B

Ai

MJ

a

b

Tc

d

We

R

A

frKldwoawntebalop

Z

D

h1

ARTICLE IN PRESSAAE-50852; No. of Pages 9

Basic and Applied Ecology xxx (2015) xxx–xxx

Kenyan endemic bird species Turdoides hindei at home innvasive thickets

ike Teuchera, Christina Fischerb, Christin Buschc, Michelle Hornd, Jana Igld,anika Kernerd, Anke Müllerd, Ronald K. Mulwae, Jan Christian Habeld,∗

Department of Cartography, Trier University, D-54286 Trier, GermanyRestoration Ecology, Department of Ecology and Ecosystem Management, School of Life Sciences Weihenstephan,echnische Universität München, D-85354 Freising, GermanyInstitute of Biology, Freie Universität Berlin, D-14195 Berlin, GermanyTerrestrial Ecology Research Group, Department of Ecology and Ecosystem Management, School of Life Scienceseihenstephan, Technische Universität München, D-85354 Freising, Germany

Department of Ornithology, National Museums of Kenya, Nairobi, Kenya

eceived 19 May 2014; received in revised form 5 January 2015; accepted 6 January 2015

bstract

Thickets along rivers in Eastern Kenya are important habitats for many endangered species. These habitats also provideundamental ecosystem services for humans. Intense anthropogenic activities during the past decades have caused a severeeduction of this vegetation and resulted in fragmentation of the remaining thicket patches. We assessed the occurrence of theenyan endemic bird species Hinde’s Babbler (Turdoides hindei) in a highly fragmented environment and performed detailed

and use mapping along the Nzeeu River in East Kenya. We measured the time the birds spent in thicket patches, whichiffered in their habitat setting: pristine versus surrogate vegetation, different habitat size and different edge-size-ratio. Further,e identified areas of potential conflicts between human activities and our target species. Four T. hindei family groups werebserved, mostly in invasive Lantana camara patches. Habitat size and edge-size-ratio of the respective thicket patches revealed

significant impact on the duration of stay of T. hindei with disproportional longer stays in small habitat patches and in patchesith larger edge-size-ratio than in rather large patches or thickets with small edge-size-ratio. The 75%- and 95%-kernels showedo overlap between family groups and only marginal overlap with the 75%-kernels of human disturbances. Our data show thathe invasive L. camara thickets (even small patches with high edge-size ratio) are a suitable surrogate habitat for the Kenyanndemic for T. hindei. The birds avoid open land likely because of higher predation pressure outside of thickets. Limited overlapetween zones of human activity and the occurrence of T. hindei may be a response either to lacking thickets in these areas,

nd/or an adaptation to elevated hunting pressure in these zones. Therefore, the transformation of thickets into open agricultural

For the preservation of the remaining T. hindei family groups in

and has a negative impact on the persistence of T. hindei.

Please cite this article in press as: Teucher, M., et al. A Kenyan endemic bird species Turdoides hindei at home in invasive thickets. Basicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.002

ur study area we suggest to establish an interconnected network of thicket patches, as the high mobility of the species allowsersistence in such patchy environments.

usammenfassung

Gebüschstrukturen entlang von Flussläufen in Ostkenia bilden wichtige Lebensräume für viele gefährdete Tierarten.iese Ökosysteme stellen außerdem wichtige Ökosystemdienstleistungen für die Menschen bereit, die sich entlang der

∗Corresponding author. Tel.: +49 0 8161 71 4861; fax: +49 0 8161 71 4427.E-mail address: Janchristianhabel@gmx.de (J.C. Habel).

ttp://dx.doi.org/10.1016/j.baae.2015.01.002439-1791/© 2015 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.

B

2

FsvuwCuVaiudSaggVwud©

K

I

t(tasSq(a

dft2sfmG2dl

Tgaei

ARTICLE IN PRESSAAE-50852; No. of Pages 9

M. Teucher et al. / Basic and Applied Ecology xxx (2015) xxx–xxx

lussläufe angesiedelt haben, bereit. Die intensive anthropogene Nutzung während der letzten Jahrzehnte führte zu einemtarken Rückgang dieser Gebüschstrukturen und damit verbunden zu einer Fragmentierung der verbliebenen Vegetation. In derorliegenden Untersuchung haben wir das Vorkommen der für Kenia endemischen Drosselart Turdoides hindei aufgenommennd eine detaillierte Landnutzungskartierung entlang des Nzeeu Flusses in Ostkenia, unserem Studiengebiet, durchgeführt. Dabeiurde die Verweildauer der Individuen in den Gebüschresten gemessen. Für die Gebüschreste haben wir weitere, detaillierteharakteristika aufgenommen: ursprüngliche und invasive Vegetation, Gebüschgröße, und das Verhältnis zwischen Habitatrandnd Habitatgröße. Außerdem wurden potentielle Konflikträume identifiziert, in denen sich menschliche Nutzung und dasorkommen dieser Vogelart überlagern. Die vier beobachteten Familiengruppen hielten sich überwiegend in dichtem Gebüschuf, das hauptsächlich aus der invasiven Pflanzenart Lantana camara bestand. Die Individuen hielten sich überproportional längern kleinen Habitatresten und in Gebüschen mit einem großen Habitatrand-Habitatgröße-Verhältnisse auf. Die berechneten 75%-nd 95%-Kernels zeigten keine Überlappung zwischen den vier Familiengruppen und nur eine geringe Überschneidung miten 75%-Störungskernels menschlicher Aktivitäten. Unsere Daten zeigen dass die invasive Pflanzenart L. camara ein wichtigesekundärhabitat für T. hindei darstellt, selbst wenn die Gebüsche klein sind und ein großes Habitatrand-Habitatgröße-Verhältnisufweisen. Die Vogelart meidet offene Flächen, da hier von einem erhöhten Prädationsdruck ausgegangen werden kann. Dieeringe Überlappung von Bereichen menschlicher Aktivität und dem Vorkommen von T. hindei könnte entweder auf daseringere Vorhandensein von Gebüschstrukturen in diesen Bereichen zurückzuführen sein, oder eine Reaktion auf die intensiveogeljagd in diesen Gebieten sein. Die Rodung von Gebüschen und deren Umwandlung in landwirtschaftlich genutzte Flächeirken sich somit negativ auf die Persistenz von T. hindei aus. Um das Überleben der verbliebenen T. hindei Familiengruppen innserem Studiengebiet sicherzustellen schlagen wir ein Netzwerk von Gebüschstrukturen vor, welche durch die hohe Mobilitätieser Vogelart ein Überleben, selbst in einer lückenhaften Umwelt ermöglicht.

2015 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.

e; Hind

so2Otu2a

aptahzb

(

M

S

eywords: Edge effects; Fragmentation; Habitat quality; Habitat siz

ntroduction

Ecosystems along East African rivers serve as impor-ant habitats for many endangered animal and plant speciesMcClanahan & Young, 1996) and in parallel provide impor-ant ecosystem services for people settling along these riversnd their livestock, such as a high ground water level, fertileoils, shade, wood and various (medicinal) plants (Enanga,hivoga, Maina-Gichaba, & Creed, 2011). Wood is fre-uently used for timber, charcoal and the burning of bricksGovernment of Kenya, 1981; Wyant & Ellis, 1990). This cre-tes a conflict between human needs and nature conservation.

Today, most of the vegetation along these rivers is highlyegraded as it was transformed into agricultural land toeed the growing human population or has been replaced byhe invasive shrub species Lantana camara (Enanga et al.,011). In consequence, the remaining vegetation exists inmall and isolated remnant patches. The increasing level ofragmentation together with the decrease in habitat qualityay negatively affect the local flora and fauna (Bosschieter,oedhart, Foppen, & Vos, 2010; Richard & Armstrong,010). This situation may finally lead to the decrease in theensity of local populations and may reduce individual fitnesseading to local extinction (Boscolo & Metzger, 2011).

The globally vulnerable Kenyan endemic bird speciesurdoides hindei, Hinde’s Babbler, occurs in small and geo-raphically isolated population clusters across the Central

Please cite this article in press as: Teucher, M., et al. A Kenyan endemicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.0

nd Eastern Kenyan high- and lowlands, preferably in thick-ts along rivers (Shaw & Musina, 2003). The agriculturalntensification caused a strong decrease of these vegetation s

e’s Babbler; Invasive plant; Land use; Single-large or several-small

tructures and subsequently has led to a population declinef this bird species during the past decades (Shaw & Musina,003; Shaw, Musina, & Gichuki, 2003; Shaw, Njoroge,tieno, & Mlamba, 2013a,b). Especially the populations at

he south-eastern distribution margin are thought to suffernder high demographic pressure (Government of Kenya,014a,b) and the effects of climatic changes in our studyrea (Jaetzold, Hornetz, Shisanya, & Schmidt, 2007).

We assessed T. hindei in a highly fragmented environmentlong the Nzeeu River south of Kitui, East Kenya, where weerformed a detailed land use mapping. We measured theime our study species spent in various environments, suchs thickets varying in size and quality and further recordeduman activities in the area. Based on this data we defineones of potential conflict between human activity and theird species. In detail we raise the following three questions:

(i) Does T. hindei show any habitat preferences in our studyarea?

(ii) Does habitat performance (size, design, quality) influ-ence the habitat suitability for T. hindei?

iii) Are overlaps between species’ occurrence and humanactivities potential zones of conflict?

aterials and methods

tudy species

bird species Turdoides hindei at home in invasive thickets. Basic02

Turdoides hindei occurs in cool and moist highland regionsuch as the foothills of Mount Kenya (Meru, Kirinyaga)

ARTICLE IN PRESSBAAE-50852; No. of Pages 9

pplied

amaa&phbgLh&N

fttth2ct(icfiL

S

2eretstosdp1qddtamtrl

L

R

hbstWaWLop

A

ucvcu1Tqactswdifaos

D

lTt&w(2mteaueel

M. Teucher et al. / Basic and A

nd east of the Aberdares Range (Nyery region), but also atedium elevations (Thika, Oldonyo Sabuk) and at the dryer

nd warmer lower elevations in eastern Kenya (Machakosnd southwards, Kitui and eastwards) (Plumb, 1979; Shaw

Musina, 2003; Shaw et al., 2003). Most of the species’opulations can be found along rivers where the individualside in dense thickets (Dowsett & Forbes-Watson, 1993). Theird species is a co-operative breeder and thus lives in familyroups differing in size (Njoroge, Bennun, & Lens, 1998).ike other representatives of the genus Turdoides, also T.indei shows territorial behaviour (Gaston, 1978a,b; Zahavi

Zahavi, 1990; Monadjem, Owen-Smith, & Kemp, 1995;joroge et al., 1998; Shaw et al., 2003).Most of the habitats suitable for T. hindei are today highly

ragmented due to clearance of the vegetation and deforesta-ion along most rivers. As a result of severe disturbances ofhese habitats, most remaining patches are now dominated byhe invasive shrub species L. camara, providing a surrogateabitat for T. hindei (Njoroge et al., 1998; Njoroge & Bennun,000; Shaw et al., 2013). The strong habitat destruction andonsequences of climatic change caused a severe decline ofhis bird species, especially at its eastern distribution marginShaw et al., 2003, 2013a,b). This negative population trendn combination with the species’ restricted distribution in theentral and eastern provinces of Kenya has led to its classi-cation as globally vulnerable according to the IUCN Redist (BirdLife International, 2012).

tudy area

All data were recorded between the 18th September and1st November 2013 along the Nzeeu River located south-ast of Kitui (1◦23′S; 38◦00′E). The bimodal precipitationegime provides two rainy seasons in our study area (Jaetzoldt al., 2007); the period of our study covered the end ofhe dry season and the beginning of the second rainy sea-on. Periodical percipitation and limited soil fertility restricthe agricultural productivity (Jaetzold et al., 2007). 97.1%f the human population in our study area depend on sub-istence crop farming (Government of Kenya, 2014a). Theemographic pressure in Kenya is very high and the humanopulation in our study area has almost doubled between999 and 2009 (Government of Kenya, 2014a). In conse-uence, fallow periods for fields are neglected. This furtherecreases soil fertility, which in turn leads to an increasedemand for ‘new’ agricultural land. In addition, the vegeta-ion along the river is frequently exploited for timber, charcoalnd brick production. The region is further affected by cli-ate change with increasing rainfall variability and rising

emperatures (Jaetzold et al., 2007). These factors lower theeliability of agricultural production and food security andead to a severe destruction of pristine habitats.

and use mapping

Please cite this article in press as: Teucher, M., et al. A Kenyan endemicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.0

Land use was mapped along a 2 km section of the Nzeeuiver, in accordance with the spatial distribution of the four T.

taa

Ecology xxx (2015) xxx–xxx 3

indei family groups observed. We mapped a 200 m zone onoth sites of the river. Mapping was performed with a Sam-ung Galaxy Tab 2 10.1 tablet with a built-in GPS device, andhe software QGIS (QGIS for Android vers. 1.9.90-Alpha).

e mapped the Nzeeu River and streets as lines. Thickets,gricultural fields and open land were mapped as polygons.e further divided the thickets into pristine and the invasive

. camara. Point data were taken for trees and various typesf human disturbance, such as houses, brickworks, charcoalroduction areas and graveyards.

ssessing Turdoides hindei

We monitored four family groups of T. hindei from 7 amntil 5 pm. The presence of family groups was detected viaontact calls. During foraging in a new habitat patch, indi-iduals increase the number and volume of these contactalls (own observation) facilitating the tracking of individ-als. GPS coordinates of individuals were determined every0 min. with a GPS device (Garmin Etrex H, Kansas, USA).o prevent potential disturbances of the animals and subse-uent displacement, bird observations were conducted witht least 10 m distance to the respective group; exact GPSoordinates were taken after the group moved to a new loca-ion. To ensure comparability among the family groups, aimilar number of observation points (ca. 250 per group)ere taken for each group. Furthermore, all observations ran-omly covered the entire day which we divided afterwardsnto the following time slots: morning (earlier than 10 am),orenoon (10 am to 12.29 pm), noon (12.30 pm to 2.59 pm),nd afternoon (after 2.59 pm). Each time period consistedf an equal time interval (2.5 h) and for each time interval aimilar number of observations were collected.

ata analysis and statistics

The 95% Minimum Convex Polygon (MCP) was calcu-ated for each T. hindei group based on GPS observations.he 95%-MCP was selected as this estimator is assumed

o be suitable for comparative analyses (Kernohan, Gitzen, Millspaugh, 2001). Furthermore, 75%- and 95%-kernelsere calculated for each group with the fixed kernel method

Worton, 1989) using the R package adehabitatHR (Calenge,006) in R vers. 3.0.2 (R Core Team, 2013). The ad-hocethod was used to estimate the smoothing parameter h for

he kernel of each group. For calculating the habitat prefer-nces of T. hindei we exported the kernels to QGIS 2.0.1nd performed an overlay analysis with the mapped landse data. We calculated the Jacobs Index (Jacobs, 1974) toxpress habitat preferences, here classified into two param-ters: “thicket” and “open space” (consisting of agriculturaland and fallow land). The Jacobs Index compares the propor-

bird species Turdoides hindei at home in invasive thickets. Basic02

ion of available habitat with the proportion of habitat usednd ranges from −1 (absolute avoidance) over 0 (no inter-ction) to 1 (absolute preference) for the respective habitat

ARTICLE IN PRESSBAAE-50852; No. of Pages 9

4 M. Teucher et al. / Basic and Applied Ecology xxx (2015) xxx–xxx

Table 1. Activity ranges of Turdoides hindei for the four family groups. The table shows smoothing parameters calculated with the ad-hocmethod for each family, 95% Minimum Convex Polygon (MCP), 95% kernel, 75% kernel, and 75% kernel for individual birds.

Group Smoothing parameter h(ad-hoc method)

95% MCP [ha] 95% kernel [ha] 75% kernel [ha] 75% kernel (individual birds)[ha]

3 26.26 5.04 7.13 3.48 1.164 28.60 4.79 7.5 4.07 1.026 48.70 13.43 23.91 12.88 2.157 7.09

tcilTapb

abh(igbhw

tba(wBi&rwgCb

swrwgw(odtol

spqat

R

L

7oomeafac1

P

it7g71((dc(to

63.18 9.38 2

ype (thicket or open space). Based on that approach, wealculated the proportion of localisations of the bird speciesn thickets compared to all observations. We further calcu-ated the proportions within the 75%- and the 95%-kernels.o analyse the preferences of T. hindei concerning the avail-bility of single large trees within the habitat we additionallyerformed an overlay analysis with the mapped land use data,ird observations and locations of trees.To quantify the overlap between T. hindei and human

ctivities we calculated 75%-kernels for human disturbanceased on factors mentioned above. As smoothing parameter

we used the mean smoothing parameter of all bird groupsh = 41.69). The calculated kernels were exported from Rnto QGIS and intersected with the 75%-kernel of the birdroups. With an intersect analysis of the 75%-kernel of theirds’ activity ranges and the calculated kernels of the 75%uman disturbances new polygons representing the overlapere produced in QGIS.To test for potential differences in the movement dis-

ances varying with the time of day, we summed distancesetween consecutive locations per 1 h using the R packagedehabitatLT (Calenge, 2006). Effects of the time of daymorning, forenoon, noon, afternoon) on movement distancesere analysed using linear mixed effects models (Pinheiro &ates, 2000) with a maximized log-likelihood implemented

n the R package nlme (Pinheiro, Bates, DebRoy, Sarkar, the R Development Core Team, 2013). To account for

epeated measurements from bird groups the factor groupIDas included as a random effect. Different variances per birdroup were modelled using the varIdent variance structure.ontrasts between different times of day were investigatedy re-ordering factor levels.Furthermore, the effects of habitat patch size and edge-

ize-ratio and the type of thicket (L. camara vs. pristine) asell as two-way interactions between both parameters on the

elative duration of stay per patch (duration of stay/patch size)ere analysed using linear mixed effects models. The factorsroupID (n = 4) and patchID (n = 20) nested within groupIDere included as random effects to account for temporal

observation of the same bird group) and spatial (observationf the same thicket) autocorrelation. To achieve normal erroristribution and to avoid heteroscedasticity movement dis-

Please cite this article in press as: Teucher, M., et al. A Kenyan endemicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.0

ance was square root-transformed and the relative durationf stay as well as habitat patch size and edge-size-ratio wereog-transformed (x + 1). Model simplification was done in a

H

f

9.25 1.32

tepwise model selection by AIC implemented in the MASSackage (Venables & Ripley, 2002) until a minimal ade-uate model was obtained. Parameter estimates, t-statisticsnd p-values for all models were assessed from the summaryables.

esults

and use

In total we conducted land use mapping for an area of9.2 ha. The major proportion of this study area consists ofpen space, fallow land and agricultural fields (59%). Mostf the agricultural plots were interspersed by fruit trees likeango, citrus and papaya. 41% of the mapped area was cov-

red with thickets, of which 91% were the invasive L. camarand only 9% could be identified as remaining pristine galleryorest without L. camara. We found 167 points of humanctivities: 39 locations for brick production, 5 sites for char-oal production, 29 houses, 88 wells, 4 small reservoir dams,

bigger water tank, and 1 sandpit.

opulation structure and activity

The four observed family groups differed in number ofndividuals, with three, four, six and seven individuals. Nei-her the MCP (Fig. 1) nor the 95% kernels (red line, Fig. 1) or5%-kernels (blue line, Fig. 1) overlapped with neighbouringroups. The space visited by one individual obtained for5%-kernels (in parenthesis for 95% kernels) ranged from.02 ha (1.87 ha) per bird for the group of four birds to 2.15 ha3.99 ha) per bird for the group consisting of six individualsTable 1). Mean movement distances decreased over time ofay, with highest values during morning (132.62 m per hour)ompared to forenoon (71.12 m; t112 = −2.86, p < 0.01), noon62.96 m; t112 = −3.15, p < 0.01) and afternoon (53.62 m;112 = −3.76, p < 0.001), with no differences between thether temporal cohorts (Fig. 2).

bird species Turdoides hindei at home in invasive thickets. Basic02

abitat use and patch quality

The main proportion of observations of T. hindei over allour family groups was restricted to thickets (72.3%), and

Please cite this article in press as: Teucher, M., et al. A Kenyan endemicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.0

ARTICLE IN PRESSBAAE-50852; No. of Pages 9

M. Teucher et al. / Basic and Applied

Fig. 1. Land use matrix with minimum convex polygons 95%(yellow), and 75% and 95% kernels for the four family groups (con-sisting of 4, 3, 6 and 7 individuals—from south to north). Black dotsrepresent occurrences of Turdoides hindei. (For interpretation of thereferences to color in this figure legend, the reader is referred to theweb version of this article.)

Fig. 2. Boxplots of Turdoides hindei movement distances (in m/h)during different times of the day (morning: n = 36; forenoon: n = 42,noon: n = 20, afternoon: n = 21). The line inside the box shows themedian. The box contains values between 25th and 75th percentile.Minimum and maximum of all of the data are included between thewhiskers. Different letters indicate significant differences betweentimes of the day, assessed from the summary table of linear mixedeffects models.

cowA9Ascssip

ps(htp(pitpmit

H

rmw(skv0

D

P

tcisfmatr

Ecology xxx (2015) xxx–xxx 5

ould further be specified to 97.9% in L. camara and 2.1%f the observations in pristine thicket. Only few observationsere in open space (17.5%) and agricultural land (10.2%).pproximately 48% of the 75%-kernels and 35% of the5%-kernels covered thickets (both, pristine or L. camara).nalysis on the relevance of the presence of single trees

howed that the majority of T. hindei observations (65.9%)an be found in thicket patches including trees, 27.8% in openpace with single trees and 14.8% in agricultural patches withingle trees. Thus, T. hindei was more frequently observedn habitat patches with single trees than without (χ2 = 8.24,

< 0.01).The scores of the Jacobs Index revealed that thickets

rovide high habitat suitability for T. hindei, while openpace like agricultural and fallow land were mainly avoidedTable 2). According to linear mixed effects models, T.indei stayed significantly longer in small habitat patcheshan in large habitat patches (Estimate = −1.63, t227 = −8.43,

< 0.001) (Fig. 3A) independent of the type of thicketL. camara thicket vs. pristine; Estimate = 0.22, t227 = 1.44,

= 0.15). Furthermore, T. hindei stayed significantly longern patches with increasing edge-size-ratio (Estimate = 1.53,227 = 9.99, p < 0.001), with higher relative duration of stayer patch in L. camara compared to pristine thicket (Esti-ate = 0.30, t227 = 2.00, p < 0.05) (Fig. 3B). There was no

nteraction of habitat patch size and edge-size-ratio withhicket type, which was excluded from both models.

uman disturbances

The ranges of the 75% disturbance kernels based on theecorded disturbance points (see above) overlapped onlyarginally with the 75% kernel of T. hindei presence,ith a mean overlap of 14% across all four family groups

Fig. 4). The group consisting of seven individuals showed thetrongest overlap of its 75% kernel with the 75% disturbanceernel (9.4%) compared to the other family groups (six indi-iduals 2.7%, three individuals 1.9%, and four individuals%).

iscussion

opulation structure

The four family groups showed distinct, spatially restrictederritories without any overlap. Accordingly, T. hindei can belassified as a sedentary, territorial bird species, which goesn line with previous studies on the population ecology of thepecies (Njoroge et al., 1998; Shaw et al., 2003). The fouramily groups differed strongly in the number of individual

bird species Turdoides hindei at home in invasive thickets. Basic02

embers (ranging from three to seven). However, the habitatrea needed per individual was relatively similar, except forhe group of six individuals inhabiting the most fragmentedegion for which the territory was almost double the size

ARTICLE IN PRESSBAAE-50852; No. of Pages 9

6 M. Teucher et al. / Basic and Applied Ecology xxx (2015) xxx–xxx

Table 2. Jacobs Index expressing the relevance of habitat structures, calculated for the two habitat types, thicket and open space, and eachfor the 75% and 95% kernel. Values range from −1 (absolute avoidance) to +1 (absolute preference).

Group Thickets Open space

95% kernel 75% kernel 95% kernel 75% kernel

3 0.844 0.788 −0.741 −0.5724 0.981 0.979 −0.988 −0.98867

cttca1iaSP2

P

t(dtTcketotMot

Pd

ofdcttlha&thTobfflva

lm

Fwl

0.921 0.864

0.941 0.890

ompared to the other groups. In this territory, only 44% ofhe 95% kernel covered thickets and the thicket patches inhis territory were scattered. Previous studies already indi-ated a positive correlation between habitat fragmentationnd the spatial extent of the territory of T. hindei (cf. Lens,992). Similar trends could also be found for other passer-ne bird species, like the Great Read Warbler Acrocephalusrundinaceus (Bosschieter et al., 2010), the Spotted Owltrix occidentalis (Carey, Reid, & Horton, 1990), or threeasserine birds in forest fragments in Brazil (Hansbauer et al.,008).

ristine or surrogate habitat

72.3% of all occurences of T. hindei were restricted tohickets, with the major proportion found in L. camara97.9%). These findings underline the high relevance ofense thickets in general, and suggest a high suitability ofhe invasive L. camara for the Kenyan endemic T. hindei.he importance of dense thickets is further underlined whenomparing the occurrence of T. hindei in the 75% and 95%ernels, with a higher proportion of thicket in the core ofach territory. The preference of T. hindei for thickets is fur-her supported by the Jacobs Index, which also suggests thatpen space like agricultural land is avoided. This is in con-rast with populations observed in other regions, i.e. close to

Please cite this article in press as: Teucher, M., et al. A Kenyan endemicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.0

t. Kenya or the Aberdares, where T. hindei can be observedn agricultural land such as coffee plantations (own observa-ions).

hta

ig. 3. Linear mixed effects model of the relative duration of T. hindei pithout effects of the type of thicket (solid line), and (B) the edge-size-ra

ine). Regression lines represent predictions (with other explanatory variab

−0.859 −0.829−0.731 −0.839

atch design, availability of trees and humanisturbance

The importance of dense thickets for T. hindei shown byur data is congruent with other studies on the habitat pre-erences of T. hindei, which indicated that the populationensity and breeding success of T. hindei decrease after thelearance of thickets (Njoroge et al., 1998). However, ourarget species spent disproportionally more time in smallhickets and in thickets with a high edge-size-ratio than inarger patches. This is in contrast to other studies indicating aigher risk of predation for bird species in small patches andlong narrow corridors (Castellon & Sieving, 2006; Vergara

Hahn, 2009). Obviously, the degree of fragmentation ofhe remaining thicket patches has no significant impact on T.indei, as long as the total cover of thickets remains the same.he presence of trees might also play a crucial role: 65.9%f T. hindei observations were made in patches characterizedy large trees. The starting point to access adjoining thicketsrequently was the top of a larger tree, crossing open land byying to another tree, and subsequent hiding in the groundegetation (own observations). Hence, large trees likely serves observation points and stepping stones between patches.

The 75% kernels of human disturbance marginally over-apped with the 75% kernels of T. hindei presence which

ay have two non-exclusive reasons: (i) T. hindei is avoiding

bird species Turdoides hindei at home in invasive thickets. Basic02

uman presence because of bird hunting in this region; (ii)hickets, the main habitat for T. hindei are more common inreas with no or limited human activities. However, we are

resence in a habitat patch in relation to (A) the habitat patch sizetio for L. camara thicket (dashed line) and pristine thicket (dottedles held at mean values) from minimal adequate models.

ARTICLE IN PRESSBAAE-50852; No. of Pages 9

M. Teucher et al. / Basic and Applied Ecology xxx (2015) xxx–xxx 7

F and hul d lilac)t

nd

T

Tait(sso

ashTs

bsetHaTwtnpTtcabi

ig. 4. Overlapping kernels of the occurrence of Turdoides hindei

ines) and the 75% kernels of disturbance by human activity (shadehe reader is referred to the web version of this article.)

ot able to distinguish the roles of these two factors with ourata set.

ranslating our data into conservation action

Our data yielded four main findings: (i) the main habitat of. hindei is dense thicket with the invasive L. camara servings surrogate habitat; (ii) T. hindei can be seen more frequentlyn small thicket patches in our study area (possibly becausehe majority of the thicket fragments are small in this area);iii) the existence of single trees plays a key role for the habitatuitability of thickets as these serve as vantage points andtepping stones; (iv) the occurrence of T. hindei does notverlap with hotspots of human activity.

In conclusion, a complete clearing would likely lead to loss of local populations of this bird species. Yet, several

Please cite this article in press as: Teucher, M., et al. A Kenyan endemicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.0

mall habitat patches already would help to maintain suitableabitats and support the persistence of T. hindei in the future.hese remaining thicket patches should be interspersed withingle trees, which are important linking elements for the

h

bt

man disturbance, based on 75% kernels of bird observations (blue. (For interpretation of the references to color in this figure legend,

irds. Thus, against the background of the single-large-or-everal-small (SLOSS) debate, our example indicates thatven fragmented thickets still provide an interconnected habi-at mosaic with large trees as important linking components.ence, the conservation strategy of several small protected

reas provides the background for a management adapted to. hindei. However, we have to note that these observationsere made at a time of year in which there is normally lit-

le breeding activity. Turdoides hindei most probably prefersesting in larger thicket patches to reduce the risk of nestredation by natural predators or anthropogenic disturbance.hus, groups may prefer to remain in the proximity of larger

hickets when they have recently fledged offspring. If so, theonservation of ‘several small’ thicket fragments might onlypply outside of the breeding season. Further work during thereeding season is required to determine the value of retain-ng at least one large thicket patch per territory (as breeding

bird species Turdoides hindei at home in invasive thickets. Basic02

abitat), in addition to several small thickets.The creation of a protected area for this Kenyan endemic

ird species is not very realistic in an environment charac-erised by food-insecurity and underdevelopment (Ramphal,

ARTICLE IN PRESSBAAE-50852; No. of Pages 9

8 pplied

1tStfit

A

sS(ttao

R

B

B

B

C

C

C

D

E

G

G

G

G

G

H

J

J

K

L

M

M

N

N

P

P

P

R

R

R

S

M. Teucher et al. / Basic and A

993). The local people should be made aware of the impor-ance of these habitats, also for ecosystem service provision.uch awareness could be raised by appropriate information

hrough the officers in charge of agriculture or even throughnancial incentives for not clearing thickets and transforming

hem into agricultural land.

cknowledgements

This work was kindly supported by the Technische Univer-ität München (Germany) the German Academic Exchangeervice (Promos–Studienreise). We thank Onesmos KiokoNational Museums of Kenya) supporting us in the field, andhe Mutua family for hosting us during our field work. Wehank Martin Husemann (MLU Halle, Germany) and twononymous referees for fruitful comments on a draft versionf this article.

eferences

irdLife International. (2012). Turdoides hindei. IUCNRed list of threatened species. BirdLife International.〈www.iucnredlist.org〉.

oscolo, D., & Metzger, J. P. (2011). Isolation determines patternsof species presence in highly fragmented landscapes. Ecography,34, 1018–1029.

osschieter, L., Goedhart, P. W., Foppen, R. P. B., & Vos, C.C. (2010). Modelling small-scale dispersal of the Great ReedWarbler Acrocephalus arundinaceus in a fragmented landscape.Ardea, 98, 383–394.

alenge, C. (2006). The package adehabitat for the R software:A tool for the analysis of space and habitat use by animals.Ecological Modelling, 197, 516–519.

arey, A. B., Reid, J. A., & Horton, S. P. (1990). Spotted Owlhome range and habitat use in southern Oregon coast ranges.The Journal of Wildlife Management, 54, 11.

astellon, T. D., & Sieving, K. E. (2006). An experimental test ofmatrix permeability and corridor use by an endemic understorybird. Conservation Biology, 20, 135–145.

owsett, R. J., & Forbes-Watson, A. D. (1993). Checklist ofbirds of the Afrotropical and Malagasy regions. Tauraco Press.Li.

nanga, E. M., Shivoga, W. A., Maina-Gichaba, C., & Creed, I. F.(2011). Observing changes in riparian buffer strip soil propertiesrelated to land use activities in the river Njoro watershed, Kenya.Water, Air, & Soil Pollution, 218, 587–601.

aston, A. J. (1978a). Demography of the jungle bab-bler, Turdoides striatus. Journal of Animal Ecology, 47,845–870.

aston, A. J. (1978b). Ecology of the common babbler Turdoidescaudatus. Ibis, 120, 415–432.

overnment of Kenya, N.E.S. (1981). Kitui District environmental

Please cite this article in press as: Teucher, M., et al. A Kenyan endemicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.0

assessment report. Nairobi: National Environment Secretariat,Ministry of Environment and Natural Resources.

overnment of Kenya. (2014a). County poverty rates 2005/06.Kenya open data. https://www.opendata.go.ke/.

S

Ecology xxx (2015) xxx–xxx

overnment of Kenya. (2014b). Proportion of households engagedin crop farming by region county estimates—2005/6. 2005/6.Kenya open data. https://www.opendata.go.ke/.

ansbauer, M. M., Storch, I., Leu, S., Nieto-Holguin, J.-P., Pimentel,R. G., Knauer, F., & Metzger, J. P. W. (2008). Movementsof neotropical understory passerines affected by anthropogenicforest edges in the Brazilian Atlantic rainforest. Biological Con-servation, 141, 782–791.

acobs, J. (1974). Quantitative measurement of food selection.Oecologia, 14, 413–417.

aetzold, R., Hornetz, B., Shisanya, C. A., & Schmidt, H. (2007).Farm management handbook of Kenya. Volume II: Natural con-ditions and farm management information. Part C: East Kenya.Subpart C1: Eastern Province. Eschborn, Nairobi: Ministry ofAgriculture; Deutsche Gesellschaft für Technische Zusamme-narbeit (GTZ).

ernohan, B. J., Gitzen, R. A., & Millspaugh, J. J. (2001). Analysisof animal space use and movements. In J. J. Millspaugh, & J.M. Marzluff (Eds.), Radio tracking and animal populations (pp.125–166). San Diego, CA: Academic Press.

ens, L. D. (1992). Variation in coherence of crested tit winterflocks—An example of multivariate optimization. Acta Oeco-logica, 13, 553–567.

cClanahan, T. R., & Young, T. P. (1996). East Africanecosystems and their conservation. Oxford: Oxford UniversityPress.

onadjem, A., Owen-Smith, N., & Kemp, A. C. (1995). Aspectsof the breeding biology of the Arrowmarked Babbler Turdoidesjardineii in South Africa. Ibis, 137, 515–518.

joroge, P., & Bennun, L. A. (2000). Status and conservation ofHinde’s Babbler Turdoides hindei, a threatened species in anagricultural landscape. Ostrich, 71, 69–72.

joroge, P., Bennun, L. A., & Lens, L. (1998). Habitat use by theglobally endangered Hinde’s Babbler Turdoides hindei and itssympatric relative, the Northern Pied Babbler T. hypoleucus. BirdConservation International, 8, 59–65.

inheiro, J., Bates, D., DebRoy, S., Sarkar, D., & R Develop-ment Core Team. (2013). nlme: Linear and nonlinear mixedeffects models. R package version 3. R Development CoreTeam.

inheiro, J. C., & Bates, D. M. (2000). Mixed-effects models in Sand S-PLUS. New York, NY: Springer Verlag.

lumb, W. J. (1979). Observations on Hinde’s Babbler Turdoideshindei. Scopus, 3, 61–67.

Core Team. (2013). R: A language and environment for statis-tical computing. Vienna, Austria: R Foundation for StatisticalComputing. URL 〈http://www.R-project.org/〉.

amphal, S. (1993). In IUCN (Ed.), Parks for life: Report of theIVth world congress on National Parks and protected areas (pp.56–58). Gland, Switzerland: IUCN.

ichard, Y., & Armstrong, D. (2010). The importance of inte-grating landscape ecology in habitat models: Isolation-drivenoccurrence of north island robins in a fragmented landscape.Landscape Ecology, 25, 1363–1374.

haw, P., & Musina, J. (2003). Correlates of abundance andbreeding success in the globally threatened Hinde’s Bab-bler (Turdoides hindei) and its congener, Northern Pied

bird species Turdoides hindei at home in invasive thickets. Basic02

Babbler (T. hypoleucus). Biological Conservation, 114,281–288.

haw, P., Musina, J., & Gichuki, P. (2003). Estimating changein the geographical range and population size of Hinde’s

ARTICLE IN PRESSBAAE-50852; No. of Pages 9

pplied

S

S

V

V

W

W

M. Teucher et al. / Basic and A

Babbler Turdoides hindei. Bird Conservation International, 13,1–12.

haw, P., Njoroge, P., Otieno, V., & Mlamba, E. (2013a). Detectingchange in the status and habitat of Hinde’s Babbler Turdoideshindei: 2000 to 2011. Ibis, 155, 428–429.

haw, P., Njoroge, P., Otieno, V., & Mlamba, E. (2013b). The range,abundance and habitat of Hinde’s Babbler Turdoides hindei:Fine-scale changes in abundance during 2000–2011 reflect tem-

Please cite this article in press as: Teucher, M., et al. A Kenyan endemicand Applied Ecology (2015), http://dx.doi.org/10.1016/j.baae.2015.01.0

poral variation in scrub cover. Bird Conservation International,24, 453–465.

enables, W. N., & Ripley, B. D. (2002). Modern applied statisticswith S (fourth ed.). New York, NY: Springer.

Z

Available online at www.s

ScienceD

Ecology xxx (2015) xxx–xxx 9

ergara, P. M., & Hahn, I. (2009). Linking edge effects and patchsize effects: Importance of matrix nest predators. EcologicalModelling, 220, 1189–1196.

orton, B. J. (1989). Kernel methods for estimating the uti-lization distribution in home-range studies. Ecology, 70,164–168.

yant, J., & Ellis, J. (1990). Compositional patterns of riparianwoodlands in the Rift Valley of northern Kenya. Vegetatio, 89,23–37.

bird species Turdoides hindei at home in invasive thickets. Basic02

ahavi, A., & Zahavi, A. (1990). Arabian babblers: The quest forsocial status in a cooperative breeder cooperative breeding inbirds. Cambridge: Cambridge University Press.

ciencedirect.com

irect