The Condor 101:299-310 0 The Cooper Ormthological Society 1999

TITS, WARBLERS, AND : FOLIAGE-GLEANING OF NEARCTIC AND PALEARCTIC BOREAL FORESTS ’

RUSSELL GREENBERG Smithsonian Migratory Center, National Zoological Park, Washington, DC 20008, e-mail: @erols.com

VLADIMIR PRAVOSUDOV~ Department of Zoology, Ohio State University, Columbus, OH 43210

JOHN STERLING Smithsonian Migratory Center, National Zoological Park, Washington, DC 20008

ANNA KOZLENKO 4075 Monticello Blvd., Apt. 305, Cleveland Heights, OH 44121

VITALLY KONTORSHCHIKOV State Darwin 57, Vasilova, Moscow, Russia 117292

Abstract. We describe two major patterns of abundance of different types of canopy foliage-gleaning birds at four sites (northwestern Canada, eastern Canada, European Russia, and central Siberia) in boreal forest. First, because of the extreme numerical dominance and breadth of habitat use of the Chaffinch (Fringilla coelebs), the European Russian sites stand as outliers to the relationships between the number of species and the abundance of all foliage-gleaning birds, as well as to more detailed patterns of the composition of the foliage- gleaner guild. The Chaffinch is one of the few canopy foliage-gleaning birds in temperate forests that is abundant in agricultural habitats during the nonbreeding season, which prob- ably allowed a rapid expansion of its range and growth in abundance with the advent of farming in Europe. The second pattern emerges when the European Russian sites are ex- cluded from analyses: the proportional abundance and number of species of warblers or warbler-like birds were positively related to overall abundance of foliage gleaners. This suggests that warblers prosper when there is high resource productivity during the breeding season. A possible resolution between the patterns found in European Russia and the other boreal forest sites may be that Fringilla are functionally equivalent to warblers and have essentially replaced warblers to achieve their high abundance. The patterns of relative abun- dance give support to both the Resource Seasonality Hypothesis (warblers versus tits), and the Ecological Replacement Hypothesis (Fringilla versus warblers) as explanations for shifts in the numerical dominance of different types of foliage gleaners in different regions. Key words: boreal forests, community convergence, community structure, continental comparison, foliage-gleaning bird, Fringilla, warblers.

INTRODUCTION is quite comparable. Many of the dominant high-

The study of similar habitats in different conti- er taxa of songbirds, however, are only distantly nents is a common approach to testing general- related between the two continents (Haila and ities in the patterns of avian species assemblag- Jtivinen 1990). Variation in the taxonomic com- es. Boreal forests are an attractive focus for position of forest avifaunas has fueled the search these types of studies because they form a con- for ecological equivalents between regions, as tinuous band acrossNorth America and Eurasia well as for biogeographic and ecological pro- of low diversity forests dominated by a few cessesthat have determined the composition of shared genera of canopy trees. Although one-to- different assemblages(Morse 1989). one correspondencebetween forest types may be The canopy foliage-gleaner guild is the most elusive, the range of habitats available to birds abundant and species-rich guild acrossall boreal forests. Various studies have compared the ecol- ogy of specific paired groups of foliage gleaners, ’ ’ Received24 February 1998. Accepted6 January particularly the Old World versus New World 1999. 2Present address: Department of Biological Scienc- warblers (Cody 1974, Winkler and Leisler es, Purdue University, West Lafayette, IN 47907-1392. 1994). The problem with this approach is that

[2991 300 RUSSELL GREENBERG ET AL. the flush of foliage insects in temperate forests types of foliage gleaners in different boreal for- often supportsthe breeding of a morphologically est habitats in light of these hypothesesand with or ecologically diverse group of species.Species an emphasis on comparing Palearctic and Ne- that forage on foliage arthropodsare often war- arctic sites. Considering survey data gathered bler or warbler-like birds, but also include tits, from sites in different boreal forest regions, we finches, and flycatchers. The relative impor- addressthe following questions: (1) What is the tance, both in terms of species richness and pattern of variation in relative abundance of dif- abundance, of these different types of foliage ferent foliage-gleaner types in different regions gleaners varies considerably between regions. of the boreal forest? (2) What is the relationship The search for ecological equivalents between between species richness and abundance of fo- sites should, therefore, be expanded beyond spe- liage-gleaning birds and different types of fo- cific taxonomic pairings to include all foliage- liage-gleaning birds? (3) Is there a predictable gleaning species. relationship between the abundance of different Two hypotheses have been put forth to ex- types of foliage-gleaning birds and the overall plain intercontinental and regional differences in abundance of foliage-gleaning birds? (4) To the dominance of different types of foliage what degree do the aforementioned relationships gleaners, particularly with respect to the relative vary between different sites and continents in importance of tits versus warblers (Sylviinae or the boreal forest? Parulinae). The Ecological Replacement Hy- pothesis posits that as a result of historical pat- METHODS terns of colonization and diversification, one STUDY SITES type of foliage gleaner usurps resourceslimiting the ecological opportunities for other types. For Fieldwork was conducted in successive sum- example, New World assemblages support a mers: central Siberia 1993, northwestern Canada large number of species of warbler or warbler- 1994, European Russia 1995, and eastern Can- like birds, whereas Old World assemblages(par- ada 1996. We selectedfour study sites organized ticularly those in southern or western Europe) into two comparable pairs (Fig. 1). The first pair often have a higher number of species(Haila is “middle taiga” sites with highly continental and Jtivenin 1990). This has led some (Morse climates: the southern MacKenzie District of the 1989) to speculate that tits replace wood war- Northwest Territories (MacKenzie) and the mid- blers to a large degree in Palearctic forests and dle section of the Yenisey River between Tun- to suggest that in southern Europe tits replace guska and Angara Rivers (Siberia). Both areas Old World warblers (Herrera 1978). were centered on a major river system (Yenisey The second hypothesis is more explicitly eco- and Mackenzie are the fourth and seventh lon- logical. The Resource Seasonality Hypothesis gest rivers in the world, respectively) and had a posits that certain types of foliage gleaners are similar range of latitudes (Siberia, 6O”lO’ to specialized on harvesting the bloom of summer 62”15’N; MacKenzie, 60”14’ to 61”52’N). Re- arthropodsand become more common where the search was conducted at three sites each in breeding season resource peak is more pro- Northwest Territories and central Siberia. The nounced. For example, Haila et al. (1987), high- Siberian sites (Mymoe, Vorogova, and Fomka) lighting the negative correlation between tit and were a maximum of 295 km apart and the max- Phylloscopus abundance in European census imum distance was 340 km for the MacKenzie data, suggestedthat the latter become more nu- sites (Fort Liard, Fort Simpson, and Fort Provi- merically dominant in more seasonal habitats. dence). The second pair of study areas was lo- Related to this hypothesis,Wiens (1975) and Ra- cated in less continental regions characterized as benold (1978) proposedthat the high abundance “South Taiga.” Both the European Russian (Eu- and diversity of wood warblers in Northeastern rope) and eastern Canadian (Ontario) sites were coniferous forests of North America is related to far more disturbed and developed than the cen- the large pulse of food during the breeding sea- tral Siberian and Mackenzie areas. Due to logis- son in these forests compared to Western and tical constraints, the Russian fieldwork was con- Southern forests. ducted primarily within 40 km of the Kostrom- This paper is an initial step toward under- skaya Taiga Field Station located on the Unja standing the relative importance of the major River (58”14’N, 55”25’N). However, 10% of the BOREAL FOREST FOLIAGE-GLEANING BIRDS 301

FIGURE 1. Map depicting location of study sites in Canada and Russia. Sites marked with a square are in northern taiga and those with a circle are in southerntaiga. survey points were conducted at the Zvenigorod poplar, alder (Ahus tenuifoliu), and willow, (4) Research Station 600 km southwest (near Mos- trembling aspen stands (Populus tremuloides), cow) with the addition of no species. The On- and (5) jack pine (Pinus bunksiunu) forests. In tario sites were primarily in the clay belt be- the European Russian site we sampled (1) tween the cities of Gogama, Timmins, Cochran, mixed, upland second-growth taiga (approxi- and Hearst, as well as White River and Mara- mately 60 years old), with spruce (Piceu ubieu), thon with a maximum intersite distance of 490 aspen, linden (Tilliu cordutu), and birch (Beth km and a latitudinal range of 47”51’N to pubescens), (2) birch-aspen forest, (3) European 4Y30’N. The lower latitudes of locations in On- pine (Pinus sylvestris) forest, and (4) riparian tario reflect the overall southern distribution of elm-alder (Urnus-Ahus incunu) woods. In On- the boreal forest in this region. tario we sampled (1) aspen stands (aspen, birch [Betulu pupyriferu], and balsam poplar), (2) HABITATS SAMPLED mixed upland taiga (white spruce, balsam fir We sampled four types of mature forest habitat [Abies bulsumeu], jack pine, birch, aspen, and in Siberia: (1) riparian alder (Alnaster frutico- poplar), (3) black spruce, and (4) jack pine sus) and willow (Sulh spp.), (2) mixed flood- stands. plain forest, with spruce-fir (Pica obovatu-Abi- es sibiricu) and birch (Beth spp.), (3) upland POINT COUNTS birch standswith some aspen (Populus tremulu), Because bird censusing techniques often differ and (4) mixed upland taiga with Siberian pine between studies and regions, it is often difficult (Pinus sibiricu), spruce, larch (Larix sibiricu), to make systematic comparisons. For the pur- fir, and birch. We surveyed five habitat types in posesof this analysis, we gatheredfield data em- MacKenzie: (1) black spruce forest (Piceu mur- ploying the same point-count methodology at all ianu), (2) mixed flood-plain forest, with white sites. Point counts (Hutto et al. 1986, Petit et al. spruce (Piceu gluucu) and balsam poplar (Po- 1995) consisted of a series of up to 15 5-min pulus bulsumiferu), (3) riparian, with balsam counts located 200 m apart. Censuseswere con- 302 RUSSELL GREENBERG ET AL. ducted 30 min after sunrise for approximately 3 black spruce near Gogama (12), Timmins (1 l), hr when wind speedsare less than 12 km hr’. Cochran (53), Marathon (6), and Obatanga (5). The number and approximate distance of all birds were recorded as well as the mode of de- PUBLISHED CENSUS DATA tection (song, call note, strictly visual). As much Point counts provide a standardized rapid as- as possible, all points on a transect were located sessment, which allows direct comparison be- in a single habitat type at a minimum of 100 m tween sites. We have examined published cen- from a habitat edge, which requires large tracts suses based mainly on territorial mapping or of habitat for surveying. Point counts were con- more rarely on repeated fixed transectsto assess ducted from 5-25 June, which correspondswith the generality of some of the qualitative results the peak singing activity of migratory species. of the analysis from our point count data. The Resident species (tits) initiate breeding earlier censuseswere selected to cover a broad habitat than migrants and so were probably vocalizing and geographic range, but are by no means ex- less than migrants during our work. Although haustive. The data set (available from first au- this may cause underestimation of tit popula- thor) includes 21 censuses from eastern North tions, we regard this as an equivalent bias for all America, 6 from western Canada, 28 from sites. In addition, a comparison of our point northern Europe (West of Urals), and 20 from count data with published census data from the Siberia. We calculated the mean value (or in the same region showed a high correlation (r = case of Breeding Bird Censuses from Canada, 0.95) between the two data sets with respect to we used published means following Erskine composition of different types of birds and no [1977]) when more than one survey was con- significant underestimation of tits (see Results). ducted in the same habitat and site.

Because of the severe problems of differential CLASSIFICATION OF FOLIAGE-GLEANER TYPE detectability between species and habitats, we We subdivided the canopy foliage-gleaning have relied upon the mean number of birds with- guild into four types, corresponding to differ- in 50 m of the point to assessfoliage gleaner ences in morphology or, in the case of flycatch- abundance (Petit et al. 1995). In order to stan- ers, large differences in foraging behavior: (1) dardize the species list to assessoverall species tits (species of Paridae or Aegithalos), (2) - richness, we used the total number of species es (Fringilla, Emberizids, ), (3) war- detected with an abundance greater than 0.02 in- blers (Parulinae, Vireonidae, and the probably dividuals per point, and standardized effort by polyphyletic [Sheldon and Gill 19961 Sylvinae randomly subsampling the data from 75 points. including Regulus and Slyvia), and (4) flycatch- The summary of individual species by site is ers (Empidonax, Ficedula, Erithacus, Phoeni- available at the Smithsonian Migratory Bird curus). The last group includes species that for- Center web site (http:Nwww.si.edu/smbc). age predominantly from leaves, twigs, and In MacKenzie, we sampled mixed white branches with a mean attack distance > 1 m, but spruce near Ft. Liard and Black Mesa (91 point not those such as Muscicapa and Contopus spp. counts, see below), Ft. Simpson (20), and Ft. which forage for aerial arthropods.Two uncom- Providence (27); black spruce near Ft. Liard mon species (Gray Jay Perisoreus canadensis, (60), Ft. Simpson (32), and Trout Creek (17); and Golden Oriole Oriolus oriolus) that were not riparian deciduous near Ft. Liard (55) and Ft. easily classified were ignored in the analyses. Simpson (35); aspen near Ft. Liard (43) and Ft. We have included some speciesthat are often Simpson (32). In Siberia, we sampled upland not considered foliage-gleaning species. Based taiga near Myrnoe (177), Volgova (45), and on our foraging observations at our study sites, Fomka (58); floodplain taiga near Myrnoe (32), these species probably do a substantial amount Volgova (64), and Fomka (57); birch near Vor- of their foraging off foliage or twigs. The non- ogova (61) and Fomka (48); and riparian near traditional speciesinclude some that also forage Mymoe (43), Vorogova (48), and Fomka (57). on the ground (European Robin Erithacus ru- In Ontario, we sampled upland mixed forest near beculu, Dark-eyed Junco Junco hyemalis, Chip- Marathon (40), Hearst (57), Cochrane (18), ping Sparrow Spizella passerina, and Yellow- White Lake (28), and Timmins (5); aspen-poplar breasted Bunting Emberiza aureola). These spe- near Timmins (91) and White River (9); and cies were commonly found in the canopy. How- BOREAL FOREST FOLIAGE-GLEANING BIRDS 303

ever, with ground-foraging birds it is often portions of finches. The New World sites had a difficult to determine how much of their forag- significantly greater proportion of warblers ing time is spent on the ground, because there (72%; H = 12.3, P = 0.006). We found no over- they are more difficult to observe. We also in- all significant difference in the proportion of tits clude some Cardueline finches, which are well and flycatchers acrosssites. However, except for known to feed on seeds extensively, even when the riparian vegetation, with its lack of cavities, rearing young (Newton 1967). In the forest hab- central Siberia had a strikingly high proportion itats we studied, siskins and bullfinches forage of tits (33%) compared to other regions. commonly for foliage arthropods during the height of the breeding season. These species SPECIES RICHNESS VERSUS ABUNDANCE OF were generally uncommon on our surveys and FOLIAGE GLEANERS AND FOLIAGE-GLEANER TYPES exclusion of these species does not greatly alter the major conclusions of this paper. The correlation between species richness and abundance of foliage gleaners across habitats RESULTS was significant and strong (r = 0.85, P < 0.001) NUMERICAL DOMINANCE, ABUNDANCE, AND with European Russia excluded, and somewhat COMPOSITION OF FOLIAGE-GLEANING weaker, but still significant, with the Russian GUILDS habitats included (r = 0.73, P < 0.001, Fig. 2); We found significantly more foliage-gleaning in- the Russian sites are brought into conformity dividuals per point in European Russia than oth- with other sites by removing the Chaffinch num- er sites, largely a result of the very high abun- bers (9 = 0.83, P < 0.001). dance of Chaffinches in all habitats (ANOVA The relationship between speciesnumber and based on mean per habitat, overall habitat effect abundance across foliage-gleaner type (warbler, F 3,,1 = 28.2, P = 0.006, Tukey’s post-hoc com- finch, etc.) also was strong for the pooled North parison was significant only for Europe versus American and Siberian sites (r = 0.96, P < other sites). The upland Siberian sites had low O.OOl), but considerably weaker if the European densities compared to similar North American Russian sites are included (r = 0.72, P < 0.001). sites (Table 1). Jack pine forest on sandy glacial Species richness and abundance of type were outwash is equivalent to European pine forests significantly correlated when considered sepa- of Siberia, which were not surveyed and so were rately for Siberia (r = 0.91), MacKenzie (0.84) not considered in this comparison. The dominant and Ontario (0.97) (all P values < O.Ol), but not species varied among habitats in the two North European Russia (0.54). American and Siberian sites (Table l), with war- blers generally dominating the former and tits ABUNDANCE OF FOLIAGE-GLEANER TYPES the latter. However, the Chaffinch was the dom- VERSUS OVERALL FOLIAGE-GLEANER ABUNDANCE inant foliage-gleaner in all four habitats at the European Russian site. The average proportion It is likely that the variation in overall abun- of Chaffinches acrosshabitats in European Rus- dance of foliage-gleaning birds reflects differ- sia was 0.44. The highest values for the other ences in food resources of different habitats. sites were Willow Tit (Poecile montanus) in Si- Based on this assumption, we have plotted the beria (0.17), TennesseeWarbler (Vermivora per- relative abundance of different foliage-gleaner egrina) in MacKenzie (0.19), and Yellow-rum- types against total foliage-gleaning bird abun- ped Warbler (Dendroica coronata) in Ontario dance (birds per point). We conducted ANCO- (0.11). VA using site and the mean total number of fo- In terms of numbers of individuals, the rela- liage gleaners per point as independent variables tive importance of different types of foliage and the mean number of individuals of different gleaners varied among the different study sites. types as the dependent variable. The analysis The proportion of finches varied significantly was first run with the site by abundance inter- (Kruskal-Wallis H = 11.0, P = 0.01) with the action term. When there was no significant in- highest proportion (56%, Table 1) in European teraction, the ANCOVA was then run without Russia, also a consequenceof the high Chaffinch the interaction term to test for the effect of site abundance. Siberia had a moderate (26%), and and abundance (Wilkinson 1990). Because of the New World sites low (15% and 9%), pro- the anomalously high abundance of Chaffinches

BOREAL FOREST FOLIAGE-GLEANING BIRDS 305

. I . . .

l Europe 0 Siberia 0 MacKenzie 1A Ontario

I 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 Mean foliage-gleaning birds per point per habitat 1 2 3 4 ? FIGURE 2. The mean number of foliage gleanersper type per habitat versusthe number of speciesper type of foliage-gleaners in 75 points in each habitat with 14 abundance> 0.02 per point. . 12 - at the European Russian site, we analyzed the data with all sites and with the European Rus- sian data excluded. When we eliminated the Eu- ropean Russian habitats from the analysis, with their overriding dominance of Chaffinches, we found a strong relationship between the abun- dance of warbler-like birds and foliage-gleaner abundance (Fig. 3A). The ANCOVA (overall r 1 2 3 4

= 0.96) showed a large effect of bird abundance Mean number of foliage-gleaning birds per point (F, 9 = 336.8, P < 0.001) and a smaller effect FIGURE 3. A. The mean number of total foliage of site (F2,9 = 17.6, P < 0.001). When European gleaners per point versus the mean number of foliage Russia was included, the correlation between gleaning birds per point for each habitat surveyed. warbler abundance and overall foliage-gleaner Note that the four points at the far right of the graph abundance was considerably weaker, but still for each type are European showing the reversal of the significant (Y = 0.69, P = 0.001). position of finches and warblers from the relationship of other sites. B. The total number of foliage gleaner We found a weak or nonexistent relationship species per type per habitat versus the mean number between the abundance of other types and over- of foliage-gleaning birds per habitat. all foliage-gleaner abundance (Fig. 3A). In terms of relative abundance (the proportion of foliage- gleaners of a given type) and with European abundance and overall bird abundance (r = Russia excluded, warblers showed a significant 0.99) similar to that found for warblers in the increase with increasing foliage-gleaner abun- three non-European communities alone. The dance (r = 0.81, P = 0.001) and tits showed a ANCOVA produced a strong effect of overall weaker, but significant, negative relationship (r foliage-gleaner abundance (F,,,2 = 394.3, P < = -0.58, P < 0.01). With European Russia in- 0.001) and a weaker effect of site (F3,,2 = 15.5, cluded, only finches showed a significant cor- P < 0.01). relation with abundance (r = 0.64, P = 0.006). The number of species of warblers also was When the relationship between warbler abun- highly correlated with the abundance of foliage- dance and overall bird abundance (Fig. 3A) for gleaning birds (r2 = 0.85, P < 0.001) when the all sites is examined, the high abundance of European site was excluded (Fig. 3B) and not Chaffinches and the lower than expected abun- significantly correlated when these data were in- dance of warblers at the European sites suggest cluded (Fig. 4, r = 0.39, P = 0.16). An AN- a possible functional relationship. Summing the COVA on warbler species number (European abundance of Fringilla and warblers produces Russia excluded) produced no significant inter- an extremely tight relationship between their action between site and overall abundance. The 306 RUSSELL GREENBERG ET AL.

1.0

l Europe x09 * 0 Siberia 0 MacKenzie go6 A $ A 0 1l Ontario

I I 10 30 50 70 90 0.05 0.15 0.25 0.35 0.45 0.55 0.65 f Mean percentage of individuals per type published

1.0 FIGURE 5. Mean percentage of individuals in dif- I 0.9 1 1,.0 SiberiaEurope ferent foliage-gleaner types for census plots from a re- B A 0 MacKenzie gion (e.g., eastern North America, northern Europe, ;a 0.6 lo A Ontario etc.) versus the mean percentage of individuals in dif- ferent foliage-gleaner types for point counts from the same region (this study).

cause of the three Siberian habitats with high tit and low warbler abundance. When European Russian sites were excluded, the dominant neg- I I 0.05 0.15 0.25 0.35 0.45 0.55 0.65 ative correlation was between warblers and tits Mean proportion of birds per point (r = -0.73, P = 0.005) with a somewhat weak- er, yet significant negative correlation between FIGURE 4. A. Proportion of finches versus warblers for each habitat surveyed with point counts. B. Pro- warblers and finches (r = -0.61, P = 0.03). In portion of tits versus warblers for each habitat sur- the case of the analyses with and without Eu- veyed with point counts. ropean Russia, we found no significant relation- ship between tits and finches (Y = 0.10 and 0.28, respectively), nor did we find any significant re- ANCOVA without interactions produced a sig- lationships between the proportion of flycatchers nificant model (Y = 0.89, P < 0.001) with only and any other type. foliage-gleaner abundance entering as a signifi- PATTERNS IN PUBLISHED CENSUS DATA cant variable (F,,g = 20.8, P < 0.01). The di- versity of other types was not significantly re- Based on the comparison with the census plots lated to bird abundance or site. or transects,the patterns of the point count data are robust (Fig. S), particularly considering that COMPLEMENTARITY OF ABUNDANCE OF the geographic and ecological range of the pub- DIFFERENT FOLIAGE-GLEANER TYPES lished data were greater than for our study. The A strong negative correlation (r = -0.79, P = proportion of finches was highest (41.2%) in 0.001, Fig. 4A) was found between the propor- northern European and western Canadian tion of finches and warbler-like birds acrossdif- (39.3%) communities and lowest in eastern ferent habitats. New World communities are North America (11 .O%). Warblers were propor- mostly representedat the high warbler/low finch tionately most common in eastern North Amer- portion of the graph, and European communities ica (75.7%) and western Canada (43.5%) and on the opposite end. Siberian communities show comparably low in the Palearctic regions (36%). the greatest scatter because these habitats sup- Tits were relatively common in Siberia (33.8%) port the greatest number of tits, which is not and uncommon in northern Europe and North reflected in this relationship. On the other hand, America (4.1-10%). Flycatchers were relatively the correlation between the proportion of war- common in Europe (14%) and rare elsewhere bler-like birds and tits was suggestive, but not (0.8-6.9%). The overall patterns of foliage- significant (Y = -0.48, P = 0.08, Fig. 4B). The gleaning guild composition were surprisingly suggestionof a relationship occurs primarily be- similar when plot and point count data were BOREAL FOREST FOLIAGE-GLEANING BIRDS 307 compared. The correlation between mean pro- PATTERNS ACROSS NON-EUROPEAN SITES portion of type in a point count region and plots When the European site was excluded, the abun- from the comparable region (e.g., eastern North dance and number of species of warblers America vs. Ontario; northern Europe vs. Kos- showed a highly significant relationship with troma, etc., with only spruce and pine habitats overall foliage gleaner abundance with only a included for western Canada) had an r of 0.95 small effect of site, whereas the abundance of with a slope of 1.10 (Fig. 5). finches and tits was unrelated to overall abun- As in the boreal point-count data set, the pro- dance of foliage gleaners. This observation sup- portion of warblers was related to the proportion ports the Resource Seasonality Hypothesis for of both finches (r = -0.61, P < 0.001) and tits the importance of warblers in different boreal (r = -0.58, P < O.OOl), values which are more forest habitats. According to this hypothesis, the similar to each other than those for the point high proportion of tits in many Siberia habitats count data (-0.79 and -0.48, respectively). is primarily a result of the reduced summer re- When only Siberia and North America data were sources and reduced warbler abundance. Tit included, we found the major negative correla- populations are generally higher in Siberia, but tion between tits and warblers (r = -0.74, P < the absolute number (overall, 0.26 per point) O.OOl), with a weaker relationship between comprises a small portion of breeding foliage- finches and warblers (r = -0.57, P < 0.001). gleaner assemblages (Fig. 4), and variation in As in the point count data, the relationship be- the proportion of tits and warblers is largely de- tween tit and finch abundance was weak (r = termined by a large increase in abundance of 0.11 and 0.10 for analysis with and without warblers with increasing foliage-gleaner density. northern Europe, respectively). None of the cor- In contrast, the high proportion of tits in south- relations between proportion of flycatchers and ern Europe (Herrera 1978, Haila and Jkvinen other types were significant. 1990) is associated with high absolute abun- Chaffinch was the most common foliage- dances and may be driven more by mild winter gleaning species in 19/25 of the northern Euro- climates and greater winter food resources. pean communities, occurring commonly in We stressthe Resource Seasonality Hypothe- broad-leafed and coniferous habitats alike, ac- sis for explaining variation in the proportion of counting for an average of 30% in northern Eu- warblers and tits in boreal forest. However, the rope (range 2-57%). Chaffinches were scarce consistently higher abundance of tits in Palearc- only in montane sites in Norway, and far north tic habitats is related to their higher regional spe- sites (northern Finland and Pechora region of cies richness and may have a historical expla- Russia). Probably the most consistently domi- nation. Still, tit abundance in Siberia is high nant New World speciesis Red-eyed (Vir- even where only one common species of tit eo olivaceus). This species was the dominant (Willow Tit) was found, which suggeststhat an speciesin only 5 of 20 censusesin eastern North ecological explanation is required as well. America, comprising 13.6% of the foliage- gleaners in thosehabitats, and was almost absent OVERDOMINANCE OF CHAFFINCHES IN from coniferous forests. EUROPEAN ASSEMBLAGES Many of the patterns discerned in the North DISCUSSION American and Siberian data disappearedor were Given the clear overdominance of Chaffinches actually reversed when the European data were in most European communities, we separatedis- included. For example, the relationship between cussion of the distribution of foliage gleaning foraging-type abundance and species richness types into three issues: (1) what were the pat- diverges strongly from those of the other sites. terns of abundance of foliage-gleaner types in The proportion of warblers actually declines the non-European communities? (2) what ex- with foliage-gleaner abundance when the Euro- plains the numerical importance of Chaffinches pean points were included in the analysis. All of in European forest bird communities? and (3) these deviations can be attributed to the strong can the discrepancies between the European over-dominance of Chaffinches in all four hab- habitats and other regions be explained by the itats surveyed in European Russia. hypothesis that Fringilla functionally replace Except where it is replaced by the warblers? (Fringilla montifringilla) in the far north, the 308 RUSSELL GREENBERG ET AL.

Chaffinch is a dominant foliage gleaning bird in ropean boreal forests. To a large degree, this pat- all European forests. It comprises from 15-45% tern can be explained by historical factors. The of the individual foliage gleaners, with the high- center of origin of Phylloscopus,the most forest- est proportions found in northern European for- adapted of the Old World warblers, was proba- ests-both coniferous and broad-leaved. No bly in Asia. In European forests, Phylloscopus North American bird approaches this level of are represented by two clades that probably in- abundance in such a wide geographic range or vaded in the Tertiary and have shown little eco- variety of forest types. Within regions as diverse logical or morphological divergence (Richman as southern France (Blonde1 et al. 1990) and 1996). It is likely that through selection or ex- northwest Russia (Morozov 1992), the Chaffinch tinction, the high loss of continuous forest cover has the broadest habitat amplitude of all forest during the Pleistocene and since human habita- birds. tion has resulted in a depauperateforest-adapted Based on MtDNA evidence, Chaffinches warbler fauna (Monkkonen and Welsh 1994). probably colonized Europe in the late Pleisto- For whatever reason, the presence of a depau- cene, as recently as 50,000-95,000 years ago perate assemblage of migratory forest insecti- (Marshall and Baker, in press). The speciescon- vores may have set the stage for the more re- tinues to spread into Siberia (Rogacheva 1992). cently expanding populations of Chaffinch to Given the current abundance of this species in find suitable resourcesin European forests. European farmlands during the nonbreeding sea- son (O’Conner and Shmbb 1986, Cramp and POSSIBLE INTERACTIONS BETWEEN Perrins 1994), its spreadinto the largely forested WARBLERS AND CHAFFINCHES areas of northern Europe was probably acceler- If we are correct in our hypothesisthat the Chaf- ated by the spread of agriculture. Such a recent finch was a relatively recent invader of Euro- spread into European farmlands has been sug- pean forests, then it suggestsan interesting ques- gested for the Dunnock (Prunella modularis; tion: How did European forests support such a Murton 1971) and (Passer do- large increase in the density of foliage-gleaning mesticus;Johnston and Klitz 1977). birds? Two possibilities are that (1) assemblages Thus far, we have presented a close relation- were not close to a food-based carrying capacity ship between both warbler abundance and rich- and that increases in winter resources supplied ness and overall foliage-gleaner abundance in by agriculture is a sufficient explanation for the non-European boreal forest, along with an expansion of Chaffinches, and/or (2) at least anomalous pattern in the European habitats. some of the increase occurred at the expense of These divergent observations can perhaps be re- the breeding population densities of warblers solved if we consider that Fringilla are func- and other foliage-gleaning birds. tionally similar to warblers in their resource use Although competition is no longer generally and that the expansion of Chaffinch in Europe acceptedas important in structuring boreal forest has occurred at the expense of warbler abun- communities (Monkkonen et al. 1997), the pos- dance which would be consistent with the Eco- sibility that competitive interactions may medi- logical Replacement Hypothesis. This idea has ate the distribution and abundance of Fringilla some support: when we clustered foliage-glean- and warblers should at least be considered.Frin- ing speciesbased on foraging behavior and site, gilla tend to arrive earlier than warblers, being Chaffinches clustered tightly with certain Phyl- able to withstand periods of cold weather by loscopuswarblers, , and Parulinae (Green- shifting to foraging for seeds (Preobrazhen- berg et al., unpubl. data). With this in mind, it skaya, pers. comm.), and hence may be able to is striking how close the relationship was be- dominate sites through prior access.In addition tween warbler abundance and biomass and over- to prior residency, Fringilla are considerably all foliage-gleaner abundance across all study larger than warblers (20-22 g vs. 7-11 g), which sites if Fringilla are pooled with warblers (Fig. also would contribute to their behavioral domi- 3A). nance. Hogstad (1975) presented evidence for HISTORICAL DYNAMICS OF BOREAL FOREST the competitive interaction of another Fringilla, ASSEMBLAGES the Brambling, and Willow Warblers (Phyllos- A high abundance of Fringilla and a low abun- copus trochilus). Interestingly, he found that the dance and diversity of warblers characterizeEu- Willow Warbler was dominant to the Brambling. BOREAL FOREST FOLIAGE-GLEANING BIRDS 309

However, other studies have not found evidence ences, in particular, Katya Preobzazhenskaya and Oleg of such an interaction (Angell-Jacobsen 1980, Bursky, for providing housing and logistical support for work in Myrnoe and Kostromomskaya Field Sta- Fonstad 1984). Given the biogeographicpatterns tions. Anwar Kerimov hosted us at the Zvenigorod described here, a close examination of Chaf- field station. The following people assisted in gather- finch-warbler interactions seems warranted. ing data for this project: Mark Amberle, Sam Droege, Todd Easterla, Jennifer Gammon, Gjon Hazard, Ri- CONCLUSIONS chard Hoyer, and Olga Romanenko. Y. Ravkin provid- Abundance of foliage-gleaning birds is largely ed census data from the data base of the Institute of Ecology, Russian Academy of Science, No- determined by abundance of migratory species vosibirsk. Sam Droege and YrjG Haila provided useful of warblers or finches (predominantly Fringillu), comments on earlier drafts. A Pew Fellowship for Bio- and these groups increase in their importance diversity and the Environment provided funding to the with increasing abundance as well. Our results senior author. suggest that the warblers and finches are re- LITERATURE CITED sponding to the relative amplitude of summer productivity and resource availability. Tits are ANGELL-JACOBSEN,B. 1980. Overlap in feeding pattern generally more diverse and abundant in Palearc- between Willow Warbler Phylloscopus trochilus and Brambling Fringillu montifringilla in two for- tic communities, suggesting a role for a histori- est habitats in western Norway. Ornis Stand. 11: cally longer association of this dominant group 146-154. of resident insectivores in Old World forests. BERMINGHAM, E., S. ROHWEK, S. FREEMAN, AND C. However, the variation in absolute abundance is WOOD. 1992. Vicariance biogeography in the Pleistocene and speciation in North American small and contributes little to the decrease in wood warblers: a icst of Mengel’s model. Proc. proportion of migrant foliage-gleaning birds be- Natl. Acad. Sci. 89:6624-6628. tween sites. Furthermore, the abundance of tits BLONDEL, J. 1990. Biogeography and history of forest was highest in Palearctic communities with only bird faunas in the Mediterranean Zone, p. 95-109. one common tit (upland taiga in Siberia), which In A. Keast [ed.], Biogeography and ecology of forest bird communities. SPB Academic Publish- suggeststhat an ecological explanation may be ing, The Hague, Netherlands. required to explain high tit abundance in Old CRAMP. S.. AND C. M. PERRINS. 1994. Handbook of World sites as well. birds of Europe, the Middle East and North Af- The northern boreal forests have developed rica. Vol. VIII: Crows to finches. Oxford Univ. recently (Haila and J%rvinen 1990), and the con- Press, Oxford. ERSKINE, A. J. 1977. Birds in boreal Canada. Can. stantly changing mixes of trees provide highly Wildl. Serv. Rep. Ser. 41. variable habitat through the Pleistocene. How- FONSTAD, T 1984. Reduced territorial overlap between ever, for the most part, recent phylogenetic an- the Willow Warbler Phylloscopus trochilus and alyses have shown considerable depth to the re- the Brambling Fringilla montifringillu in heath forests: competition of different habitat preferenc- Dendroica lationships in such groups as (Ber- es. Oikos 42:3 14-322. mingham et al. 1992), Phylloscopus and Regulus GILL, E B., A. MOSTROM, AND A. L. MACK. 1993. Spe- (Richman 1996), and Parus (Gill et al 1993). ciation in North American chickadees. I. Patterns This suggeststhat current species or closely re- of MtDNA genetic divergence. Evolution 47: 195- lated ancestral species have interacted in tem- 212. HAILA, Y., AND 0. JARVINEN. 1990. Northern conifer perate forest since at least the late Tertiary. In forests and their bird species assemblages, p, 61- this context, the apparent recent invasion of tem- 97. In A. Keast [ed.], Biogeography and ecology perate and boreal forestsby the Chaffinch stands of forest bird communities. SPB Academic Pub- out. The factors that allowed a less specialized lishing, The Hague, Netherlands. insectivorous bird to achieve numerical abun- HAILA, Y., 0. J;~RVINEN,AND S. RAIVIO. 1987. Quan- titative versus qualitative distribution patterns of dance are unclear. Perhaps the high productivity birds in the western Palearctic taiga. Annales of European forests coupled with a depauperate Zoologici Fennici 24: 179-195. fauna of more specialized insectivores causedby HERRERA, C. M. 1978. On the breeding distribution high levels of forest loss during the Pleistocene pattern of European migrant birds: MacArthur’s (M6nkkiinen 1994, MBnkkiinen and Welsh theme reexamined. Auk 95:496-509. HOGSTAD, 0. 1975. Interspecific relations between 1994) set the stage for this invasion. Willow Warblers (Phylloscopus trochilus) and Brambling (Fringilla ~ontifr~rzgillu) in subalpine ACKNOWLEDGMENTS forests. Norwegian J. Zool. 23:223-234. We thank the Institute of Evolutionary Morphology HUTTO, R. L., S. I%. PLETSCHET, AND P HENDRICKS. and Animal Ecology of the Russian Academy of Sci- 1986. A fixed-radius point count method for non- 310 RUSSELL GREENBERG ET AL.

breeding and breeding seasonuse. Auk 103:593- NEWTON,I. 1967. The adaptive radiation and feeding 602. ecologv of some British finches. Ibis 109:33-108. JOHNSTON,R. E, AND W. J. KLITZ. 1977. Variation and O’CONNO~~R.J., AND M. SHRURR.1986. Farming and evolution in a granivorousbird: the House Spar- birds. Cambridge Univ. Press, Cambridge. row, p. 15-5 1. In S. C. Kendcigh and J. Pinowski PETIT,D. R., L. J. PETIT,V. A. SAAB,AND T. E. MARTIN. [eds.], Granivorous birds in ecosystems. Cam- 1995. Fixed-radius point counts in forests: factors bridge Univ. Press, Cambridge. influencing effectivenessand efficiency, p. 49-56. MACARTHUR,R. 1959. On the breeding distribution In C. J. Ralph, S. Droege, and J. Sauer [eds.], pattern of North American migrant birds. Auk 76: Monitoring bird populationtrends by point counts. Gen. Tech. Rep. No. PSW-GTR 149. USDA For. 318-325. Serv. Albany, CA. MARSHALL,H. D., AND A. J. BAKER.In press. Mito- RABENOLD,K. N. 1978. Foraging strategies,diversity, chondrial DNA control-region sequencesindicate and seasonality in bird communities of Appala- a sequentialrange expansionand low current lev- chian spruce-fir forests. Ecol. Monogr. 48:397- els of gene flow among European populationsof 424. chaffinches (Fringilla coelebs). Mol. Ecol. RICHMAN,A. D. 1996. Ecological diversification and M~NKK~NEN,M. 1994. Diversity patternsin Palearctic communitv structurein the Old World leaf war- and Nearctic forest bird assemblages.J. Bioge- blers (genus Phylloscopus): a phylogenetic per- ography 21:183-195. socctive. Evolution 50:2461-2470. M~NKK~NEN,M., I? HELLE,G. J. NIEMI, AND K. MONT- ROGA&EVA, H. 1992. The birds of central Siberia. GOMERY.1997. Heterospecific attraction affects Hussum Druck- und Verlagsggesellschaft,Husum, community structure and migrant abundancesin Germany. northern breeding communities. Can. J. Zool. 75: SHELDON,E H., ANDI? B. GILL. 1996. A reconsidera- 2077-2083. tion of songbird phylogeny, with emphasison the evolution of titmice and their Silvioid relatives. M~NKKONEN,M., AND D. WELSH. 1994. A biogeo- Syst. Biol. 45:473-495. graphic hypothesison the effects of humancaused WIENS,J. A. 1975. Avian communities,energetics, and landscapechanges on the forest bird communities functions in coniferous forest habitats, p. 226- of Europe and North America. Annales Zoologici 265. In Proceedings of the symposium on man- Fennici 3 I :61-70. agement of forest and range habitatsfor nongame MORSE,D. H. 1989. American warblers: an ecological birds. Gcn. Tech. Rep. WO-1. U.S. For. Serv., and behavioral perspective. Harvard Univ. Press, Washington, DC. Cambridge, MA. WILKINSON,L. 1990. The system for statistics.Systat MURTON,R. K. 1971. Man and birds. Collins, London. Inc., Evanston, IL.