j. Raptor Res. 28(3):154-157 ¸ 1994 The Raptor Research Foundation, Inc.

NEST-BOX VERSUS NATURAL-CAVITY NESTS OF THE EASTERN SCREECH-OWL: AN EXPLORATORY STUDY

FREDERICK R. GEHLBACH Departmentof Biology,Baylor University, Waco, TX 76798 U.S.A.

ABSTI•½T.--The use of nest boxesversus natural tree cavitiesby easternscreech-owls (Otus asio) in centralTexas was exploredfor 9 yr. Box sitesresembled natural-cavity sites vegetatively and physically exceptthat the boxeswere positionedsomewhat lower. The three box sizesprovided spanned the size rangeof natural cavitiesused by screech-owls.Box sizemade no differencein nestsite selection,clutch size, or nestingsuccess. Three kinds of wood usedin box constructionmade no differenceeither. The largestboxes were usedmore often for replacementnests, and the smallestones tended to crowdbroods, contributingslightly to mortality and early fledging.Overall, nestboxes produced data on frequencyof use, clutch size, and fledgling productivity that were equivalent to data from natural cavities. KEY WORDS: breedingsuccess; eastern screech-owl; nest boxes; nest-site use; Otus asio;tree-cavity nests.

Nidos en cajasanideras versus cavidades naturales de Otusasio: un estudioexploratorio RESUMEN.--Seexplor6 por nueveaftos el uso de cajasanideras versus cavidades naturales ubicadas en firbolespor parte de Otusasio en el centro de Texas. Las cajasanideras se asemejaronalas cavidades naturalestanto vegetativacomo fisicamente, excepto que las cajasfueron ubicadasalgo mils abajo.E1 tamafiode las cajasutilizadas correspondla al espectrode tamafiosde las cavidadesnaturales usadas por O. asio;no se registrarondiferencias en la selecci6nde sitio de nidificaci6n,tamafio de la nidada o •xito de lospolluelos. Tres tiposde maderasse utilizaron en la construcci6nde las cajasy tampocose registraron diferencias.Las cajasmls grandesfueron utilizadasa menudopara reemplazode nidosyen las mils pequefiasse tendla a agrupar las crlas, aumentandoligeramente la mortalidad y un desarrolloprecoz. En general,las cajasanideras producen datos sobre la frecuenciade uso,tamafio de nidaday productividad de volantonesque fueron equivalentesa los obtenidosdesde cavidades naturales. [Traducci6n de Ivan Lazo]

Many raptor researchersemploying artificial owls, only the boreal or Tengmalm'sowl (Aegolius structures like nest boxes or nesting platforms do funereus) has been studied in a somewhat similar not utilize or report simultaneouscomparisons with manner; clutchsize and number of fledglingsof that natural structuresor distinguishsite suitability or specieswere determinedin both nestboxes and nat- availability in studying habitat selection.Moller ural cavities but not in controlled fashion (Korpi- (1989, 1992) and Clobert and Lebreton (1991) note mSki 1984). possibledeficiencies in suchstudies using nest boxes, METHODS includingthe potentialfor greaternesting success in The study was conductedin 135 ha of Woodway, a boxes than natural cavities. Here, I focus on the suburb of Waco, TX, U.S.A., during 1967-75. This area simultaneous comparison of eastern screech-owls had 508 human residents/km2 and 25.9% green space (Otusasio) nesting in boxesand natural tree cavities averaging 327-1504 trees/ha in lawns and wooded ra- to discover if there was a difference in the use of and vines,respectively. Eastern screech-owls and the natural tree cavitiesthey usedwere locatedby mappingcavity- breedingsuccess in the two site types. advertisementsongs and making cavity inspectionsin Field studiesof birds may need to be conducted Decemberthrough early March before nestingbegan. I over severalyears during which investigativebiases continuedto searchfor used and unusedbut apparently canmultiply errors.Consequently, I testedfor biases suitable cavities (as large or larger than used cavities; McCallum and Gehlbach [1988]) until the cumulative that might result from using nest boxes prior to number found versus cumulative search effort indicated conductinga 16-yr populationstudy of the eastern that essentiallyall were known. screech-owl(Gehlbach 1994). Amongcavity-nesting Nine nest boxes were constructed,three each of exterior

154 SEPTEMBER 1994 SCREECH-OWL BOX AND CAVITY NESTS 155 plywood,solid pine, and solidcedar (1.9-cm-thick wood), RESULTS and were painted dark brown on the outside.Cavity size has beenshown to influenceclutch size in owls (Korpim5ki Nest Cavities and Habitat. Of the 23 suitable 1985), and to test its affect on screech-owlsI built one box of each wood type with 225, 400, and 625 cm2 bottoms. natural cavities that I found, eastern screech-owls All boxes had a 6.8 cm entrance hole 25 cm above the box nestedin 15 that were deeper(>25 cm) with larger bottom.The bottomareas of theseboxes spanned those of floors(> 10 cm minimum dimension)and had small- natural cavities.Depth of the boxeswas the averageof er entrances (<15 cm maximum dimension) than 12-58-cm-deep natural cavities,and the entrancediameter was the mean minimum dimension of the natural entranc- the others(MANOVA F = 2.9, P = 0.04). The used es. siteswere 3.7 m (SD = 1.6) above ground in nat- Box locationsdepended on landowner permissionand urally rotted, hollow limbs or tree trunks. At least wereplaced 70-300 m fromnatural cavities used by screech- 10 of thesecavities had beenenlarged by fox squir- owls and other boxeson straight tree trunks. Trunk di- rels. The red-bellied woodpecker(Melanerpes car- ameters at these positionswere equal to or larger than box width. Boxes faced nine different directions and were olinus),the largestlocal woodpecker species, did not 3-4 m above ground. Five boxeswere paired with the excavatecavities large enoughfor screech-owls. closestpreviously used natural cavitiesmost like them in Paired box- and natural-cavitysites were 72-280 orientation, height, and volume. The other four could not m apart and similar in their vegetationaland phys- be paired becauseit requiredseveral years to discoverall natural sitesused by screech-owls.I evaluated only first- ical features (MANOVA F = 1.1, NS). However, nest data, sincereplacement nests are so different (Gehl- the boxes tended to be lower (• -- 3.1 vs. 3.7 m, bach 1994). ANOVA F = 3.5, P = 0.06) and in smaller diameter A 75-m, 5-point,20-quarter transect, randomly aligned trees (• = 26.7 vs. 32.2 cm, F = 2.4, P = 0.09). througheach box and cavitytree, was employedto assess While all nine nestboxes were placedlower in small- vegetationalfeatures according to Gehlbach(1988, 1994). These and physicalmeasurements were made before the er trees than the 15 used natural cavities (ANOVA boxeswere positioned,so that box sites would resemble F = 3.2, P < 0.05), both site-typeswere equally natural cavity sites. Measurementsand synthesescon- distributedamong the sevenmost commontrees of cernedtree and shrubdensity, height, and diversity,the the canopy(X 2 = 1.7, NS). Cedar elm (Ulmus cras- evergreenfraction, canopy coverage, tree speciesrelative sifolia) was the dominant tree and had the most nat- importance, nest tree diameter, nest height and bottom area, and distanceto the nearesthouse, permanent water, ural cavities(48%) and boxes (44%). and suitable/availablecavity or box. The frequencieswith which screech-owlsused Severalhandfuls of dried leaveswere placedin the bot- paired boxesand natural cavitiesfor nestingwere tom of each box to simulate an old fox squirrel (Sciurus quite similar, and similar to all nine boxesand nine niger) nest becauseall but one of the usednatural cavities had them. Thereafter, box and cavity contentswere un- cavitieswith at least 5-yr records(Table 1). Type disturbedexcept in a minor way during nestinspections. of wood and box size made no difference in nest box Fox squirrelsand other cavity usersand nest predators selection;useage relative to availability rangedfrom were not disturbedduring the weeklysurveys of nestcon- 60% in pine and 67% in small boxesto 70% in cedar tentswhich extendedthrough June, and nest debriswas not removed between years. and 73% in large boxes(X 2 = 1.7, NS). Multifactor box versus natural-cavity environments, Preferred nesting habitat was distinguishedas high- versuslow-use boxes,and used versusunused nat- having more evergreensin the canopy, lower tree ural cavitieswere evaluatedwith multivariateanalyses density, closer alternate nest site, and lower shrub employingtransformed data. Individual parameterswere densityin first-to-lastorder of importance(F-- 4.0, then testedwith univariate analysisof variance.Use and reproductivedata for the box/cavitypairs were assessed P < 0.07). This combination of features describesa with Wilcoxon signed-rankstests, while unpaired data shady,park-like landscapewith large, cavity-prone were subjectedto Mann-Whitney U-tests.Spearman rank trees at low densities. These habitat characteristics correlationswere used to test relationshipsbetween box were identifiedby subjecting12 high- and 12 low- sizeand nestcontents. Chi-square analyses were madeof nests in boxes versus cavities, use of and results from the use nest sites(employing the 60% median use rate different box sizesand woodtypes, and box versuscavity of all nine boxesplus 15 cavities)to stepwisedis- dispersionsamong dominant trees. criminant analysis. Means,standard deviations, sample sizes (if notobvious Reproduction.Clutch size and numberof fledg- from the designs),and exactprobabilities are givenexcept lings/clutchin the nestboxes were statisticallylike for those>0.10, which I considernon-significant (NS) in two-tailedtests. Probabilities of <0.10 havepotential bi- those in natural cavities (Table 1). Clutch size in ologicalmeaning in view of my confirmatorystudy (Gehl- the 29 nests in boxes were not related to bottom area bach 1994). of the box (rs = 0.17, NS) unlike the situation in 156 FREDERICK R. GEHLBACH VOL. 28, No. 3

Table 1. Eastern screech-owluse of and breedingsuccess in the study of eastern screech-owls.By contrast, xn nest boxes and natural cavities, 1967-75. Means ___ KorpimSki(1984) foundmore eggsand fledglings standarddeviations and samplesizes (in parentheses)are of boreal owls in boxes than natural cavities, al- of first nestsonly. a thoughhe did notstudy both site-types concurrently. Southern(1970 and pets.comm.) made same-season PARAMETERS comparisonsof cavity- and box-useby tawny owls CAVITIES NEST BOXES NATURAL (Strixaluco), but did not enumeratecavity-nest con- Site use (%) tents. He found that 33-75% of the owls used boxes Pairsb 64.2 + 13.0 (5) 75.8 + 19.1 (5) annually. Most studentsof cavity-nestingraptors Allb 70.5 +_15.3 (9) 77.5 + 16.2 (9) assumethat nest boxes provide natural data, even Successfulnests (%)c though KorpimSki (1984) suggestsotherwise (also see Gauthier 1988, Robertsonand Rendell 1990). Pairsb 67.8 +_21.2 (5) 81.1 + 12.4 (5) Allb 70.7 + 18.6 (9) 72.8 + 15.7 (9) The overall 67% userate of boxesin my studyis considerablyhigher than 4-13% valuesreported for Clutch size 3.9 + 0.5 (29) 3.8 + 0.6 (16) eastern screech-owlsby Van Camp and Henny Fledglings/eggs(%) 51.3 + 16.8 (29) 57.0 + 19.4 (16) (1975), McComb and Noble (1981), and Fowler and a Wilcoxon Z ( 0.9, NS, for all paired data comparisons;Mann- Dimmick (1983). Perhapsthis is becauseI placed Whitney U ( 19, NS, for all unpaireddata comparisons. b Pairs= fiveboxes each paired with a naturaltree cavity having boxesin sitessimilar to natural cavitiesafter making similar environmentalfeatures; all = the paired boxesand cavities environmentalmeasurements. The high usewas not plus four more boxesand four cavitieswith at least 5-yr records. due to an unusuallydense or protectedsuburban c At leastone chickfledged. population,or to a scarcityof suitablenatural cav- ities which outnumbered the boxes. In fact, a much sparserrural populationwith twice as many natural borealowls (KorpimSki 1985). But when chicklosses nestsites per breedingpair of owlshad box-userates occurred,they were 21% higherin the smallestboxes averaging61% concurrentlywith a 70% usein sub- versuslarger onesand slightly more frequent there urbia (Gehlbach 1994). (31% vs. 20%; x2 = 1.7, NS). Also, chicks fledged Nest boxesclosely matching used natural cavities somewhat sooner from smaller boxes (rs = 0.39, P furnish vital information on nest-site selection and -- 0.06). Successfulnests tended to be more frequent the influenceof cavity size on chick mortality and in paired natural cavitiesbut not significantlyso, fiedging. Korpim/iki (1984, 1985) obtainedresults and the larger samplesof nine siteseach were es- different from mine using different size boxesbut sentiallyalike (Table 1). his studydid not attemptto differentiateamong the Possible Inspection Biases. Screech-owlsused variety of available natural sites.When boxesare only 65.6% (SD = 18.6) of the availablepaired boxes usedto compensatefor the scarcityof natural tree and tree cavitiesin 1968-71 comparedto 78.1% (SD cavities to manage owl populations(e.g., Saurola = 20.9) in the following 4 yr (Wilcoxon Z -- 2.0, P 1989), I suggestthat one simulatethe range of nat- = 0.04). Demand for nestingspace was apparently ural nestcavities, since cavity size can influence some not a factor because the mean use of 20 suburban aspectsof reproduction. boxeswas not significantlydifferent over the next Comparingenvironmental features used by a spe- 16 yr despitea nearly two-fold flux in population ciesversus randomly selected environmental features density (:e = 75.6%, SD -- 16.1; Mann-Whitney U is incorrectly called habitat selection(e.g., Cody = 11, NS; Gehlbach 1994). Moreover, productivity 1985). Habitat that is suitablefor a speciesmust be in the paired boxesplus cavitieswas only 45.9% (SD distinguishedfrom all availablehabitat. In fact, some = 39.3) in 1968-71 but rose to 55.6% (SD = 40.7) randomlychosen sites may be unsuitable,and rap- in 1972-75 (Wilcoxon Z = 1.4, NS) and remained tors may not use all possiblesites unless forced to about the same in the following 16 yr (53.8%, SD by population pressure.Studies of habitat selection = 15.6; Mann-Whitney U = 19, NS). must include environments that have been and could be used besidesthose now used, and thus furnish a DISCUSSION spectrumof conditionsfor selectionby birds under Clearly, nest boxesare essentiallyequivalent to variouspopulation densities. Measurements are based natural tree cavitiesand henceare a legitimate tool on criteria like the mean-minimum dimensions of SEPTEMBER 1994 SCREECH-OWL Box AND CAVITY NESTS 157

past and current use (McCallum and Gehlbach Amosand F.R. Gehlbach[EDs.], Edwards plateau veg- 1988). etation:plant ecologicalstudies in central Texas. Bay- Although the nest-boxmethod proved to be valid lot Univ. Press, Waco, TX U.S.A. for studying eastern screech-owls,I believe boxes 1994. The easternscreech owl: life history, attracted deleterioushuman attention early in my ecology,and behaviorin the suburbsand countryside. Texas A. and M. Univ. Press,College Station, TX study when they were novel. In addition,my own U.S.A. learning of inspectiontechniques may have inad- KORPIM•.KI,E. 1984. Clutch size and breedingsuccess vertentlyfocused on the lower and hencereadily of Tengmalm'sowl Aegoliusfunereusin natural cavines accessedboxes despite attempts to treat all paired and nest boxes. Ornis Fenn. 61:80-83. sitesequally. This couldhave produced the disparate 1985. Clutch size and breeding successin re- early- versuslate-study results, because frequency lation to nest-box size in Tengmalm's owl Aegohus of site use and productivitydid not changeappre- funereus. Holarctic Ecol. 8:175-180. ciablyduring populationflux over the next 16 yr. MCCALLUM, D.A. AND F.R. GEHLBACH. 1988. Nest- Thus, I advisethat populationinvestigations be pre- site preferencesof fiammulated owls in western New Mexico. Condor 90:653-661. cededby explorationsthat refine all investigative MCCOMB, W.C. AND R.E. NOBLE. 1981. Nest-box and approaches,not just the use of substitutehabitats. natural cavity use in three midsouth forest habitats.J. Wildl. Manage. 45:93-101. ACKNOWLEDGMENTS MOLLER,A.P. 1989. Parasites,predators and nestboxes I thankthe manypersons mentioned in Gehlbach(1994), facts and artefacts in nest box studies of birds? Oikos especiallyGeorge Newman and Claudia and the late C.W. 56:421-423. Davis, whosenest boxesI watched beforedesigning this 1992. Nest boxes and the scientific rigour of study.Jeffrey Marks and anonymousreviewers provided experimental studies.Oikos 63:309-311. constructivecomments on the manuscript. ROBERTSON,R.J. AND W.B. RENDELL. 1990. A com- parisonof the breedingecology of a secondarycavity- LITERATURE CITED nestingbird, the tree swallow (Tachycinetabicolor) •n CLOBERT,J. ANDJ.D. LEBRETON. 1991. Estimation of nestboxesand natural cavities. Can. J. Zool. 68:1046- demographicparameters in bird populations.Pages 1052. 75-104 in C.M. Perrins, J.D. Lebreton, and G.J.M. SAUROLA,P. 1989. Breeding strategyof the Ural owl, Hirons [EDs.], Bird population studies:relevance to Strix uralensis.Pages 235-240 zn B.-U. Meyburg and conservationand management.Oxford Univ. Press, R.D. Chancellor lEDs.I, Raptors in the modern world. New York, NY U.S.A. World Working Group on Birds of Prey. Berlin, Ger- COPY, M.L. [ED.]. 1985. Habitat selection in birds. many. Academic Press, New York, NY U.S.A. SOUTHERN,H.N. 1970. The natural controlof a pop- FOWLER,L.J. AND R.D. DIMMICK. 1983. Wildlife use ulation of tawny owls (Strix aluco).J. Zool. (Lond.). of nest boxes in eastern Tennessee. Wildl. Soc. Bull. 11: 162:197-285. 178-181. VAN CAMe, L.F. AND C.J. HENNY. 1975. The screech GAUTHIER,G. 1988. Factors affecting nest-box use by owl: its life historyand populationecology in northern buffieheadsand other cavity-nestingbirds. Wildl. Soc. Ohio. U.S. Fish and Wildl. Serv. N. Am. Fauna No. Bull. 16:132-141. 71. GEHLBACH, F.R. 1988. Forests and woodlands of the northeasternBalcones Escarpment. Pages 57-77 in B.B. Received3 December1993; accepted18 July 1994