The Condor 87:205-216 0 The CooperOmithological Society 1985

FORAGING AND HABITAT RELATIONSHIPS OF INSECT-GLEANING IN A SIERRA NEVADA MIXED-CONIFER FOREST

DANIEL A. AIROLA AND REGINALD H. BARRETT

ABSTRACT. -Foraging habits and relative abundancesof 12 birds comprising the insect-gleaning guild in a Sierran mixed-conifer forest were studied during two breeding seasonsto determine: (1) foraging habitat preferences,(2) the extent to which speciesdiffer in their use of various components of the foraging niche, (3) patterns of relative abundance vs. niche breadth, and (4) differences between resident and migrant species.Comparisons of proportional availability and use of foliage height classesand tree speciesshowed that tree speciesand, to a lesserextent, heightswere used selectively by the guild. Incense-cedar(Calocedrus decurrens)was consistentlyavoided by all species;other tree specieswere generally used in a complementary manner by different birds. Of four measuredcomponents of foraging niche, the use of foraging site (consisting of air or tree part) showed the greatest difference between species,followed by tree species,foraging tech- niques, and foraging height. We found no correlation between niche breadth and speciesabundance for all guild members; however, a significant positive corre- lation existed for the five resident species.Resident and migrant speciesgroups showed few fundamental differences in foraging patterns, except that migrants tended to use a greater proportion of deciduous foliage than residents.Our results suggestthat to provide for this guild, land managers should maintain natural levels of tree speciesdiversity in the mixed-conifer forest type.

Many studieshave shown that syntopic insec- 1980). We studied foraging substrate prefer- tivorous birds differ in their methods of for- ences, foraging behaviors and relative abun- aging. In forested habitats, birds tend to use dances of the 12 most common members of different foragingtechniques, foraging sites, tree an insect-gleaning guild. We sought to: (1) species,and heights. The relative importance compare the guild’s use of tree species and of these foraging niche components in distin- vertical foliage layers with the availabilities of guishing species has received less attention. these habitat components; (2) describe and Differences in importance of niche compo- compare foraging sitesand techniquesused by nents reported in previous studies may reflect each species;(3) evaluate the relative impor- either real differences that exist in different tancesof foraging heights,tree speciesuse, for- habitats and geographicareas (Balda 1969) or aging sites, and foraging techniques in differ- incomplete analysis of all potentially relevant entiating speciesecologically; (4) determine if factors (Holmes and Robinson 198 1). In par- resident and migrant speciesdiffered in their ticular, the importance of differential plant foraging and abundance patterns; and (5) sug- species use has not been fully appreciated gest management recommendations to miti- (Holmes and Robinson 198 1). Additional in- gate the effects of habitat manipulation on formation is needed before sound generaliza- members of the insect-gleaningguild. tions can be made regarding patterns of dif- ferentiation among insect-gleaningbirds. METHODS In managed forests, plant speciescomposi- tion and vegetation structure may be altered STUDY AREA by logging, other silvicultural activities, and We worked at Blodgett Forest Research Sta- disruption of natural fire regimes(Kilgore 197 1, tion, a 1,186-ha mixed conifer-oak forest lo- Franzreb and Ohmart 1978, Szaro and Balda cated at 1,350- to 1,450-m elevation in the 1979). Identification of habitat preferences of central Sierra Nevada, El Dorado County, Cal- forest birds can suggestrecommendations to ifornia. The Forest contains five conifer tree mitigate impacts of manipulation. speciesthat are typical of the mixed-conifer The mixed-conifer forest of the Sierra Ne- foresttype(Rundeleta1. 1977:563,Vemerand vada, California, supports a large number of Boss 1980:4), and a substantialamount of Cal- arboreal insectivorous birds (Verner and Boss ifornia black oak (seeFig. 1 for scientific names 12051 206 DANIEL A. AIROLA AND REGINALD H. BARRETT

(Dedon and Barrett 1982). Species composi- tion was determined by visually estimating percent canopy cover of each plant specieson each subplot (Mueller-Dombois and Ellenberg 1974:63). Becausethe birds we studied foraged almost exclusively in trees, we excluded other plant speciesfrom preference analyses(see be- low). Vertical vegetation structure was deter- mined by estimating on each subplot the total HEIGHT (ml TREE SPECIES cover (all speciescombined) in each of eight FIGURE 1. Proportions of foliage: (a) in height classes, height classes:O-O.5 m, 0.5-2 m, 2-5 m, 5- and (b) contributed by various tree species, at Blodgett 10 m, lo-15 m, 15-20 m, 20-30 m, and >30 Forest. Tree speciescodes and scientific names are PP = m. To represent the Forest as a whole, tree ponderosa pine (Pinus ponderosa),SP = sugar pine (P. lambertiana), WF = white fir (Abies concolor), DF = species and height class cover values for all Douglas-fir (Pseudotsugumenziesiz], IC = incense-cedar 820 subplotswere averaged to obtain a mean (Calocedrusdecurrens), BO = California black oak (Quer- cover for each speciesand height class.Using cuskelloggi& and OT = other tree species. weightedaverages, the valuesfor the eightheight strata were assignedto the five strata used to of tree species).Dominant vegetation consists record bird foraging heights (see below). The of an overstory of older trees left from the means for each tree species and height class original loggingof the area in 1900-l 9 13, and were then converted to proportions of total an understory of trees and brush that subse- cover in all classesof each variable for com- quently regenerated (Fig. 1). All tree species parison with relative use of each classby bird are present in all vertical layers. In some areas, species. trees that grew following logging have grown Bird relativeabundance. On each plot used rapidly owing to high soil fertility, and now to characterize habitat, we recorded the num- form an overstory nearly as tall as residual bers of individuals of each bird species seen trees. Various small (0.5-10 ha) experimental or heard during each of 20 consecutive 10-min treatments (clearcutting, selective logging, periods, beginning 0.5 h after sunrise on one planting, brush control) have been performed morning between late April and early July (De- over the last 15 years, demonstrating many of don and Barrett 1982). Boundaries of plots the management practicesthat occur in mixed- were checked with a range finder and either conifer forests of the Sierra Nevada. marked with surveyor’s flaggingor recognized We defined the forest insect-gleaningguild from natural landmarks. The observer re- as the group of birds that gleaned from forest mained at the center of the plot during surveys, trees either from a stationary perch or while except for brief forays to check bird distances. hovering during >50% of observations. We We calculatedrelative frequency of each species did not subdivide this group into foliage- and by summing the number of individuals re- bark-gleaners because of the high degree of corded per period. Relative abundanceof each overlap in use of bark, branches, twigs, and specieswas estimated by dividing the species’ foliage by most guild members. Excluded from relative frequency by the sum of all species the group are speciesthat gleaned insectspre- frequencies. In this study, we use only bird dominantly from successional and riparian abundance data from 1977, since the method shrubs (cf. Airola and Barrett 198 1). of plot selectionwas randomized for the whole Forest only in this year. DATA COLLECTION One source of bias may have affected our We sampled vegetation, bird species abun- relative abundance values. We counted birds dance, and foraging activities on a set of 0.28- over a 2.5-month period and sampled each ha plots located throughout the Forest. In 1977, plot on only one day. Detectability possibly we randomly selected 41 sample plots from a differed owing to behavioral changesthrough set of 70 plots that had been previously estab- the breeding season.Since plots were selected lished. The original plots were randomly se- randomly and only relative abundances are lected from a set of coordinates within each of considered, however, we believe bias is rela- 70 5-l 5-ha management compartments at the tively slight. Forest. In 1978, data were collected on 80 plots Foragingdata. Foraging behaviors and sub- which were systematicallylocated 100 m apart stratesused by birds were recorded on sample within five of the 70 compartments. plots during 1977 and 1978, and in “oppor- Vegetationsampling. Vegetation was mea- tunistic” observations made in transit to and sured on 20 circular 10-m* subplots laid out from plots and in transectswalked in various in a 10-m grid pattern on each sample plot parts of the Forest. Although starting points FORAGING OF SIERRA NEVADA BIRDS 201 and directions of transects were not selected i, and qi is the proportion of the same class at random, we attempted to sample a variety that is available in the environment. The index of geographicareas, irrespective of vegetation ranges from + 1 to - 1, with positive values composition. Most observations were made indicating selection for a substrateclass by the from sunriseto lo:30 and from 16:OOto 19:00, bird, 0 indicating non-selective use, and neg- from mid-April to early August. When we saw ative values indicating avoidance. a bird attempting to take food, the following To test for significanceof L, 95% confidence data were recorded: bird species, foraging intervals (C.I.s), based on the estimated sam- height, tree species or other substrate used, pling variance of L (Strauss 1979), were con- foraging site, and foraging technique. Foraging structedby Dunn’s (196 1) method. This meth- heights were assigned to one of five height od setsa procedure-wise error rate for a series classes:O-l m, l-3 m, 3-7 m, 7-20 m, >20 of comparisons (e.g., a bird species’ prefer- m. Substratesrecorded other than tree species ences for six different tree species)so that the were shrub species,snags, and ground. “For- chance that a single type 1 error occurs within aging site” indicates the tree part or other lo- the seriesof comparisonsis equal to (Y.A pref- cation from which the bird attempted to take erence value was consideredto be significantly food: trunk, branch (> 1 cm diameter), twig different from non-selective useif the 95% C.I. (< 1 cm), foliage, and air. Foraging techniques of L did not overlap the L value of 0 (where were recorded as: (flying from a perch use equals availability). to take aerial insects), hovering(taking insects Niche breadth.Patterns of use within and from foliage while in flight), lunging(leaping between niche axes were compared using sin- from a stationary position to take insectsmov- gle-axisniche breadths.Niche breadthsof birds ing inside the tree crown), and gleaning(taking on individual axes were calculated by the per- stationary insects from a substrate while cent similarity (PS) method (Feinsinger et al. perched). Foraging sites and techniques were 1981). recorded only during 1978. Becausebirds were R often difficult to see owing to foliage height PS = 2 min(Pi, 4,) and density, we recorded consecutive foraging i=l observations for some individuals to increase = 1 - 0.5 i Ip, - qil, the amount of information collected. To re- i=l duce the bias that multiple records may create where R is the number of defined niche axis (Wagner 198 1, Morrison 1984) we treated each states.The PS value sums the absolute values individual bird as an observation when deter- of L, and indexes the sum to between 0 (total mining frequencies and sample sizes for sta- dissimilarity) and 1 (use equals availability). tistical tests. Thus, when n consecutive obser- No availabilities were defined for the for- vations of an individual were recorded, each aging site and technique axes. Percent simi- observation contributed to the species’ total larity was used to evaluate niche breadth on frequency by a value of l/n, and all observa- these two axes, however, by setting availabil- tions of the individual in the period contrib- ities as equal for all variable states on each n axis. Thus, PS measureddeviation in use from uted 2 l/n = 1 to the species’ frequencies.By a uniform distribution for these axes. Calcu- i-1 lation of all niche breadthswith the same mea- this method, we incorporated all data taken, sure permitted comparison of breadths be- without biasing speciesfrequencies toward in- tween axes. Niche breadths on the tree species dividuals which contributed multiple obser- axis were based only on the birds’ use of the vations. We report sample sizes for both the dominant tree species,which included 94% of number of individuals observed and the total foraging observations. number of observations obtained. We could not calculate complete multi-di- mensional niche breadths,as recommended by ANALYSIS May (1975) and Inger and Colwell(l977), be- Preference.“Preference” refers to the differ- cause foraging sites and techniques were re- ence between the proportional use of a sub- corded in only one year. We calculated partial strate class(tree speciesor foliage height class) composite measures for two components of and the proportion available in the environ- the multi-dimensional niche, one based on use ment. Preferenceis describedby Strauss’ (1979) of height class and tree species(two years of index: data), and the other on foraging site and tech- L = pi - qi nique. The use frequencies for each combi- nation of the niche statesin the two combined where: L is the preference value, pi is the pro- axes(height/tree and site/technique) were used portion of all usesof substrates,that is of class to calculate two-dimensional niche breadths. 208 DANIEL A. AIROLA AND REGINALD H. BARRETT

SOLIT. WARB. MTN. R.-B. BRN. GOLXR. VI REO CHICK. NUTH. CREEP. KINGL.

3463’ 32: 58 124:214 54.93 32: 58 25.51 15:31

i -.2 0 .2 -.2 0 .2 .4 -.2 0 .2 A

NASHV. YEL:RUM. HERM. WEST. BL:H. WARB. WARB. WARB. TANAGER GROS. 86138’ 35:54 55:86 56.86 61:76 599:1008

t 12 0 .2 0 .2

PREFERENCE VALUE

FIGURE 2. Foraging height preferencesof insect-gleaningbirds at Blodgett Forest. Each preferencevalue equalsthe proportion of foraging observationsin a height classminus the proportion of total foliage available in the same height class. Horizontal lines indicate 95% confidence intervals (see text). Sample sizes are reported under speciesnames, indicating (before colon) the number of individuals observed which was used in statistical analyses,and (after colon) the total number of observations made. The latter value includes multiple observations of single individuals, incor- porated into speciestotals as described in Methods.

These niche breadths were calculated by the Niche overlap.We calculated niche overlap equation H = C [PJn(PJ] (Shannon and for each pair of species using Horn’s (1966) Weaver 1949, Levins 1968) scaled from 0 to information theory index, 1 by dividing by the maximum H value achieved when all categoriesare equally used. O, = [(Pik+ PjJ”dPik + PjJ H was used because it does not require re- - Pikl”g Pik - Pjkl% Pjdj2 log 2 source availability, and thus permits all used substratesto be included. We approximated where 0, equals the overlap of species i and the composite foraging niche breadth (all axes) j, pjk eqUdS the prOpOI%On Of niche CategOq

by averagingbreadths on the two-dimensional k used by species i, and pjk equals the pro- axes. We considered this method preferable to portion of k used by speciesj. Overlaps were either calculating multi-dimensional breadths calculated for each pair of specieson individ- using only one year’s data, or averaging ual foraging niche axes, and on the two-di- breadths calculated independently for each of mensional axes (as described for niche the four axes. Correlations of relative abun- breadths). We approximated total overlap dancesof birds and niche breadth values were among species pairs on all foraging axes by calculated to test hypothesesconcerning abun- multiplying the two-dimensional axes. This dance in relation to specialistvs. generalistfor- procedure is appropriate when the axes are in- aging habits. dependent (May 1975); in fact, they showed a FORAGING OF SIERRA NEVADA BIRDS 209

TABLE 1. Common and scientific names, seasonalresidency, relative abundances,and niche breadth values for members of the insect-gleaningguild at Blodgett Forest.

Niche breadths’ Resi- R&it& Singleaxes Combinedaxes dencyb abundance HT TR SI TE HT:TR SI:TE

Mountain Chickadee (Parus gambeh] P 0.072 0.88 0.59 0.50 0.25 0.68 0.46 Chestnut-backedChickadee (Parus rufescens) P 0.068 0.80 0.71 0.43 0.35 0.66 0.51 Red-breasted (Sitta canadensis) P 0.162 0.78 0.72 0.73 0.34 0.80 0.75 Brown Creeper (Certhia americana) P 0.018 0.76 0.84 0.20 0.25 0.83 0.00 Golden-crowned (Regulussatrapa) P 0.159 0.70 0.76 0.41 0.38 0.64 0.51 Solitary Vireo (Vireo solitarius) M 0.059 0.74 0.78 0.64 0.45 0.72 0.75 Warbling Vireo ( Vireo gilvus) M 0.028 0.70 0.51 0.48 0.33 0.58 0.46 Nashville Warbler (Vermivora rujkapilla) M 0.107 0.76 0.73 0.52 0.29 0.72 0.53 Yellow-rumped Warbler (Dendroica coronata) M 0.035 0.74 0.63 0.64 0.50 0.66 0.82 Hermit Warbler (Dendroicaoccidentalis) M 0.180 0.78 0.51 0.56 0.27 0.63 0.52 Western Tanager (Piranga ludoviciana) M 0.060 0.78 0.85 0.55 0.62 0.71 0.74 Black-headedGrosbeak (Pheucticusmelanocephalus) M 0.053 0.76 0.68 0.45 0.53 0.68 0.65 Mean 0.76 0.69 0.51 0.38 0.69 0.56 * HT = foragingheight, TR = tree species,SI = site, TE = technique. b P = permanentresident, M = migrant. weak, but significant,negative association(r = 0.00). A summary of the individual species’ -0.27, P = 0.01). height use patterns (Fig. 2) shows that of 60 Similarity patterns within the guild were possible comparisons, 26 (43%) were signifi- shown by cluster analysis (Dixon 198 1:448, cant and positive, 15 (25%) were significant Program lM), using the product of foraging and negative, and 19 (32%) did not differ sig- overlap values of species pairs as similarity nificantly. This degree of selectivity indicates measures. Groups were combined based on nonrandom use of foliage layers. average similarity (Dixon 198 1). Individual species showed a variety of fo- The magnitude of overlaps between each liage height use patterns. The birds can be clas- speciesand the rest of the guild on each axis sified into five categoriesof height preference: was approximated by averaging the pairwise (1) high- Chestnut-backedChickadee (seeTa- overlaps of each species with all other guild ble 1 for scientific names), Black-headed Gros- members. This index is designated “guild beak; (2) medium and high-Yellow-rumped overlap.” Using the Mann-Whitney test (Sokal Warbler, Hermit Warbler, Western Tanager; and Rohlf 1969), we compared the magnitudes (3) medium-Red-breasted Nuthatch, Brown of guild overlap values of all specieson each Creeper, Golden-crowned Ringlet, Solitary pair of niche axes to assessthe degreeto which Vireo, Warbling Vireo; (4) low and medium- each axis distinguishedguild members in for- Nashville Warbler; and (5) no preference- aging. Mountain Chickadee. Foraging heightsdid not differ between resident and migrant groups (Table 1; Wilcoxon rank sum tests, compari- RESULTS sons at O-l m [P = 0.501, l-3 m [P = 0.381, FORAGING HEIGHT 3-7 m [P = 0.441, 7-20 m [P = 0.441, 20+ m As a group, the guild foraged selectively by [P = 0.321). height (Fig. 2). The two lowest canopy classes were avoided (L = -0.11 and -0.05 for O-l TREE SPECIES m and l-3 m heights,respectively), the middle The guild used the six dominant tree species and highestlayers were preferred (L = 0.14 and in 94% of our observations. Other substrates 0.04, respectively), and the upper middle layer used were snags(3%) and tanoak (Lithocarpus was used in proportion to its availability (L = denszflora), deerbrush (Ceanothus integerri- 210 DANIEL A. AIROLA AND REGINALD H. BARRETT

MTN. CR-BKD. R-B. BRN. SOLIT. WARBLING CHICK. CHICK. NUTH. CREEP. VIREO VIREO

33:60 31:79 118.205 55'91 31:57 25.49 16:31

I1711I PP T- - -+I + *I SP tc i- WF DF + F 5 LA 0 -.2 0 .2 -.21.L 0 .2 -.2 0 .2 .4 .6 u iif rn NASHV. YEL:RUM. HERMIT WEST. EL.- H. ALL w WARB. WARB. WARBLER TANAGER G ROS. SF! ~il--;::-c,~~

-.2 0 .2 -.2 0 .2 .4 -2 0 .2 .4 -.2 0 .2 -.2 0 .2 -.2 0

PREFERENCE VALUE FIGURE 3. Tree speciespreferences of insect-gleaningbirds at Blodgett Forest. Preference values and confidence intervals are calculatedas described in Figure 2. Sample sizes are shown as described in Figure 2.

mu@, and ground (1% each). Certain birds negative, and 39 (54%) were not significantly made substantial use of these minor foraging different. These results indicate non-random substrates.Snags were used by the Red-breast- use of trees by foraging birds. ed Nuthatch and the Brown Creeper in 9% and We assignedthe birds to seven categoriesof 7% of observations, respectively. The Nash- tree speciespreference, basedon their selection ville Warbler used deerbrush 10% of the time, of certain combinations of the three most im- and the Solitary Vireo used tanoak in 9% of portant tree types, pine (ponderosa and sugar), observations. white fir, and oak: (1) pine-Hermit Warbler, The guild as a whole showed marked differ- Red-breasted Nuthatch; (2) white fir-Gold- ences in preferences for the six major tree en-crowned Ringlet, Yellow-rumped Warbler, species(Fig. 3). Ponderosa and sugarpine, and Western Tanager; (3) oak- Warbling Vireo, black oak were preferred (Ls = 0.11,0.05, and Nashville Warbler; (4) pine-fir- Mountain 0.05, respectively). White fir was used in pro- Chickadee; (5) oak-pine-Black-headed Gros- portion to its availability, whereas Douglas-fir beak; (6) oak-fir-Chestnut-backed Chicka- was weakly but significantly avoided (L = dee; and (7) no preference-Brown Creeper, -0.03). Incense-cedar was strongly avoided Solitary Vireo. (L = -0.17). Analysis of tree speciespreferences based on Individual bird speciesdiffered considerably residency indicated that, as a group, migrant in their preferences for tree species. Overall, species (seven species, Table 1) used greater their preferenceswere somewhat complemen- amounts of hardwood foliage (28%), and less tary, except that all birds avoided incense-ce- conifer foliage (72%), than did resident species dar (significantly so for nine species).Ponder- (12% hardwood, 88% conifers; Wilcoxon rank- osa pine, white fir, and black oak each were sum test, P = 0.02). This pattern, however, was preferred significantly by four species,and sug- not uniform within groups: the Chestnut- ar pine was preferred by three. Douglas-fir was backed Chickadee, a resident, used a high not preferred by any bird. Of 72 possiblebird/ amount of hardwood foliage (23%), whereas tree comparisons, 15 (2 1%) were significant the Hermit Warbler, a migrant, seldom used and positive, 18 (25%) were significant and hardwoods (8%).

212 DANIEL A. AIROLA AND REGINALD H. BARRETT

TABLE 2. Correlations between relative abundanceof bird speciesand niche breadthson four singleniche axes, two two-dimensional axes, and the mean of the two-dimensional axes.

All species (n = 12) Residents (n = 5) Migrants (n = 7) Niche axis r P I P r P

One-dimensional Height 0.04 0.44 -0.39 0.26 0.64 0.06 Tree -0.15 0.31 -0.15 0.40 -0.25 0.29 Site 0.35 0.12 0.72 0.08 0.02 0.48 Technique -0.27 0.19 0.75 0.07 -0.58 0.08 Two-dimensional Ht:Tr -0.05 0.43 -0.26 0.33 0.02 0.48 Si:Te 0.19 0.26 0.82 0.04 -0.42 0.17 Four dimensional x 2-d axes 0.20 0.26 0.81 0.05 -0.31 0.25

dimensional axes indicated no significant re- ilarity level, four groupsare evident: (1) a trunk- lationships, using either all guild members or gleaner (Brown Creeper); (2) a branch and only migrants (Table 2). Correlations ap- trunk-gleaner (Red-breasted Nuthatch); (3) proached significance, however, between mi- three species that preferred oaks and either grant abundance and both height (P = 0.06) foraged low or usedtwigs ( and Nashville and technique (P = 0.08) breadths. We con- Warbler); and (4) seven upper canopy foliage- sider the marginal relationship between mi- feeders. Within group 3, the Warbling Vireo grant abundanceand height breadth to be spu- was distinguished by its greater use of twigs rious, given the similarity of breadth values and higher foraging. Within group 4, the for this group (Table 1). Residentsshowed pos- Mountain Chickadeeand Hermit Warbler sub- itive relationships between abundance and group was distinguishedby preference for pine niche breadths that were: (1) nearly significant and avoidance of fir, greater use of branches for individual foraging site (P = 0.08) and and twigs, and dependence on gleaning. The techniques (P = 0.07) axes; (2) significant for Chestnut-backed Chickadee and the Golden- the multi-dimensional site-techniquesaxis (P = crowned Ringlet were segregated by their 0.04); and (3) significant for the four-dimen- greater use of foliage as a foraging site and sional niche breadth (P = 0.05, Table 2). consistentlyhigh similarity on other axes. The Yellow-rumped Warbler, Western Tanager, NICHE OVERLAP and Black-headed Grosbeak sub-group was Insect-gleaning birds overlapped to different distinguishedby use of high foliage and use of extents on various components of the foraging a greater variety of foraging sites and tech- niche (Table 3). Basedon comparisonsof guild niques. Migrant and resident specieswere not overlaps on pairs of single-dimension axes, separated distinctly. species’ overlapped least in their use of for- aging sites (Mann-Whitney tests, P < 0.05 for DISCUSSION comparisons of site and each other axis). Tree The strong selectivity shown by the insect- species use showed the next-to-lowest guild gleaning birds, both as a guild and as separate overlap (P < 0.05), and overlaps on the heights species,indicates that vegetation structure and and techniquesaxes were higher and of similar tree speciescomposition are important habitat magnitude (P > 0.05). attributes to them. Our preference measures The pattern of overlap between specieson incorporate habitat selection by birds at two the two-dimensional axes differed slightly from levels: the selection of breeding habitat, and that on individual axes. The height-tree over- the use of occupied habitat. We suggestthat laps were lower than those on the site-tech- foraging preferences offer reasonable expla- niquesaxis (Mann-Whitney test, P = 0.05; Ta- nations as to which vegetation conformations ble 3). The relative magnitudes of overlap on may be occupied by the various guild mem- the two-dimensional axes (calculated from bers, provided that other needs (e.g., nest sites multi-dimensional niche states)were opposite and song perches) are met. from what would have resulted simply from One interpretational problem with prefer- multiplying single-axesoverlaps. encesis that values may underestimate the im- Similarities in species’ foraging niches are portance of abundant food resources.For ex- summarized in a clusterdiagram basedon four- ample, the 7-20-m foliage layer contained 45% dimensional overlaps (Fig. 5). At the 50% sim- of available foliage, and was used by the guild FORAGING OF SIERRA NEVADA BIRDS 213

TABLE 3. Mean niche overlaps among each speciesand other guild members. Two-dimensional overlaps are cal- culated from multi-dimensional niche categories.Four-dimensional overlaps are means of the productsof the overlaps on the two-dimensional axes (see text).

FOLK- One-dimensionaloverlaps Two-dimensional dimensional TIXe Height: Ht:Tr x Height species Site Technique tree sp. Site: tech. Si:Te

Mountain Chickadee 0.95 0.83 0.89 0.92 0.65 0.76 0.49 Chestnut-backedChickadee 0.94 0.90 0.88 0.95 0.67 0.77 0.52 Red-breasted Nuthatch 0.96 0.88 0.69 0.94 0.71 0.60 0.43 Brown Creeper 0.94 0.89 0.08 0.92 0.73 0.08 0.06 Golden-crowned Ringlet 0.93 0.85 0.79 0.94 0.70 0.77 0.53 Solitary Vireo 0.89 0.86 0.78 0.94 0.64 0.76 0.48 Warbling Vireo 0.95 0.78 0.66 0.92 0.58 0.62 0.36 Nashville Warbler 0.90 0.83 0.81 0.95 0.63 0.78 0.49 Yellow-rumped Warbler 0.95 0.88 0.80 0.92 0.66 0.78 0.52 Hermit Warbler 0.97 0.75 0.80 0.94 0.62 0.78 0.48 Western Tanager 0.96 0.84 0.71 0.80 0.73 0.68 0.49 Black-headedGrosbeak 0.93 0.87 0.80 0.87 0.68 0.72 0.49 Mean 0.95 0.85 0.72 0.92 0.67 0.68 0.45

in 45% of our observations. Despite the lack foliage aided their aerial maneuverability. of a positive preference value, we cannot con- Gleaners frequently used high foliage soon af- clude that the layer was not important to the ter it was exposed to sunlight. Balda (1969) guild. Similarly, white fir foliage was abundant also reported early use of high foliage, which and was used in proportion to its availability. he attributed to increasedinsect activity in the Birds may selectively use foliage at different newly-warmed foliage. heights for several reasons. Avoidance of the Preferencesof birds for various tree species two lower foliage layers was expected, as fo- reflect the abundance, type, and accessibility liage here consistsprimarily of shrubs, herbs, of prey items on trees, plus the morphological forbs, and grasses,whereas the guild was de- and behavioral characteristics of the birds fined as those insect-gleaners that predomi- (Franzreb 1978, Holmes and Robinson 198 1, nantly use forest trees. The lower layers are Sabo and Holmes 1983). The guild’s consistent used mainly by ground-feedinggranivores and avoidance of incense-cedar foliage may have insectivores,and by other foliage-gleaningbirds been because: (1) the small scale-like needles characteristic of shrubs. The Nashville War- may mature quickly, reducing the period of bler, althoughprimarily a gleanerof forest trees, susceptibility to phytophagous insects, and, used low shrub foliage more than the other guild members. Several factors may explain why the middle foliage layer was strongly preferred. This layer MT. CHICKADEE incorporates the understory of stands with a HERMIT WARBLER sparse overstory favored by the Nashville C.-B. CHICKADEE Warbler and Solitary Vireo, and the overstory G. CR. KINGLET of younger stands favored by many other species.The preference by the Brown Creeper Y.-R. WARBLER and the Red-breasted Nuthatch for this layer W. TANAGER reflects their use of furrowed bark, which is B.-H. GROSBEAK more abundant on the older, lower portions of SOLITARY VIREO trees (Travis 1977, Jackson 1979); moreover, NASHV. WARBLER becausewe defined the middle height classas including a greater vertical extent, it contained WARBLING VIREO more trunk surface area than the two lowest R. BR. NUTHATCH layers. BROWN CREEPER The highest layer was preferred to a mod- I I I I I I I I erate extent by the guild asa whole, and strong- .Bo .SD .a 20 0 ly by certain species. The Yellow-rumped SIMILARITY Warbler and Western Tanager both hawk in- FIGURE 5. Dendrogram of foraging similarities of in- sectsfrequently and may have been attracted sect-gleaningbirds at BlodgettForest. Similaritiesare based to the upper layer becauseits small amount of on 4-dimensional niche overlaps (see text). 214 DANIEL A. AIROLA AND REGINALD H. BARRETT

thus, lowering insectdensities; (2) essentialoils, striking. Alatalo (1982) found similar results which make incense-cedar wood resistant to in a Finnish conifer forest, whereas Robinson decay (Anderson 1963) and are present in fo- and Holmes (1982) found that hovering was liage (Zarvarin 1958), may also reduce foliage used much more frequently by foliage insect- digestability and, hence, insect abundance; (3) feeders in a northeastern United States hard- cedar’s densegrowth form and hangingfoliage woods forest. These differencesmay reflect bird may make it difficult for birds to search for responses to fundamental differences in the insects;and (4) much of the cedar in the study structure of trees in the two forest types (cf. area grew in dense understory thickets, which Sabo and Holmes 1983). Many conifers have may have hindered the birds’ foraging maneu- foliage arrayed along rigid horizontal branch- verability. The relatively low use of Douglas- es. In our study area, this configuration is char- fir may be a result of its physiognomy: much acteristic of white fir, sugar pine, and, to a of the foliage grows on small hangingbranches lesser extent, ponderosa pine. This structure that provide poor support for the less maneu- allows birds to hop along branches and twigs, verable guild members. Preference for the de- and gleaninsects from a relatively large amount ciduous black oak is probably due to the abun- of foliage. In contrast, hardwood foliage in the dance of insects available on emergent spring eastern U.S. generally occurs at the ends of foliage. Further study of insect productivity branches, where birds must expend more en- and bird foraging efficiency in white fir and ergy hovering to take insects (Robinson and pines will be needed before one can explain Holmes 1982). The only deciduous hardwood these preferences. in our area, black oak, bears leaves on sturdy, The pattern of tree speciespreferences shown upright twigs. Hence, birds have little difficulty by the insect-gleanersindicates the importance gleaning from its foliage. of tree species in determining habitat suita- The foraging techniques of the tanager and bility. Except for the consistent avoidance of grosbeakdiffered from those of the other guild incense-cedar, the birds generally used differ- members, owing to attempts by these large ent tree speciesin a complementary way. We birds to capture large prey. Both are “variable- suggestthat the greater diversity of tree species distance searchers” (sensu Robinson and in the mixed-conifer forest, in comparisonwith Holmes 1982) that not only take insects of all other habitats in the Sierra Nevada (Rundel et sizes by gleaning, but also search for larger al. 1977, Vemer and Boss 1980:4), promotes insectsat greater distances.The tanager hawks a higher diversity of insect-gleaningbirds. More large, slow insectsfrom exposedperches, often speciesof forest insect-gleanersbreed in this between bouts of gleaning. Although it usually type of forest than in any other forested habitat gleans,the grosbeakalso appears to search for in the Sierra (summarized from Vemer and larger insects at greater distances within the Boss 1980:86-90). tree crown below it. Upon finding larger flying Preferencesby insect-gleaningbirds for cer- or stationary prey up to 3 m away, the grosbeak tain tree specieshave been reported for other lungesdown upon it. No other speciesat Blod- geographic areas (Hartley 1953, Balda 1969, gett Forest used this technique regularly. Holmes and Robinson 198 1). Studiesof breed- The relative rankings of overlap values on ing birds in Arizona, which included some of various foraging axes among insect-gleaning the bird and tree speciesthat we studied, have speciesat the Forest are similar to results of reported different patterns. Franzreb (1978) other recent studies. At Blodgett, speciesdif- found that the three most common foliage- fered from one another most often in their use gleaning speciesin a mixed-conifer forest pre- of foraging sites,then followed by tree species, ferred Douglas-fir and white fir, and avoided and least (and nearly equally) by height and ponderosapine. Balda (1969) found that pines technique. That foraging technique differen- (Pinus leiophylla) and oaks were preferred by tiated these speciesminimally is in part a triv- foliage feeders in an oak-juniper woodland. In ial result of our use of gleaning as a criterion a forest dominated by ponderosapine, he found to define the guild. Alatalo’s (1982) study of pine and Douglas-fir used according to their this guild in a Finnish conifer forest ranked availabilities. In a pine-juniper ecotone, pon- foraging site most important, followed in de- derosa pine was under-used by insect-gleaners creasing order by tree species,technique, and (Laudenslayer and Balda 1976). More work is height. In oak woodlands in Mexico and Cal- needed to describe fully the regional and sea- ifornia, Landres and MacMahon (1983) found sonal patterns of tree speciespreferences and technique and food site to be most important, to describethe effect of tree speciesavailability and height least important, in differentiating on preference within regions. between members of the foliage- and bark- The predominant use of stationary gleaning, gleaning guilds. (They did not quantify plant rather than hovering, by all guild members is speciesuse.) FORAGING OF SIERRA NEVADA BIRDS 215

Other studieshave shown different rankings Species abundance has been positively corre- of niche overlaps within similar guilds. Root lated with niche breadth in a number of other (1967) found foraging technique, plant species, bird communities (e.g., Balda 1969, Szaro and and foraging site to be decreasinglyimportant Balda 1979). Among foliage-gleaning birds, in distinguishingfoliage-gleaners in mixed oak- Alatalo (1982) found, as we did, a stronger chaparral in California. Tits (Parus spp.) in an positive relationship between niche breadth Englishwoodland differed decreasinglyby plant and abundance of residents than that of mi- speciesuse, height, and foraging site (Hartley grants. Also, the relationship was stronger for 1953). Hutto (198 1) found that four riparian residentsin winter than in summer, suggesting western wood warblers differed most in for- that population regulation may occur at that aging heights and least in technique and site time. during both summer and winter. Our results have implications for manage- The above-mentioned differences in pat- ment of the Sierra Nevada type of mixed- terns of niche segregationprobably reflect real conifer forest. Insect-gleaning birds appear differences among habitats and geographic especially likely to be affected by forest man- areas. This conclusion remains somewhat un- agement practices that alter tree species di- certain, however, owing to differences in the versity. Fire suppressionand early logginghave way various workers have defined guilds and altered tree species composition in much of niche components. More comparative studies the mixed-conifer type forests, increasing in- using standardized methodologies are clearly cense-cedar and white fir, and reducing the needed. extent of pines (Kilgore 197 1, Rundel et al. Our analysis revealed few fundamental dif- 1977). Use of prescribed fire and other tech- ferencesin the foraging niches of resident and niques to re-establish natural speciescompo- migratory speciesduring the breeding season. sition may be beneficial to this guild. Current Most striking was the greater use of conifers timber management and fuelwood cutting by residents and the greater use of oak by mi- practices have also substantially affected tree grants. Most residents avoided deciduous fo- speciesdiversity and other habitat conditions liage, and used substratesthat were available in this forest type (Vemer 1980). Based on our year-round (conifer foliage, branches, and tree results, we believe that efforts should be in- trunks). Presumably, this reflects ancestralad- creased to maintain a high diversity of tree aptations of the palearctic-derived residents speciesin standsmanaged for timber, by using (Mayr 1976), and these species’ need to retain both even- and uneven-aged forest manage- abilities to forage efficiently on substrates ment techniques to encouragea variety of tree available during the winter when prey are scarce species,by planting a mixture of trees in clear- (Sabo and Holmes 1983). In contrast, most cuts, and by retaining oaks and managing for migrants preferred the productive, seasonal, their replacement over time. hardwood foliage more characteristic of neo- tropical areas from where these speciesare de- ACKNOWLEDGMENTS rived (Mayr 1976). We emphasize, however, We especiallythank C. J. Ralph for his advice and support, that differencesin foliage-type preferenceswere M. G. Raphael for his discussionof many ideas and his only relative. Migrant specieshave adapted to review of the manuscript, and P. Hall for her assistance with analysis.Y. Cohen, R. Heald, P. V. Brylski, W. Stem, make substantialuse of conifer foliage. Among and R. C. Colwell contributed importantly to various as- his foliage-gleaning birds, Alatalo (1982) also pects of the study. S. Hejl, J. Vemer, M. L. Morrison, N. found that migrants used greater amounts of K. Johnson, R. T. Holmes, and an anonymous reviewer deciduous foliage than did residents. provided valuable comments on various drafts of the manuscript. The Department of Forestry and Resource Residents and migrants also differed in the Management, University of California, Berkeley,provided relationship of speciesforaging-diversity and facilities at the study site. Financial support was provided abundance. The positive correlation between by McIntire-Stennis project 3501; USDA Forest Service, resident species’ abundances and niche Region 5; and USDA Pacific SouthwestForest and Range breadths is consistentwith the idea that birds Experiment Station; and the Wollenberg Foundation. Will Augsburgerprepared the manuscript. with greater behavioral flexibility may achieve higher populations.We do not suggestthat such population regulation is likely to occur during the breeding season;rather, if it occurs at all, LITERATURE CITED it is more likely to do so in winter. The lack AIROLA,D. A., AND R. H. BARRETT. 1981. Vertebrates of a relationship between abundanceand niche of BlodgettForest ResearchStation, El Dorado Coun- breadth in migrants supports the widely held ty, California-an annotated specieslist. Univ. Calif. Div. Agric. Sci., Special Publ. 3267, Berkeley. view that a more complex set of factors reg- ALATALO,R. V. 1982. Multidimensional foraging niche ulates the abundanceof migrant specieson the organizationof foliage-gleaningbirds in northern Fin- breeding grounds (Keast and Morton 1980). land. Omis Stand. 13:56-7 1. 216 DANIEL A. AIROLA AND REGINALD H. BARRETT

ANDERSON,A. B. 1963. The chemistry of decay resis- the diversity of life: selected essays.Belknap Press, tance and its decrease with heartwood aging in in- Cambridge, MA. cense-cedar (Libocedrus decurrens Torrey). Holz- MORRISON,M. L. 1984. Influence of sample size and forschung 17:l-5. sampling design on analysisof avian foraging behav- BALDA,R. P. 1969. Foliage use by birds of the oak- ior. Condor 86: 146-150. juniper woodland and ponderosapine forest in south- MUELLER-D• MBOIS,D., AND H. ELLENBERG.1974. Aims eastern Arizona. Condor 71:399-412. and methods of vegetation ecology. John Wiley and DEDON,M., AND R. H. BARRETT. 1982. An inventory Sons, New York. system for assessingwildlife habitat relationships in ROBINSON,S. K., AND R. T. HOLMES. 1982. Foraging forests. Cal&Neva. Wildl. Trans. 55-60. behaviorof forestbirds: the relationshipsamong search DIXON, W. J. [ED.]. 1981. BMDP Statistical Software. tactics, diet and habitat structure. Ecology 63: 1918- Univ. of California Press, Berkeley. 1931. DUNN, 0. J. 1961. Multiple comprisons among means. ROOT,R. B. 1967. The niche exploitation pattern of the J. Am. Stat. Assoc. 56:52-64. Blue-gray Gnatcatcher. Ecol.-Monogr. 37:3 17-349. FEINSINGER,P., E. E. SPEARS,ANLI R. W. POOLE. 1981. RUNDEL.P. W.. D. J. PARSON.AND D. T. GORDON. 1977. A simple measure of niche breadth. Ecology 62:27- Montane and subalpine vegetation of the Sierra Ne- 32. vada and CascadeRanges, p. 559-599. In M. G. Bar- FRANZREB,K. E. 1978. Tree species used by birds in bour and J. Major [eds.], Terrestrial vegetation of loggedand unloggedmixed coniferousforests. Wilson California. John Wiley and Sons, New York. Bull. 90:221-238. SABO,S. R., AND R. T. HOLMES. 1983. Foraging niches FRANZREB,K. E., AND R. D. OHMART. 1978. The effects and the structureof forest bird communities~incon- of timber harvesting on breeding birds in a mixed- trasting montane habitats. Condor 85: 12 1-I 38. conifer forest. Condor 80:43 l-44 1. SHANNON,2. E., AND W. WEAVER. 1949. The mathe- HARTLEY,P. H. T. 1953. An ecological study of the matical theory of communication. Univ. of Illinois feeding habits of the English titmice. J. Anim. Ecol. Press, Urbana. 22:261-288. SOKAL,R. R., AND F. J. ROHLF. 1969. Biometry. W. H. HOLMES,R. T., AND S. K. ROBINSON.1981. Tree species Freeman and Co., San Francisco. preferencesof foraginginsectivorous birds in a north- STRAUSS, R. E. 1979. Reliability estimates for Ivlev’s em hardwoods forest. Oecologia (Berl.) 48:31-35. electivity index, the forageratio and a proposedlinear HORN, H. S. 1966. Measurement of “overlap” in com- index of food selection. Trans. Am. Fish. Sot. 108: parative ecologicalstudies. Am. Nat. 100:419-424. 344-352. Hurro. R. L. 1981. Seasonalvariation in the foraaina SZARO,R. C., AND R. P. BALDA. 1979. Bird community behavior of some migratory western wood warbiers. dynamics in a ponderosa pine forest. Stud. Avian Auk 98~765-777. Biol. 3. INGER,R. F., AND R. K. COLWELL. 1977. Organization TRAVIS,J. 1977. Seasonalforaging in a Downy Wood- of contiguouscommunities of amphibians and rep- pecker population. Condor ?9:371-375. _ tiles in Thailand. Ecol. Monogr. 47:229-253. VERNER.J. 1980. Bird communities of mixed-conifer JACKSON,J. A. 1979. Tree surfacesas foragingsubstrates forests of the Sierra Nevada, p. 198-224. In R. M. for insectivorous birds, p. 69-94. In J. G. Dickson, DeGraff [tech. coor.], Management of western forests R. N. Conner, R. F. Fleet, J. C. Knoll, and J. A. and grasslandsfor nongame birds. U.S. Dep. Agric. Jackson[eds.], The role ofinsectivorousbirds in forest For. Serv. Gen. Tech. Rep. INT-86. ecosystems.Academic Press, New York. VERNER,J., AND A. S. Boss [tech coor.]. 1980. California KEAST,A., AND E. S. MORTON[EDS.]. 1980. Migrant birds wildlife and their habitats: western Sierra Nevada. in the Neotropics: ecology,behavior, distribution and U.S. Dep. Agric. For. Serv. Gen. Tech. Rep. PSW- conservation. Smithsonian Institution Press, Wash- 37. ington, DC. WAGNER,J. L. 1981. Visibility and bias in avian foraging K~LGORE.B. M. 1971. Resoonsesof breedina birds to data. Condor 83:263-264. habitat changesin a giant sequoia forest. irn. Midl. ZAVARIN,E. 1958. The extraneoussubstances ofincense- Nat. 85:135-152. cedar (Heyderia decurrensTorrey). California For- LANDRES,P. B., AND J. A. MACMAHON. 1983. Com- estrv and Forest Products. No. 5. Univ. California munity organization of arboreal birds in some oak Ag&ultural Experiment Station, Berkeley. woodlands of western North America. Ecol. Monogr. 53:183-208. Department of Forestryand ResourceManagement, Uni- LAUDENSLAYER,W. F., JR., AND R. P. BALDA. 1976. versitv of California. Berkelev. California 94720. and Breedingbird useof a pinyon-juniper-ponderosapine USDA Firest iervice, PacificS&hweit Forest and Range ecotone. Auk 93:571-586. Experiment Station, Arcata, California 95521. Addressof LEVINS,R. 1968. Evolution in changingenvironments. second author: Department of Forestry and Resource Princeton Univ. Press, Princeton, NJ. Mangement, Universityof California, Berkeley,California MAY, R. M. 1975. Some notes on estimating the com- 94720. Presentaddress ofjirst author: USDA Forest Ser- petition matrix, (Y.Ecology 56:737-741. vice, Lassen National Forest, Chester,California 96020. MAYR, E. 1976. History of the North American bird Received 11 January 1984. Final acceptance26 November fauna, p. 565-588. Zn E. Mayr [ed.], Evolution and 1984.