The Condor91:166-776 Q The CooperOrnithological Society 1989
WINTER CACHING ECOLOGY OF DECIDUOUS WOODLAND BIRDS AND ADAPTATIONS FOR PROTECTION OF STORED FOOD ’
DANIEL R. PETIT AND LISA J. PETIT Department of Zoology, Universityof Arkansas,Fayetteville, AR 72701
KENNETH E. PETITE Department of BiologicalSciences, Kent State University,Kent, OH 44242
Abstract. We quantified cache-site selection by White-breasted Nuthatches (Sitta caro- linensis),Tufted Titmice (Parus bicolor), and Black-capped (P. atricapillus) and Carolina (P. carolinensis)chickadees storing sunflower seeds in four deciduous woodlands during winter, 1983-1988. All speciesexhibited significant interpopulational variation in caching ecology.This may be related to, among other factors, differencesin the floristic and physiog- nomic characteristicsof vegetation on each site. Nuthatches, titmice, and chickadeeseach demonstrateda distinct pattern of seedstorage that, in general,corresponded with published reports of foraging ecology. Those differencesdid not appear to be related to the need for protection of stored food from snow and ice cover, but rather for concealmentfrom potential cache robbers. In protecting those cache sites, titmice and chickadeesrelied mainly upon inaccessibilityby placing seedson small branchesand twigs, frequently in the outer portion of tree canopies.Nuthatches, on the other hand, usedboth difficult-to-reach (far under bark or deep in furrows)and cryptic(covered with bark or other material)locations to protect storedseeds. Disparity between parids and the nuthatchin cachingecology may be due to morphologicaldifferences and by the number of cacherobbers likely to encountereach speciescache’ sites. Key words: White-breastedNuthatch; Sitta carolinensis;Tufted Titmouse;Pants bicolor; Black-cappedChickadee; P. atricapillus; Carolina Chickadee;P. carolinensis;caching; win- ter. INTRODUCTION three of which may be important to small pas- Food-storing behavior of passerineshas been ex- serinesthat store seedsduring winter at northern amined in detail, especially aspectsof spatial dis- latitudes: (1) cacherobbers may steal food before tribution (e.g., Cowie et al. 1981, Sherry et al. an individual can relocate its stored food; (2) 198 1, James and Verbeek 1985) and recovery environmental factors,such as snow and ice, may (e.g., Tomback 1980, Sherry et al. 198 1, Vander prevent birds from retrieving their caches;and Wall 1982) of cached food in laboratory or con- (3) an individual may lose a stored food item by trolled field situations. Relatively less detailed forgetting its location. study has been made of the actual cache sites Most studies of avian caching behavior have used by free-ranging populations and the factors been conducted with pat-ids in Europe and cor- that may affect cache-site selection. Given that vids in North America. Yet, food storing in win- stored food can potentially influence fitness of ter bird assemblagesin eastern North America an individual (e.g., through improved reproduc- is common and, surprisingly, no one to date has tive successor through increased survival during described in detail the caching ecology of many periods of food shortage;Roberts 1979), ability of those species.The purposesof this paper are to prevent cache loss is crucial to the evolution to (1) describe quantitatively caching behavior of food-storing behavior (Andersson and Krebs of several winter populations of White-breasted 1978, Smith and Reichman 1984). Vander Wall Nuthatches (Sit& carolinensis), Tufted Titmice and Smith (1987) listed five sourcesof cacheloss, (Parus bicolor), and Black-capped (P. atricapil- lus) and Carolina (P. carolinensis) chickadeesin Ohio and Arkansas, and (2) assessmechanisms by which those speciesprotect stored seedsfrom I Received26 September1988. Final acceptance24 May 1989. potential cache robbers. Most parids and sittids z Presentaddress: 348 ChurchSt., Doylestown,OH are found in mixed-species flocks during winter, 44230. typically storing seedsin closeproximity to other
[7661 CACHING ECOLOGY 767 individuals. Consequently, we postulate that sitions several times during each hour of obser- speciesshould store food in locations that cannot vation. Becauseof the intense caching behavior be found easily by other flock members. exhibited by woodland birds (e.g., see Cowie et al. 198 I), we could not monitor all birds at once, STUDY AREA AND METHODS but rather randomly chosea focal individual when We observed cachingbehavior of nuthatches,tit- it flew to the feeder. For eachcase of food storage, mice, and chickadeesat four locations: (1) Darby we recorded 12 characteristics associated with Creek Metropolitan Park, Franklin County, Ohio the cache site: (1) distance from feeder; (2) sub- (Darby); (2) Findley State Park, Lorain County, strate- substrateon which seedwas cached (tree Ohio (Findley); (3) Kent State University cam- species,dead branch on ground, or grape vine); pus, Portage County, Ohio (Kent); and (4) Lake (3) height; (4) microsubstrate-if on tree, twig Fayetteville Environmental Center, Washington (5 1 cm diameter at cache site), branch (> 1 cm County, Arkansas (Fayetteville). Each study area at cache site), or trunk, (5) diameter of micro- was part of a larger (> 10 ha) woodland. White- substrate at cache site- I 1.0, > 1.0-2.5, >2.5- breasted Nuthatches and Tufted Titmice were 8.0, >8.0-15.0, >15.0-23.0, >23.0-38.0, present on all study sites,but the speciesof chick- >38.0-53.0, or >53.0 cm; (6) horizontal posi- adees differed between the two more northern tion-inner 33%, middle 33%, or outer 33% of sites(Black-capped Chickadee at Kent and Find- tree canopy; (7) tree diameter-diameter at breast ley) and those to the south (Carolina Chickadee height (same categories as [5]); (8) cache loca- at Darby and Fayetteville). Becauseof the taxo- tion-(a) furrow, crack, or rotted area; (b) under nomic, morphological, and ecological similarity bark; (c) end of broken branch or in a bud; (d) between the two chickadee species,we grouped between two branches; (e) leaf on tree; (f) other; (a priori) both speciestogether in some analyses. (9) side of branch-top (25%) side (50%), or bot- Birds were observed between 08:OO and 14:00 tom (25%); (10) direction on tree trunk-if cached on 4-l 2 days during December-February 1983- on trunk, the compass direction at the site of 1984 (Darby), 1985-1986 (Kent, Findley, and seed storage;(11) cover- whether an individual Fayetteville), and 1986-1988 (Fayetteville). We coveredthe cachewith bark, lichens,snow, leaves, did not attempt to restrict weather conditions or other materials; and (12) shell-whether the under which data were collected, but wind ve- seedwas shelled or unshelled when cached.Vari- locity was always low (Beaufort scale 52) and ables (9) and (10) were not compared acrosspop- temperatures between -4” and 15°C. ulations because of small sample sizes at some At each site, one 0.04-ha circular plot was ran- locations. Instead, those two variables were used domly positioned (15-60 m from feeder) in each only in assessingnonrandom placement of seeds. of four 90” arcs centered on the N, S, E, and W Binoculars (7 x 42) were used to obtain details compass directions. All tree and shrub stems of cache sites. within each circle were identified to speciesand Differences among populations and among measured (diameter at breast height; dbh). This species were assessedwith log-likelihood ratio provided a measure of vegetation structure and (G) tests for categorical variables and with one- composition on the four study areas. way analysis of variance (ANOVA) for contin- We observed caching behavior within a 70-m uous data. Probability levels 10.05 indicated radius ( 1.5 ha) circle centered on a single feeding significant differences. station (1.5 m above ground on a pole) stocked Because we dealt with small populations of with shelled (X length = 8.2 +- 1.7 mm SD, X unmarked individuals, we were compelled to re- maximum width = 4.4 f 0.7 mm; n = 10) and cord multiple observations for each individual unshelled (X length = 11.2 + 0.7 mm; x = max- and treat them as independent samples. If those imum width = 5.0 f 0.3 mm; n = 10) sunflower observations are statistically dependent, how- (Helianthus) seeds(approximately 50% of each ever, then pseudoreplication has occurred and type). Rarely (< 10% of observations for each our reportedcy-levels are tenuous (Hurlbert 1984). species) did birds fly farther than 70 m when However, in this study, we believe that multiple caching seedsand those observations were dis- samples per individual were not necessarilycor- carded from analyses. One to three observers related because,for every cache, birds flew back positioned themselves at predetermined com- to the original point source of food to gather passdirections from the feeder, shifting their po- another seed.Thus, with each trip, birds initiated 768 D. R. PETIT, L. J. PETIT AND K. E. PETIT
TABLE 1. Characteristicsof vegetationstructure for eachofthe four deciduouswoodlands in Ohio and Arkansas where caching behavior was observed.
Site Darby Findley Kent Fayetteville Vegetationcharacteristic R SD R SD R SD R SD
Stems/ha 4,800 934 8,119 2,138 5,606 1,493 4,050 679 Basal area (mz)/ha 45 7 38 10 50 20 50 25 Canopy height (m) 19 2 13 1 19 2 23 1 a new caching bout, and there was no obvious (n = 300), and chickadees(n = 5 12). Most (60%) constraint placed on birds due to their previous observations were collectedat Fayetteville, while caching location. Limited observations on sev- the remainder were approximately equally dis- eral color-marked titmice revealed no apparent tributed among Darby (15%) Findley (13%), and dependence among cache sites chosen on 20-30 Kent (12%). Numbers of caching observations consecutive caching bouts. (with sample sizesin parentheses)for eachspecies (White-breasted Nuthatches,Tufted Titmice, and RESULTS chickadees, respectively) at each location were: COMPARISON OF VEGETATION ON Darby (77, 83, 1 l), Findley (22, 23, 120), Kent STUDY SITES (41,36,75), and Fayetteville (263, 158,306). On All study sites were located in deciduous wood- all days at all feedersthere were two to four nut- lands with canopies 13-23 m tall and well-de- hatches, two to six titmice, and four to eight veloped understories (many saplingswith trunks chickadees.Figures 2-4 and Tables 2 and 3 sum- ~2.5 cm dbh), but tree size class distributions marize the caching ecology of nuthatches, tit- and tree speciescomposition varied among lo- mice, and chickadees. cations (Table 1, Fig. 1). Kent and Fayetteville White-breasted Nuthatch. Nuthatches cached were dominated by large (> 38 cm dbh) trees, most seedswithin 30 m of the feeding station (K whereas Darby and Findley had a better repre- = 20.3 f 14.4 m SD). Seedswere always taken sentation of medium-sized (8-38 cm) trees. singly and, if not already shelled, usually wedged When tree species composition was consid- into a bark furrow on a tree trunk and hammered ered, even more striking site-specificdifferences open with three to four blows. The entire seed became evident. On all sites stem densities were then was carried to a large-diametered (usually dominated by a miscellaneous collection of > 8 cm) trunk (69%) or branch (30%) and wedged understory shrubs and saplings, and each area under bark or in a furrow or crack in the bark. was unique in tree species composition when The seed was covered in about 50% of the ob- measured by basal area. Red (Quevcusrubru) and served cacheswith bark/rotted wood (81%), li- white (Q. alba) oaks were the dominant canopy chens (15%), snow (3%) or moss (1%). Cache speciesat Fayetteville, while white oak and, to sites averaged 8.0 (k5.6) m above ground, al- a lesser extent, red oak and white ash (Fraxinus though height varied from 1 to 23 m. Large (>23 americana) were well represented at Kent. (For cm dbh) white, red, and post oaks comprised this and all subsequent analyses, the “red oak” >75% of the substratesupon which seedswere group may contain not only Q. rubra, but also a stored. Nuthatches cached seedson bottoms of small [ < 5%] proportion eachof post [Q. stelluta], branchesmore than would be expectedby chance, pin [Q. palustris], and black [Q. velutina] oaks.) but they used compass aspects on tree trunks Together, shagbark hickory (Carya ovata) and randomly. white ash comprised >60% of the basal area at Tufted Titmouse. Titmice stored most seeds Findley, while these same species, along with within 40 m of the feeding station (X = 27.3 + black walnut (Jugluns nigru), were dominant 17.5 m). Only one seed was taken per trip and overstory trees at Darby. it was usually shelled before being stored (80% of the cases).When shelling seeds,titmice flew GENERAL CACHING ECOLOGY to a l- to 2.5-cm branch, typically ~20 m from We recorded 1,2 15 instancesof cachingby White- the feeder, and placed the seedbetween both feet. breasted Nuthatches (n = 403), Tufted Titmice After cracking open the shell with several strikes CACHING ECOLOGY 769
Darby cl Stems
Findlc
.rrdhm Kent a 1 5L-d4L4_-u Fayetteville 1
STEM SIZE CLASS
FIGURE 1. Comparison of vegetationstructure and composition on the four study sitesin Ohio and Arkansas. Stem size classes:A = ~2.5, B = >2.5-8, C = >8-15, D = > 15-23, E = >23-38, and F = >38 cm diameter at breast height. of the bill, the entire seed was cached; only oc- equally in inner, middle, and outer portions of casionally (approximately 15% of the cases)was limbs, and seedswere positioned disproportion- the seed partially consumed before being stored. ately on the top sides of branches. Titmice did Parus bicolor stored seedson small- to medium- not discriminate among different sides of tree sized (< 8 cm) branches and twigs on medium- trunks. Height of cache sites ranged from the to large-diametered (> 8 cm dbh) trees. As with ground to 25 m, but averaged7.8 (57.2) m. Tuft- nuthatches, titmice frequently were observed in ed Titmice used a diversity of specific cache lo- oaks (40%) but they also cached seedsin grape cations, but most often wedgedseeds under loose vines (15%) and dead branches on the ground bark (46%). Furrows, cracks, and rotted areas (10%). When caching in trees, titmice were seen (15%) miscellaneous sites, such as the ground 110 D. R. PETIT, L. J. PETIT AND K. E. PETIT
HORIZONTAL n WHITE-BREASTED soy MICROSUBSTRATE POSITION NUTHATCH IJTuFTED TITMOUSE
qCHICKADEE sw.
I TRUNK BRANCH TWIG INN MID OUT
loorCOVER 100 SHELL
80
60
20
0 dl1 COVERED NOT SHELLED UNSHELLED COVERED FIGURE 2. Comparison of the caching ecologyof White-breasted Nuthatches, Tufted Titmice, and Carolina and Black-cappedchickadees storing sunflower seedsin deciduouswoodlands during winter.
n WHITE-BREASTED NUTHATCH qTUFTED TITMOUSE
SUBSTRATE qCHICKADEE sw. L ear i w 0‘ - ii? l&l *O- o:JlEil -II, mm DEAD GRAPE HICKORY MISC. RED/ WHITE OAK BRANCH POST OAK
00- DIAMETER
< ’ >l-2.6 ,2.6-S >a-13 .15-23 >23-38 > 33-63 763 FIGURE 3. Comparison of the caching ecology of White-breasted Nuthatches, Tufted Titmice, and Carolina and Black-cappedchickadees storing sunflower seedsin deciduouswoodlands during winter. CACHING ECOLOGY 171
n W&l-EE~~ HASTED E qTUFTED TITMOUSE CACHE LOCATION IBI~ICKADEE SPP.
FURROW/ UNDER END OF BETWEEN LEAF ON OTHER CRACK BARK BRANCH TWO TREE OR IN BUD BRANCHES
80 e 1 TREE DIAMETER
41 > l-2.6 > 2.5-3 >3-16 >lS-23 z-23-33 >38-63 763 FIGURE 4. Comparisonof the cachingecology of White-breastedNuthatches, Tufted Titmice, and Carolina and Black-cappedchickadees storing sunflower seedsin deciduouswoodlands during winter.
(12%) between two branches (12%), and at the store it separatelyfrom the first. Also, in contrast end of broken branches and twigs (12%) also to titmice and nuthatches, both speciesof chick- were recorded regularly as cache sites. Titmice adees often (approximately 50% of our obser- rarely covered stored food. vations) ate one-third to one-half of the seedbe- Black-capped and Carolina chickadees.Like fore caching the remaining portion. Chickadees titmice, chickadeescached seedswithin 40 m of storedmost seedson branches(48%) twigs (25%) the feeder (X = 26.5 + 17.0 m), opened unshelled and small (~2.5 cm dbh) trunks (21%). More seedsin a similar fashion, and shelled them be- than 85% of those substrates were ~8 cm in fore storing. In about 5% of visits to feeding sta- diameter and chickadeesdemonstrated a strong tions, chickadeestook two (always shelled) seeds. preference for caching on bottom sides of One of those seedswas placed on a branch (or branches. Chickadees showed the same vertical occasionally on snow) and the other seed stored. The bird then would take the second seed and TABLE 3. Comparisonof expected(random) usewith observed use of trunk aspect (compass direction) for TABLE 2. Comparisonof expected(random) usewith White-breasted Nuthatches, Tufted Titmice, and observed use of sides of branches for White-breasted chickadees.Sample sizes in parentheses. Nuthatches, Tufted Titmice, and chickadee species. Sample sizes in parentheses. Observed“se (%) White- breasted Tufted Observeduse (%) Expected Nuthatch Titmouse Chickadees ExpectedWhite-breasted Tufted Ch;y6$e Aspect (%) (263) (33) (104) Branchside (%) Nuthatch (96) Titmouse (85) 316-45” 25.0 24.7 15.2 21.2 Bottom 25.0 38.5 24.7 55.7 46-l 35” 25.0 19.8 24.2 23.0 Sides 50.0 33.3 36.5 17.6 136-225” 25.0 30.4 39.4 34.6 Top 25.0 28.1 38.8 26.7 226-315” 25.0 25.1 21.2 21.2 G 12.4 52.0 49.2 G 6.0 4.0 4.8 P <0.005
TABLE 4. Interpopulational differences in caching SPECIES COMPARISONS ecology of White-breasted Nuthatches, Tufted Tit- Distancefrom feeder. On average, White-breast- mice, and chickadee species across four locations in Ohio and Arkansas. ed Nuthatches cached closer to the feeder than either titmice or chickadees;the latter two species
White- weresimilar(P’=21.8;df=2, 1,208;P = G-test except for hei t and distance (ANOVA); * = P c 0.05, ** = ed Titmice cachedseeds at similar heights, while P 5 0.01, *** = P 5 0.I? 01, ns = not significant. nuthatches stored seedshigher in trees than did chickadees;titmice and chickadeesdid not differ (F = 3.45; df = 2, 1,207; P = 0.032; LSD test). range in cache sites as titmice; on average those Microsubstrate.There was a strong preference siteswere at a similar height as those of congeners for trunks by nuthatches, branches by titmice, (X = 7.0 + 6.2 m). Oak trees were used (53%) and branchesand twigs by chickadees.All species when storingsunflower seeds, but chickadeeswere were statistically distinct in their use of micro- observed on a wide variety of substrates.All size substrates. classesof treeswere used and, like titmice, chick- Diameter of microsubstrate.Chickadees and adeescached at all distancesfrom the trunk. When titmice were statistically indistinguishable in se- on trunks, chickadeesshowed no preference for lection of branch and trunk diameters, but both any compass direction. Most (56%) seedswere parids differed from White-breasted Nuthatches. wedged under bark, but many of the same lo- Cover.Only nuthatches covered their stored cations used by Tufted Titmice also were selected seeds. by chickadees.Carolina and Black-cappedchick- Shell.All speciesstored shelled sunflower seeds adeeswere unique, however, in their propensity 79-86% of the time, percentagesthat were not to hide seedsin dead leaves still hanging from statistically heterogeneous. trees (8%). We never observed chickadeescov- Horizontal position. Whereas titmice and ering seedsafter storage. chickadeesused all distancesfrom the trunk when caching seeds,nuthatches concentrated on inner POPULATION COMPARISONS portions. White-breastedNuthatch. Nine of 10 cache-site Treediameter.Both Tufted Titmice and chick- characteristicsdiffered significantlyacross the four adees were found in slightly smaller trees than populations studied. Only horizontal position was were White-breasted Nuthatches. similar among populations (Table 4). Cachelocation. Nuthatches were highly selec- TuftedTitmouse. Seven of 10 variables were tive in their use of cache sites. In contrast, chick- significantly different among the four popula- adeesand titmice used a variety of sitesalthough tions. Horizontal position, height, and covering they also differed statistically from each other of cachesvaried little among locations (Table 4). (Table 5 summarizes speciesdifferences). Black-cappedand Carolinachickadees. Nine of 10 comparisons showed statistical differences. Only cover was similar acrossstudy sites (Table DISCUSSION 4). Differences were not attributable to our com- INTERPOPULATIONAL VARIATION bining both species of chickadees for analysis because differences existed both within and be- Nuthatches, titmice, and chickadeesstored food tween specieswhen separateanalyses were con- in locations generally similar to those described ducted. in the literature as foragingsites (e.g., Grubb 1975, CACHING ECOLOGY 773 1982; Watt 1975; Pierce and Grubb 1979; Whit- TABLE 5. Paitwise speciescomparisons of caching ing 1979; pers. observ.), although some impor- ecology of White-breasted Nuthatches, Tufted Tit- mice, and chickadee species. tant differences do occur (Petit, unpubl.). Each speciesdemonstrated a distinct pattern of seed Commxison storageand there was considerablevariation from Chicka- Chicka- Tit- deel dee/ mouse/ population to population. Interpopulational dif- Variable titmouse nuthatch nuthatch ferences may be attributable to at least four Distance nsa *** sourcesof variation. First, individuals may alter *** *** *** cache-site selection because of the local pool of Substrate Height ns ** ns potential cacherobbers. On our study sites,how- Microsubstrate *** *** *** ever, there were few differences in bird species Microsubstrate diameter ns *** *** composition (e.g., 75% of all nonraptorial species Cover ns *** *** seenwere common to all four study plots). White- Shell ns ns ns Horizontal position ns *** *** breastedNuthatches, Tufted Titmice, and chick- Tree diameter ns *** *** adees used all feeders and their numbers were Cache location *** *** similar at all sites. Thus, although interspecific a G-test exceptfor distanceand height (ANOVA); ** = P 5 0.01, *** interaction doesnot appear to have substantially = P 5 0.001, ns = not significant. altered bird caching ecologyamong populations, more detailed study of this possibility is neces- caching behavior of marked individuals under sary before this question can be addressedsat- various weather and social situations. isfactorily. Within the group of focal species, Finally, habitat differencesamong study areas however, interaction among heterospecificsmay could influence caching ecology. Although all have influenced cache-site selection. Most no- study siteswere situated within deciduouswood- tably, nuthatches stored food closer to the feeder lands, there were several substantial differences than did the parids. Our observations suggest among them in vegetation structure and floristic that this relationship was dominance-related, as composition, and that heterogeneity could ac- nuthatcheswere aggressivetoward chickadeesand count for the interpopulational variation in cache- titmice near (< 10 m) the feeding stations (also site selection documented here. If habitat had a see Waite and Grubb 1988). significant impact on caching ecology, then we Secondly, genetic or learned differencesin se- would expect that the more similar two plots lection of cache sites may exist among popula- were in vegetation composition and structure, tions. Individual Marsh Tits (Pam pafustris) the more closely the two avian populations in- showeddistinct preferencesfor certain cachesites, habiting those siteswould be in their use of cache at least on a short-term basis (Cowie et al. 198 1). sites.To test this idea, we calculated overlap val- Becausewe probably dealt with only two to 16 ues (Schoener 1968) for all pairwise combina- individuals of a speciesat each location, simple tions of (1) habitat structure on the four study variation among individuals (or populations) in sites, and (2) habitat use by the four populations cache-siteselection could explain the overall lack of each species.The habitat variables used were of agreement among populations. A comparison those shown in Figure 1; that is, both percentage of intra- vs. interpopulational variation in cach- of stems and percentageof basal area (1) in dif- ing ecology or experimental relocation of indi- ferent size classesof trees, and (2) in different viduals among populations may provide some categories of tree species. The corresponding resolution to the question of population-specific cache-sitevariables used were simply the percent cache-site selection. use of different tree size classesor tree species Thirdly, site-specificelements, such as weath- categories by a given population. Finally, we er, presence of con- and heterospecifics, and computed Spearman’s correlation coefficient (r,) overall abundance of other food sources,could for each set of habitat overlap indices (six pair- act randomly to produce variation among pop- wise combinations) with the correspondingover- ulations. Certainly, some of those factors affect lap indices of habitat use. Correlation coefficients how birds forage (e.g., Grubb 1975, 1977; Peters were computed separately for each species.Re- and Grubb 1983), so we would expect those pro- sults (Table 6) showed that there was little re- cessesto act on cachingbehavior, as well. Afruit- lationship between similarity of the four sites in ful approach to this problem might be to quantify their distribution of stemsacross size classesand 114 D. R. PETIT, L. J. PETIT AND K. E. PETIT TABLE 6. Spearman’s correlation coefficients com- storms, but only the side facing the prevailing paring similarity in vegetation structure and compo- wind direction becomes covered (pers. observ.). sition on the four study areas with similarity in selec- Thus, we would predict that birds would avoid tion of tree speciesand size classesby the bird species. Correlation coefficientswere computedfor eachspecies those sides of trunks that were most likely to be by comparingthe six pairwise combinationsof overlap covered by snow. However, all speciesused trunk indices for habitat availability with the six pairwise aspect randomly. Thus, in our study, selection combinations of overlap indices for habitat use. See of a branch side or trunk aspectdoes not appear text for further details. to be a responseto potential snow and ice cover. White- HeterospeciJics.Chickadees, titmice, and nut- breasted Tufted hatches travel in mixed-species flocks during Habitat variable Nuthatch Titmouse Chickadees winter and the potential benefits gained by that (1) Tree size classes behavior are well-known (e.g., Morse 1980). (a) stems -0.21 -0.64 -0.33 Flocking may be disadvantageousto individuals, (b) basal area 0.71 0.71 0.60 however, when a large group is involved in stor- (2) Tree species ing food taken from a single source. By hoarding (a) stems -0.01 -0.36 0.07 seedsin close proximity to many individuals, a (b) basal area 0.71 0.66 1.oo bird may jeopardize the secrecyof its cachesites. To protect their cachesfrom cacherobbers, then, tree species and the similarity in the three species’ birds can employ severalprotective measures(see use of those habitat features. Conversely, when Vander Wall and Smith 1987): (1) aggressively habitat was expressedin terms of basal area, there defend the hoarded food, (2) space seedsat dis- were consistent trends for populations to cache tances that reduce the probability of a cacherob- on more similar-sized trees and on more similar ber finding a cache site based upon a previously tree specieswhen the local habitats were in closer located site, (3) alter the cache site so as to in- conformity with one another. This analysis sug- crease the crypticity of the cache, and (4) store geststhat local habitat structure may have influ- food in locations which limit their accessability enced selection of cache sites. Use of different- to other species.All speciesstudied here are scat- sized treesor tree speciesby birds may alter other ter-hoarders (Morris 1962, Stapanian and Smith characteristicsof cache sites becauseof the var- 1978); that is, they store food in numerous, dis- ied opportunities encountered on different types persed locations. Although some scatter-hoard- of substrates. ing speciesaggressively defend their stores, this behavior is not practiced by smaller species,such CACHE PROTECTION as the ones in this study. Environmental factors. To increase the proba- There is some evidence that Marsh Tits space bility of recovering stored food, birds may avoid cached seedsat nearest neighbor distances that certain types of storagesites vulnerable to snow significantly reduce the chances that cache rob- and ice cover. For example, Haftorn (1974) sug- bers will locate a given seed based upon the lo- gested that Boreal Chickadees (Parus hudsoni- cation of another seed(s).Cowie et al. (198 1) and cus) in Alaska alleviated this potential problem Sherry et al. (198 1) found that individual Marsh by storing food low in trees,but above the ground Tits spacedsunflower seedsat a nearest neighbor and on the bottom sides of branches. Although distance of approximately 7 m. Although our this trend was not observed by Haftom (1954, study was not designed to address the question 1956) for two European parids, it is theoretically of optimal spacing of cache sites, observations one means by which birds can retain accessto of several color-marked individuals suggestthat stored food regardlessof environmental condi- titmice spacecaches 5-15 m apart. tions (Vander Wall and Smith 1987). In this study, Cache concealment by passerines is usually nuthatches and chickadeesstored a greater pro- achieved by either actively covering the food re- portion of seeds on bottoms of branches than serve with suchmaterials as bark, lichens, or soil, would be expected by chance. However, seeds or by placing seedsin naturally concealed loca- were cached on tops of branches in proportions tions, e.g., under bark or in furrows on tree trunks at least as large as that expected (i.e., few seeds (Haftom 1954, Cowie et al. 1981, James and were stored on sides of branches). Likewise, tree Verbeek 1983, pers. ObSeN.). In this study, only trunks often are covered with ice and snow after White-breasted Nuthatches covered their stored CACHING ECOLOGY 115 seeds on a regular basis. All speciesfrequently foraging ecology (e.g., Schoener 1965). The rel- chosecache locations that were naturally cryptic, atively short bills of parids cannot cache, and at least to the human eye: furrows, cracks,under then retrieve, seedsplaced deep into furrows or bark on trees, and in curled dead leaves. Even under large strips of bark, but can efficiently ma- here, though, nuthatches(98%) utilized the above nipulate a seed into a small crack in a twig or in naturally concealed sites more than did either a small bud. titmice (65%) or chickadees(77%). By combining The local pool of potential cache robbers also concealed sites with covering of stored seeds, may dictate where birds storefood and the degree nuthatchesclearly concealeda higher proportion of concealment that is necessary.In our study, of their cachesthan did the parids. White-breasted Nuthatches apparently relied All speciesselected cache sites that were some- upon cryptic (both natural and altered) cachesites what inaccessible to other birds, although the to a greater extent than did either titmice or type of site varied among species. Chickadees chickadeesas a defense against kleptoparasites. and titmice placed seedson smaller branchesand This result may be due to more specieson our twigs, frequently on the outer portion of tree can- study sites foraging in ways comparable to nut- opies. Both species, but especially chickadees, hatches than to chickadeesand titmice. We re- reduced the size of the seed before caching it. corded nine species that are similar to White- This may be a response to the relatively small breasted Nuthatches in their foraging ecology: cracks and strips of loose bark that are available Pileated (Dryocopus pileatus), Red-bellied on small twigs and branches as compared to larg- (Melanerpes carolinus), Red-headed (M. eryth- er branches and trunks used by nuthatches and, rocephalus),Hairy (Picoidesvillosus), and Downy to some extent, titmice. The practice of storing woodpeckers (P. pubescens),Northern Flicker only portions of seedsalso may reduce the prob- (Colaptes auratus), Yellow-bellied Sapsucker ability that wind might dislodge the seedsfrom (Sphyrapicus varius), Brown Creeper (Certhia shallow cracksor from under small piecesof loose americana), and Red-breasted Nuthatch (Sitta bark. Nuthatches used their long bills to wedge canadensis).Alternatively, only two specieswere whole seedsinto deep furrows or far under loose seenthat foragein locations frequented by chick- bark. adees and titmice: Golden-crowned Kinglets How do White-breasted Nuthatches, Tufted (Regulus satrapa) and Purple Finches (Carpod- Titmice, and Black-capped and Carolina chick- acus purpureus). Kilham (1974) postulated that adees protect their stored food from heterospe- White-breasted Nuthatches covered their stored cifics?It appearsthat chickadeesand titmice pri- food becauseof local competitors. marily rely upon inaccessability of cache sites, Our results suggestthat caching behavior may while nuthatches utilize both difficult-to-reach be shapedby local environmental conditions. The locations and crypticity to prevent seeds from importance of protecting stored food, both from being stolen by other species. Differences be- other birds and from the environment, may have tween parids and the nuthatch in cache-site se- caused each species in this study to occupy a lection may be determined by both morpholog- distinct caching niche. In addition, morphologi- ical constraints placed on each species and by cal constraints placed on speciesmay limit the the number of potential cache robbers likely to number of sites suitable for effective storageand encounter each species’ cache sites during nor- recovery of food. mal foraging bouts. For example, morphology of ACKNOWLEDGMENTS the bill and legs restricts a speciesin the types of substratesthat it can exploit (e.g., Richardson M. L. Morrison,K. G. Smith, and an anonymousref- ereemade numerousconstructive comments on this 1942, Schoener 1965, Partridge 1976). The po- work. We thank BuckeyeFeeds, Inc. of Dalton, Ohio sitioning of the legs of titmice and chickadees for donatingthe sunflowerseeds, and R. E. and J. M. allows for mobile maneuvering (most notably, Petit for invaluable financial and logisticalsupport. hanging) on small substrates, such as twigs, This work is dedicatedto the memory of our father, whereas the nuthatches’ more posteriorly posi- RichardE. Petit. tioned legsrestrict this speciesto larger branches LITERATURE CITED and trunks (Richardson 1942, Osterhaus 1962, ANDERSON,M., AND J. KREBS. 1978. On the evo- Partridge 1976). Bill length and shape,too, prob- lution of hoarding behaviour. Anim. Behav. 26: ably influence caching behavior as they do birds’ IOI-Ill. 116 D. R. PETIT, L. J. PETIT AND K. E. PETIT COW~E,R. J., J. R. KREBs,AND D. F. SHERRY. 1981. perimental analysisof sex-specificforaging in the Food storing by Marsh Tits. Anim. Behav. 29: Downy Woodpecker, Picoidespubescens. Ecology 1252-1259. 64:1437-1443. GRUBB, T. C., JR. 1975. Weather-dependentforaging PIERCE,V., AND T. C. GRUBB,JR. 1979. Laboratory behavior of some birds wintering in a deciduous studies of foraging in four bird speciesof decid- woodland. Condor 17:175-182. uous woodland. Auk 98:307-3207 GRUBB, T. C., JR. 1977. Weather-dependentforaging RICHARDSON.F. 1942. Adantive modification for tree behavior of some birds wintering in a deciduous trunk foraging in birds.-Univ. Calif. Publ. Zool. woodland: horizontal adjustments. Condor 79: 46:317-368. 27 l-274. ROBERTS,R. C. 1979. The evolution of avian food- GRUBB,T. C., JR. 1982. On sex-specificforaging be- storing behavior. Am. Nat. 114:418-438. havior in the White-breasted Nuthatch. J. Field SCHOENER,T. W. 1965. The evolution of bill size Omithol. 53:305-314. differencesamong sympatric congenericspecies of HAF~ORN,S. 1954. Contribution to the food biology birds. Evolution 19:189-2 13. of tits, especiallyabout storing surplusfood. P&t SCHOENER,T. W. 1968. The Anolis lizards of Bimini: I. The Crested Tit (Parus c. cristatusL.). K. Nor. resourcepartitioning in a complex fauna. Ecology Vidensk. Selsk. Sk;. 1953 Nr. 4:1-122.’ 491704-726. HAF~ORN,S. 1956. Contribution to the food biology SHERRY,D. F., J. R. KREBS, AND R. J. COWIE. 1981. of tits, especiallyabout storing surplusfood. Part Memory for the location of stored food in Marsh III. The Willow Tit (Pam atricupillusL.). K. Nor. Tits. Anim. Behav. 2911260-1266. Vidensk. Selsk. Skr. 1956 Nr. 3:1-79. SMITH,C. C., AND 0. J. REICHMAN. 1984. The evo- HAFTORN,S. 1974. Storage of surplus food by the lution of food caching by birds and mammals. BorealChickadee Purus hudsonicus in Alaska, with Annu. Rev. Ecol. Syst. 15:329-35 1. some records on the Mountain Chickadee Purus STAPANIAN,M. A., ANDC. C. SMITH. 1978. A model gumbeli in Colorado. Omis Stand. 5: 145-16 1. for seed scatterhoarding:coevolution of fox squir- HURLBERT,S. H. 1984. Pseudoreplication and the rels and black walnuts. Ecology 59:884-896. design of ecological field experiments. Ecol. TOMBACK,D. F. 1980. How nutcrackers find their Monogr. 54:187-211. seed stores. Condor 82: 10-19. JAMES,P. C., ANDN.A.M. VERBEEK.1983. The food VANDERWALL, S. B. 1982. An experimental analysis storagebehaviour of the Northwestern Crow. Be- of cache recovery in Clark’s Nutcracker. Anim. haviour 851276-291. Behav. 30:84-94. JAMES,P. C., ANDN.A.M. VERBEEK.1985. Clam stor- VANDERWALL: S. B., ANDK. G. SMITH. 1987. Cache- age in a Northwestern Crow (Corvus cuurinus): protecting behavior of food-hoarding animals, p. dispersion and sequencing.Can. J. Zool. 63:857- 61 l-644. In A. C. Kamil, J. Krebs, and H. R. 860. Pulliam [eds.], Foraging behavior. Plenum Publ. KILHAM, L. 1974. Covering of stores by White- Corp., New York. breastedand Red-breastednuthatches. Condor 76: WAITE, T. A., AND T. C. GRUBB,JR. 1988. Copying 108-109. of foraging locations in mixed-species flocks of MORRIS,D. 1962. The behaviour ofthe greenacouchi temperate-deciduouswoodland birds: an experi- (Myoproctupruttz] with specialreference to scatter mental study. Condor 90: 132-l 40. hoarding. Proc. Zool. Sot. (Lond.) 139:701-732. WATT, D. J. 1975. Comparison of the foraging be- MORSE,D. H. 1980. Behavioral mechanismsin ecol- havior of the Carolina Chickadee and Tufted Tit- ogy. Harvard Univ. Press, Cambridge, MA. mouse in northwestern Arkansas. M.Sc.thesis. OSTERHAUS,M. B. 1962. Adaptive modifications in Univ. Arkansas, Fayetteville. the leg structure of some North American war- WHITING,R. M., JR. 1979. Winter feeding niche par- blers. Am. Midl. Nat. 68:474-486. titionment by Carolina Chickadees and Tufted PARTRIDGE,L. 1976. Some aspectsofthe morphology Titmice in east Texas, p. 33 l-340. In J. G. Dick- of Blue Tits (Purur cue&em) and Coal Tits (P. son, R. N. Connor, R. R. Fleet, J. A. Jackson,and uter) in relation to their behavior. J. Zool. 179: J. C. Kroll [eds.], The role of insectivorousbirds 121-133. in forest ecosystems.Academic Press,New York. PETERS, W. D., AND T. C. GRUBB,JR. 1983. An ex-