flight, morphology, and temperature of Neotropical rain-

forest butterflies. American Naturalist 135: 748-765Chardon JP, Adriaensen F & Matthysen E (2003) Incorporating landscape elements into a connectivity

measure: a case study for the speckled wood butterfly (Pararge aegeria L.). Landscape Ecology 18: 561-573

Chown SL & Nicolson SW (2004) Insect Physiological Ecology: Mechanisms and Patterns. Oxford University PressClark RA, Fox CJ, Viner D & Livermore M (2003) North Sea cod and climate change – modelling the effects

of temperature on population dynamics. Global Change Biology 9: 1669-1680Clench HK (1966) Behavioral thermoregulation in butterflies. Ecology 47: 1021-1034Clobert J, Danchin E, Dhondt AA & Nichols JD (2001) Dispersal. Oxford University PressClobert J, Ims RA & Rousset F (2004) Causes, mechanisms and consequences of dispersal. In: Hanski IA &

Gaggiotti OE (eds.) Ecology, Genetics, and Evolution of Metapopulations. Academic Press, London, pp 307-355Collinge SK (2000) Effects of grassland fragmentation on insect species loss, colonization, and movement

patterns. Ecology 81: 2211-2226Collinge SK & Forman RTT (1998) A conceptual model of land conversion processes: predictions and

evidence from a microlandscape experiment with grassland insects. Oikos 82: 66-84

Connor EF & McCoy ED (1979) Statistics and biology of the species-area relationship. American Naturalist 113: 791-833

Conradt L, Bodsworth EJ, Roper TJ & Thomas CD (2000) Non-random dispersal in the butterfly Maniola jurtina: implications for metapopulation models. Proceedings of the Royal Society of London Series B 267: 1505-1510

Conradt L, Roper TJ & Thomas CD (2001) Dispersal behaviour of individuals in metapopulations of two British butterflies. Oikos 95: 416-424

Conradt L, Zollner PA, Roper TJ, Frank K & Thomas CD (2003) Foray search: an effective systematic dispersal strategy in fragmented landscapes. American Naturalist 161: 905-915

Coyne JA, Bryant SH & Turelli M (1987) Long-distance migration of Drosophila. 2. Presence in desolate sites and dispersal near a desert oasis. American Naturalist 129: 847-861

Crespi BJ (1988) Adaptation, compromise, and constraint: the development, morphometrics, and behavioral basis of a fighter-flier polymorphism in male Hoplothrips karnyi (Insecta: Thysanoptera). Behavioral Ecology and Sociobiology 23: 93-104

Darwin C (1859) On the Origin of Species by means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life. John Murray, London

Davies NB (1978) Territorial defence in the speckled wood butterfly, Pararge aegeria: the resident always wins. Animal Behaviour 26: 138-147

Debinski DM & Holt RD (2000) A survey and overview of habitat fragmentation experiments. Conservation Biology 14: 342-355

Dempster JP (1991) Fragmentation, isolation and mobility of insect populations. In: Collins NM & Thomas JA (eds.) The Conservation of Insects and their Habitats. Academic Press, London, pp 143-153

Dempster JP, King ML & Lakhani KH (1976) The status of the swallowtail butterfly in Britain. Ecological Entomology 1: 71-84

Dennis RLH (1993) Butterflies and Climate Change. Manchester University PressDennis RLH & Williams WR (1987) Mate locating behaviour of the large skipper butterfly Ochlodes venata:

flexible strategies and spatial components. Journal of the Lepidopterist’s Society 41: 45-64Dennis RLH & Shreeve TG (1988) Hostplant-habitat structure and the evolution of butterfly mate-locating

behaviour. Zoological Journal of the Linnean Society 94: 301-318Dennis RLH & Shreeve TG (1989) Butterfly wing morphology variation in the British Isles – the influence

of climate, behavioural posture and the hostplant habitat. Biological Journal of the Linnean Society 38: 323-348

D’Eon RG, Glenn SM, Parfitt I & Fortin MJ (2002) Landscape connectivity as a function of scale and organism vagility in a real forested landscape. Conservation Ecology 6(2): art. 10

Desrochers A & Hannon SJ (1997) Gap crossing decisions by forest songbirds during the post-fledging period. Conservation Biology 11: 1204-1210

Didham RK, Hammond PM, Lawton JH, Eggleton P & Stork NE (1998) Beetle species responses to tropical forest fragmentation. Ecological Monographs 68: 295-323

Digby PSB (1955) Factors affecting the temperature excess of insects in sunshine. Journal of Experimental Biology 37: 186-212

Dingle H (1996) Migration: the Biology of Life on the Move. Oxford University PressDouwes P (1976) Activity in Heodes virgaureae (Lepidoptera: Lycaenidae) in relation to air temperature, solar-

radiation, and time of day. Oecologia 22: 287-298Dover JW (1996) Factors affecting the distribution of satyrid butterflies on arable farmland. Journal of

Applied Ecology 33: 723-734Dover JW & Sparks T (2000) A review of the ecology of butterflies in British hedgerows. Journal of

Environmental Management 60: 51-63Dover JW & Fry GLA (2001) Experimental simulation of some visual and physical components of a hedge

and the effects on butterfly behaviour in an agricultural landscape. Entomologia Experimentalis et Applicata 100: 221-233

{ 102 }

{ References }

{ 103 }

Dreisig H (1995) Thermoregualtion and flight activity in territorial male graylings, Hipparchia semele (Satyridae), and large skippers, Ochlodes venata (Hesperiidae). Oecologia 101: 169-176

Dudley R (2000) Biomechanics of Insect Flight: Form, Function, Evolution. Princeton University PressEhrlich PR (1961) Intrinsic barriers to dispersal in a checkerspot butterfly. Science 134: 108-109Ellers J & Boggs CL (2003) The evolution of wing color: male mate choice opposes adaptive wing color

divergence in Colias butterflies. Evolution 57: 1100-1106Ellers J & Boggs CL (2004) Functional ecological implications of intraspecific differences in wing

melanization in Colias butterflies. Biological Journal of the Linnean Society 82: 79-87Fahrig L (1997) Relative effects of habitat loss and fragmentation on population extinction. Journal of Wildlife

Management 61: 603-610Fahrig L (1998) When does fragmentation of breeding habitat affect population survival? Ecological Modelling

105: 273-292Fahrig L (2001) How much habitat is enough? Biological Conservation 100: 65-74Fahrig L (2002) Effect of habitat fragmentation on the extinction threshold: a synthesis. Ecological Applications

12: 346-353Fahrig L (2003) Effects of habitat fragmentation on biodiversity. Annual Review of Ecology Evolution and

Systematics 34: 487-515Fairbairn DJ & Roff DA (1990) Genetic correlations among traits determining migratory tendency in the

sand cricket Gryllus firmus. Evolution 44: 1787-1795Fischer K & Fiedler K (2000) Sex-related differences in reaction norms in the butterfly Lycaena tityrus

(Lepidoptera: Lycaenidae). Oikos 90: 372-380Fischer K & Fiedler K (2001) Resource-based territoriality in the butterfly Lycaena hippothoe and

environmentally induced behavioural shifts. Animal Behaviour 61: 723-732Flather CH & Bevers M (2002) Patchy reaction-diffusion and population abundance: the relative importance

of habitat amount and arrangement. American Naturalist 159: 40-56Forsman A (1999) Temperature influence on escape behaviour in two species of pygmy grasshoppers.

Ecoscience 6: 35-40Fownes S & Roland J (2002) Effects of meadow suitability on female behaviour in the alpine butterfly

Parnassius smintheus. Ecological Entomology 27: 457-466Fraser GS & Stutchbury BJM (2004) Area-sensitive forest birds move extensively among forest patches.

Biological Conservation 118: 377-387Fric Z & Konvicka M (2002) Generations of the polyphenic butterfly Araschnia levana differ in body design.

Evolutionary Ecology Research 4: 1017-1032Gelman A, Carlin JB, Stern HS & Rubin DB (1995) Bayesian Data Analysis. Chapman and Hall, LondonGilchrist GW (1990) The consequences of sexual dimorphism in body size for butterfly flight and

thermoregulation. Functional Ecology 4: 475-487Gilchrist GW (1995) Specialists and generalists in changing environments. 1. Fitness landscapes of thermal

sensitivity. American Naturalist 146: 252-270Gillis EA & Nams VO (1998) How red backed voles find habitat patches. Canadian Journal of Zoology 76: 791-794Goodwin BJ (2003) Is landscape connectivity a dependent or independent variable? Landscape Ecology 18:

687-699Goodwin BJ & Fahrig L (2002) How does landscape structure influence landscape connectivity? Oikos 99:

552-570Gorissen D, Merckx T, Vercoutere B & Maes D (2004) Veranderd bosgebruik en dagvlinders. Landschap 21:

85-95Gotthard K, Nylin S & Wiklund C (1994) Adaptive variation in growth rate: life history costs and

consequences in the speckled wood butterfly, Pararge aegeria. Oecologia 99: 281-289Gulinck H, Walpot O, Janssens P & Dries I (1991) The visualization of corridors in the landscape using

SPOT data. In: Saunders DA & Hobbs RJ (eds.) Nature Conservation 2: the Role of Corridors. Surrey Beatty, Chipping Norton, New South Wales, pp 9-17

Guppy CS (1986) The adaptive significance of alpine melanism in the butterfly Parnassius phoebus F. (Lepidoptera, Papilionidae). Oecologia 70: 205-213

Gustafson EJ (1998) Quantifying landscape spatial pattern: what is the state of the art? Ecosystems 1: 143-156Haddad NM (1999) Corridor use predicted from behaviors at habitat boundaries. American Naturalist 153:

215-227Hanski IA (1999) Metapopulation Ecology. Oxford University PressHanski IA (2001) Population dynamic consequences of dispersal in local populations and in metapopulations.

In: Clobert J, Danchin E, Dhondt AA & Nichols JD (eds.) Dispersal. Oxford University Press, pp 283-298Hanski IA & Gilpin ME (1997) Metapopulation Biology: Ecology, Genetics and Evolution. Academic Press, LondonHanski IA & Ovaskainen O (2000) The metapopulation capacity of a fragmented landscape. Nature 404:

755-758Hanski IA, Kuussaari M & Nieminen M (1994) Metapopulation structure and migration in the butterfly

Melitaea cinxia. Ecology 75: 747-762Hanski IA, Breuker CJ, Schöps K, Setchfield R & Nieminen M (2002) Population history and life history

influence the migration rate of female Glanville fritillary butterflies. Oikos 98: 87-97Hanski IA, Erälahti C, Kankare M, Ovaskainen O & Sirén H (2004) Variation in migration propensity

among individuals maintained by landscape structure. Ecology Letters 7: 958-966Hansson BS (1995) Olfaction in Lepidoptera. Experientia 51: 1003-1027Hardy ICW (1998) Butterfly battles: on conventional contests and hot property. Trends in Ecology and Evolution

13: 385-386Harris RJ & Reed JM (2002) Behavioral barriers to non-migratory movements of birds. Annales Zoologici

Fennici 39: 275-290Harrison S (1989) Long-distance dispersal and colonization in the Bay checkerspot butterfly, Euphydryas

editha bayensis. Ecology 70: 1236-1243Heinrich B (1975) Thermoregulation in bumblebees. II. Energetics of warm-up and free flight. Journal of

Comparative Physiology B 96: 155-166Heinrich B (1986) Thermoregulation and flight activity of a satyrine, Coenonympha inornata (Lepidoptera:

Satyridae). Ecology 67: 593-597Heinrich B (1993) The Hot-Blooded Insects. Strategies and Mechanisms of Thermoregulation. Harvard University

Press, CambridgeHeinrich B & Casey TM (1978) Heat transfer in dragonflies: fliers and perchers. Journal of Experimental Biology

74: 17-36Heinrich B & Mommsen TP (1985) Flight of winter moths near 0°C. Science 228: 177-179Hill JK, Thomas CD & Lewis OT (1996) Effects of habitat patch size and isolation on dispersal by Hesperia

comma butterflies: implications for metapopulation structure. Journal of Animal Ecology 65: 725-735Hill JK, Thomas JA & Lewis OT (1999a) Flight morphology in fragmented populations of a rare British

butterfly, Hesperia comma. Biological Conservation 87: 277-283Hill JK, Thomas CD & Blakeley DS (1999b) Evolution of flight morphology in a butterfly that has recently

expanded its geographic range. Oecologia 121: 165-170Hill JK, Thomas CD & Huntley B (1999c) Climate and habitat availability determine 20th century changes

in a butterfly’s range margin. Proceedings of the Royal Society of London Series B 266: 1197-1206Hill JK, Collingham YC, Thomas CD, Blakeley DS, Fox R, Moss D & Huntley B (2001) Impacts of landscape

structure on butterfly range expansion. Ecology Letters 4: 313-321Hill JK, Thomas CD, Fox R, Telfer MG, Willis SG, Asher J & Huntley B (2002) Responses of butterflies to

twentieth century climate warming: implications for future ranges. Proceedings of the Royal Society of London Series B 269: 2163-2171

Hoffmann G (1983) The search behavior of the desert isopod Hemilepistus reaumuri as compared with a systematic search. Behavioral Ecology and Sociobiology 13: 93-106

Honnay O, Verheyen K & Hermy M (2001) Permeability of ancient forest edges for weedy plant species invasion. Forest Ecology and Management 5554: 1-14

{ 104 }

{ References }

{ 105 }

Hughes CL, Hill JK & Dytham C (2003) Evolutionary trade-offs between reproduction and dispersal in populations at expanding range boundaries. Proceedings of the Royal Society of London Series B 270: S147-S150

Ide JY (2002) Mating behaviour and light conditions cause seasonal changes in the dispersal pattern of the satyrine butterfly Lethe diana. Ecological Entomology 27: 33-40

Ihaka R & Gentleman R (1996) R: a language for data analysis and graphics. Journal of Computational and Graphical Statistics 5: 299-314

Ims RA & Yoccoz NG (1997) Studying transfer processes in metapopulations: emigration, migration and colonization. In: Hanski IA & Gilpin ME (eds.) Metapopulation Biology: Ecology, Genetics, and Evolution. Academic Press, London, pp 247-265

IPCC (2001) Climate Change 2001: the Scientific Basis. Cambridge University PressJennings S, Reynolds JD & Mills SC (1998) Life history correlates of responses to fisheries exploitation.

Proceedings of the Royal Society of London Series B 265: 333-339Jiggins CD, Naisbit RE, Coe RL & Mallet J (2001) Reproductive isolation caused by colour pattern mimicry.

Nature 411: 302-305Johnson CG (1969) Migration and Dispersal of Insects by Flight. Methuen and Co, The Chaucer Press, LondonJohnson CJ, Parker KL, Heard DC & Gillingham MP (2002) Movement parameters of ungulates and scale-

specific responses to the environment. Journal of Animal Ecology 71: 225-235Jonsen ID & Fahrig L (1997) Response of generalist and specialist insect herbivores to landscape spatial

structure. Landscape Ecology 12: 185-197Jonsen ID & Taylor PD (2000a) Fine-scale movement behaviors of calopterygid damselflies are influenced

by landscape structure: an experimental manipulation. Oikos 88: 553-562Jonsen ID & Taylor PD (2000b) Calopteryx damselfly dispersions arising from multiscale responses to

landscape structure. Conservation Ecology 4 (2): art. 4Kapan DD (2001) Three-butterfly system provides a field test of mullerian mimicry. Nature 409: 338-340Kareiva P (1985) Finding and losing host plants by Phyllotreta: patch size and surrounding habitat. Ecology

66: 1809-1816Karlsson B (1994) Feeding habits and change of body composition with age in three nymphalid butterfly

species. Oikos 69: 224-230Karlsson B & Van Dyck H (2005) Does habitat fragmentation affect temperature-related life-history traits? A

laboratory test with a woodland butterfly. Proceedings of the Royal Society of London Series B (in press)Kemp DJ & Krockenberger AK (2002) A novel method of behavioural thermoregulation in butterflies.

Journal of Evolutionary Biology 15: 922-929Kemp DJ & Wiklund C (2004) Residency effects in animal contests. Proceedings of the Royal Society of London

Series B 271: 1707-1711Kingsolver JG (1983) Thermoregulation and flight in Colias butterflies: elevational patterns and mechanistic

limitations. Ecology 64: 534-545Kingsolver JG (1985a) Thermoregulatory significance of wing melanization in Pieris butterflies (Lepidoptera,

Pieridae) – physics, posture, and pattern. Oecologia 66: 546-553Kingsolver JG (1985b) Butterfly thermoregulation: organismic mechanisms and population consequences.

Journal of Research on the Lepidoptera 24: 1-20Kingsolver JG (1987) Predation, thermoregulation, and wing color in Pierid butterflies. Oecologia 73: 301-306Kingsolver JG (1995) Fitness consequences of seasonal polyphenism in Western white butterflies. Evolution

49: 942-954Kingsolver JG & Moffat RJ (1982) Thermoregulation and the determinants of heat transfer in Colias

butterflies. Oecologia 53: 27-33Kingsolver JG & Watt WB (1983) Thermoregulatory strategies in Colias butterflies: thermal stress and the

limits to adaptation in temporally varying environments. American Naturalist 121: 32-55Kingsolver JG & Watt WB (1984) Mechanistic constraints and optimality models – thermoregulatory

strategies in Colias butterflies. Ecology 65: 1835-1839

Kingsolver JG & Wiernasz DC (1987) Dissecting correlated characters: adaptive aspects of phenotypic covariation in melanization pattern of Pieris butterflies. Evolution 41: 491-503

Kingsolver JG & Wiernasz DC (1991) Seasonal polyphenism in wing-melanin pattern and thermoregulatory adaptation in Pieris butterflies. American Naturalist 137: 816-830

Kingsolver JG & Huey RB (1998) Evolutionary analyses of morphological and physiological plasticity in thermally variable environments. American Zoologist 38: 545-560

Kinnison MT & Hendry AP (2001) The pace of modern life II: from rates of contemporary microevolution to pattern and process. Genetica 112: 145-164

Krebs JR & Davies NB (1997) Behavioural Ecology – an Evolutionary Approach, 4th edn. Cambridge University Press.Kuussaari M, Nieminen M & Hanski I (1996) An experimental study of migration in the Glanville fritillary

butterfly Melitaea cinxia. Journal of Animal Ecology 65: 791-801Laird NM & Ware JH (1982) Random effects models for longitudinal data. Biometrics 38: 963-974Lees E & Tilley RJD (1980) Influence of photoperiod and temperature on larval development in Pararge

aegeria (L.) (Lepidoptera: Satyridae). Entomologist’s Gazette 3: 3-6Leimar O & Norberg U (1997) Metapopulation extinction and genetic variation in dispersal-related traits.

Oikos 80: 448-458Leimar O, Norberg U & Wiklund C (2003) Habitat preference and habitat exploration in two species of

satyrine butterflies. Ecography 26: 474-480Lima SL & Zollner PA (1996) Towards a behavioral ecology of ecological landscapes. Trends in Ecology and

Evolution 11: 131-135Lord JM & Norton DA (1990) Scale and the spatial concept of fragmentation. Conservation Biology 4: 197-262Lorenzon P, Clobert J & Massot M (2001) The contribution of phenotypic plasticity to adaptation in Lacerta

vivipara. Evolution 55: 392-404MacArthur RH & Wilson EO (1967) The Theory of Island Biogeography. Princeton University PressMaes D & Van Dyck H (1999) Dagvlinders in Vlaanderen: Ecologie, Verspreiding en Behoud. Antwerp: Stichting

Leefmilieu, Brussels: Instituut voor Natuurbehoud en Vlaamse Vlinderwerkgroep.Maes D & Van Dyck H (2001) Butterfly diversity loss in Flanders (north Belgium): Europe’s worst case

scenario? Biological Conservation 99: 263-276 Malcolm JR (1998) A model of conductive heat flow in forest edges and fragmented landscapes. Climate

Change 39:487-502Marden JH & Chai P (1991) Aerial predation and butterfly design: how palatability, mimicry, and the need

for evasive flight constrain mass allocation. American Naturalist 138: 15-36Marino PC & Landis DA (1996) Effect of landscape structure on parasitoid diversity and parasitism in

agroecosystems. Ecological Applications 6: 276-284Masters AR, Malcolm SB & Brower LP (1988) Monarch butterfly (Danaus plexippus) thermoregulatory

behavior and adaptations for overwintering in Mexico. Ecology 69: 458-467Mate BR, Gisiner R & Mobley J (1998) Local and migratory movements of Hawaiian humpback whales

tracked by satellite telemetry. Canadian Journal of Zoology 76: 863-868Matlack GR (1993) Microenvironment variation within and among forest edge sites in the eastern

United States. Biological Conservation 66: 185-194McNab BK (1994) Energy conservation and the evolution of flightlessness in birds. American Naturalist 144: 628-642Meffe GK & Carroll CR (1997) Principles of Conservation Biology. Sinauer Associates, Sunderland, MassachusettsMennechez G, Schtickzelle N & Baguette M (2003) Metapopulation dynamics of the bog fritillary

butterfly: comparison of demographic parameters and dispersal between a continuous and a highly fragmented landscape. Landscape Ecology 18: 279-291

Mennechez G, Petit S, Schtickzelle N & Baguette M (2004) Modelling mortality and dispersal: consequences of parameter generalisation on metapopulation dynamics. Oikos 106: 243-252

Merckx T & Van Dyck H (2002) Interrelations among habitat use, behavior, and flight-related morphology in two cooccurring satyrine butterflies, Maniola jurtina and Pyronia tithonus. Journal of Insect Behavior 15: 541-561

{ 106 }

{ References }

{ 107 }

Merckx T & Van Dyck H (2005) Dichotomous variation in male mate-location behaviour of the butterfly Pararge aegeria fades away in fragmented landscape. Animal Behaviour (in press)

Merckx T, Van Dyck H, Karlsson B & Leimar O (2003) The evolution of movements and behaviour at boundaries in different landscapes: a common arena experiment with butterflies. Proceedings of the Royal Society of London Series B 270: 1815-1821

Meyer CL & Sisk TD (2001) Butterfly response to microclimatic conditions following Ponderosa pine restoration. Restoration Ecology 9: 453-461

Miner BG & Vonesh JR (2004) Effects of fine grain environmental variability on morphological plasticity. Ecology Letters 7: 794-801

Morales JM (2002) Behavior at habitat boundaries can produce leptokurtic movement distributions. American Naturalist 160: 531-538

Morales JM & Ellner SP (2002) Scaling up animal movements in heterogeneous landscapes: the importance of behavior. Ecology 83: 2240-2247

Morecroft MD, Taylor ME & Oliver HR (1998) Air and soil microclimates of deciduous woodland compared to an open site. Agricultural and Forest Meteorology 90: 141-156

Müller M & Wehner R (1994) The hidden spiral: systematic search and path integration in desert ants, Cataglyphis fortis. Journal of Comparative Physiology Series A 175: 525-530

Niemela J & Spence JR (1991) Distribution and abundance of an exotic ground beetle (Carabidae): a test of community impact. Oikos 62: 351-359

Norberg U (2003) Evolution of Dispersal and Habitat Exploration in Butterflies. Thesis (PhD), Stockholm University

Norberg U & Leimar O (2002) Spatial and temporal variation in flight morphology in the butterfly Melitaea cinxia (Lepidoptera: Nymphalidae). Biological Journal of the Linnean Society 77: 445-453

Norberg U, Enfjäll K & Leimar O (2002) Habitat exploration in butterflies: an outdoor cage experiment. Evolutionary Ecology 16: 1-14

Norton MR, Hannon SJ & Schmiegelow FKA (2000) Fragments are not islands: patch vs landscape perspectives on songbird presence and abundance in a harvested boreal forest. Ecography 23: 209-223

Noss RF (2001) Beyond Kyoto: forest management in a time of rapid climate change. Conservation Biology 15: 578-590

Noss RF & Csuti B (1997) Habitat fragmentation. In: Meffe GK & Carroll CR (eds.) Principles of Conservation Biology, 2nd edn. Sinauer Associates, Sunderland, Massachusetts, pp 269-304

Nylin S & Gotthard K (1998) Plasticity in life-history traits. Annual Review of Entomology 43: 63-83Nylin S, Wickman P-O & Wiklund C (1989) Seasonal plasticity in growth and development of the speckled

wood butterfly, Pararge aegeria (Satyrinae). Biological Journal of the Linnean Society 38: 155-171Nylin S, Wiklund C, Wickman P-O & Garcia-Barros E (1993) Absence of trade-offs between sexual size

dimorphism and early male emergence in a butterfly. Ecology 74: 1414-1427Nylin S, Wickman P-O & Wiklund C (1995) Life-cycle regulation and life-history plasticity in the speckled

wood butterfly – are reaction norms predictable. Biological Journal of the Linnean Society 55: 143-157Olivieri I, Gouyon P-H (1997) Evolution of migration rate and other traits – the metapopulation effect.

In: Hanski IA & Gilpin ME (eds.) Metapopulation Biology – Ecology, Genetics and Evolution. Academic Press, San Diego, California, pp 293-323

Optimas (1999) Optimas 6.5. – User’s Guide and Technical Reference, 9th edn. Optimas Corporation, WashingtonOttenheim MM, Wertheim B, Holloway GJ & Brakefield PM (1999) Survival of colour-polymorphic Eristalis

arbustorum hoverflies in semi-field conditions. Functional Ecology 13: 72-77Palmer JO & Dingle H (1989) Responses to selection on flight behavior in a migratory population of

milkweed bug (Oncopeltus fasciatus). Evolution 43: 1805-1808Palumbi SR (2001) Evolution – Humans as the world’s greatest evolutionary force. Science 293: 1786-1790Parker GA (1978) Evolution of competitive mate searching. Annual Review of Entomology 23: 173-196Peng R, Sutton SL & Fletcher CR (1994) Distribution patterns of some species of Scatopsidae (Insecta,

Diptera) and the effect of microclimate on their flight activity. Journal of Zoology 232: 585-594

Pigliucci M (1996) How organisms respond to environmental changes: from phenotypes to molecules (and vice versa). Trends in Ecology and Evolution 11: 168-173

Pigliucci M, Whitton J & Schlichting CD (1995) Reaction norms of Arabidopsis. I. Plasticity of characters and correlations across water, nutrient and light gradients. Journal of Evolutionary Biology 8: 421-438

Pimm SL & Raven P (2000) Biodiversity – Extinction by numbers. Nature 403: 843-845Pivnick KA & McNeil JN (1986) Sexual differences in the thermoregulation of Thymelicus lineola adults

(Lepidoptera: Hesperiidae). Ecology 67: 1024-1035Postma E & van Noordwijk AJ (2005) Gene flow maintains a large genetic difference in clutch size at a

small spatial scale. Nature 433: 65-68Price T, Qvarnström A, Irwin DE (2003) The role of phenotypic plasticity in driving genetic evolution.

Proceedings of the Royal Society of London Series B 270: 1433-1440Rawlins JE (1980) Thermoregulation by the black swallowtail butterfly, Papilio polyxenes (Lepidoptera:

Papilionidae). Ecology 61: 345-357Redpath SM (1995) Impact of habitat fragmentation on activity and hunting behavior in the tawny owl,

Strix aluco. Behavioral Ecology 6: 410-415Reznick DN & Ghalambor CK (2001) The population ecology of contemporary adaptations: what empirical

studies reveal about the conditions that promote adaptive evolution. Genetica 112: 183-198Ricketts TH (2001) The matrix matters: effective isolation in fragmented landscapes. American Naturalist

158: 87-99Ricketts TH, Daily GC, Ehrlich PR & Fay JP (2001) Countryside biogeography of moths in a fragmented

landscape: biodiversity in native and agricultural habitats. Conservation Biology 15: 378-388Ries L & Debinski DM (2001) Butterfly responses to habitat edges in the highly fragmented prairies of

Central Iowa. Journal of Animal Ecology 70: 840-852Roland J, Keyghobadi N & Fownes S (2000) Alpine Parnassius butterfly dispersal: effects of landscape and

population size. Ecology 81: 1642-1653Rundle HD & Nosil P (2005) Ecological speciation. Ecology Letters 8: 336-352 Rutowski RL (1981) Sexual discrimination using visual cues in the checkered white butterfly (Pieris protodice).

Zeitchrift für Tierpsychologie – Journal of Comparative Ethology 55: 325-334Rutowski RL (1984) Sexual selection and the evolution of butterfly mating behaviour. Journal of Research on

Lepidoptera 23: 125-142Rutowski RL (1991) The evolution of male mate-locating behavior in butterflies. American Naturalist 138:

1121-1139Rutowski RL (2003) Visual ecology of adult butterflies. In: Boggs CL, Watt WB & Ehrlich PR (eds.) Butterflies:

Ecology and Evolution Taking Flight. University of Chicago Press, pp 9-25Rutowski RL, Demlong MJ & Leffingwell T (1994) Behavioral thermoregulation at mate encounter sites by

male butterflies (Asterocampa leilia, Nymphalidae). Animal Behaviour 48: 833-841SAS (2001) SAS® Proprietary Software Release 8.2. SAS Institute, Cary, NC, USAScheiner SM (1993) Genetics and evolution of phenotypic plasticity. Annual Review of Ecology and Systematics

24: 35-68Schlichting CD (1986) The evolution of phenotypic plasticity in plants. Annual Review of Ecology and Systematics

17: 667-693Schmitz H (1994) Thermal characterization of butterfly wings: I. Absorption in relation to different color,

surface structure and basking type. Journal of Thermal Biology 19: 403-412Schooley RL & Wiens JA (2003) Finding habitat patches and directional connectivity. Oikos 102: 559-570Schtickzelle N & Baguette M (2003) Behavioural responses to habitat patch boundaries restrict dispersal

and generate emigration-patch area relationships in fragmented landscapes. Journal of Animal Ecology 72: 533-545

Schtickzelle N & Baguette M (2004) Metapopulation viability analysis of the bog fritillary butterfly using RAMAS/GIS. Oikos 104: 277-290

{ 108 }

{ References }

{ 109 }

Schultz CB & Crone EE (2001) Edge-mediated dispersal behavior in a prairie butterfly. Ecology 82: 1879-1892Scott JA (1974) Mate-locating behavior of butterflies. American Midland Naturalist 91: 103-117Shapiro AM (1976) Seasonal polyphenism. Evolutionary Biology 9: 259-333Shelly TE & Ludwig D (1985) Thermoregulatory behavior of the butterfly Calisto nubila (Satyridae) in a

Puerto-Rican forest. Oikos 44: 229-233Shipp JL, Grace BW & Schallje GB (1987) Effects of microclimate on daily flight activity of Simulium arcticum

Malloch (Diptera, Simuliidae). International Journal of Biometeorology 31: 9-20Shreeve TG (1984) Habitat selection, mate location, and microclimatic constraints on the activity of the

speckled wood butterfly Pararge aegeria. Oikos 42: 371-377Shreeve TG (1986a) Egg-laying by the speckled wood butterfly (Pararge aegeria): the role of female behaviour,

host plant abundance and temperature. Ecological Entomology 11: 229-236Shreeve TG (1986b) The effect of weather on the life cycle of the speckled wood butterfly Pararge aegeria.

Ecological Entomology 11: 325-332Shreeve TG (1987) The mate location behaviour of the male speckled wood butterfly, Pararge aegeria, and the

effect of phenotypic differences in hind-wing spotting. Animal Behaviour 35: 682-690Shreeve TG (1992a) Adult behaviour. In: Dennis RLH (ed.) The Ecology of Butterflies in Britain. Oxford University

Press, pp 22-45Shreeve TG (1992b) Monitoring butterfly movements. In: Dennis RLH (ed.) The Ecology of Butterflies in Britain.

Oxford University Press, pp 120-138Shreeve TG & Dennis RLH (1992) The development of butterfly settling posture: the role of predators,

climate, hostplant-habitat and phylogeny. Biological Journal of the Linnean Society 45: 57-69Sibly RM, Winokur L & Smith RH (1997) Interpopulation variation in phenotypic plasticity in the speckled

wood butterfly, Pararge aegeria. Oikos 78: 323-330Silberglied RE (1984) Visual communication and sexual selection among butterflies. In: Vane-Wright RI &

Ackery PR (eds.) The Biology of Butterflies. Academic Press, London, pp 207-223Soderstrom B, Svensson B, Vessby K & Glimskar A (2001) Plants, insects and birds in semi-natural pastures

in relation to local habitat and landscape factors. Biodiversity and Conservation 10: 1839-1863Southwood TRE (1962) Migration of terrestrial arthropods in relation to habitat. Biological Reviews 37: 171-214Srygley RB (1994) Shivering and its cost during reproductive behavior in Neotropical owl butterflies, Caligo

and Opsiphanes (Nymphalidae, Brassolinae). Animal Behaviour 47: 23-32Srygley RB & Chai P (1990a) Flight morphology of Neotropical butterflies: palatability and distribution of

mass to the thorax and abdomen. Oecologia 84: 491-499Srygley RB & Chai P (1990b) Predation and the elevation of thoracic temperature in brightly colored

Neotropical butterflies. American Naturalist 135: 766-787Stauffer JR & Gray EV (2004) Phenotypic plasticity: its role in trophic radiation and explosive speciation

in cichlids (Teleostei: Cichlidae). Animal Biology 54: 137-158Steffan-Dewenter I & Tscharntke T (1997) Early succession of butterfly and plant communities on set-aside

field. Oecologia 109: 294-302Steffan-Dewenter I & Tscharntke T (1999) Effects of habitat isolation on pollinator communities and seed

set. Oecologia 121: 432-440Steffan-Dewenter I & Tscharntke T (2000) Butterfly community structure in fragmented habitats. Ecology

Letters 3: 449-456Steffan-Dewenter I & Tscharntke T (2002) Insect communities and biotic interactions on fragmented

calcareous grasslands – a mini review. Biological Conservation 104: 275-284Steffan-Dewenter I, Munzenberg U & Tscharntke T (2001) Pollination, seed set and seed predation on a

landscape scale. Proceedings of the Royal Society of London Series B 268: 1685-1690Steffan-Dewenter I, Munzenberg U, Burger C, Thies C & Tscharntke T (2002) Scale-dependent effects of

landscape context on three pollinator guilds. Ecology 83: 1421-1432Stevens DJ (2004) Pupal development temperature alters adult phenotype in the speckled wood butterfly,

Pararge aegeria. Journal of Thermal Biology 29: 205-210

Stevens VM, Polus E, Wesselingh RA, Schtickzelle N & Baguette M (2004) Quantifying functional connectivity: experimental evidence for patch-specific resistance in the Natterjack toad (Bufo calamita). Landscape Ecology 19: 829-842

Stevenson RD & Josephson RW (1990) Effects of operating frequency and temperature on mechanical power output from moth flight muscle. Journal of Experimental Biology 149: 61-78

Stockwell CA, Hendry AP & Kinnison MT (2003) Contemporary evolution meets conservation biology. Trends in Ecology and Evolution 18: 94-101

Sultan SE (1987) Evolutionary implications of phenotypic plasticity in plants. Evolutionary Biology 21: 127-178Sutcliffe OL, Thomas CD & Peggie D (1997) Area-dependent migration by ringlet butterflies generates a

mixture of patchy population and metapopulation attributes. Oecologia 109: 229-234Talloen W, Van Dyck H & Lens L (2004) The cost of melanization: butterfly wing coloration under

environmental stress. Evolution 58: 360-366Taylor PD & Merriam G (1995) Wing morphology of a forest damselfly is related to landscape structure.

Oikos 73: 43-48Thies C & Tscharntke T (1999) Landscape structure and biological control in agroecosystems. Science 285:

893-895Thomas CD (1994) Extinction, colonization, and metapopulations: environmental tracking by rare species.

Conservation Biology 8: 373-378Thomas CD (2000) Dispersal and extinction in fragmented landscapes. Proceedings of the Royal Society of

London Series B 267: 139-145Thomas CD & Harrison S (1992) Spatial dynamics of a patchily distributed butterfly species. Journal of

Animal Ecology 61: 437-446Thomas CD & Hanski IA (1997) Butterfly metapopulations. In: Hanski IA & Gilpin ME (eds.) Metapopulation

Dynamics: Ecology, Genetics and Evolution. Academic Press, London, pp 359-386Thomas CD, Hill JK & Lewis OT (1998) Evolutionary consequences of habitat fragmentation in a localized

butterfly. Journal of Animal Ecology 67: 485-497Thomas CD, Bodsworth EJ, Wilson RJ, Simmons AD, Davies ZG, Musche M & Conradt L (2001) Ecological

and evolutionary processes at expanding range margins. Nature 411: 577-581Thomas CD and 18 others (2004) Extinction risk from climate change. Nature 427: 145-148Thornhill R & Alcock J (1983) The Evolution of Insect Mating Systems. Harvard University Press, Cambridge,

MassachusettsTischendorf L & Fahrig L (2000) How should we measure landscape connectivity? Landscape Ecology 15:

633-641Tolman T & Lewington R (1997) Butterflies of Britain and Europe. Harper Collins, LondonTscharntke T, Steffan-Dewenter I, Kruess A & Thies C (2002) Characteristics of insect populations on

habitat fragments: a mini review. Ecological Research 17: 229-239Turcotte Y & Desrochers A (2003) Landscape-dependent response to predation risk by forest birds. Oikos

100: 614-618Van Dyck H (2003) Mate-location: a matter of design? Adaptive morphological variation in the speckled

wood butterfly. In: Boggs CL, Watt WB & Ehrlich PR (eds.) Butterflies: Ecology and Evolution Taking Flight. University of Chicago Press, pp 353-366

Van Dyck H & Matthysen E (1998) Thermoregulatory differences between phenotypes of the speckled wood butterfly: hot perchers and cold patrollers? Oecologia 114: 326-334

Van Dyck H & Matthysen E (1999) Habitat fragmentation and insect flight: a changing ‘design’ in a changing landscape? Trends in Ecology and Evolution 14: 172-174

Van Dyck H & Wiklund C (2002) Seasonal butterfly design: morphological plasticity among three developmental pathways relative to sex, flight and thermoregulation. Journal of Evolutionary Biology 15: 216-225

Van Dyck H & Baguette M (2005) Dispersal behaviour in fragmented landscapes: routine or special movements? Basic and Applied Ecology (in press)

{ 110 }

{ References }

{ 111 }

Van Dyck H, Matthysen E & Dhondt AA (1997a) The effect of wing colour on male behavioural strategies in the speckled wood butterfly. Animal Behaviour 53: 39-51

Van Dyck H, Matthysen E & Dhondt AA (1997b) Mate-locating strategies are related to relative body length and wing colour in the speckled wood butterfly Pararge aegeria. Ecological Entomology 22: 116-120

Van Dyck H, Matthysen E & Wiklund C (1998) Phenotypic variation in adult morphology and pupal colour within and among families of the speckled wood butterfly Pararge aegeria. Ecological Entomology 23: 465-472

Van Swaay CAM & Groenendijk D (2004) Vlinders en libellen onder de meetlat: jaarverslag 2003. Rapport VS2004.011, De Vlinderstichting, Wageningen

Vane-Wright RI & Boppré M (1993) Visual and chemical signalling in butterflies: functional and phylogenetic perspectives. Philosophical Transactions of the Royal Society of London Series B 340: 197-205

Verbeke G & Molenberghs G (2001) Linear Mixed Models for Longitudinal Data. Springer Verlag, New YorkVielmetter W (1958) Physiologie des Verhaltens zur Sonnenstrahlung bei dem Tagfalter Argynnis paphia L.: I.

Untersuchungen im Freiland. Journal of Insect Physiology 2: 13-37Wahlberg N, Klemetti T, Selonen V & Hanski IA (2002) Metapopulation structure and movements in five

species of checkerspot butterflies. Oecologia 130: 33-43Warren MS and 14 others (2001) Rapid responses of British butterflies to opposing forces of climate and

habitat change. Nature 414: 65-69Waser PM, Creel SR & Lucas JR (1994) Death and disappearance: estimating mortality risks associated with

philopatry and dispersal. Behavioral Ecology 5: 135-141Wasserthal LT (1975) The role of butterfly wings in regulation of body temperatures. Journal of Insect

Physiology 221: 1921-1930Watanabe M & Imoto T (2003) Thermoregulation and flying habits of the Japanese sulfur butterfly Colias

erate (Lepidoptera: Pieridae) in an open habitat. Entomological Science 6: 111-118Watt WB (1968) Adaptive significance of pigment polymorphisms in Colias butterflies. I. Variation of

melanin pigment in relation to thermoregulation. Evolution 22: 437-458Watt WB (2003) Mechanistic studies of butterfly adaptations. In: Boggs CL, Watt WB & Ehrlich PR (eds.)

Butterflies: Ecology and Evolution Taking Flight. University of Chicago Press, pp 319-352Wedell N, Gage MJG & Parker GA (2002) Sperm competition, male prudence and sperm-limited females.

Trends in Ecology and Evolution 17: 313-320Weibull AC, Bengtsson J & Nohlgren E (2000) Diversity of butterflies in the agricultural landscape: the

role of farming system and landscape heterogeneity. Ecography 23: 743-750West-Eberhard MJ (1989) Phenotypic plasticity and the origins of diversity. Annual Review of Ecology and

Systematics 20: 249-278West-Eberhard MJ (2003) Developmental plasticity and evolution. Oxford University PressWickman P-O (1985) The influence of temperature on the territorial and mate locating behavior of the

small heath butterfly, Coenonympha pamphilus (L.) (Lepidoptera: Satyridae). Behavioral Ecology and Sociobiology 16: 233-238

Wickman P-O (1988) Dynamics of mate-searching behaviour in a hilltopping butterfly, Lasiommata megera: the effects of weather and male density. Zoological Journal of the Linnean Society 93: 357-377

Wickman P-O (1992) Mating systems of Coenonympha butterflies in relation to longevity. Animal Behaviour 44: 141-148

Wickman P-O & Wiklund C (1983) Territorial defence and its seasonal decline in the speckled wood butterfly (Pararge aegeria). Animal Behaviour 31: 1206-1216

Wiens JA (1997) Metapopulation dynamics and landscape ecology. In: Hanski IA & Gilpin ME (eds.) Metapopulation Biology: Ecology, Genetics, and Evolution. Academic Press, London, pp 43-62

Wiens JA (2001) The landscape context of dispersal. In: Clobert J, Danchin E, Dhondt AA & Nichols JD (eds.) Dispersal. Oxford University Press, pp 96-109

Wiens JA, Stenseth NC, Van Horne B & Ims RA (1993) Ecological mechanisms and landscape ecology. Oikos 66: 369-380

Wiklund C (2003) Sexual selection and the evolution of butterfly mating systems. In: Boggs CL, Watt WB & Ehrlich PR (eds.) Butterflies: Ecology and Evolution Taking Flight. University of Chicago Press, pp 67-90

Wiklund C & Fagerström T (1977) Why do males emerge before females? A hypothesis to explain incidence of protandry in butterflies. Oecologia 31: 153-158

Wiklund C & Persson A (1983) Fecundity, and the relation of egg weight variation to offspring fitness in the speckled wood butterfly Pararge aegeria, or why don’t butterfly females lay more eggs? Oikos 40: 53-63

Wiklund C & Forsberg J (1991) Sexual size dimorphism in relation to female polygamy and protandry in butterflies: a comparative study of Swedish Pieridae and Satyridae. Oikos 60: 373-381

Wiklund C, Persson A & Wickman P-O (1983) Larval estivation and direct development as alternative strategies in the speckled wood butterfly, Pararge aegeria in Sweden. Ecological Entomology 8: 233-238

Wiklund C, Wickman P-O & Nylin S (1992) A sex difference in the propensity to enter direct/diapause development – a result of selection for protandry. Evolution 46 519-528

Wilcove DS, McLellan CH & Dobson AP (1986) Habitat fragmentation in the temperate zone. In: Soulé ME (ed.) Conservation Biology: the Science of Scarcity and Diversity. Sinauer Associates, Sunderland, Massachusetts, pp 237-256

Williamson MH (1981) Island Populations. Oxford University PressWillmer PG (1986) Microclimatic effects on insects at the plant surface. In: Juniper BE & Southwood TRE

(eds.) Insects and Plant Surfaces. Edward Arnold, London, pp 65-80Willmer PG (1991) Thermal biology and mate acquisition in ectotherms. Trends in Ecology and Evolution 6:

396-399Wilson RJ & Thomas CD (2002) Dispersal and the spatial dynamics of butterfly populations. In: Bullock J,

Kenward R & Hails R (eds.) Dispersal Ecology. Blackwell Science, Oxford, pp 257-278Wilson RJ, Thomas CD, Fox R, Roy DB & Kunin WE (2004) Spatial patterns in species distributions reveal

biodiversity change. Nature 432: 393-396With KA & Crist TO (1995) Critical thresholds in species’ responses to landscape structure. Ecology 76:

2446-2459With KA, Gardner RH & Turner MG (1997) Landscape connectivity and population distributions in

heterogeneous environments. Oikos 78: 151-169With KA, Cadaret SJ & Davis C (1999) Movement responses to patch structure in experimental fractal

landscapes. Ecology 80: 1340-1353Wolff JO, Schauber EM & Edge WD (1997) Effects of habitat loss and fragmentation on the behavior and

demography of gray-tailed voles. Conservation Biology 11: 945-956Yeh PJ & Price TD (2004) Adaptive phenotypic plasticity and the successful colonization of a novel

environment. American Naturalist 164: 531-542Yeomans SR (1995) Water-finding in adult turtles: random search or oriented behaviour? Animal Behaviour

49: 977-987Zollner PA (2000) Comparing the landscape level perceptual abilities of forest sciurids in fragmented

agricultural landscapes. Landscape Ecology 15: 523-533Zollner PA & Lima SL (1997) Landscape-level perceptual abilities in white-footed mice: perceptual range

and the detection of forested habitat. Oikos 80: 51-60Zollner PA & Lima SL (1999) Search strategies for landscape-level interpatch movements. Ecology 80: 1019-1030Zschokke S, Dolt C, Rusterholz HP, Oggier P, Braschler B, Thommen GH, Ludin E, Erhardt A & Baur B

(2000) Short-term responses of plants and invertebrates to experimental small-scale grassland fragmentation. Oecologia 125: 559-572

Summ

ary

{ 113 }

Summary

This thesis has an evolutionary ecological focus, investigating how organisms cope with changing environments. More specifically, this work addresses the intriguing and time-ly issue of differing life history traits (including functional morphology and behaviour) in landscapes where an organism’s habitat gets fragmented. Habitat fragmentation is widely recognised as a key problem for biodiversity at the global and local scale. The speckled wood butterfly (Pararge aegeria L.) was chosen as a model of a flying ectotherm (more precisely a heliotherm). Although this temperate-zone satyrid is primarily a woodland butterfly, it also occurs in more fragmented landscapes which include wood-land aspects; such as agricultural landscapes with hedgerows and/or small patches of woodland. A primary interest is to evaluate to what extent landscape structure influ-ences movement behaviour, such as tracking resources and dispersal between popula-tions. Movements are closely linked with the behaviour and properties of butterfly ther-moregulation. In order to test a series of hypotheses, we compared the behaviour, mor-phology and thermal properties of individuals from continuous woodland landscapes to those from highly fragmented, agricultural landscapes. In the latter, the environment between suitable habitat fragments is an intensively managed and hostile landscape matrix. Using several experimental approaches – including reciprocal transplant ex-periments and observations in ‘common garden’ designs – we provide evidence for land-

{ 114 }

Summ

ary

{ 115 }

scape-related differences in behaviour, functional morphology and eco-physiology. This suggests fast micro-evolutionary changes. From an adaptive point of view, these differ-ences are in line with the assumption of different dispersal rates that may result from altered microclimatic conditions and from changed distributions (and hence encounter rates) of resources between landscape types.

In evolutionary time, varying environments may lead to different morphs as a re-sult of genetic adaptation and divergence, or phenotypic plasticity; the latter allows in-dividuals to deal with varying or changing environments more rapidly than the former. Results of the first experiment (i.e. reciprocal transplant with offspring of females from differently fragmented landscapes), suggest the existence of landscape-related pheno-typic plasticity in butterfly flight morphology. To the best of our knowledge, this is the first evidence for this kind of plasticity. The observed change in flight morphology can be interpreted in terms of altered ability to cover distances among resources, in line with altered distributions and encounter rates of resources in the landscape associated with habitat fragmentation. In addition, we found significant genotype-by-environ-ment (G x E) interactions for some of the traits. This suggests that the observed land-scape-related phenotypic plasticity can respond to natural selection, and, hence, that it may contribute to evolutionary changes in dispersal (and probably in other flight-relat-ed behaviours that need further research) with changed landscape structure. The cues causing these responses have not yet been studied, but microclimatic parameters (such as temperature) are most likely involved. In this context, we hypothesised that species with high levels of plasticity in morphology and life history have a greater capacity to deal with changing and variable landscapes than other species. The experiment also re-vealed new data on differential larval mortality between the different landscapes, and showed an additional role of landscape of origin to this mortality. This demonstrates that evolutionary ecological studies on habitat fragmentation should not be exclusively limited to adult ecology.

Altered microclimates may cause direct effects on flight and costs of flight in small ectotherms such as butterflies, independent of changes in the distribution of resources (e.g. as host plants or feeding sources). Agricultural landscapes were typified by higher ambient temperatures and wind speeds compared to woodland landscapes. This was found to lead to a larger variation in heating rate of P. aegeria in agricultural landscapes, using butterfly dummies (i.e. dead butterfly specimens) on thermoprobes. We simulated flight also using these dummies to measure the distance required to cool thorax flight temperatures from optimal (33°C) to a given (25°C) suboptimal temperature under dif-ferent environmental/landscape conditions. Butterflies may compensate for the higher level of convective cooling in agricultural landscapes by flying more frequently in the vicinity of shelter providing vegetation structures, especially when they do so close to the substrate. We were then able to prove that living individuals fly more frequently at lower heights in agricultural landscapes than in woodlands. Besides these behav-

ioural aspects, we found that individuals in agricultural landscapes had more fur than woodland butterflies. This can be interpreted as a morphological adaptation to reduce the impact of convective cooling by its thermal isolation effect. At the landscape level, these results draw attention to the potential importance of vegetation structures as corridors for movement across fragmented landscapes. Moreover, we predict that the significance of such structural characteristics will be weather or climate dependent. If the butterflies were only able to make relatively short flights, a finer grained ecologi-cal network of corridors would be required. Next, we experimentally tested whether flight ability differed between laboratory-reared offspring from a woodland landscape and those from an agricultural landscape. Apart from the observation that P. aegeria is able to fly at much lower ambient temperatures than other previously studied butterfly species, individuals of woodland origin flew for longer at suboptimal and optimal tem-peratures than did progeny of females from the agricultural landscape. In summary, all results point to temperature-related, adaptive differences in cooler, but sheltered environments of continuous woodland versus warmer, but more exposed environments of fragmented, agricultural landscapes.

Selection regimes on movement behaviour, through the costs associated with flight activity, may be related to the degree of habitat fragmentation as a result of microcli-matic properties together with the distribution of resources. Here, we studied whether and to what extent P. aegeria butterflies from contrasting landscape types differed in movement behaviour or related traits, such as mate location behaviour, flight activity, dispersal propensity, and the ability to find forested habitat. We found significant dif-ferences in mate location behaviour with landscape: males showed higher levels of ag-gressive take-offs (an indicator of territorial perching), and made more displacements (an indicator of patrolling) in agricultural landscapes compared to woodland. Similarly, in an independent census, there was a much higher frequency of behaviour interme-diate between perching and patrolling in agricultural landscapes than in woodland. These results suggest that the clear dichotomy of perching versus patrolling as typically observed in woodland, fades away in highly fragmented, agricultural landscapes. Al-though alternative mate location strategies have attracted much attention among behav-ioural ecologists, and habitat fragmentation provides a typical framework for studies on changes in dispersal behaviour, the study of mate location in the context of habitat fragmentation has only rarely been addressed. Our experimental approach to compare behavioural responses among adult offspring of field-collected females in large, outdoor cages, revealed that woodland butterflies were more mobile than those from agricul-ture: more frequently at flight, covered longer distances, were more prone to cross open-shade boundaries, and travelled more frequently between ‘woodland’ parts in the cages. In another field experiment, we released butterflies form both types of landscape at several distances from suitable habitat to evaluate differences in detection ability. But-terflies from the agricultural landscape oriented themselves towards forested habitat

{ 116 }

from a distance twice as far as did conspecifics from woodland. These results agree with heritable differences in flight activity, dispersal propensity and habitat-finding ability among the landscapes pointing to evolutionary consequences of habitat fragmentation. We discuss several new ideas that emerge from the results, such as about the variation between populations from different landscapes to deal with coarse-grained ecological infrastructures. It may be valuable to test in other species, particularly those of high conservation interest, for different optima in the geometry of corridor or stepping stone networks for the same species in different landscape types.

In addition to the several lines of evidence for evolutionary changes in thermoregu-lation and movement behaviour with landscape, we found that the effects were often sex-specific. These sexual differences can be understood relative to the multifunctional nature of butterfly flight, which differs between males and females as they differ in behavioural repertoires and ecological roles. This study on the evolutionary ecology of movement, behaviour, functional morphology and thermoregulation in P. aegeria adds sig-nificantly to the diverse ecological and evolutionary effects that habitat fragmentation may have on life history traits of organisms dealing with our man-made landscapes.

Samenvatting

{ 117 }

Samenvatting

Vanuit een evolutionair ecologisch gezichtspunt focust dit doctoraatsproefschrift op de vraag hoe organismen omgaan met veranderende omgevingen. Dit werk behandelt meer bepaald het intrigerende en actuele punt van veranderende ‘life history’ aspecten (zoals functionele morfologie en gedrag) in landschappen waarin de habitat van een soort gefragmenteerd raakt. Habitatfragmentatie wordt algemeen erkend als een belan-grijk probleem voor biodiversiteit, en dat zowel op lokale als op meer globale schaal. Als studiemodel voor een vliegend ectotherm organisme (en meer specifiek een heliotherm organisme) kozen we het Bont zandoogje (Pararge aegeria L.). Hoewel dit zandoogje van gematigde streken in hoofdzaak een bosvlinder is, komt het ook voor in meer gefrag-menteerde landschappen met een zeker bosaspect, zoals in landbouwlandschappen met hagen en houtkanten en/of lapjes bos. Onze interesse gaat in hoofdzaak uit naar het bepalen van de mate waarin de structuur van het landschap een invloed uitoefent op verplaatsingsgedrag (bv. hulpbronnen zoeken, dispersie tussen populaties). Verplaats-ingen bij vlinders zijn sterk gelinkt met thermoregulatie. Met het oog op het testen van een reeks hypotheses vergeleken we het gedrag, de morfologie en de thermale ei-genschappen van individuen uit grote boscomplexen met die uit sterk gefragmenteerde landbouwlandschappen. In deze laatste is de hoeveelheid geschikt habitat beperkt tot snippers in een intensief beheerde, vijandige landschapsmatrix. Aan de hand van ver-

{ 118 }

Samenvatting

{ 119 }

scheidene experimentele benaderingen – waaronder wederkerige transplantatie experi-menten en observaties in ‘common garden’ opstellingen – leveren we bewijsmateriaal voor landschapsgerelateerde verschillen in gedrag, functionele morfologie en eco-fysi-ologie. Dat wijst op snelle micro-evolutionaire veranderingen. Vanuit een adaptief ge-zichtspunt komen die verschillen overeen met de assumptie van een andere mate van dispersie, die mogelijk resulteert uit veranderde microklimatologische omstandigheden en uit een andere spreiding (en dus een andere trefkans) van de hulpbronnen tussen landschapstypes.

In een evolutionair perspectief kunnen variërende omgevingen leiden tot ver-schillende morfologische varianten die een gevolg zijn van genetische adaptatie en di-vergentie, of van fenotypische plasticiteit; dit laatste laat individuen toe sneller om te gaan met variërende of veranderende omgevingen. De resultaten van een eerste experi-ment (i.e., wederkerige transplantatie met nakomelingen van wijfjes uit verschillend gefragmenteerde landschappen) suggereren het bestaan van landschapsgerelateerde fenotypische plasticiteit in vlieggerelateerde morfologie bij vlinders. Voor zover we weten, is dit de eerste maal dat zo’n plasticiteit aangetoond wordt. De waargenomen verschillen in vlieggerelateerde morfologie kunnen geïnterpreteerd worden vanuit een veranderd vermogen om afstanden tussen hulpbronnen te overbruggen. Dat is in overeenstemming met veranderde spreidingen en trefkansen van hulpbronnen in het landschap als een gevolg van habitatfragmentatie. Bovendien vonden we significante genotype-omgevingsinteracties bij enkele morfologische kenmerken. Dat wijst erop dat de waargenomen landschapsgerelateerde fenotypische plasticiteit in staat is op natuurl-ijke selectie te reageren. Ze draagt op die manier bij tot evolutionaire veranderingen in dispersie (en waarschijnlijk ook in ander vlieggerelateerd gedrag dat verder onderzoek vereist) met een veranderde structuur van het landschap. Welke prikkels die reacties veroorzaken, is nog niet bestudeerd, maar microklimatologische parameters (zoals tem-peratuur) zijn zeer waarschijnlijk betrokken partij. In dat verband formuleerden we een hypothese die stelt dat soorten met een hoge mate aan plasticiteit in morfologie en ‘life history’ meer troeven bezitten om het hoofd te bieden aan veranderende en vari-abele landschappen. Het experiment onthulde ook nieuwe gegevens over differentiële larvale mortaliteit tussen de verschillende landschappen en toonde ook aan dat het landschap van oorsprong een bijkomende rol vervult in deze mortaliteit. Dat laat zien dat evolutionair ecologische studies rond habitatfragmentatie best niet beperkt blijven tot ecologie van adulte individuen.

Veranderde microklimaten kunnen onafhankelijk van veranderingen in de configuratie van hulpbronnen, zoals waardplanten of voedselbronnen, ook rech-tstreekse effecten sorteren op het vliegen (en de kosten ervan) in kleine ectotherme or-ganismen zoals vlinders. Landbouwlandschappen werden gekenmerkt door een hogere omgevingstemperatuur en windsterkte vergeleken met boslandschappen. Dat verschil vertaalt zich in een grotere variatie in opwarmsnelheid van P. aegeria in landbouwland-

schappen, wat we bestudeerden met dummy’s (i.e., dode vlinders) op een thermosonde. Met die dummy’s simuleerden we ook vluchten om de afstand te schatten die kan over-brugd worden om de thorax van een optimale vliegtemperatuur (33°C) af te koelen tot een bepaalde suboptimale temperatuur (25°C), en dat onder verschillende omgevings-/landschapscondities. Vlinders zouden kunnen compenseren voor de hogere graad van convectieve koeling in landbouwlandschappen door frequenter in de nabijheid van beschuttingverlenende vegetatiestructuren te vliegen, vooral wanneer ze dat op lage vlieghoogte doen. We toonden aan dat levende individuen inderdaad frequenter op lage hoogte vliegen in landbouwlandschappen dan in bossen. Behalve gedragsverschillen vonden we ook morfologische verschillen: individuen in landbouwlandschappen had-den meer beharing dan bosindividuen. Omdat beharing isoleert, kan dat geïnterpret-eerd worden als een morfologische adaptatie om de impact van convectieve koeling te verminderen. Die resultaten vestigen de aandacht op het potentiële belang van vegetati-estructuren als corridors voor verplaatsingen doorheen gefragmenteerde landschappen. Bovendien voorspellen we dat het belang van die structuren weers- of klimaatsafhan-kelijk is. Mochten vlinders enkel in staat zijn om relatief korte vluchten te maken, zou een meer fijnkorrelig ecologisch netwerk van corridors vereist zijn. Daarnaast testten we op een experimentele manier of het vliegvermogen verschilt tussen in het labora-torium gekweekte nakomelingen van wijfjes uit een boslandschap versus een landbou-wlandschap. Los van de observatie dat P. aegeria in staat is te vliegen bij veel lagere om-gevingstemperaturen dan andere dagvlindersoorten die tot nu toe werden bestudeerd, vlogen vlinders van bosoorsprong langer bij suboptimale en optimale temperaturen dan nakomelingen van wijfjes uit een landbouwlandschap. Samengevat kunnen we stellen dat alle resultaten temperatuurgerelateerde adaptieve verschillen suggeren tussen de koelere, maar beschutte omgevingen van grote boscomplexen en de warmere, maar meer open omgevingen van gefragmenteerde landbouwlandschappen.

De selectieregimes op verplaatsingsgedrag, door de aan vliegactiviteit ver-bonden kost, zouden gerelateerd kunnen zijn aan de graad van habitatfragmentatie en dat zowel als een gevolg van microklimatologische eigenschappen als van de configura-tie van de hulpbronnen. Hier bestudeerden we of en in welke mate P. aegeria vlinders van contrasterende landschapstypes verschillen in verplaatsingsgedrag of in ermee samenhangende kenmerken, zoals partnerzoekgedrag, vliegactiviteit, dispersieneiging, en het vermogen om bebost habitat te vinden. We vonden een significant effect van landschap op partnerzoekgedrag: mannetjes vertoonden grotere aantallen agressieve ‘take-offs’ (i.e., indicator van territoriaal ‘perching’) en ze maakten meer verplaatsingen (i.e., indicator van ‘patrolling’) in landbouwlandschappen dan in boscomplexen. Evenzo toonde een onafhankelijke census aan dat er in landbouwlandschappen een veel hogere frequentie voorkomt van een gedrag intermediair tussen ‘perching’ en ‘patrolling’. Die resultaten suggereren dat de duidelijke grens tussen ‘perching’ en ‘patrolling’, zoals die typisch terug te vinden is in een boslandschap, vervaagt in sterk gefragmenteerde

{ 120 }

landbouwlandschappen. Hoewel alternatieve partnerzoekstrategieën heel wat aandacht kregen bij gedragsecologen, en hoewel habitatfragmentatie een typisch kader biedt voor studies over veranderingen in dispersiegedrag, is er slechts zelden onderzoek gedaan naar partnerzoekgedrag in de context van habitatfragmentatie. Onze experimentele benadering om buiten, in grote kooien gedragsresponsen te vergelijken tussen volwas-sen nakomelingen van in het veld verzamelde wijfjes, bracht aan het licht dat vlinders van het bos mobieler waren dan deze van het landbouwlandschap; ze waren vaker aan het vliegen, ze legden grotere afstanden af, ze waren sneller geneigd om open-schaduw grenzen over te steken, en ze verplaatsten zich frequenter tussen de ‘bos’delen in de kooien. In een ander veldexperiment lieten we vlinders van beide landschapstypes los op meerdere afstanden van geschikt habitat om verschillen in detectiecapaciteit na te gaan. Vlinders van het landbouwlandschap waren in staat om zich te oriënteren naar bebost habitat vanaf een dubbel zo grote afstand dan hun soortgenoten van het bos. Die verschillen tussen individuen van andere landschapstypes stemmen overeen met erfelijke verschillen in vliegactiviteit, dispersieneiging en vermogen om habitat te de-tecteren; ze wijzen op evolutionaire gevolgen van habitatfragmentatie. We bespreken een aantal nieuwe ideeën die uit de resultaten ontspruiten. Bijvoorbeeld de idee over de variatie tussen populaties van verschillende landschappen om het hoofd te bieden aan een grofkorrelige ecologische infrastructuur. Het kan nuttig zijn om bij andere, en dan vooral bedreigde soorten, na te gaan of er verschillende optima bestaan inzake de geometrie van corridor- en stapsteennetwerken voor eenzelfde soort in verschillende landschapstypes.

Aansluitend bij al die argumenten voor evolutionaire veranderingen in ther-moregulatie en verplaatsingsgedrag als gevolg van de structuur van het landschap be-merkten we dat de effecten vaak geslachtsspecifiek zijn. Die seksuele verschillen kunnen begrepen worden in het licht van het multifunctionele karakter van het vlieggedrag bij vlinders. De functie van vliegen verschilt tussen mannetjes en wijfjes, in overeenstem-ming met hun verschillende gedragsrepertoires en ecologische functies. Dit onderzoek naar de evolutionaire ecologie van verplaatsingen, gedrag, functionele morfologie en thermoregulatie in P. aegeria levert een betekenisvolle bijdrage tot de diverse ecologische en evolutionaire effecten die habitatfragmentatie kan hebben op ‘life history’ kenmerk-en van organismen in cultuurlandschappen.

Dankw

oord

{ 121 }

Dankwoord

Het doctoraat zit er bijna op. En dat vind ik erg spijtig! Weet je, het geeft veel voldoening aan natuurstudie te doen. Bovendien heeft evolutionaire biologie iets van ‘een open-baring’, die je toelaat een omvattend zicht te krijgen op de mechanismes achter geob-serveerde patronen. Een echte ‘waarheidservaring’ noem ik het, zonder opium, zonder geloof… Werken aan dit doctoraat was dan ook enorm deugddoend! In een ongedwongen en ontspannen werksfeer kon ik van het ene experiment in het andere duiken. En alle hadden ze iets te vertellen. Naast een al dan niet duidelijk antwoord op de geteste hy-potheses, riepen ze steeds een heleboel nieuwe vragen op. Aan onderzoeksvragen was er dus nooit een gebrek!

Ik hoop dat Professor Doctor Hans Van Dyck [Hans(ki) voor de vrienden] enigszins kan inschatten hoe dankbaar ik hem ben. Hans, je fijne persoonlijkheid en je weten-schappelijke honger werken enorm motiverend! Al vanaf m’n licentiaatsthesis (over vlindertjes natuurlijk) stond je paraat om me verder te helpen, en dat ben je blijven doen… Dank je wel Hans! Erik, ook jou wil ik bedanken; ik was welkom in je onder-zoeksgroep, je tekende mee de lijnen van m’n onderzoek uit, je zette me op weg voor de IWT-verdediging, er kon zelfs een thomasmobiel af (grijsgroene SuperRcinq – RIP) en je hield goedlachs een oogje in het zeil. De Onderzoeksgroep Dierenecologie en de aan-lendende groep De Morfo’s wil ik in de bloemetjes zetten voor de zeer aangename sfeer.

{ 122 }

Dankw

oord

{ 123 }

Die hing er gewoon altijd in de lucht en ondanks de geregelde wissels hangt die er nog altijd. Er viel altijd wel iets te bekokstoven, te beleven, te bediscussiëren, te grapjassen, te schransen, te beloeren, te uit bochten te wringen, te troosten, te ondervinden en te zwanzen met (in alphabetical order) Ann (en een lelijke gang werd mooi), Anne (de gedroomde labo-aanvulling), Anthony (zelfs Anthony kan plots ladderzat worden), Au-rora (está incréible la soltura de tu boca), Bieke (in vele dingen geïnteresseerd), Diederik (een van de mezenfreaks), Diederik-halsbandparkiet (met een‘twien je veilig de Brus-selse bossen kan afstruinen op zoek naar Amerikaanse bommen in de verborgen holen van vreemde holenbroeders), Els (’t was goed samenwerken), nog eens Erik (dé broer van Hugo Matthysens), Evie (ofte bipedaal badmintonnen), Frank (voor het zwaardere werk), Frans (trekt er nog steeds gezwind op uit), Geert (de eeuwige student), Goedele (de eeuwige doctoraatsstudente), nog eens Hans (zelfs z’n broer Tom nam er een voor-beeld aan), Hans Scheers (probeer eens de notebooks van IBM, die zijn steviger Hans), Jan (da mister fix-it man), Joeri (de getaande vliesvleugelige), John (a British globetrot-ter mad about caecilians, cae-ci-lians?, what? Yes indeed. Thanks for converting the summary into decent English), Joke (biologie is dan toch niet je van het), Joris (dat was kort), Katleen (punk is not dead, it just became a bit cute), Kirsten (de kers op Willem), Koen Coockie (jeanine), Koen Borghmans (droom je er nog steeds van erkend digitali-sateur te worden? Koen, vooral bedankt voor je energie bij het volgen van ontketende Bonte zandoogjes. Alhoewel je aan de haal ging met m’n thesisstudente, bleef je inzet on-voorwaardelijk!), Kris (statistische analyses kunnen heel lang duren), Kristiaan (voor droge humor: één adres), Liesbeth (ja, inderdaad, ja hoor), Luc (‘Gást!’, ‘de begeleiders wachten’, ‘de biologie is goe bezig’ en menig ander jolig woord), Luis (murcielagos y badminton, y basta), Maarten (hét voorbeeld dat je je zelfs kan overwerken door je in te laten met het vangen der gewoone beestschens), de Matheve (‘gij moet mij nog eens uwe gsm geven gij’), Mwangi (A discussion is a piece of art), Peter Galbusera (de rustige populatiegeneticus met Italiaanse roots), Peter Aerts (ondanks gegarandeerd verlies is het tof badmintonnen met Peter), Raoul (attractiviteit, dáár draait het om), Raphael (het rasechte instrumenten-, zang- en talenwonder, dat me zelfs zover kreeg gaita te spelen), Sofie (waer bestu bleven?), Stefan (statistisch zeer significant, maar verslaafd aan coke), Tom Van de Casteele (de perfectionist), Tom Van Dooren (fenotypisch koken met Tom, gezond voor je geest), Tine (een labo heeft er minstens één nodig), Vladimir (à la recherche d’un petit coupé sport), Wayaki (hm), Willem (de kakelverse papa), Wouter (niet zomaar een tuinkabouter, gewoon omdat het zo mooi rijmt wouter, wees maar blij dat ik niets anders schreef: hoogvlieger bv.) en Ximena (spectacled bears are sóóó cute). Ook voor de ethologen Denitza (Bulgarije dichtbij), Tinne (hartritme en zenuwbelasting: een experiment), Leen (doctoreren in recordtempo onder moeilijke omstandigheden: chapitô!) en Guy (de medestudent uit Leuven): een dikke dank jullie wel!

Voorts had ik ook ongelooflijk veel aan Dirk Maes (de vlinderman op het Instituut), aan de Zweedse collega’s Olof Leimar en Bengt Karlsson (thanks for the collaboration

and hospitality), aan Paule en Pascale voor de vlotte hulp op het secretariaat, aan Jorge (quedas viviendo en mis ropas) en aan de interessante ontmoetingen en steunbetui-gingen van vlinderaars allerhande (Bob Srygley, Camille Parmesan, Christer Wiklund, Christine Schneider, Dries Gorissen, Karin Enfjäl, Larissa Conradt, Lizzie Cant, Michel Baguette (je vous remercie de tout coeur de vouloir faire parti de mon jury), Mikko Kuussaari, de mensen van de Nederlandse Vlinderstichting, Nicolas Schtickzelle, Paul Brakefield, Paul Howe, Peter Waser, Philippe Goffart, Rinny Kooi, Ron Rutowski (a ‘big thank you’ for being so kind to take part in the examination committee), Ruth Walther, Sofie Van de Woestijne, Tim Shreeve, en nog vele andere die ik me niet onmiddellijk voor de geest haal.

Voor extra-professionele steun en plezier ben ik volgende (groepjes) mensen van harte dankbaar: Ana & Ben (uitzonderlijk veel dank voor jullie vriendschap en gast-vrijheid; Ana, weet je nog dat ik van jou de uitdrukking ‘un beso de mariposa’ leerde?), Raquel (muito obrigado para ser estado a mea borboleta), de ‘mannen van ‘t collègge’, de riskers van Leuven, de badmintonners voor de gezonde middagen, de vele danspartners en vrienden uit het folkmilieu, de ‘Galiciërs’ van A Contrabanda, Marieke (voor de lief-kozingen) en m’n familie, met speciale en heel emotionele woorden van dank voor papa, moeke en Johan.

De fotografen Jeroen Mentens en Rollin Verlinde prijs ik om mijn onmogelijke vra-gen gewoon als een uitdaging te zien. Sebastiaan Barbé bespaarde me heel wat lay-outgevloek. Voorts ook dank voor de vele steunbetuigingen van mensen die anomien wensten te blijven.

‘Nothing in biology makes sense, except in the light of evolution’ (T. Dobzhansky, 1973)

{ grafi c design : A

rban }

and evolutionary ecology of movement behaviour in the...

Documents