A field study of larval development in a dragonfly assemblage in African desert

ponds (Odonata)

Frank Suhling1,2, Kamilla Schenk1,2, Tanja Padeffke1 & Andreas Martens1,31Zoologisches Institut, Technische Universitat Braunschweig, Fasanenstraße 3, D-38102 Braunschweig, Germany2Present address: Institut fur Geookologie, Technische Universitat Braunschweig, Langer Kamp 19c,D-38106 Braunschweig, Germany (E-mail: f.suhling@tu-bs.de)3Present address: Abteilung Biologie, Padagogische Hochschule Karlsruhe, Bismarckstraße 10, D-76133 Karlsruhe,Germany (E-mail: martens@tu-bs.de)

Received 26 November 2002; in revised form 30 April 2004; accepted 23 June 2004

Key words: Odonata, growth rate, colonisation sequence, emergence, Namibia, temporary waters, migration

Abstract

Aquatic animals distributed along a ‘habitat-permanence’ gradient (HPG), differ in life history (Wellbornet al., 1996. Annual Revue of Ecology and Systematics 27: 337–363). Dragonflies that occur in hot aridregions often occur in temporary waters and consequently perform direct and rapid development. Drag-onfly species of the Namibian desert do differ in their selection of habitats along the HPG and thereforemay also differ in life cycle. Here, we attempt to monitor colonisation, larval growth and emergence in atemporary pond of known history. We studied the development of dragonfly species that laid eggs inartificial ponds constructed by us in March 2001. The assemblage consisted of species that originate fromdifferent habitats along the HPG. To obtain data on larval development we took samples from the ponds at10-day intervals. Most species showed rapid development. By regressing the maximum sizes attained bylarvae on each sampling date against time we estimated growth rates for five species and were thereby ableto estimate that total duration of development from oviposition to emergence ranged between 38 and70 days. Observation of first oviposition and first emergence for three of these species corroborated ourestimates based on growth rate. Of few species, which laid eggs in the ponds no larvae or adults were found.For some this may have been the result of predation whereas others may not have grown fast enough toemerge before the ponds dried up. Our results indicate that dragonflies cannot recognise whether a pondwill retain water long enough for full larval development and oviposit in waters that will not allow larvaldevelopment.

Introduction

Aquatic systems can be classified with respect tothe probability of drying. At one side of such acontinuum are permanent waters and at the otherare temporary waters (Williams, 1987; Wellbornet al., 1996; Stoks & McPeek, 2003). Whereas inpermanent waters distribution of taxa is con-strained by biotic factors such as predation andcompetition, in temporary waters it is mostlylimited by physical conditions, mainly drying

(Williams, 1987, 1996; Wellborn et al., 1996). Onemajor attribute of organisms of temporary watersis rapid development; organisms of permanentwaters often develop more slowly (Wellborn et al.,1996). Odonata occupy almost all kinds of habitatsalong the HPG ranging from permanent runningwaters and lakes to small temporary rain pools(Corbet, 1999). Comparative studies by Kumar(1976) in India revealed that the duration of larval

Hydrobiologia 528: 75–85, 2004.� 2004 Kluwer Academic Publishers. Printed in the Netherlands. 75

development correlates with the duration andseasonality of breeding habitats. Consequently,dragonflies show wide variation in life-cycleduration, ranging from multivoltine, having up tofour or five generations per year, to partivoltinespecies requiring three or even more years forcomplete one generation (Corbet, 1999, p. 218).

In tropical and subtropical arid regions manywater bodies contain water only for part of theyear and Odonata species which breed in suchtemporary waters exhibit rapid larval growth(Hodgkin & Watson, 1958). Corbet (1999, p. 220)placed the Odonata into categories according tolife-cycle features. Most widespread species ofAfrican arid regions (cf. Suhling et al., 2003) be-long to type A.2.2 of Corbet (1999, p. 220). Suchspecies are typically multivoltine and breed inephemeral ponds during the rainy season. Noma-dic adults may be carried by rain-bearing weather-fronts. A typical species of this type is the obligatecircum-tropical migrant (sensu Corbet, 1999)Pantala flavescens in which the duration of larvaldevelopment times can be as short as 43 days(Kumar, 1984). In other species with a similarecology this may even be 20 days (see review inCorbet, 1999, p. 630). A further common type,A.1., is facultatively multivoltine, its aquatic hab-itats being continuously available. Many speciesmay inhabit several habitat types, and may adopta life-cycle of either type A.2.2. or A.1., dependingon the type of habitat available. In both typesembryonic and larval development is brief, lastingless than 120 days. A few such species are uni-voltine, either having an egg diapause lasting forseveral months or siccatating as adults (Corbet’stype 2.1.2) (Suhling et al., 2003). There is no evi-dence to suggest that partivoltine species are to befound in desert regions (Suhling et al., 2003).

The aim of our study was to compare the larvaldevelopment of some widespread African drag-onfly species that occur in different habitats in theNamibian semi-desert. We assumed that all specieswould show rapid larval development. However,we predicted that migrants like P. flavescens, whichregularly colonise rain-fed pools, would havehigher growth rates than species that normallycolonise other types of habitats. Such differences ingrowth rate should have consequences for repro-ductive success in temporary waters depending onhabitat duration. We studied larval development

in the field by sampling larvae from artificialponds. To estimate growth rates from field datadifferent methods have been used (Benke, 1970;Benke & Benke, 1975; Krishnaraj & Pritchard,1995; Pritchard et al., 2000). In our case thesemethods did not yield useful data (see Methods);so we developed an alternative method. To obtainaccurate data regarding duration of larval devel-opment we recorded colonisation sequence, ovi-position times and emergence at our ponds.

Methods

Study area

The study was carried out at Tsaobis LeopardNature Park (22� 22¢ S, 15� 44¢ E; altitude740 m a.s.l.), 50 km S of Karibib, Namibia. Tsa-obis is a private nature reserve with an area of330 km2 in the semi-desert/savanna transitionzone (cf. Barnard, 1998) bordering the Namib-desert to the east. The climate is hot and arid mostof the year. The mean monthly maximum tem-peratures range from 29 �C during winter (July) to35 �C during summer (January), mean minimumtemperatures being 10 and 19 �C, respectively. Thetemperatures at Tsaobis are usually somewhathigher (Cowlishaw & Daviers, 1997). Mean annualrainfall, which mainly occurs during the rainyseason from January to the end of April, is 85 mm(Cowlishaw & Daviers, 1997; Barnard, 1998). Thearea is characterised by a saturation deficit of)3400 to )3800 mm per year (Barnard, 1998). Allstreams and springs in the park are ephemeral, butsome artificial waterholes provided with groundwater are permanent. Most of them are situatedclose to the ephemeral Swakop River, which bor-ders Tsaobis to the north. At Tsaobis, the riverbedof the Swakop is normally dry. During the rainyseason the river may run for between a few hoursto some days (cf. Jacobson et al., 1995; F. Suhling,own observ.). For a detailed description of thelandscape and vegetation at Tsaobis see Cowli-shaw & Daviers (1997).

Artificial ponds

As part of a study on macroinvertebrate commu-nity composition of temporary ponds (cf. Padeffke

76

& Suhling, 2003) we constituted eight ponds on 31March 2001 using black polythene sheet as liner.Each pond had an area of 3.0 · 2.5 m and wasfilled with groundwater from the Swakop River,obtained by pumping from about 30 m below theriverbed. Initially the depth of water was 0.25 mand evaporated to zero within 75 days. The waterhad an initial conductivity of 2.2 mS cm)1, whichincreased to of 9.1 mS cm)1 just before the pondsdried up. The ponds were situated about 50 mfrom the Swakop River and were surrounded byshrubs and trees, so each was shaded for about 1 hduring the day. Each pond was furnished with onetuft of rushes and one tuft of grass, and addi-tionally one tree-branch and one old tree root wasintroduced as perches for adult dragonflies andoviposition substrates for dragonflies with endo-phytic oviposition. The bottom was covered with a2-cm layer of sand taken from above the level ofthe riverbed of the Swakop. To monitor emer-gence, the perimeter of each pond was fenced withgauze (mesh width 1 mm, 30 cm high), whichserved as an emergence support.

Immediately after setup, the ponds were col-onised by a number of aquatic insects, amongthem Chironomidae, Culicidae (both Diptera) andCorixidae (Hemiptera). Their larvae occurred atdensities of >2000 per m2 in Chironomidae (esti-mated from sweep net samples) and probablyserved as main food source for the dragonfly lar-vae. Hence, we assume that prey did not limitgrowth of the dragonfly larvae in the ponds.Normally, in natural ephemeral ponds in southernAfrica several Crustaceans like cladocerans, co-pepods, ostracods, and fairy shrimps develop fromegg banks and these may also serve as food fordragonfly larvae (cf. Brendonck et al., 2002). Inour artificial ponds no such crustaceans developed,probably because the sediment consisted fromsand taken from a terrestrial site, where apparentlyno egg bank existed.

To obtain data on water temperature weintroduced an automatic data logger (Van-Essen-Instruments) to the bottom of one of the ponds tomonitor water temperatures at 1-h intervals (pre-cision of 0.1 �C). Water temperatures varied be-tween 37 and 12 �C. Minimum temperaturesdecreased as the season progressed. However, dueto high daytime air temperatures and decreasingwater level of the ponds, maximum the water

temperatures were always on a similar high level(Fig. 1).

Colonisation sequence of adult Odonata

To monitor colonisation by adult Odonata werecorded the presence of adults at each pond everyday. We also counted number of males and fe-males per pond during a 10-min observation per-iod at 12:00 h local time every day, which is11:03 h solar time. As opportunity permitted, werecorded the presence of oviposititing females overseveral hours per day. We watched the pondssufficiently frequently each day that we consider itunlikely that we could have overlooked the time offirst arrival of any species at the ponds.

Larval growth

To monitor growth of dragonfly larvae sampleswere taken at five 10-day intervals beginning on 14April 2001 using a hand-net (mesh size: 0.5 mm).Ten sweeps of the net of 30 cm in length and13 cm in width were taken from each pond and oneach occasion of sampling. The samples weretransferred to a sieve 35 cm in diameter (mesh size0.25 mm), and all dragonfly larvae seen were re-moved manually. All aquatic animals were sortedand identified to species if possible. Among thedragonfly larvae the only gomphid was easy toseparate by its flat labium compared to the spoon-shaped labium of the libellulids. The larvae ofPantala, Crocothemis, Sympetrum and Orthetrumhave distinct shape of the head, body-colour andpresence of dorsal and lateral spines. The two

10

15

20

25

30

35

40

31/3 4/4 8/4 12/4 16/4 20/4 24/4 28/4 2/5 6/5 10/5 14/5

Tem

pera

ture

[C]

Figure 1. Water temperatures measured at 1-h intervals from

31 March to 17 May 2001 in one of the artificial ponds using an

automatic data logger.

77

Trithemis spp. in the ponds could not be reliablydistinguished in the first few stadia. However, sincein the later stadia most Trithemis larvae belongedto T. kirbyi we will hereafter arbitrarily assign alllarvae only to T. kirbyi. We counted and size-measured all larvae under a dissecting microscopeand ocular micrometer to the nearest 0.05 mm. Wemeasured larval head width, which is a preferreddimension for specifying size of odonate larvae(Benke, 1970; Corbet, 2002). We constructed size-frequency histograms for each sampling occasionand determined larval stages according to themethod of Benke (1970) using the terminology ofCorbet (2002): F-0 ¼ final stadium, F-1 ¼ penul-timate stadium, etc.

One way of estimating rate of developmentfrom field data is to calculate the mean ordinalstadium number per sampling occasion (Benke,1970). This method is useful for comparing specieswith longer life-cycles, i.e. univoltine or partivol-tine species (cf. Corbet, 1999). In species with ashort larval duration and long reproductive peri-od, a pattern that is common in species fromtemporary waters (e.g. Fincke, 1992, 1994), thismethod will greatly underestimate the growth rateif oviposition and therefore recruitment of smalllarvae continues during the time that other larvaeare developing. Therefore we used a differentmethod, assuming that the largest larva per sam-pling indicates the maximum rate of growth.Accordingly, we plotted maximum head widthagainst time, namely against days after the estab-

lishment of the ponds. Development curves werefitted using linear regression. The replicates are theoccasions at which larvae of the respective speciesoccurred in the samples, which was four times inOrthetrum chrysostigma, T. kirbyi and Paragom-phus genei, and three times in Crocothemis eryth-raea and Sympetrum fonscolombii. In Pantalaflavescens, though represented at all sample occa-sions, the number of replicates was three sinceemergence of this species began already after thethird sample occasion.

Emergence

We monitored emergence daily at all ponds byexamining all possible supports until the pondsbecame dry. All exuviae found were removeddaily. We assume that no exuvia was overlookedsince the ponds were small and each pond wasfenced with gauze.

Results

Colonisation sequence

The colonisation of the ponds by adult dragonflieswas rapid (Table 1). Already by the evening of 31March 2001, when we started inundation, the firstegg laying Trithemis kirbyi was observed. By 1April at least one adult of most other species hadarrived, although Sympetrum fonscolombii and

Table 1. Colonisation sequence of Odonata at artificial ponds in the Namibian semi-desert

Species First arrival First Maximum numbers

oviposition Males Females

Trithemis kirbyi 31 March 31 March 31 4

Pantala flavescens 1 April 1 April 4 3

Anax imperator 1 April 1 April 1 1

Trithemis annulata 1 April 1 April 12 1

Crocothemis erythraea 1 April 12 April 3 1

Paragomphus genei 1 April 14 April 6 2

Orthetrum chrysostigma 1 April 15 April 7 1

Sympetrum fonscolombii 8 April 16 April 19 3

Africallagma glaucum 13 April 23 April 5 4

Dates of first observation of individuals of the species and of first observation of oviposition in 2001 are given. The maximum numbers

refer to the pooled numbers of individuals from all eight ponds recorded during one single record (see Methods).

78

particularly the zygopteran Africallagma glaucumappeared later (Table 1). Whereas in some speciesoviposition was recorded on the day of first arri-val, in other species oviposition was first witnessedmuch later (Table 1). From the day of their firstarrival most species were permanently present atthe ponds, with the exception of Anax imperator.Also, oviposition occurred almost every day. InPantala flavescens, however, oviposition was re-corded only during a brief period between 1 Marchand 5 March, and then again on 11 March.

Larval development and growth rates

On the first sampling date, 14 April 2001, only afew larvae of P. flavescens were caught, althoughthese had already approximated stadium F-5(Fig. 2). On 4 May, 34 days after establishment ofthe ponds, the first F-0 larvae P. flavescens werefound. From 24 April to 11 May two distinct size-cohorts of this species were present, probably de-rived from the two oviposition-occasions describedabove. Larvae of Orthetrum chrysostigma, T. kir-byi and Paragomphus genei were detected first on24 April (Fig. 3a–c). The two former taxa hadreached F-2 by 4 May and F-0 was first encoun-tered on 22 May. F-2 P. genei were first collectedon 12 May and F-0 were not encountered(Fig. 3c). In all three species small larvae domi-nated in every sample, presumably reflectinguninterrupted oviposition. Crocothemis erythraeaand Sympetrum fonscolombii were not encounteredbefore 4 May (Fig. 4a and b). Although larvae ofthe former were collected in low numbers, deve-

lopment could be followed and resembled that ofO. chrysostigma and T. kirbyi (Fig. 4a). In con-trast, in S. fonscolombii, which reached com-paratively high numbers, almost no change in thesize-frequency curve could be discerned (Fig. 4b).

Linear regression of maximum size against timerevealed high coefficients (R2 > 0.95 in all cases),which were significant for all species, except forP. flavescens, in which, however, there was a trend(Fig. 5). Hence, linear curves fitted well withmaximum size of larva against time in all species,and the slope of the curves can therefore be used toestimate the daily maximum growth rates. Theestimated growth rate was highest in P. flavescens,followed by O. chrysostigma, C. erythraea, P. geneiand T. kirbyi (Fig. 5). For S. fonscolombii noestimate was possible.

0 5 10 15 200

1

2

3

4

5

6

7

Hea

dw

idth

[mm

]

14 April 01N=2

0 5 10 15 20

24 April 01N=31

0 5 10 15 20

04 May 01N=39

0 5 10 15 20

12 May 01N=32

0 5 10 15 20

22 May 01N=12

100%

F-0 F-0 F-0

F-1 F-1 F-1 F-1

F-2 F-2F-2

Percent of total number of larvae

Figure 2. Larval head width frequency diagram for Pantala

flavescens derived from total numbers of larvae sampled from

eight artificial ponds (see Methods).

Hea

dw

idth

[mm

]H

ead

wid

th[m

m]

N=1715 N=1300 N=904

0

1

2

3

4

5

6 N=176

F-1

F-2

F-0

F-1

F-2F-2

0 10 20 30

Percent of total number of larvae0 10 20 30 0 10 20 30

0

1

2

3

4

5

6

Hea

dw

idth

[mm

]

N=12 N=112 N=77 N=97

F-1

F-2F-2

0

1

2

3

4

5

N=121 N=99

F-0

N=16

F-1F-1

N=162

F-2 F-2 F-2

6

0 5 10 15 20 25 30 35 40

24 April 01 04 May 01 12 May 01 22 May 01(a)

(b)

(c)

Figure 3. Larval head width frequency diagram for (a) Orthe-

trum chrysostigma, (b) Trithemis kirbyi, and (c) Paragomphus

genei derived from total numbers of larvae sampled from eight

artificial ponds (see Methods).

79

Emergence

Considering the high number of larvae sampled,the emergence from the ponds was low. Totalnumber of exuviae was 81 in P. flavescens, with thefirst emergence on 12 May. Of O. chrysostigma andC. erythraea only one specimen emerged on 4 Juneand 8 June, respectively.

Discussion

Seven species of dragonflies developed in ourartificial ponds. Our larval samples indicate that atleast five of these undergo rapid development, andthat most attained the final larval stadium. Onlythree species emerged before the ponds fell dry.Larval growth rate differed between species andrates ranged from 0.252 mm (head width) per dayin Pantala flavescens to 0.094 mm per day inTrithemis kirbyi (see Fig. 5).

We assumed linear growth rates, an assumptionsupported by regression analyses in studies onother species (e.g. Krishnaraj & Pritchard, 1995).Hence, the results from our growth-rate analyses

allow some more rough estimates. We assumedhead width of the adult to represent the maximumsize to be reached by an individual. Using thelinear growth rate, we can divide adult size by thedaily growth rate to get an estimate for the numberof days needed to complete development fromhatching to emergence. Using the rates shown inFigure 5 we estimated the duration of larvaldevelopment at 33 days for P. flavescens, 45 daysfor Orthetrum chrysostigma, 56 days for Croco-themis erythraea, 58 days for T. kirbyi and 60 daysfor Paragomphus genei, respectively. Taking theduration of embryonic development into account,which is on average 5 days for P. flavescens, 6 daysfor O. chrysostigma and C. erythraea, 8 days forT. kirbyi and 10 days for P. genei (own unpub-lished data), the total duration of developmentshould be somewhat longer (see values in Table 2).These estimates are corroborated by our fieldobservations. In P. flavescens the time betweenfirst oviposition and first emergence was 42 days,which is only 4 days more than the estimate. ForO. chrysostigma and C. erythraea, respectively, wearrived at 50 and 57 days, 1 and 5 days less thanthe estimates. These underestimates may be ex-plained by the fact that both species emerged justbefore the ponds became dry and larvae may in-crease growth under time constraints (cf. Johans-son & Rowe, 1999).

Corbet (1999, p. 220) distinguished differentmodes of development in Odonata. Direct andrapid development (‘unregulated development’)lacking diapause in the larval or egg stage isprobably the most common mode of developmentof arid zone Odonata (Suhling et al., 2003). Thedevelopment times derived from our study aresimilar to those found for other desert and sa-vanna species (cf. Hodgkin & Watson, 1958). Ta-ble 2 gives an overview for such species that occurin Africa. Two of the species studied by us havebeen studied several times so that a comparison ispossible. For P. flavescens all studies reveal a brieflarval period ranging from 38 to 65 days, even atthe southern limit of its range in Australia(Hawking & Ingram, 1994). In C. erythraea alsomost studies show rapid development with theexception of that by Wenger (1955), who rearedlarvae from a French population over winter in aheated room, but without ‘optimal temperatures’and with limited food. These conditions may have

Hea

dw

idth

[mm

]H

ead

wid

th[m

m]

0

1

2

3

4

5

6 N=7 N=9 N=10

F-1

F-2

F-0

F-1

F-2

04 May 01 12 May 01 22 May 01

0 10 20 30 0 10 20 30 0 10 20 300

1

2

3

4

5

6 N=23 N=48 N=34

F-2

Percent of total number of larvae

(a)

(b)

Figure 4. Larval head width frequency diagram for (a) Croc-

othemis erythraea and (b) Sympetrum fonscolombii derived from

total number of larvae sampled from eight artificial ponds (see

Methods).

80

prolonged the larval stage. The comparatively longduration of development found for S. fonscolombii(Gardner, 1951) may have been prolonged simi-larly. Anax imperator needs 1–2 years for devel-opment in Britain (Corbet, 1957) and 1 year inSwitzerland (Robert, 1958), but may develop fas-ter in Namibia, though we do not have resultsfrom our study (see below). Interestingly, our dataindicate rapid development in the gomphid P. ge-nei. Most species of this family are riverine andprevious studies have indicated that even tropicalmembers of the family need at least about 1 yearto complete larval development (Dudgeon, 1989;Suhling & Muller, 1996). However, P. genei isunivoltine in Spain (Testard, 1975) and it probablyhas more than one generation per year in Tunisia(Jodicke, 2001).

In dragonfly species that undergo direct devel-opment temperature and availability and qualityof food are the main factors affecting the durationof development (Hassan, 1976; Lawton et al.,1980; Krishnaraj & Pritchard, 1995). Growth in-creases with temperature to a species-specific limit(Krishnaraj & Pritchard, 1995). In our ponds allspecies developed under the same conditions oftemperature and food. However, growth ratesdiffered between species. High growth rate is oftencorrelated with high activity and rate of preycapture (e.g. Maurer & Sih, 1996), although con-version efficiency of food ingested may also con-tribute to growth rate (McPeek et al., 2001).Activity has been suggested as one of the mostimportant traits the changes in species’ attributesalong the gradient from temporary to permanent

012345678

Y = -0.977 + 0.101XR2 =0.995, P = 0.046

Crocothemiserythraea

012345678

Y = -2.377 + 0.139XR2= 0.995, P < 0.001

Orthetrumchrysostigma

0 20 40 60 80 100

12345678

Y = -1.597 + 0.252XR2 = 0.982, P = 0.087

Pantalaflavescens

Hea

dw

idth

[mm

]

0 20 40 60 80Successive days

9

100

0

Successive days

Y = -1.688 + 0.099XR2 = 0.988, P = 0.006

Paragomphusgenei

Y = -0.659 + 0.094XR2 = 0.997, P = 0.002

Trithemiskirbyi

Figure 5. Regression plots for maximum larval size (head width) per sampling occasion against time elapsed (days) since the first day

of existence of the artificial ponds for five species. The solid dots and lines are the regression plot. The slopes of the curve formulas

derived from these curves represent the growth rates of the species. Dotted lines are projections from the regression line with the size of

the adult and the second-stadium larva, which in all species is the first free-living stadium after the pro larva, as end points (open

circles). These lines represent a rough estimate for the total duration of larval development.

81

Table

2.Minim

um

durationofdevelopmentfrom

ovipositionto

emergence

ofdragonfliesthatoccurin

Africa,derived

from

laboratory

(a)andfieldstudies(b),e.g.by

estimatingdurationbymeasuringtimebetweenfloodingofnaturalpondsandfirstem

ergence

oftherespectivespecies

Species

Method

Min.duration(days)

Country

Reference

Aeshnidae

Anaxephippiger

(Burm

eister)

a60

Africa

Stortenbeker

(1967)

A.ephippiger

b76

Greece

Schnapauffet

al.(2000)

A.ephippiger

ba

100ca.

Nigeria

Gambles(1960)

AnaxtristisHagen

ba

100ca.

Nigeria

Gambles(1960)

Gynacanthavesiculata

Karsch

ba

70ca.

Nigera

Gambles(1960)

Gomphidae

Paragomphusgenei

(Selys)

b70

Namibia

Thisstudy

Libellulidae

Acisomapanorpoides

Rambur

a77

Nigeria

Hassan(1976)

A.panorpoides

97

Pakistan

Chowdury

&Jashim

uddin

(1994)

A.panorpoides

b109

Nigeria

Hassan(1976)

Crocothem

iserythraea

(Brulle)

b59

Namibia

Thisstudy

C.erythraea

?60

Egypt

ElAmin

&ElRajah(1981)(inCorbet,1999,p.630)

C.erythraea

b76

Greece

Schnapauffet

al.(2000)

C.erythraea

a285

France

Wenger

(1955)

Orthetrum

chrysostigma(Burm

eister)

b49

Namibia

Thisstudy

Orthetrum

julia(Longfield)

ab

51

Nigeria

Gardner

(1956)

Palpopleura

lucia(D

rury)

a34

Nigeria

Hassan(1975)

P.lucia

a51

Nigeria

Gardner

(1956)

P.lucia

b58

Nigeria

Hassan(1976)

Pantala

flavescens(Fabricius)

b38

Namibia

Thisstudy

P.flavescens

a43

India

Kumar(1984)

P.flavescens

b51

Australia

Hawking&

Ingram

(1994)

P.flavescens

a65

Hawaii

Warren

(1915)

Sympetrum

fonscolombii(Selys)

c66

Greece

Schnapauffet

al.(2000)

S.fonscolombii

a217

Malta

Gardner

(1951)

Trithem

isannulata

(Beauvois)

?52

Egypt

ElRayah&

ElDin

AbuShama(1978)

Trithem

iskirbyi(Selys)

b61

Namibia

Thisstudy

Urothem

isassignata

(Selys)

a124

Nigeria

Hassan(1976)

U.assignata

?125

Ivory

Coast

Forge(1981)

U.assignata

b152

Nigeria

Hassan(1976)

SomestudieswerecarriedoutonpopulationsthatoccuroutsideAfrica.aGambles(1960)expressed

hisresultsasnumber

ofmonthsplusextradays.

bTim

eto

finalstadium

only.?=

Noinform

ationavailable.

82

waters because activity is associated with energyintake (Skelly, 1995; Maurer & Sih, 1996; Well-born et al., 1996; Wissinger et al., 1999). Althoughmost species we studied obviously develop intemporary ponds, they can nevertheless be placedalong an HPG. In Namibia P. flavescens and S.fonscolombii are typical rain-pool breeders,whereas O. chrysostigma and C. erythraea mainlyoccur in longer-lived, though temporary, waters.Trithemis kirbyi and P. genei are primarily river-ine, but also breed in temporary pools or ponds inriverbeds (own unpublished data). It is consistentwith these considerations that activity, capturerate, and growth rate of early stadia of P. flaves-cens has been shown to be higher than therespective values for C. erythraea and T. kirbyi(Johansson & Suhling, 2004). The growth ratesderived from our study agree with that relation-ship. Because our ponds became dry after 75 days,a consequence of the growth rates is that manyspecimens of the rain pool breeder P. flavescensemerged successfully from our ponds, whereasonly one adult emerged of O. chrysostigma andC. erythraea, respectively. Trithemis kirbyi andP. genei have lower growth rates and did notemerge at all. However, had our ponds retainedwater for a few more days or weeks, all species inthem might have produced adults.

One may expect that larvae survive droughtburied in wet sand and continue development whenthe ponds refill in the next rainy season. In Odo-nata several such cases of siccatation are reported(see review in Corbet, 1999, p. 190). Interestingly,most such observations are from temperate zones,not deserts, as one would expect. Reports of livinglarvae of Trithemis arteriosa (Burmeister, 1839)and O. chrysostigma in dry mud of seasonal poolsin the Sahara (Dumont, 1982), do not constituteevidence that these species are able to survive aslarvae until the end of the dry season (Corbet, 1999,p. 191). Our study does not contribute to thisquestion because the plastic lining of our pondsprevented digging. However, at least in Africandeserts siccatation of dragonflies in the larval stageseems to be unusual (Suhling et al., 2003).

In two species that laid eggs in our ponds, Anaximperator and Africallagma glaucum, no larvaewere recorded. Both species are endophytic egglayers, and the oviposition substrate (rushes, oldtree roots; see Methods) offered may have been not

adequate, although it was used for egg laying.Both species may also have suffered from intra-guild predation, which is habitual in dragonflypopulations (Wissinger, 1992; Fincke, 1999; Suh-ling & Lepkojus, 2001, Padeffke & Suhling, 2003).This was probably also the case in S. fonscolombii,for which we collected small larvae but no largerlarvae. Sympetrum fonscolombii and A. glaucumcolonised the ponds late and due to this they mayhave particularly suffered from predation by largerlarvae of earlier colonisers (cf. Padeffke & Suhling,2003). Another common temporary pond dwellingdragonfly in Africa, Anax ephippiger, did not layeggs in our ponds, although we encountered fewadults (Martens et al., 2003). The set up of ourponds was perhaps too late in the season for A.ephippiger. In 2002, we observed A. ephippiger inthe study area only in February (F. Suhling,unpublished data).

We predict that Dragonfly species that occur inregions where permanent waters are rare will beopportunistic when selecting oviposition sites.Ovipositing dragonflies cannot determine whetherthe water in a pond will persist long enough toallow larval development. For species that developslowly, ephemeral ponds will be traps (or ‘deadends’), and will fail to produce offspring or willproduce very few. Our study indicates that thissituation may occur frequently.

Acknowledgements

We thank Tana and Andre Burger of TsoabisLeopard Nature Park for pleasant hospitality andthe Namibian Ministry of Environment and Tour-ism for a research permit (423/2001). Philip S.Corbet and Carsten Schutte kindly corrected draftversions of the manuscript and provided valuablecomments. The study was supported by GermanMinistry of Science (BMBF 01LC0024) and is partof BIOTA (Biodiversity Transect Analysis in Afri-ca).

References

Barnard, P. (ed.), 1998. Biological Diversity in Namibia – A

Country Study. Namibian Biodiversity Task Force, Wind-

hoek, 332 pp.

83

Benke, A. C., 1970. A method for comparing individual growth

rates of aquatic insects with special reference to the Odonata.

Ecology 51: 328–331.

Benke, A. C. & S. S. Benke, 1975. Comparative dynamics and

the life histories of coexisting dragonfly populations. Ecol-

ogy 56: 302–317.

Brendonck, L., E. Michels, L. De Meester & B. Riddoch, 2002.

Temporary pools are not ‘enemy free’. Hydrobiologia 456:

147–159.

Chowdury, S. H. & M. Jashimuddin, 1994. Morphometric

studies on the larva of Acisoma panorpoides panorpoides

(Rambur). In: Srivastava, V. K. (ed.), Advances in Oriental

Odonatology. Cherry Publications, Allahabad: 69–76.

Corbet, P. S., 1957. The life-history of the Emperor Dragonfly

Anax imperator Leach (Odonata: Aeshnidae). Journal of

Animal Ecology 26: 1–69.

Corbet, P. S., 1999. Dragonflies. Behaviour and ecology of

Odonata. Harley Books, Colchester, 829 pp.

Corbet, P. S., 2002. Stadia and growth ratios of Odonata: a

review. International Journal of Odonatology 5: 45–73.

Cowlishaw, G. & J. G. Daviers, 1997. Flora of the Pro-

Namib Desert Swakop River catchment, Namibia:

community classification and implications for desert vege-

tation sampling. Journal of Arid Environments 36: 271–

290.

Dudgeon, D., 1989. Gomphid (Odonata: Anisoptera) life cycles

and production in a Hong Kong forest stream. Archiv fur

Hydrobiologie 114: 531–536.

Dumont, H. J., 1982. Relict distribution patterns of aquatic

animals: another tool in evaluating late pleistocene climate

changes in the Sahara and Sahel. Palaeoecology of Africa

and Surrounding Islands 14: 1–24.

El Rayah, E. A. & F. T. El Din Abu Shama, 1978. Notes on

morphology and bionomy of the dragonfly, Trithemis an-

nulata scortesii Nielsen (Odonata: Anisoptera), as a predator

on mosquito larvae. Zeitschrift fur angewandte Entomologie

85: 81–86.

Fincke, O. M., 1992. Behavioural ecology of the giant dam-

selflies of Barro Colorado Island, Panama (Odonata. Zy-

goptera: Pseudostigmatidae). In Quintero, D. & A. Aiello

(eds), Insects of Panama and Mesoamerica: Selected Studies.

Oxford University Press, Oxford: 449–462.

Fincke, O. M., 1994. Population regulation of a tropical dam-

selfly in the larval stage by food limitation, cannibalism,

intraguild predation and habitat drying. Oecologia 100: 118–

127.

Fincke, O. M., 1999. Organization of predator assemblages in

Neotropical tree holes: effects of biotic factors and priority.

Ecological Entomology 24: 12–23.

Forge, P., 1981. Development et rendement de croissance des

larves d’Urothemis assignata Selys (Odonata: Libellulidae)

das la region de Lamto (Cote-d’Ivoire). Acta Oecologia

Generalis 2: 213–226.

Gambles, R. M., 1960. Seasonal distribution and longevity in

Nigerian dragonflies. Journal of the West African Scientific

Association 6: 18–26.

Gardner, A. E., 1951. The life-history of Sympetrum fonsco-

lombii Selys. Odonata – Libellulidae. Entomologist’s Gazette

2: 56–66.

Gardner, A. F., 1956. The biology of dragonflies. Proceedings

and Transactions of the South London Entomological and

Natural History Society 1554–1955: 109–134 + pls I–VII.

Hassan, A. T., 1975. Studies on the larval development of

Palpopleura lucia lucia, Acisoma panorpoides inflatum and

Urothemis assignata (Anisopetra: Libellulidae) in a semi-

natural environment. Nigererian Journal of Entomology 1:

143–146.

Hassan, A. T., 1976. The effects of food on the larval devel-

opment of Palpopleura lucia lucia (Drury) (Anisoptera: Li-

bellulidae). Odonatologica 5: 27–33.

Hawking, J. H. & B. A. Ingram, 1994. Rate of larval devel-

opment of Pantala flavescens (Fabricius) at its southern limit

of range in Australia (Anisoptera: Libellulidae). Odonato-

logica 23: 63–68.

Hodgkin, E. P. & J. A. L. Watson, 1958. Breeding of dragon-

flies in temporary waters. Nature 181: 1015–1016.

Jacobson, P. J., K. M. Jacobson & M. K. Seely, 1995.

Ephemeral rivers and their catchments. Desert Research

Foundation of Namibia, 160 pp.

Jodicke, R., 2001. Saisonale Einnischung von Paragomphus ge-

nei in Tunesien (Odonata: Gomphidae). Libellula 20: 13–22.

Johansson, F. & L. Rowe, 1999. Life history and behavioural

response to time constraints in a damselfly. Ecology 80:

1242–1252.

Johansson, F. & F. Suhling, 2004. Behaviour and growth of

dragonfly larvae along a permanent to temporary water

habitat gradient Ecological Entomology 29: 196–202.

Krishnaraj, R. & G. Pritchard, 1995. The influence of larval

size, temperature, and components of the functional re-

sponse to prey density on growth rates of the dragonflies

Lestes disjunctus and Coenagrion resolutum (Insecta: Odo-

nata). Canadian Journal of Zoology 73: 1672–1680.

Kumar, A., 1976. Biology of Indian dragonflies with special

reference to seasonal regulation and larval development.

Bulletin of Entomology, New Dehli 3: 37–47.

Kumar, A., 1984. On the life history of Pantala flavescens

(Fabricius) (Libellulidae: Odonata). Annals of Entomology

2: 43–50.

Lawton, J. H., B. A. Thompson & D. J. Thompson, 1980. The

effects of prey density on survival and growth of damselfly

larvae. Ecological Entomology 5: 39–51.

Martens, A., R. Jodicke & F. Suhling, 2003. An annotated

checklist of the Odonata of Namibia. Cimbebasia 18: 139–

160.

Maurer, E. F. & A. Sih, 1996. Ephemeral habitats and variation

in behavior and life history: comparisons of sibling sala-

mander species. Oikos 76: 337–349.

McPeek, M. A., M. Grace & J. M. L. Richardson, 2001.

Physiological and behavioural responses to predators shape

the growth/predation risk trade-off in damselflies. Ecology

82: 1535–1545.

Padeffke, T. & F. Suhling, 2003. Temporal priority and intra-

guild predation in temporary waters: an experimental study

using Namibian desert dragonflies. Ecological Entomology

28: 340–347.

Pritchard, G., L. D. Harder, A. Kortello & R. Krishnaraj, 2000.

The response of larval growth rate to temperature in three

species of Coenagrionid dragonflies with some comments on

84

Lestes disjunctus (Odonata: Coenagrionidae). International

Journal of Odonatology 3: 105–110.

Robert, P.-A., 1958. Les Libellules (Odonata). Delachaux et

Niestle, Neuchatel, 364 pp.

Schnapauff, I., K. Ullmann & F. Suhling, 2000. Die Libellen-

Lebensgemeinschaft griechischer Reisfelder: Auswirkungen

von Habitatdauer, Anbaumethode und Vegetationsdichte.

Libellula Supplement 3: 63–80.

Skelly, D. K., 1995. A behavioural trade-off and its conse-

quences for the distribution of Pseudacris treefrog tadpoles.

Ecology 76: 150–164.

Stoks, R. & M. A. McPeek, 2003. Predators and life histories

shape Lestes damselfly assemblages along a freshwater

habitat gradient. Ecology 84: 1576–1587.

Stortenbeker, C. W., 1967. Observations on the population

dynamics of the red locust, Nomadacris septemfasciata

(Serville), in its outbreak areas. Agricological Research Re-

port, Wageningen 694, 118 pp.

Suhling, F. & S. Lepkojus, 2001. Different growth and behav-

iour influences asymmetric predation among early instar

dragonfly larvae. Canadian Journal of Zoology 79: 854–860.

Suhling, F. & O. Muller, 1996. Die Flußjungfern Europas

(Gomphidae). Die Neue Brehm-Bucherei, Westarp, Magde-

burg, 237 pp.

Suhling, F., R. Jodicke & W. Schneider, 2003. Odonata of

African arid regions – are there desert species? Cimbebasia

18: 207–224.

Testard, R., 1975. Note sur l’emergence, le sex ratio et l’activite

des adultes de Mesogomphus genei Selys, dans le sud de

l’Espagne (Anisoptera: Gomphidae). Odonatologica 4: 11–

26.

Warren, A., 1915. A study on the food habits of Hawai-

ian dragonflies. Bulletin of the Collection of Hawaii 3: 4–

45.

Wellborn, G. A., D. K. Skelly & E. E. Werner, 1996. Mecha-

nisms creating community structure across a freshwater

habitat gradient. Annual Revue of Ecology and Systematics

27: 337–363.

Wenger, O.-P., 1955. Uber die Entwicklung von Crocothemis

erythrea Brulle (Odonata – Libellulidae). Bulletin de la So-

ciete entomologique Suisse 28: 280–281.

Williams, D. D., 1987. The Ecology of Temporary Waters.

Croom Helm, London & Sydney, 193 pp.

Williams, D. D., 1996. Environmental constraints in temporary

fresh waters and their consequences for the insect fauna.

Journal of the North American benthological Society 15:

634–650.

Wissinger, S. A., 1992. Niche overlap and the potential for

competition and intraguild predation between size-struc-

tured populations. Ecology 73: 1431–1444.

Wissinger, S. A., H. H. Whiteman, G. B. Sparks, G. L. Rouse &

W. S. Brown, 1999. Foraging trade-offs along a predator-

permanence gradient in subalpine wetlands. Ecology 80:

2102–2116.

85