J. Field Ornithol., 63(3):276-285 A COMPARISON OF THREE TECHNIQUES FOR ANALYZING THE ARTHROPOD DIET OF PLAIN TITMOUSE AND CHESTNUT-BACKED CHICKADEE NESTLINGS PAULA K. KLEINTJES AND DONALD L. DAHLSTEN Divisionof BiologicalControl Universityof California Berkeley,California 94720 USA Abstract.--Photography, fecal sacand gut analysiswere comparedfor their effectivenessin quantifying the compositionof arthropod prey in Plain Titmouse (Parus inornatus)and Chestnut-backedChickadee (P. rufescens)nestling diets. Photographyproduced the most quantitative and taxonomicinformation. Fecal sac and gut analysiswere less reliable for quantitative and taxonomicwork but were adequatefor determiningthe presenceof prey items. Of the prey identifiedon film, the Plain Titmouse diet containeda large percentage of Lepidopteralarvae (88%), whereasthe Chestnut-backedChickadee diet containedlarge percentagesof pamphiliid sawfly larvae (63%) and rhaphidophoridcamel crickets(17%). Lepidopteralarvae and Orthoptera were the mostcommon prey in both the Plain Titmouse fecal sacsand guts, whereasOrthoptera and Hymenopterawere the most abundantprey in the Chestnut-backedChickadee fecal sacsand guts. Compared with either fecal sac or gut analysis,photography was consideredto be the mosteffective and completemethod for determiningthe diet of cavity nestingyoung. COMPARACI•)N DE TRES Tl•CNICAS PARA ANALIZAR LA UTILIZACI(•N DE ARTR•PODOS EN LA DIETA DE PICHONES DE PARUS INORNATUS Y P. RUFESCENS Sinopsis.--Last•cnicas de fotografia,anAlisis de los sacosfecales y anAlisisde contenido estomacal,fueron comparadas para determinarsu efectividaden cuantificarla composici6n de artr6podoscomo parte de la dieta de pichonesde Parusinornatus y P. rufescens.E1 m•todo fotogrAficoprodujo la mejorinformaci6n taxon6mica y cuantitativa.E1 anAlisis del contenido estomacaly de lossacos fecales, fue menosconfiable para anAlisistaxon6mico y cuantitativo, pero result6 adecuadopara determinar la presenciade presasparticulares. De las presas identificadasen la dietade P. inornatus,las larvas de lepid6pterosresultaron ser las de mayor consumo(88% de la dieta), mientras queen la dieta de P. rufescenspredominaron las larvas de pamfilidos(Hymenoptera) con un 63% y grillos(Orthoptera) con un 17%. Las larvas de lepid6pterosy losort6pteros resultaron ser la presamAs comfin en lossacos fecales y los contenidosestomacales de individuosde P. inornatus,mientras que los ort6pterosy los himen6pterosresultaron ser las presasmAs comunes en los sacosfecales y el tracto digestivo de P. rufescens.AI compararselos tres m•todosentre si, la fotografiaresult6 ser el m•todo mAscompleto y efectivopara determinar la dietade pichonesde avesque anidan en cavidades. Numerous methodshave been used to assessthe compositionof insec- tivorousbird diets. Techniquesused for passerinebirds have included visual observations,emetics, artificial nestlings,ligatures, gut contents, fecal contentsand automatic photography (Calver and Wooler 1982, Otvosand Stark 1985, Rosenbergand Cooper1990, Royama 1970). Such techniqueshave been useful for assessingnestling and adult nutrition, foragingbehavior, inter- and intra- specificcompetition for food,and the impact of avian predatorsupon arthropodpopulations (Cowie and Hins- ley 1988, Crawford and Jennings1989, Tinbergen1960). Rosenbergand Cooper (1990) recently reviewedthe advantagesand 276 Vol.63, No. 3 AnalyzingArthropod Diet [277 disadvantagesof variousapproaches to avian diet analysis,but their eval- uations were based upon previous studiesusing single methods. Few studieshave simultaneouslycompared the resultsof more than one tech- nique to quantify the compositionof a birds' diet (but seeJenni et al. 1990). Our objectivewas to comparethree methodsof quantifyingar- thropodcomposition of nestlingdiets. The studywas unique for we used empiricalevidence to make a comparisonof fecalsac analysis, gut analysis and automatic photography. We examinedthe nestlingdiet of two insectivorous,secondary cavity- nestingbirds, the Plain Titmouse (Parus inornatus)and the Chestnut- backedChickadee (P. rufescens).Both speciescoexist in the California coastaloak woodlands (Dixon 1954). Their behavioral interactionshave beenthe subjectof previousstudies but little has beendone on the com- positionof their nestlingdiets (Dixon 1954, Hertz et al. 1976, Root 1964, Rowlett 1972). MATERIALS AND METHODS The study was conductedon the northeasternslope of the Berkeley Hills in the East Bay Municipal Utilities District, Contra Costa County, California, during the monthsof April and May, 1988-1990. The slope encompassesmature standsof planted Monterey Pine (Pinus radiata), adjacentto standsof Coast Live Oak (Quercusagrifolia) woodland. The majority of understoryvegetation consists of poisonoak (Toxicodendron diversilobum)and blackberry(Rubus ursinus). Fifty-three artificial nestboxeswere establishedin this area in 1978 (Gold and Dahlsten 1989). They were randomlyplaced along a seriesof trails at intervals of 25-50 m, 1.5 m above the ground. The nestboxes were constructed of sawdust and cement and contained a removable front (Schweglerand Sons,Munich, Germany). The diameterof the nesthole was 33 mm with a box heightof 25 cm and diameterof 11.5 cm. Nestboxes were checkedweekly from March-May and increasedto bi-weekly once egg laying had begun. During the Springs of 1988 and 1989, we collectedfresh fecal sacs from young during routine box checks.Once nestlingswere considered old enoughto be handledsafely (about 8 d) we gently pressedthe sides of the cloacauntil a sac was produced.The proceduredid not always guaranteea sacso we obtainedadditional fecal sacswhen handlingnest- lings during banding. We consideredsamples independent as the sacs were collectedat variable times and datesthroughout each nestingseason from a minimum of three nests and with a minimum of 24 h between collectionsfrom the samenest. We placedeach fecal sac in a small labelled plastic vial. Nest disturbancedue to weather, predation, human interferenceor unknown factors produceda number of abandoneddead nestlings.We examinedthe gut contentsof the dead nestlingswith the understanding that some digestionof gut contentswould have continuedafter death. The nestlingswere collectedwithin 24 h of deathexcept for two nestlings 278] P. K. Kleintjesand D. L. Dahlsten J.Field Ornithol. Summer 1992 collected within 48 h. Their bodies were frozen 1-2 h after collection and their gizzard and intestineswere later removedand placed in vials of 70% ethyl alcohol. Fecal sacswere thawed and placed in plastic petri dishes(60 x 12 mm). Each sacwas teasedapart in 70% ethyl alcoholwith the hardened nitrogenouswastes removed. All identifiable pieceswere sorted,coded and placedin a 1-dr shell vial pluggedwith cotton.Each shell vial was then was placed in a larger screw cap vial containingethyl alcoholand unidentifiable material. Gizzard and intestine contents were flushed out with ethyl alcohol and treated as above. We identifiedarthropod body parts to the lowesttaxonomic level pos- sible. Identificationswere made by comparingsamples to insectparts from a referencecollection and to referencesin the literature (Borror et al. 1981, Essig 1926, Furniss and Carolin 1977, Peterson1948, Ralph et al. 1985). Identifiable parts were then matchedto approximatethe numberof insectsoccurring in eachsample (i.e., 2 Lepidopteramandibles -- 1 larvae, 3 cicadalegs -- 1 cicada).As the majority of insectpieces were too fragmentedto be countedand identifiedto Order, we calculated percent compositionof the contentsfrom the identifiablematerial only (Table 1). These percentagesare not true representationsof the diet compositionof Plain Titmouseand Chestnut-backedChickadee nestlings but are proportionsof the identifiedmaterial that remainedintact through- out the digestiveprocess. Extraneous material was notedbut not counted. On averageit took 30-60 min to processeach individual fecal or gut sample. This was dependentupon observerexperience and amount of contents.A 2 x 5 contingencytable (x 2 testa = 0.05, StatviewII, Abacus Concepts,Inc.) was used to compare relative frequenciesof the prey Orders (Orthoptera, Homoptera, Lepidoptera,Hymenoptera and other Orders (in total lessthan 12%)) amongthe two methods. To assessadequacy of samplesize of numbersof fecal sacsand guts, we useda methodfor estimatingprey diversitydescribed by Pielou (1975) and applied to stomachanalysis by Sherry (1984). We randomizedthe individual fecal sacor gut sampleswithin our 1988-1990 collectionsfor each speciesof bird. We then calculatedthe diversity of prey in each individualfecal sac or gut startingwith sample1, then pooledthe contents of eachsuccessive sample 1 and 2, then 1, 2, and 3 and so on up to the total number of samplesin the collection.We usedthe Brillouin diversity index (H) at each step:H = (1/N)ln(N!/nl!, n2! . nt!) where there are nl, n2 ... nt prey items in each of x different prey categories(insect Orders), with N total prey times per cumulativesample. If enoughfecal sacsor gutswere collected,saturation curves of prey diversitywould result in plateausat X number samplesbecause additional gut or feceswould add little dietary informationto increaseprey diversity. We obtained photographswith a camera box recording apparatus patternedafter Royama (1970). Each nestboxwas fittedwith a Minolta• super 8 mm movie cameraand Vivitar• flash unit placedin the back of the box (Dahlsten and Copper 1979). Photocellswere placedon opposite sidesof the entrance hole. As an adult entered the box, the photocell Vol.63,