Breeding and Foraging Behaviour and Habitat Characteristics of the Scaly Ground-Roller Geobrasfes Sgua Migerus in Madagascar

Ostrich 2004, 75(1&2): 25-31 Copynght @ NISC Pty Ltd Printed in South Afnca All ights reseNed OSTRICH /ssN 0030 6525

Breedingand foragingbehaviour and habitatcharacteristics of the ScalyGround-roller Geobrasfes sgua migerus in Madagascar

Jean Eric Rakotoarisoal,2'3*and Berthin Bel 1 ThePeregrine Fund, PO Box4113, Antananaivo 101, Madagascar ' Facultedes Sclences,PO Box 906,Antananarivo 101 , Universityof Antananarivo,Madagascar 3 Currentaddress: l inois StateUniversitv. PO Box4120. Normal. lL 61761.United States of America * Correspondingauthor, e-mail: [email protected]

The endemicScaly Ground-roller Geobrasfes squamigerus was studiedduring two breedingseasons from October1997 to January1999 in MasoalaNational Park, Madagascar. Several vocalisations were associatedwith territorialdefence, contact, excitement,and aggressivenesstowards intruding conspecifics. Of the 269 prey itemsobserved, 71.5o/o werc invertebrates, 7.5%vertebrates, and 21 .0Vo unidentified. Earthworms and centipedesrepresented the mostnumerous prey-types taken, rep- resenting55% and 21o/oof the identifiedprey, respectively. Three nests were locatedin valleysand nearstreams. Nests were placedin groundburrows with tunnelsthat measuredless than 1Ocmin diameterand less than 1m in depth.A singleegg clutchwas laidin each nest,and incubationand the nestlingperiods lasted 18 daysand 24 days,respectively. Nesting began in late October,and one youngfrom each nest successfullyfledged in mid-December.Only the femaleincubated and brood- ed the young.Both male and female provided food for the nestling,and feeding rates did not differbetween the sexes.Nesting habitatdiffered significantly from randomplots sampled. Herbaceous coverage density was higherin nestingareas than random Dlots.

c'l+ Ln (^, lntroduction

The Scaly Ground-rollerGeobrasfes squamigerus (sensu of slash-and-burnclearings, secondary growth, and largely Kirchmanet al. 2001) is a burrow-nestingspecies (Benson intactforests. The maturelowland rain forest of the Masoala et al. 1976,Goodman 1994) whose rangeextends from the Peninsulahas a canopy height less than 30m with few northto the centreof the easternrain forest of Madagascar emergenttrees, high floristic diversity, and steep mountain- (Langrand1990). lt belongsto the endemicfamily of ous topography(Guillaumet 1984), and elevationsrange Brachypteraciidaeand was formerly placed in the genus from 0-1 200m. Average rainfall recorded at AFS from Brachypteraciasuntil a recent phylogeneticstudy of the fam- 1992-1996was 6 049mm(Thorstrom et al. 1997),making it ily (Kirchmanet al. 2001).The Scaly Ground-rolleris cate- one of the wettestregions on the island.Monsoon rains and gorisedas vulnerable(a taxonfacing a highrisk of extinction cyclones occur between December and April (Donque in the wild in the medium-termfuture) in the recentlyupdat- 1972).September through November is normallythe driest ed Red Data List (Collaret al. 1994).The speciesis known periodof the year.Annual temperatures vary from a low of to inhabitundisturbed evergreen primary rain forest with a 18"Cto a highof 33"C. preference for relatively dark undergroMh that offers fairly We located Scaly Ground-rollersby their calls. low herbaceousvegetation and a carpetof dead leavesand Vocalisationswere recordedand analysedwith Spectrogram branches(Langrand 1990). Little information is availableon 3.2.1software (Horne 1994). Behaviours associated with the its behaviouralecology and breedingbiology and this paper callwere notedfor furtheridentification of eachcall type. We presentsthe firstsystematic study of this species. capturedadufts with mist nets (2.6mx 12.2m,36mmmesh) installednear nestsor where oairshad often been encoun- Methods and study area tered. Capturedground-rollers were measuredwith vernier calipers(nearest 1 .Omm), weighed with Pesola3009 spring ScalyGround-rollers were studiedduring two breedingsea- scale (nearest1 .09), and fittedwith colouredplastic rings. sons, coveringthe periodsOctober 1997 to January 1998 Morphologicalmeasurements include bill lengthtaken from and October1998 to January1999, in the MasoalaNational the end of featheringon the top mid-lineto the tip (Palmer Park, northeasternMadagascar. The study was conducted 1962);unflattened wing chord taken from the front of the at two sites,both below600m above sealevel. The firstsite folded wrist to the tip of the longestprimary (Biggs et ai. '1978); was situatednear AndranobeField Station(AFS) on the tarsus lengthtaken from the posteriorcentre of the west coast of the Masoala Peninsula(15'41'S, 49"57'E). tibiotarsal-tarsometatarsaljoint to the dorsalbase of the cen- The secondsite, also on the peninsula,was 3km east and tre toe (Biggs et al. 1978)iand tail length taken from the up slope from the villageof Ambanizana,7km north of the pointof insertionof the centralfeather to the tiD. AFS. The peninsulais road lessand composedof a mosaic Scalv Ground-rollernests were located in two wavs. Rakotoarisoaand Be

First,in order to solicitresponse from breedingadults, we Results playedtape recordingsof their calls.We followedadults to the oresumednest areas and searchedfor Dotentialnest We located three nests for which observationtime totalled sites. We then monitoredDotential burrows and recorded 483h,with an averageof 10h + SD of 1.8h per day in 1997 observationsof Scaly Ground-rollernesting activitiesat (n = 22 days),and 7h t SD of 1.3hper day in 1998(n = 39 these sites. Second, we attached a 2.19 back-mounted days). transmitter(Holohil Systems Ltd, Canada) to a captured adultand monitoredits movementsand behavioursDrior to Vocalisations the nestingactivities. The transmitterweighed less than 3% Several vocalisationswere identified.These calls were of the weight of the individualand lastedtwo weeks. The associatedwith differentbehaviours. Territorial calls were 'whoop' nesting behaviourof Scaly Ground-rollerswas observed represented by a sequence of notes that were from blinds or other camouflagedlocations at about 30m uttered with various frequencies according to circum- from nest. Observationswere made using 10 x 40 binocu- stances.Frequency of notes(number of notesper time unit) lars and datawere recordedin a continuouswritten log that was seen to increasewhen two neighbouringbirds were in includedthe time of day, frequency,and durationof their close proximityto each other while defendingtheir respec- nestingactivities at and aroundnest sites. tive territories(from 7-10 call min-1,duration = 22min).Most We recordeddata for each pair usingthe colouredplas- territorial calis were given early in the morning tic rings and assignedsex based on behaviourduring the (05h00-06h00)and late in the afternoon(16h00-17h00) nestingperiod. As only one individualwas seen to perform (Figure1). Territorial calls were broadcastfrom percheswith all the incubationduties and brooding in each pair, we an averageheight of 1.2m (range0.8-2m) and drameterof assumedthat individualwas the female. 2.5cm(range 1-4cm, n = 5 perches). For nestingphenology, egg layingwas assumedto have A second type of vocalisation,the contact call, was occurredand incubationbegan when the femalestarted to associatedwith communicationbetween a pair as they spend more than three hours a day in the nest. Hatching movedthrough the forest.The callconsisted of a hollowsin- 'whoop' was assumedto have occurredinside and broodingbegun gle and deep note similar to the territorialcall. when the male startedto deliverprey to the nest while the Spectrogram analyses of the territorial and contact calls femaleremained inside. Nests were checked about every 10 showeda differencebetween them in frequencyand dura- daysto followthe developmentof the young.Prey itemsfed tion of notes.Territorial call notes had a seeminglyhigher to the nestlingby adultswere visuallyidentified and sizes frequency(mean = 43.23kH2,SD = 0.04kH2,n = 10 notes) estimated.Nest chamberswere excavatedand measured and lasteda shorterperiod of time (mean= 312.9ms,SD = the followingbreeding season after confirmationthat they 67.8ms, n = 10 notes) than the contact call (25.73k{z. had been abandoned. 372ms,n=1call). 'kwek' Severalparameters were consideredwithin an area of A third type of vocalisation,a sharp,harsh alarm 100m2around the nestto characterisethe site.Saplings and call was made when the bird becameexcited (e.9. during herbaceouscoverage were categorisedinto ranges of size. preycapture) or when threatened(e.9. when trappedin mist Data recordedincluded herbaceous coverage density, tree density,and distancefrom water, slope, and canopydensity. The same parameterswere measuredon 10 randomlysam- pled plotsof 100m'within the same forestblocks. Location Territorialcalls of the plots(distance from the nest,direction and angle)was Contactcalls determinedby randomlydrawing numbers in a bag contain- I ing numberedpapers. We then comparednest sites with randomplots in orderto detectany uniquecharacteristics of the nest site. Data referring to nest site characteristicswere first test- ed for normalityusing the Shapiro-Wilktest (Shapiroand Wilk 1965).All variablesexcept number of tree species,distance from water, and distancefrom valley bottom were nor- mallydistributed. The Student'st-test was usedto compare nest sites and random plots for herbaceousvegetation, saplingdensity, tree density,slope of the nest,slope of the site, number of liana, ferns, tree height,tree diameterat (dbh), foliagedensity. The Wilcoxontwo breastheight and 99| sampletest (Sokaland Rohlf1995) was usedfor numberof vqtqt ,.Fo.q*s*$glss*6$*lr--d,_dS${)*v*b*dS tree species,distance from water,and distancefrom valley. We used the sequentialBonferonni correction (Sokal and HOURS Rohlf1995) to adjustthe rejectioncriteria of the analysesin orderto decreasethe probabilityof makinga type I error,as Figure 1 : Periodicityof two call types made by Scaly Ground-rollers some of the comparisonswere not totallyindependent. All during two breeding seasons on the Masoala Peninsula, the analyseswere donewith the SAS package(SAS 2001). Madagascar Ostrich 2004, 75(1&2Jt25-31

nets). This call is similar to that describedby Langrand prey itemsper day (range9-20, n = 34 days)was delivered (1990).A fourth type of call was made when the bird was to the nestlingin each nest. Prey items had a mean length threatenedand consistedof a sharpraspy hissing kwish-sh of 6cm (range2-30cm, n = 269). Preytaken exclusively on accompaniedby spreadingits tail, fluffingits feathersand the groundrepresents 80.3% of the items observed.Scaly facingthe observer.This call was madewhen the bird was Ground-rollersuse their beak to dig and overturnleaf litter upset(e.9. in the presenceof a humanin closeproximity) or when searchingfor food.While digging, they use theirbeak startled(e.9. during an encounterwhen a pairof birdswere to toss the soil and groundcover debris aside and pick up with theiryoung) and was presumablyintended to intimidate the food items. Foragingstrategies seemed to be based or distractthe intruder. mainlyon the detectionof noiseand movement.A foraging ground-rollerwould remainin one place for some seconds Measurements slightlyinclining its head, listeningand checkingcarefully Six rndividualswere trapped, measured and weighted (Table for movement.Upon locationof a prey item,the bird would 1). Trappedpaired birds appeared to be sexuallydimorphic quickly hop forward, cocking the head for capture, and with males being larger than females (mean wing length: would dispatchthe prey with a flick of the head if the latter male125.5mm, female 118.5mm; mean weight: male 1559, triedto resist. female 137.59,n = 2 males and 2 females).For the third oair. the sex of each individualcould not be determined Habitatuse becausenesting behaviour was not observeddue to the late Scaly Ground-rollerswere observedon the ground during discovery of the nest. nearlyall of its mainactivities and occasionallyon low perches while callingor resting.On severaloccasions, the study Feeding and foraging behaviour pairs were seen foraging in valley bottoms with creek We made 483h of observationsduring the period when drainagesand from time to time on valley ridges.No inten- nestlingswere beingfed at three nests.There was no sig- sive use of treeswas recordedduring this study.Trees were nificantdifference in feedingrates of nestlingby adultmales used only as roosts(n = 2 observations),as perchesfrom and femalesof each pair (t-test,P = 0.35, n = 18 days for which territorialcalls were vocalised(n = 7 observations), pair 1; P = 0.16,n = 16 daysfor pair2). Of the 269 prey and finallyas restingperches (n = 5 observations). items observed, 71.5o/owerc invertebrates,7.5% verte- The radiotagged individual was locatedtwice at two dif- brates,and 21.0% unidentified.Disregarding the unidenti- ferentnight roost sites that were abovelhe ground.The {irst fied prey category,earthworms (Pheretima sp.) (5a.7%) roostwas a smalland slightlyinclined branch (15'from hor- and centipedes(Scolopendra morsltons) (20.9%) made up izontal)situated 0.27m from the centreof the tree,2.Scm in the most numerousprey taken (Table2). An averageof 12 diameter,and 3.13mabove the ground.The roostsite was a

Table 1: Measurementsof six individualsof ScalvGround-rollers from this studv

Billlength (mm) Wing length(mm) Tarsuslength (mm) Taillength (mm) Mass (g) Pair 1 male 47.4 120 165 female 30 120 49.6 90.2 140 Pai 2 male 34.2 124 51 98.2 145 female 29.6 116 49 93.5 '135 Pair3 Unknown 3'1.3 118.7 45.'l 95.2 158 Unknown 28 115 48.2 95 128

Table2: Preyobserved delivered to nestsby ScalyGround-rollers from 1997to 1999(n = 269 prey items,n = 3 pairs)

'1997 Preytype (Class,Order, Family or Genus) 1998 Average Mean Others Number(%) Number(%) Number(%) Size (cm) (preycapture location, proportion) Earthworm(Pheretima sp., l\4egascoleidae) 34(32.6) 83 (76.8) 117(55.4) 7.5 Ground Centipede(Scolopendra morsitons, Myriapodes) 37 (35.5) 7 (6.3) 44 (20.8) 7.0 Ground lnsects 16(1s.5) 9 (8.3) 25 (11.8) 3.0 25% Coleoptera 19% Hymenoptera 19% Lepidoptera 37% Unidentified Larya 4 (3.8) 4 (3.71 I (3.7) 2.5 Ground or rotten tree trunk Frog 3 (2.8) 4 (3.7) 7 (3.21 5.0 Ground Spider 4 (3.8) 0 (0) 4 (1.e) 2.0 Between branches Lizatd (Zonosaurussp.. Phelsuma sp.) 3 (2.8) 0 (0) 3 (1.4) 3.5 Treetrunk or ground Millipede(Diplopoda) (0.s) 1(0,9) (0.e) Ground Shrew tenrec (Microgalesp., Tenrecidae) (0e) 0 (0) (0.4) 6.5 Ground Snail (Sphaeroterlurnsp., Gastropoda) I (0.9) 0 (0) (0.4) 2 (diameter) Treetrunk or ground \

Rakotoarisoaand Be small tree of 6cm dbh, 70% foliage density,and 4.2m in Nest description height.The secondroost site was a lianasituated 6.5m off Nest burrowswere locatedin the ground,in closeproximity ground. Both roost sites were locatednear valley bottoms to water,on eitherflat or inclinedvalley bottoms. Nests con- and within20m of wateron a slightlyinclined slope. sisted of tunnelsthat measured8.8cm in diameter(range 6.5-10cm) and 62.5cm in depth (range 57-77cm, n = 3) Breeding cycle (Table3). The entrancetunnels were angled and it was Courtship behaviour and nest building impossibleto see into the nestingcavity from outsidethe Courtshipfeedings were observed.On one occasion,the entrance.The oosteriorend of the tunnel was wide and malecaught a 7cm earthwormand gave a contactcall. The formedthe nestingchamber. Distance between two neigh- female respondedthree times with the same call, then bouringnests was 230m and distancebetween nests that approached,and took the prey from the male's beak.The belongedto the same pair duringtwo successivebreeding male allopreenedthe female'shead for a few seconds,and seasonswas 50m (n = 1 pair). then they separated. We did not observeany nest excavationor indicationof Egg-laying, incuhation, and hatching reuse of other animal burrowsfor nest during this study. EggJayingoccurred from mid-Octoberto early November Duringincidental encounters, an individualaccompanied by (Figure 2). On one occasion, a distinct behaviourwas its matewas seencarrying dead leaves,probably to its nest, observedwhen both the male and female remainedwithin which suggestedthat both sexes participatedin the prepa- 40m of the nest and vocalisedexcitedly. Shortly afterwards, rationof the nest. Dead leaveswere a majorcomponent of they moved towardsthe nest, and the femaleentered the the nest lining. Neither dust nor piles of dirt were seen nest (duration:12min) while the male remainednearby and aroundthe nest entrancebefore the nestingperiod. continuedto call almostcontinuously. Both individualsthen left the nestingarea and movedtowards their roostingsite. The completebehavioural sequence commenced at 15h00 and lasted25min. This behaviourmight be associatedwith egg layinggiven its date of occurrenceand circumstances. E Only a single whitishegg was observedin each nest. 935 --i Reconstitutionof a Scaly Ground-rollereggshell showed <30 that egg dimensionswere 45mm x 30mm. L! z We made 128hof observationduring the incubationperi- <25 ods that lasted 18 days. Based on our observations,the t femalesoerformed all the incubationduties and spent the )20 Terminationof Darentalcare night in the nest. No prey deliverieswere observedduring N the incubationoeriod. r.r- 15 Hatching occurred around mid-November(Figure 2). lncubation o The femalestayed in the nestalmost all the time duringthe Uro brooding period, which lasted 7 days. However,on one t occasion,the femalewas observedto leavethe nest during ||t E the fifthday of the broodingperiod (duration female-off nest: 2h 10min) and brought food to the nestling.During the M M broodingperiod, the malewould visit the nestand bringfood at an averagerate of 21 prey itemsa day (range16-23, n = Figure 2: Breedingphenology of the study pairsof ScalyGround- 4 days).Prey deliveriesrecorded during the nestlingperiod rollers plottedwith averagedaily rainfalldata obtainedfrom 1997 to (day 7 post-hatchingto fledgling)averaged 12 itemsa day 1998in AFS, MasoalaPeninsula (n = 15 days).Disparity in prey deliveryrates recorded dur-

Table3: Nestand nestsite characteristics of threebreeding pairs of ScalyGround-rollers from 1997to 1999

Parameters Pair 1 Pair 2 Pair3 Mean Nestentrance diameter (cm) 10.00 10.00 6.50 8.80 Depth from nest entrance (cm) 68.00 57.00 77.00 62.50 Verticaldistance between nest entrance and receptacle(cm) 42.00 37.50 55.00 44.80 ReceptacleDimensions Height(cm) 12.00 11.00 11.50 11.50 Length(cm) 26.00 23.00 22.00 23.70 '19.80 Width(cm) 22.00 20.00 17.50 Slopeof the ground(") 38.00 59.00 45.00 47.30 Slope at the nest entrance (') 8,00 6-00 12.00 8.70 Foliagedensity (%) 32.00 67.00 93.24 64.10 Distancefrom the flat valley bottoms (m) 20.00 0.00 63.00 27.70 Distancefrom water (m) 53.00 49.00 61.00 54.30 Distancefrom the nearest human disturbance(km) 1.73 1.50 1.78 1.67 Ostrich2004, 75('l&2): 25-31

Table4: Statisticalcomparisons conducted on neslsites and random plots

Parameters Statisticaltests Significance(6r' = 0.05/o Herbaceouscoverage density (height <50cm) Studentt-test P = 0.0023 ..- (c' = 0.0055) Saplingsdensity (height >50cm and dbh <2.scm) Studentt-test P = 0.0008 ... (6r'= 0.0041) Treedensity Studentt-test P = 0.30 Non-significant Numberof species Wilcoxontwo sampletest P = 0.10 Non-significant Treeheight Studentttest P = 0.89 Non-signiticant dbh Studentt-test P = 0.64 Non-significant Foliagedensity Studentt-test P = 0.0020 ..' (o' = 0.0045) Slope/nest Studentt-test P = 0.0001 ...' (o' = 0.0038) Slope/ground Studentt-test P = 0.026 Non-significant Distancefrom water Wilcoxontwo sampletest P = 0.01 Non-significant(cr' = 0.0062) Distancefrom valley bottoms Wilcoxontwo sampletest P = 0.30 Non-significant Numberof liana Studentttest P = 0.09 Non-signiflcant Numberof ferns Studentt-test P = 0.41 Non-significant

* Level of significance

3). Thereafter,the nest site entrancewas covered,filled, and hiddenwith dead leavesand sticks(n = 3 nests). The female always stayed with the young during the first post-fledglingweek whereas the male ofren left the family groupfor a variableamount of time (meanduration: 25min, range 15-40min, n = 3). Parental care of fledglings appeared to cease at 60 days of age.

Aresf sifes' chara cteristics Statisticalanalyses showed that nest sites are mainly characterisedby a higherherbaceous cover and saplingdensity, a lowerfoliage density, and a less inclinedslope at the nest entranceas comparedto randomlyplaced study plots (Table 4). However,herbaceous cover was not uniformlydistributed within the nest site, as at least 16m2around each nest entrance had open understory(n = 3 nests).

Discussion

The Scaly Ground-rollerappeared to be very common in the study sites located below 600m above sea level. Eight pairs were locatedwithin an area of approximately100ha around AFS, and four pairsin Ambanizana.Thorstrom and Watson (1997) reported that the species was detected in all but one of the eight sites visited in Masoala Peninsula.The site where it was not detected was in Ambohitsitondroina Ambanizana(15"34'5, 50"00'E), located to the east of the villageof Ambanizana.In contrastto other areassurveyed, Ambohitsitondroinais an area characterisedby mid to high- Figure 3: Photoof a nestlingat 21 days old (threedays before (60G-1 fledgling),The nestlingtravelled up to the nestentrance gaping for elevationmoist montianeforest 000m). These obser- foodafter we imitatedthe female'scall vations suggested the species might be more common in sites at lower elevations. Such sites would have a certain level of human disturbancedue to their easy accessibility. ing broodingand nestlingperiods probablyrelates to the But as shown by this study, Scaly Ground-rollersmay not provisioningof the femaleduring the broodingperiod. require completely intact forests because trees were used only for roosting and perching; nests were on the ground, Development of young and fledgling period and prey items taken on tree trunks representedonly 2o/oof At the end of the brooding period, nestlingshad almost the prey captures observed. Moreover, we had incidental acquiredtheir full subadultplumage and were able to move observations of the species trom tavy (cleared areas for to the nest entrance to receive food. Young fledged 24 days swidden agriculture)and secondary growth. Therefore, the after the date of hatching and definitelyleft the nest (Figure Scaly Ground-rolleris not restrictedto primary forests, and Rakotoarisoaand Be this speciesmay be more adaptableto or tolerantof forest Scaly Ground-rollerand Short-legged Ground-roller degradationand alterationthan previouslysuspected. BrachypteraciasIeptosomus are two sympatricspecies that The exactdates of nest building(excavation) and reuse belongto the familyof Brachypteraciidaeand are found in of nests need furtherstudy. Although the study pairswere the MasoalaPeninsula. The ShortJeggedGround-roller is not observed in nest buildingactivities, nest dimensions an arborealspecies that nestsin talltrees and forages main- suggestthis activitycould last severalweeks. Observations ly under the canopy level with an average height above in Mananara(pers. obs.), a sitelocated about 50km south of groundfor prey captureof 6.1m + 4.0 (range0-18, n = 69) AFS, suggested nest building may even occur several (Thorstromand Lind 1999).On the other hand, the Scaly monthsbefore the breedingseason as a pairwas seentwice Ground-rolleris a terrestrialspecies that nestsand forages on 14 and 16 March1999 digging a fresh hole,probably its mainlyon the ground (80.3%of prey were taken from this nestfor the upcomingbreeding season. Another suggestion substrate).Consequently, difference in verticaluse of the is thatthe speciesmay use pre-existinganimals' ground bur- same habitat allows these tlvo sympatricspecies to coexist rows that need only to be enlarged.Such burrowsare very and to avoidcompetition. an adaptationthat would allow the commonin the studysites and are madeby terrestrialcrabs coexistenceof morphologicallysimilar birds (e.9. Pimm and (Sesarmasp.). The caseof cohabitationbetween the rodent Pimm 1982,Nudds ef a/. 1994).Further, the lackof intensive Eliurus webbi (Nesomyinae)and Scaly Ground-rollerin a use of trees by ScalyGround-rollers allows the latterto tol- ground burrow reportedby Goodman(1994) supports this eratea certainlevel of humanhabitat disturbance. hypothesis.Nest entranceswere alwayscovered and filled Rainfallwas the mostinfluential factor to exolainthe start with dead leavesand sticksat the end ofthe nestingperiod. of the nesting season. Nest site characteristicssuch as This behaviourwas intendedto hide the burrow entrance foliagedensity at and above the nest entrancemade bur- and possiblyretain the nestingsite for the followingbreed- rows more exposed to flooding during heavy rains. ing season.However, no nest reuse was observedduring Septemberthrough November are the driestmonths in this thisstudy. Moreover, dead leavesthat had beenbrought into regionand nestingactivities coincide with this dry period. the nest to protectthe egg and insulatethe nestlingproba- Low rainfallduring this periodwould facilitate finding dry and bly decayedthe followingyear and may haveprevented any dead leaves for the incubation,and food for the nestling nest reuse. Further investigationsbased on a long-term assumingthat a high rainfallmight decreasetheir foraging study and a larger sample size is needed to verify these efficiency. observations. Herbaceouscover and saplingdensity, foliage density, The ScalyGround-roller is a terrestrialspecies that feeds and slope at nest entranceare the variablesthat differed chieflyon earthwormsand centipedes.Prey itemsobserved betweennest sites and the randomplots. However, previous in this study were based mainly on food deliveredto the observationsreported the specieshas been seen in habitat nestlings.This studyand previousfindings (Langrand 1990, with dense herbaceouscover (Langrand1990, Thompson Thompsonand Evans 1992,Thorstrom and Watson 1997) and Evans 1992, Evans et al. 1992). Dense herbaceous suggestthe ground-rollermight show a preferencefor earth- cover would play two key roles:at the nest sites it would worms. Langrand(1990) reported the speciesfeeds rarely make the nest more difficultto find and withinthe habitatit on vertebrates. However, our observationsshowed a rela- wouldhelp individualsescape from predators. tively importantproportion of vertebrateprey items (9.1%) The slope at the nest entrancewas less inclinedthan taken by the species.lt was seen feedingon a largeshrew those of the random plots.This slope differenceprobably tenrec, perhaps Microgale talazaci, plated-lizard preventednest floodingduring heavy rains and water run- Zonosaurussp., and the day gecko Phelsuma sp. off. Nestswere locatednear streamsand in valleyssimilar Observationsin other sites of the Masoala Peninsula to the species preferencefor humid habitat as reported by indicatedthe Scaly Ground-rollerwas observedon several Thompsonand Evans (1992).The existenceof a perma- occasions feeding on dwarf chameleons Brookesia sp. nentlyhumid soil, associatedwith a relativelyhigh annual (Thorstromand Watson1 997). Powzyk (1995) also reported temperatureprobably contribules to the densityor availabil- that the specieswas seen feedingon a baby Raffussp. in ity of prey (e.9. earthworms) in the study sites. Mantadia. Consequently,climatic factors would be the mainfactors that All foragingwas from the ground. This species rarely limitthe distributionof this speciesby playinga key role in flew.Occasional flights were associated with riverand creek the dishibutionof the preferredprey species. crossingsand enteringand exitingthe nestsites, which may Habitat loss is the main threat for this species. have been to avoid leading potential predators to the nest Deforestationoccurs mostly in the low and mid-elevation sites. Flightswere short in duratron,low, and noisy (mean area of the eastern rain forest due to its relatively easy distance11 m rangingfrom 7-18m, n = 16). accessibilityto humans (Green and Sussman 1990). No territorialcalls were madearound the immediatenest Despiteits toleranceto habitatdegradation suggested by vicinitythough pairs were extremelyvocal shortly before the this study,Andriamasimanana et al. (2001) classifiedthe beginningof theirnesting activities. This lackof activitynear Scaly Ground-rolleramong the most vulnerablespecies the nest was probablyto avoid attractingpredators. No case associatedwith habitatfragmentation and loss based on of egg or nestlingpredation in the nestwas recordedduring their presencein the seven forest fragmentsthey studied. this study.Potential predators would includethe Ringtailed The apparenttolerance to habitatdegradation contrasted by Mongoose Galidia elegans and the Brown{ailed Mongoose its vulnerabilityto habitatloss and fragmentationsuggests Salanoia concolor,which are common in the area. that its susceptibilitymight be due to a patchydistribution Ostrich 2004, 75\1&2): 25-31

that causes sampling effects. The species might be locally Goodman SM 1994.A descriptionof the ground burrow of E/iurus common but regionally rare. Due to the rapid pace at which webbl (Nesomyinae)and case of cohabitationwith an endemic habitat loss and degradation is occurring in eastern bird (Brachypteraciidae,Brachypteracias). Mammalia 58: Madagascar and the lack of detailed biological data on the 670-672 Green GM and Sussman RW 1990. Deforestationhistory of the range of forest dependent species, conservation effofts eastern rainforestof lvladagascarfrom satellite images. Science often rely on area size as the main criteria for selecting and 248: 212-215 designing reserves. However, this study suggests that addi- Guillaumet J-L 1984.The Vegetation:an extraordinarydiversity. tional factors, such as microhabitat patterns play a key role In: Jolly A, Oberl6P and AlbignacR (eds) Key Environments: in designing an effective reserve network. Conservation of Madagascarpp 27-54. PergamonPress, New York species that have patchy distributions requires protection of Horne RS 1994.Spectrogram Version 3.2.1. Copyright 1994-1997 specific sites where the species are locally common. by R.S.Horne. All rightsreserved. FFT Code by PhilipVan Baren Therefore, further studies of species distributionsare among Kirchman JJ, Hackett SJ, Goodman SM and Bates JM 2001. the priorities for effective conservation of the entire avifauna Phylogeny and systematics of the ground-rollers (Brachypteraciidae) of the eastern rainforest of Madagascar. of Madagascar.Auk 118:849-863 Langrand O 1990. Guide to the Birds of Madagascar.Yale UniversityPress, New Havenand London Acknowledgements- We thank Rick Watson and Bill Burnhamfor Nudds TD, Sjoberg K and Lundberg P 1994.Ecomorphological making this study possible.Special thanks to Rick Watson and relationshipsamong Palearcticdabbling ducks on Balticcoastal RussellThorstrom for reviewingearlier drafts of the manuscript.We wetlandsand comparisonwith the Nearctic.Oikos 69: 295-303 thank the Directiondes Eaux et For6ts,Commission Tripartite, Palmer R 1962. Handbookof the NorthAmerican Birds, Vol. 1: Projet Masoala, and ANGAP for their collaborationwith the Loonsthrough Flamingos. Yale University Press, New Havenand PeregrineFund project in Madagascar.We also thank Russell London Thorstromand LilyArison Ren6 de Rolandfor theirhelp in the field Pimm SL and Pimm JW 1982. Resourceuse, competitionand and the Peregrine Fund project and staff for their logislic support resourceavailability in Hawaiianhoneycreepers. Ecology 63: and collaboration.Finally, we are grateful to Steve Goodman for 1468-1480 reviewingthe manuscriptand for his insightfulcomments. Powzyk J 1995, Exceptionalobservations in MantadiaNational Park, Working group on Birds of the Madagascar Region References Newslefter5(2): 4 SAS 2001.SAS User'sGuide: Statistics. 8.1 Edition.SAS institute. Andriamasimanana RH, RabenandrasanaMN, Raminoarisoa Cary,NC, USA VTHS, Virginie MC, Ratelolahy FJ and Rakotonirainy OE Shapiro SS and Wilk MB 1965.An analysisof variancetest for nor- 2001.Effets de la fragmentationde la forethumide sur les popu- mality(complete samples). Biometrika 52(3&4): 591-611 lations d'oiseaux et lemuriens dans le corridor [.4antadia- '18-22 Sokal RS and Rohlf FJ 1995, Biometry:The Principlesand Zahamena.Lemur News 6: Practice of Statisticsin Biological Research. 3d edn. WH Benson CW Colebrook-RobjentJFR and Williams A 1976. Freemanand Company,New York Contributiona l'ornithologiede Madagascar.L'Oiseau et RFO46: Thompson PM and Evans Ml 1992. The threatenedbirds of 209-242 Ambatovaky Special Reserve, Madagascar. Bird Conservation Biggs HC, Biggs R and Kemp AC 1978.Measurements of raptors. lnternational2: 221-237. In: Kemp AC (ed) Proceedingsof the Symposiumon African Thorstrom R, AndrianarimisaA and Ren6 de Roland L-A 1997. Predatory Birds. pp 77-82. Northern Transvaal Ornithological Weatherat AndranobeField Station, western Masoala Peninsula. Society,Pretoria In: Watson RT (ed) Masoala Project. Progress Report lV pp Collar NJ, Crosby MJ and Stattersfield AJ 1994. Birds to Watch 127-130.The PeregrineFund, Boise, ldaho 2: The World List of Threatened Birds. Birdlife Conservation Thorstrom R and Lind J 1999. First nest description,breeding, SeriesNo. 4, BirdlifeInternational, Cambridge rangingand foraging behaviourof the Short-leggedGround-roller Donque G 1972.The climatologyof Madagascar.In: BattistiniR Brachypteraciasleplosomus in lvladagascar.lbis 141: 569-576 and VindardR (eds)Biogeography and Ecologyin Madagascar. Thorstrom R and Watson RT 1997. Avian inventory and key pp 87-144.Junk Publishers,The Hague,Netherlands species of the Masoala Peninsula, Madagascar. Bird Evans Ml, Duckworth JW Hawkins AFA, Safford RJ, Sheldon ConservationInternational 7: 99-115 BC and Wilkinson RJ 1992. Key bird species of lvarojejy Strict Nature Reserve, Madagascar.Bird ConservationInternational 2: 201-220

ReceivedApril 2003, acceptedSeptember 2003 Editor:TB Oatley,standing in for WRJ Dean Osttich2004, 7 5( 1 &2) : 32-38 Copytight @ NISC fty LA I Pinted in SouthAtica - A rightsrcserwd OSTRICH /ssN 0030-6525 :'

ood niche segregationbetween the MalachiteKingfisher, Alcedo fa, and the Pied Kingfisher,Ceryle rudis, at LakeNokou6, B6nin

Roland Libois* and Arnaud Laudelout Unftede zoogaographiques,lnstitut de Zoologie,Universite de Liege, Quai VanBeneden, 22, &4020 Liege,Belgium " Conespondingauthor, email: [email protected]

Several species of occur on Lake Nokou6, southern B6nin, includingMalachite (Alcedo cristata)and Pied Kingfishers(Ceryle Here,we comparetheir diet and estimatethe degreeof overlapin food niche by analysingcon- tents of regurgitated collected near nesting sites of Pied Kingfishers or inside the nest chambers of Malachite Kingfishers. Characteristic skull bones were identified usinq a reference collection of local fish skeletons. Malachite Kingfishersfeed most on fish that occur around floating vegetation,mainly Kribia sp. (560/o),Hemichromis fascia- tus (28%\ and Sarotherodon (8%). lmportant differences were found between different pairs, and between adultsand nestlings,the latter fed almost exclusivelyon Kribia sp. Larger fish are fed to nestlingsthan are eaten by the adults.Pied Kingfishersprey upon !ifferentfish species,some of lhem beingcaught in the pelagicregion of the lake,par- ticularlyclupeids taken by hovering. comparisonwith MalachiteKingfishers, Pied Kingfishersfeed on a wider diversityof prey,and take largerfish, so that the d overlapbetween the speciesis relativelylow (O = 0.181).

lntroduction

The MalachiteKingfisher (A/cedo crsfafa) is one of the numerous,the mostwidespread and the mostlikely to com- common kingfishersin Africa,and is abundantin pete with the MalachiteKingfisher, as observed by Reyer et equatorialand subtropicalsavannas, where it inhabits a/. (1988). lt thereforeseems of interestto comparethe or papyrusfringes and the bank vegetationof ponds,lakes composit;onof the diet of both speciesin the same habitat and rivers(Bannerman 1953, Fry ef a/. 1992). Throughout its nd to analyse the extent of overlap in their respectivefood large distributionarea - almostthe largestamong African kingfishers- it is easilyobserved, hunting for prey from a this paper we report on variationsin the food of low perch on small branchesor emergentrocks. However, ite Kingfishersin relationto localitiesof nest sites its feeding habits have been rarely studied. Bannerman ano age of nestlings.Detailed results of analysesof the (1953)and Bouet (1961)mention small fish and freshwater food Pied Kingfishers are discussed elsewhere crustaceans(crabs and prawns) as its main food. Van Libois2003). Someren(1956) and Newman(1974) found the diet domi- natedby aquaticinsects, mainly damselflies and dragonflies Study area (nymphsand adults)as well as water beetles(Dytiscidae) and water boatmen(Notonectidae) but also includingsmall SouthernB6nin a subequatorialclimate zone (Figure 1 ), fish (e.9. Barbus sp., Tilapia sp.), frogs, tadpoles and occa- with a high midity (77-93%) and a high mean sionally small lizards and terrestrialinsects such as monthly temperatu ranging from 22.4-32.9'C. Annual grasshoppersand mantises.On a pollutedriver where there rainfallis about 1 distributedover a long rainysea- were no fish, Meadows(1977) observed an individualfeed- son from March/Aprilto and a short rainvseason from ing exclusivelyon insects (water beetles,water boatmen Septemberto iya 1980). and adult Odonata)during a period of at least 47 days. Lake Nokou6 (6"23'-6 . 2'22'-2'33'E\is a shallow Reyer(in Fry et al. 1988)reported an exclusiveconsumption lagoonnot exceeding2.50m i .In 1990and 1991, its of frsh (Tilapia grahami) and estimated the daily require- meandepth ranged from 1.07m the end of the drv season mentsat 15-20 small fish,30-40 when nesting,and 60 or (April)to 1.72m during the Seotember).lts waters morefed dailyto five nestlings. are relativelyturbid, especiallyd the floods: Secchi Manyfish-eating birds occur on the lagoonsand associ- deothvaries between 50-120cm in V in the vicinityof the ated swamps of southern B6nin, including cormorants, studysites. Salinity also fluctuates from 25-30m9 Fl egrets, herons, terns and five kingfishers; Pied (Ceryle in April-June to 0-5mg F1 in Aug ber (Laleye rudis), Woodland (Halcyon senega/e/lsis), Giant 1995).lts north-westernedges correspond the deltaof the (Megaceryle maxima), Malachite and African Pygmy-king- River56, occupiedby villagesbuilt on piles, fields fisher (/spldrnaplcfa) (Schockert1998). Among these poten- and Paspalum vaginatummeadows. lial competitors,the Pied Kingfisheris probablythe most Thefish community comprises at least78 from