The Occupation of Small Islands by Passerine Birds

THE OCCUPATION OF SMALL ISLANDS BY PASSERINE BIRDS

DOUGLASS H. MORSE

Comparisons of island and mainland faunas STUDY AREA AND METHODS have provided great insight into how com- Most of the 12 islands censused are partially or totally munities are structured (see MacArthur 1972). forested by white spruces (Picea g&x), with small In particular, islands permit natural compari- but varying amounts of balsam fir ( Abies balsumeu), sons of how animals under varying degrees of red oak (Quercus rubru), and white birch (Betulu fauna1 impoverishment exploit their resources. pupytifem). Two of the islands, however, contain To date, most islands studied from this point substantial amounts of mountain maple (Acer spicu- turn), striped maple (A. pensyltxznicum), and yellow of view are sufficiently large to support size- birch (Bet&z Idea). In addition, some have sizeable able populations of many or all of the species open areas sparsely covered by various grasses, rasp- that occupy them. Further, island popula- berry ( Rubus sp. ), bayberry ( Myricu pennsyluunicu), tions typically studied are either of unknown or goldenrod (Solidago spp. ). Since my primary concern lay with the forest-dwelling birds, I have genetic relationship to their mainland counter- included typically open-country species only when parts, or they are easily distinguished from they used the forested areas of these islands. These them, often being considered separate sub- islands are described in detail elsewhere (Morse species or even species. 1971a, 1973a). No visible successional change oc- In this paper I report upon the patterns curred during the period of the study. Records of breeding and visiting species were made by which passerine birds occupy extremely during intensive studies upon the warblers (Paruli- small forested islands and adjacent large main- dae) of the islands (Morse 1971a, 1973a). These land habitats in and around Muscongus Bay, studies provided information on the fledging suc- Maine, thus expanding upon earlier observa- cess of the warblers during the 1967-1970 seasons. During that period, several of these birds were either tions (Morse 1971a). I attempt to answer individually marked or identified by their songs. Ob- the following questions: From which main- servations of other species were designed to give an land habitats do the inhabitants of these accurate measure of presence or absence, even though islands come? Which characteristics of the abundances are in some cases estimates where many pairs were present (e.g., Song Sparrow, Melospizu islands best predict the number of species melodiu). After 1970, I made at least six censuses that will occupy them? What factors best each year between 10 June (after the last spring account for the presence or absence of any migrants) and 15 July (after the latest establishment of breeding individuals) (Morse 1971a, 1973a). given species on the islands? How stable are These censuses ranged from 15-20 min per visit on the island assemblages in time? How does the smallest islands (under 0.2 ha) to an hour or the total density of birds and the densities of more on the largest islands. Minimum criteria for colonization were satisfied if a species was seen on individual species compare with the densities half or more of the censuses and if both a male and on the mainland? Then, where appropriate, I a female were detected (i.e., settlement of a mini- relate my findings to island biogeographic mum breeding unit). Year-by-year lists of breeding species are available on request. theory (MacArthur and Wilson 1963, 1967, Several aquatic species (herons, ducks, gulls) Diamond 1975). nested on these islands but foraged elsewhere. Since Small islands of this sort are excellent sys- I have no evidence that any of these birds competed for nest sites or food resources with fully resident tems for answering such questions. By virtue species or that they preyed upon these residents of their small size (0.14.0 ha), it is often pos- regularly, I will not consider them further. sible to determine the minimal conditions To determine food supplies available to the foliage- inhabiting species, I measured the biomass of foliage- (space, etc.) required for a species to estab- inhabiting insects on these islands durinn the 1968- lish itself. Potentially key variables (physical 1970 seasons, using the same technique as in my characteristics of the forest, food supply, etc.) other studies (Morse 1973a, 1976). Weekly or bi- weekly I collected 20 live branches at random (ap- can be monitored readily. A variety of species proximately 0.2 m” and 6.0 m3) with their associated compositions provides the opportunity to eval- arthropods on the islands and mainland, attempting uate the effects of certain species upon others. to lose as few arthropods as possible. The branches were placed in plastic bags, fumigated, and then Lastly, the birds of any given species inhabit- beaten over a screen of hardware cloth. Data on ing these islands almost certainly belong to arthropods so obtained were pooled for each two- the same gene pool as those on the adjacent week period, permitting a direct comparison of island mainland (Morse 1971a), so one can eliminate and mainland insect stocks. Although this technique extracts over 99% of the arthropods on the branches evolutionary change as the basis for such dif- at the time that they are placed in the bags (Morse ferences. 1973a), it underestimates the abundance of strongly- K3991 The Condor 79:399-412, 1977 400 DOUGLASS H. MORSE flying insects. However, since most of the foliage- TABLE 1. Contributions of spruce and deciduous- foraging birds were primarily gleaners, this potential edge components to island populations of birds dur- problem should not bias the results seriously. ing 1967-1975 period.

Total pairs RESULTS Species-year@ of individuals THE SPECIES POOL Deciduous- DeCi~iUS- Island Spruce edge Spruce One can construct a mainland species pool Strawberry 0 8 0 8 (Appendix I) with considerable accuracy from Byers’ several sources of information. Mainland is Crotch I: 118 I: 1116 here defined as that land physically connected Crow 15 11 15 38 25 to the continental shoreline, plus islands of Jim’s 12 25 42 Ram 18 12 18 26 These large islands are over 25 ha in size. Crane 20 31 20 93 partly or totally covered with spruces. Earlier Indian 28 12 28 27 (Morse 1966) I compiled lists of species nest- Thief 41 20 45 92 ing in spruce forests, deciduous forests, hur- Franklin 0 31 0 174 Haddock 15 26 116 ricane-damaged areas, and edge situations on 15 Wreck 22 29 28 163 Hog Island, a large island (132 ha) separated at points from the mainland by not over 200 TOTALS 191 223 201 806 m. This compilation was based upon personal PERCENTAGE 46.1 53.9 20.0 80.0 field work and the breeding-bird censuses of n One species-year represents the occupation of an island Cadbury and Cruickshank from 1937 to 1958. for one year by <,neor nxxe breeding pairs of a species. Subsequent intensive field work has added but two more species to these lists. Thus, the forest group. This characteristic was espe- updated lists presented as Appendix I ap- cially striking in the case of the Song Sparrow, pear to be largely complete. which made up 44.8% of the total number Another group of species breeds on the of birds. Not suprisingly, the Song Sparrow mainland within 2 km of the shoreline, but is an extremely successful colonizer in other does not occur on the Hog Island lists (Ap- areas as well (Beer et al. 1956, Yeaton and pendix I). I compiled this list from personal Cody 1974). The spruce-forest component observations made since 1957 and from Cruick- was much more evenly distributed than that shank (no date). Over half of these species of the deciduous-edge component, because occur in open-country or brushy areas, and the former was comprised of species that most of the remaining species typically occur seldom attained a density greater than one in deciduous forests. pair per island (Fig. 1).

CONTRIBUTIONS OF THE DIFFERENT FOOD SUPPLY COMPONENTS The foliage sampling for arthropods indicated Seventeen species of small passerine birds a rather constant average biomass of insects bred on forested parts of the small islands through the season (Fig. 2). During the first during the 1967-1975 period. They came en- part of the season, the average biomass was tirely from the first two components of the similar to that of a white spruce forest on the mainland species pool: (1) the spruce forest mainland, while later it exceeded that of the and (2) the adjacent deciduous, disturbed, mainIand. Therefore, the abundance of food and edge areas (henceforth referred to as on the foliage does not appear to account for deciduous-edge areas). These two compo- the absence of many species from these is- nents made similar contributions to the island lands. Other possible food sources (from species pool (nine from the spruce forest ground, litter, etc.) were not systematically group, eight from the deciduous-edge group). measured. The two components also contributed rather similar numbers of species-years (defined as NUMBER OF SPECIES PRESENT the occupation of an island by a species during one breeding season) on the islands (Ta- Using stepwise multiple regression, I tested ble 1). However, the relative abundance of six potentially important factors (the inde- the deciduous-edge species was much greater pendent variables of Table 2) for their rela- than that of the coniferous species (Table 1)) tionship to the number of species that oc- because several members of the deciduous- cupied an island. I used three different edge group established considerably more measures of species numbers-cumulative list dense populations than did those of the spruce of species over 1967-1975 period, maximum BIRDS ON SMALL ISLANDS 401

TABLE 2. Dependent and independent variables for multiple regression analysis.

Independent variables

Dependent variables Ground SpNX cover Deciduous- Forest Overall Distance forest Forest Total “E%~- edge area area to main- area height fo% Island species species species (ha) vegetated land (m) (ha) Cm) (ha)

Crane 9,9, 5.7” 5,5,2.5” 4,4,3.2” 0.53 2.81 230 0.52 14 0.12 Jims’ 10,6,4.6 6,5,3.1 4,4, 1.5 0.49 0.49 500 0.44 14 0.03 Indian 9,6,5.0 6, 5, 3.5 3,3, 1.5 0.69 0.69 230 0.59 16 0.07 Thief 13,9, 7.6 8,7,5.1 5,4,2.5 1.50 1.64 1,070 1.49 15 0.02 Crow 6,4, 3.3 3, 3, 1.9 3, 2, 1.4 0.35 0.60 360 0.21 13 0.11 Ram 5, 5, 4.3 3, 3, 2.6 2,2, 1.7 0.39 0.39 640 0.35 14 0.08 Crotch 3,3, 1.9 2,2,0.9 1, 1, 1.0 0.16 0.16 720 0.16 12 0.02 Wreck 13, 10,7.3 6, 5, 3.1 7,5,4.2 3.86 3.86 1,200 2.12 14 1.05 Haddock 9,8,5.1 5, 5, 1.9 4, 4, 3.2 1.86 2.57 2,160 1.13 14 0.37 Franklin 5,5,4.4 0, 0, 0.0 5,5,4.4 1.77 3.49 1,160 1.77 14 1.75 Byers’ 2,2, 1.4 0, 0, 0.0 2, 2, 1.4 0.11 0.11 280 0.11 12 0.02 Strawberry 1, 1, 1.0 0, 0 ,o.o 1, 1, 1.0 0.10 0.10 360 0.10 14 0.00

a The three figures in order are cumulative list of species over 1967-75 period, maximum number of species in a year during this period, and mean number of species per year. number of species in any year, and mean tated were the most important variables for number of species per year (Table 2). None the maximum numbers. of the statements in this section should be I then divided the species pools into spruce taken to imply a cause-and-effect relationship and deciduous-edge components (Table 4). between the variables and the birds present. The pattern seen in the spruce component Only four variables (forest size, ground closely resembled that of the total species cover, forest height, and area vegetated) con- pool. Forest size and forest height made sig- tributed significantly to this variance in any nificant contributions to the variance of both of the measures (Table 3). These variables the cumulative and mean numbers of species, had different effects on the three measures and area vegetated contributed most to the of species numbers. Cumulative and mean variance of the maximum yearly count. numbers of species showed considerably more The pattern of the deciduous-edge compo- similarity to each other in this regard than nent differed somewhat from that of the total either did to the maximum count per year. pool and the coniferous component. Again, Forest size contributed significantly to the forest size made a significant contribution to variance in the cumulative and mean num- both the cumulative and mean numbers of bers, but not to maximum numbers. Height species, but the effect was much larger for of forest made the next largest contribution the cumulative list of species than for the to the variance in cumulative and mean num- mean number of species. However, the area bers, though attaining statistical significance only in the case of the mean species number. The amount of ground cover and area vege-

3 . t-15 I-15 16-31 June July July

FIGURE 2. Biomass of arthropods throughout the FIGURE 1. Number of times that species nested at season ( 21 SE). Data are pooled from 1968-1970 various densities (pairs per island) during 1967- seasons, each point representing a minimum of five 1975 period. samples. Mainland date from Morse (1976). 402 DOUGLASS H. MORSE

TABLE 3. Relative contributions of independent variables to the multiple correlation coefficient, R,’ for the total number of bird species.

Cumulative Maximum/year Meall

Independent variable R” F R2 F R2 F

Forest area .4351 7.70* .0316 1.91 .5007 10.03** Overall area vegetated .0499 1.63 .5433 11.90** .0836 3.15 Distance to mainland .0039 0.11 .oooo 0.00 .0009 0.02 Area of spruce forest .0002 0.01 .0002 0.01 .0042 0.14 Forest height .1243 3.77 .0512 2.78 .1950 5.77* Area in forest with ground cover .1771 4.12 .2585 11.73** .0357 1.06

TOTAL Ra .7907 - .8847 - .8200 -

* = significance at 0.05 level. ** = significance at 0.01 level. vegetated made the greatest contribution to ties, 0.23 for maximum number of species, the variance of the mean number of species, 0.20 for cumulative number of species). The but little to the cumulative number of spe- area of these three islands falls below the cies. Also, area vegetated made the only territory size of most of the island inhabitants significant contribution to the variance in (see below). the maximum number of species recorded in any one year. BASIS FOR THE PRESENCE OR ABSENCE OF CERTAIN SPECIES ON THE ISLANDS RELATION OF NUMBER OF SPECIES TO AREA In several cases I can explain why certain The three smallest islands (under 0.2 ha) di- species occupied the islands and others did verged markedly from a straight line in a log- not. One potentially important consideration log plot of area and number of species. The is whether the size of the island forests corresponding slope of the regression, z (Mac- equalled or exceeded the size of species ’ ter- Arthur and Wilson 1967), was relatively high, ritories on the mainland. Measures of aver- falling in the range of 0.5 (0.48 for mean age territory size on the adjacent mainland number of species per year, 0.51 for maximum existed for 9 of the 17 species found on the number of species in any one year, 0.55 for small islands, and 8 other potentially colo- the cumulative number of species over the nizing species from the spruce-forest and de- years of the study). However, when the three ciduous-edge categories that were not re- smallest islands were removed from the anal- corded on the small islands (Table 5). I ysis and the z values recalculated, they ap- could compare these territory sizes with the proached 0.2 (0.21 for mean number of spe- area of the island forest where only one pair

TABLE 4. Relative contributions of independent variables to the multiple correlation coefficient, R”, for the number of bird species in the coniferous and deciduous-edge components.

Spruce-forest species Deciduous-edge species

Cumulative Maximum/yr Meall Cumulative Maximum/yr Meal3 Independent variable R ’ F Rg F R” F R” F RZ F RZ F

Forest area .1952 5.29 .0290 0.83 .2140 5.93* .7628 32.16** .0017 0.07 .0134 7.04*

Overall area vegetated .0356 0.98 .2215 6.51* .0056 0.14 .0503 2.43 .7392 38.35** .9694 316.90**

Distance to mainland .OOOl 0.00 .0021 0.06 .0063 0.14 .0227 1.11 .oooo 0.00 .OOoo 0.02

Area of spruce forest .0925 1.54 .1036 1.63 .0016 0.03 .oooo 0.00 .0038 0.13 .0036 2.15

Forest height .3660 5.77* .3243 4.80 .4070 6.86* .0153 0.72 .0057 2.44 .0002 0.09

Area in forest with ground cover .0952 1.71 .1124 1.95 .0904 1.62 .0048 0.20 .oooo 0.00 .oOOO 0.00

* = significance at 0.05 level. ** = significance at 0.001 level. BIRDS ON SMALL ISLANDS 403

TABLE 5. Sizes of territories in and about mainland spruce forests.”

Area &Ilple Species (ha 5 1 SE) Range size Habitat

SPECIES OCCURRING ON SMALL ISLANDS Winter Wren 3.64 _ 2 red spruce Golden-crowned Kinglet 0.72 _ 6 white spruce Northern Parula 1.33 _ 2 white spruce Northern Parula 0.40 2 0.04 0.30-0.51 5 spruce edge Magnolia Warbler 0.72 ? 0.06 0.51-1.05 9 white spruce Yellow-rumped Warbler 0.81 % 0.03 0.60-1.03 17 white spruce Black-throated Green Warbler 0.43 2 0.01 0.35-0.55 28 white spruce Yellowthroat 0.29 - 11 damaged spruce White-throated Sparrow 1.67 - 3 white spruce White-throated Sparrow 0.35 _ 9 damaged spruce White-throated Sparrow 0.49 - 6 spruce edge Song Sparrow 0.18 18 damaged spruce Song Sparrow 0.49 _ 6 spruce edge

SPECIES NOT OCCURRING ON SMALL ISLANDS Blue Jay 4.00 _ 2 red spruce Red-breasted Nuthatch 1.34 - 3 white spruce Brown Creeper 2.22 - 2 red spruce American Robin 4.00 - 2 white spruce Hermit Thrush 2.00 - 2 white spruce Nashville Warbler 2.22 - 2 red spruce Blackburnian Warbler 0.46 f. 0.01 0.36-0.56 22 white spruce Dark-eyed Junco 1.67 - 6 white spruce

a Territories with standard errors and ranges were determined frwn observntions made in 1969-1972 and in 1974 (from Morse 1976). Other data are from unpublished observations in which the census area has been divided by the number of pairs present, so that no means or extremes are available. of any given species was present (as in the Warblers occupied more islands than would spruce-forest warblers), or with the area di- be predicted on the basis of mainland terri- vided by the number of pairs present in other tory size. Body size (Appendix II), another cases. While the latter measure does not ac- measure of energetic demand, accurately count for the possibility of territorial over- predicted the presence of Northern Parulas lap, I assume it to be small, relative to the on more islands than any other warbler, but overall space occupied. Territorial overlap the relatively large Yellow-rumped Warbler between conspecifics was low or nonexistent appeared far more often than it should have where observed in the mainland situations according to this measure. However, the (Morse 1976) that were used for comparison. social dominance ranking of these warblers Mainland territory size (Table 5) accurately (Morse 1971a, 1974, 1977) inversely matched predicted the presence or absence of many, the frequency with which they occupied the though not all, species. Six of the nine colo- islands relative to each other: Northern Par- nizing species occurred only in island forests ulas > Yellow-rumped > Black-throated Green larger than their mean territory size on the > Magnolia. Only in 1974 and 1975, when mainland (Winter Wren, Golden-crowned Yellow-rumped Warblers nested upon Crow Kinglet, Magnolia Warbler, Black-throated Island in the absence of Northern Parulas, Green Warbler, Yellowthroat, and White- did an exception to this relationship occur. throated Sparrow), while the other three spe- The ground-foraging guild may be orga- cies occupied island forests smaller than their nized similarly to the spruce-forest warblers mean territory size on the mainland (Northern of the foliage-gleaning guild. This was sug- Parula, Yellow-rumped Warbler, and Song gested by the presence on several islands of Sparrow). The smallest islands occupied by the Swainsons’ Thrush, a species that appar- Northern Parulas and Yellow-rumped War- ently is socially subordinate to the Hermit blers were also smaller than their minimum Thrush ( Morse 1972). Swainsons’ Thrushes territory sizes on the mainland. Of the eight occupied a wider range of habitats in the ab- non-colonizing species, all but the Blackburn- sence of Hermit Thrushes than where sym- ian Warbler had territory sizes larger than patric with them (Morse 1971b, 1972). The most or all of the island forests censused. distribution of White-throated and Song spar- Considering the foliage-gleaning guild in rows on these islands might have a like ex- detail, Northern Parulas and Yellow-rumped planation, since Song Sparrows expanded 404 DOUGLASS H. MORSE their activities into the midst of these island TABLE 6. Cumulative pairs per year on islands. spruce forests where White-throated Spar- rows were absent or infrequent (unpubl. Species 1967-1971 1972-197s data). SPRUCE-FOREST SPECIES Species whose presence would not be pre- Swainson’s Thrush 4.2 dicted on the basis of mainland territorial Northern Parula 8.8 :.; size often were low in the social hierarchy; Yellow-rumped Warbler 6.8 4:o Black-throated Green Warbler 2.6 0.7 they underwent ecological release on the Purple Finch 4.0 3.0 small islands in the absence of their domi- White-throated Sparrow 3.0 0.3 nants. This release took place in at least DECIDUOUS-EDGE SPECIES” three ways: more intensive use of a given Yellow Warbler” 1.2 4.0 territory (associated with decreased territory American Redstartb 0.6 1.7 size), changes in foliage-usage patterns, and Catbird 0.6 0.7 changes in habitats occupied. Size of main- Yellowthroat 1.6 2.7 land territory was a good predictor of presence a Does not include Song Sparrows. b Does not include sizeable populations of these species on or absence of species high in the hierarchy, Crane, Franklin, Haddock, and Wreck islands. These popu- but a poor predictor for species low in the lations (and Song Sparrows) were not included because, during some or all years, counts were estimates. However, hierarchy. they showed no signs of decrease during 1972-1975 period.

ADDITIONAL COMMENTS ON THE CHANGES DURING THE PERIOD OF STUDY DISTRIBUTION OF CERTAIN SPECIES Results reported thus far suggest a highly Magnolia Warblers first bred on the small predictable pattern of colonization on the islands in 1975, when a pair raised young islands. However, in light of two extremely successfully on Thief Island. Judging from poor breeding seasons associated with inclem- their large territory size on the mainland ent weather (1972 and 1973; Morse 1976), (Table 5) and their high dominance rank- I compared the frequency with which species ing, they would be predicted only on Thief occupied these islands from 1967 to 1971 with Island. Nevertheless, their frequency there their frequency from 1972 to 1975. Species was unexpectedly low, since that forest is closely associated with spruce showed an twice as large as their average mainland ter- overall marked drop in density, while species ritory. However, four of the five vagrant of the deciduous-edge component increased Magnolia Warblers seen between 1967 and simultaneously (Table 6). The decrease in 1975 were on Thief Island. These sightings numbers of spruce-forest birds was significant suggest that this island had characteristics when considered at the level of the species making it more attractive to this species, than (P = 0.05 in a two-tailed Wilcoxon test), and the other islands. When compared with the the increase among deciduous-edge species number expected by chance alone, this small at this level showed a consistent trend, though sample approaches significance (P = 0.062 the number of species available for analysis in a one-tailed Fisher test). (n = 4) was too small for statistical treatment. Blackburnian Warblers occupied territories The results suggest that: (1) the inclement on the mainland that were considerably conditions substantially decreased populations smaller than several of the forests on the of spruce-forest species, but not the decidu- small islands (Table 5) ; however, they did ous-edge species; and (2) deciduous-edge not nest on any of these islands. They were species occupied the space left by the spruce- the only species that concentrated their ac- forest species. This pattern closely parallels tivities on the mainland in parts of the forest one seen in large adjacent forests (Morse that were generally not represented on the 1976). In that case, numbers of Black- island (the tops of trees taller than those throated Green Warblers decreased abruptly usually found on the tiny islands; Morse 1968, in white spruce forests, though not in red 1971a). However, four of the six vagrant spruce forests. Simultaneously, a marked in- Blackburnian Warblers found on these islands crease of the American Redstart, a predomi- between 1967 and 1975 occurred on Indian nantly deciduous-forest species, occurred in Island, the only small island containing spruce the only forest censused that contained both trees approaching the height that these birds species. frequent in the mainland forests. This small sample also approaches significance (P = TURNOVER OF SPECIES AND INDIVIDUALS 0.072 in a one-tailed Fisher test) when com- Rates of extinction and rates of new coloniza- pared with the expected. tion of species differed widely but were gen- BIRDS ON SMALL ISLANDS 405

TABLE 7. Colonization and extinction rates of bird years. Since the mean period of occupation faunas. for the warbler species on these islands was 4.2 years, the turnover rate of individuals was Percentage of Percentage of species disappearing species appearing 3.2 times that of the species. from one year that were not present Island to the next” in preceding year* INVASION RATES Crane 15.7b (8) 15.7 ’ (8) Jims’ 26.3 (10) 26.3 ( 10) I noted vagrant warblers wherever I saw them. Indian 20.0 (8) 15.0 (6) These nonresident individuals occurred fre- Thief 18.6 (11) 18.6 ( 11) quently enough to saturate the islands upon Crow 16.7 (6) 16.7 (5) Ram 10.0 (3) 6.7 (2) occasion, or at least to breed upon them more Crotch 13.3 (2) 20.0 (3) often than they were observed to do (Table Wreck 15.7 (8) 17.6 (9) 8). At least some wandering birds were po- Haddock 17.1 (7) 19.5 (8) tentially reproductive, as they established Franklin 3.8 (1) 3.8 (1) Byers’ 18.2 (2) 27.3 (3) themselves and successfully raised young in Strawberry 0.0 (0) 0.0 (0) certain cases (Morse 1971a). I must empha- size, however, that these data dealt with male Cumulative 16.7 (66) 16.7( 66) birds only, which were conspicuous because 3 Numbers of extinctions or colmizations in parentheses. b Calculated from the equation e = xE/t(Zs), where e = of their almost continual singing. I seldom extinction rate, x = number of years censused, E = the num- saw vagrant females, though I assume that ber of species that became extinct, t = number of years in- tervening, and s = number of species recorded in a yearly they were frequently present, since late-arriv- census. e Calculated from the equation c = xC/t( 2s); where c = ing males found females. colonization rate; C = number of species that colonized; and x, t, and s are as above. OVERALL DENSITY AND BIOMASS erally high (Table 7). Taken as a whole, ex- With a few conspicuous exceptions, the over- tinctions balanced colonizations, so I conclude all density of birds changed inversely with L that these islands were in equilibrium (Table forest size (Fig. 3A). However, the smallest 7). I assume that the pattern observed on in- islands (0.10 and 0.11 ha) did not have den- dividual islands can be attributed largely to sities as high as would be predicted from the random processes. general trend. This difference probably re- The actual turnover of individuals probably sulted from their peculiar size, where two pairs was even considerably higher than the fore- of birds could seldom, if ever, be supported, going data suggest. I can document this in even though more space was available than one case. When several known individual required by a single pair of birds. Densities spruce-forest warblers on these islands were from two other islands were considerably followed from 1967 through 1969, considerable higher than those of other comparably sized turnover occurred in these birds, even when forests (circled data points in Fig. 3A). This islands were constantly occupied by the spe- difference may be associated with the sizeable cies in question (Morse 1971a). The mean amount of adjacent non-forested area on these length of occupation by an individual was islands, which was used to varying degrees 1.3 breeding seasons, though in extreme cases by some of the forest-nesting birds in question. birds occupied an island for as many as three This explanation is consistent with the fore-

TABLE 8. Frequency of vagrant birds and occupation patterns of small islands during the 1967-1969 seasons ( 10 June-15 July).

NO. NO. No. island-years NO. Corrected no. island- island- vacant on islands observations observations years yfXr.T occupied by species Species of vagrants” of vagrantsb occupied vacant at least oncee

Northern Parula 6 (2) 18 22 5 3 Yellow Warbler 2 (1) 6 12 15 12 Magnolia Warbler 4 (0) 12 0 27 3 Yellow-rumped Warbler 5 (3) 15 17 10 Black-throated Green Warbler 6 (6) 18 8 19 : Blackburnian Warbler 5 (0) 15 0 27 0 Chestnut-sided Warbler 2 (0) 6 0 27 0 American Redstart 6 (1) 18 10 17 12

a Data either previously unpublished or from Morse (1971a). Figures in parentheses refer to individuals observed on islands already occupied by other individuals of the same species. b I visited the islands once every three days on the average, so that multiplying by 3 produces a daily total. c During the years of 1967 through 1975. 406 DOUGLASS H. MORSE

A 1000 B

Density 900. Biomass 2 800. 800. . 0 $700. ’ ‘;i 700. . 2600. ’ 0 . 2 l f 600. ’ . wt500.0 ; 500 . . *zs 400.a ’ . . . g400. . = 300. -2 300 g 200 ’ f g IOO- a! ____._ 0 0.5 10 1.5 2.0 2.5 3.0 35 4.0 0 0.5 1.0 I.5 20 25 3.0 3.5 4.0 - Size of area (ha) h i 2 219 . 20- D Metabolic c 3 19. Mean demand weight E ‘. .

-; 18.,17. ’ , . . 8‘ 16. . s Q) 15. I 0‘ 0 . I4 0 t 0 13. . 12. 60 1 i II- T Y c __I_ 0 0.5 1.0 I.5 2.0 25 3.0 3.5 4.6~--- 0 0.5 I.0 I.5 2.0 2.5 3.0 3.5 4.0 Size of area (ha)

FIGURE 3. (A) Mean density, (B ) biomass, ( C) estimated metabolic demand, and ( D ) mean weights of birds on islands of varied size. Circled symbols represent forests with considerable unforested area about them. Symbols to right of 4.0 ha line represent maximum and minimum values from plots in large mainland forests. Metabolic demand (M) calculated using the equation M = KW,’ where K = 129, W = weight in grams, and b = 0.724 ( Lasiewski and Dawson 1967 ) going observation that size of vegetated area larger than that on the larger islands (Fig. (forested and unforested) was the best in- 3D) ; this results primarily from the predomi- dicator of the numbers of deciduous-edge nance of Song Sparrows on the former. species present. Population density, biomass, metabolic de- The biomass (g/ha) and estimated meta- mand, and average weight of birds (Table 9, bolic demand (Kcal * day-l * ha-l) of nest- Fig. 3) on the smallest islands were all greater ing birds also changed inversely with forest than those of mainland populations (I ’ < size (Fig. 3B and C). The biomass, but not 0.01 in one-tailed Mann-Whitney U tests). the metabolic demand, of the two high-density The second largest island of the group, Had- populations fell nearer to that of other plots dock Island (1.86 ha), had a population den- than in the preceding comparison. The mean sity, biomass, and metabolic demand that size of the birds on the smallest island was approached those of the highest-density large BIRDS ON SMALL ISLANDS 407

TABLE 9. Density and biomassof tree foragers”and ground-understoryforagers.b

Density ? SE Biomass ? SE Metabolic demand ? SE (pr/43’ ha) (g/ha 1 (Kcal day-l. ha-l)

Al%3s Ground-understory Tree Ground-understory Tree Ground-understory Tree

Tiny islands 354 + 41.9 223 t 34.8 357 * 39.2 118 -c- 19.5 138 t 13.3 57 & 11.1

Large forests 63 & 14.5 201 & 15.1 94 % 26.6 89 * 7.5 31 4 8.0 41.3 f 3.1

p Red-breasted Nuthatch, Brown Creeper, Golden-crowned Kinglet, Ruby-crowned Kinglet, Red-eyed Vireo, Black-and- white Warbler, Northern Panda, Yellow Warbler, Magnolia Warbler, Yellow-rumped Warbler, Black-throated Green War- bler, Blackbumian Warbler, Canada Warbler, American Redstart, Purple Finch. b Winter Wren, Catbird. American Robin, Hermit Thrush, Swainsons’ Thrush, Yellowthroat, Dark-eyed Junco, White- throated Sparrow; Song S&row. spruce forest censused, while on the largest Warbler, and Song Sparrow). The data avail- island, Wreck Island, (3.86 ha), these variable clearly did not completely document the ables approached those of most large spruce extent of the ability of the Song Sparrow to forests. However, the forest of these islands reduce its area, as two pairs nested on an did not closely resemble the undisturbed large island of 0.16 ha each year, suggesting that spruce forests because the islands had much the figure of 0.10 ha can be reduced to 0.08 deciduous, as well as coniferous, growth ha, at least. Further, this species nested each ( Morse 1973a). year on an island of 0.10 ha, as well as on an The difference between the small islands island of 0.11 ha, suggesting that this level and the large forests resulted mostly from an of crowding was readily tolerated. increase in density, biomass, and metabolic The differences in density of species pres- demands of ground and understory foragers ent both on the small islands and in the large on the small islands (Table 9). All three mea- forests appeared to result from the disappear- sures of the ground and understory component ance of potential competitors encountered in were significantly greater on these islands the large forests. In the case of the three spe- than in the large forests (P < 0.001 in one-tailed cies for which relatively accurate measures Mann-Whitney U tests). While the density, of substantial ecological release exist, changes biomass, and metabolic demand of tree- in density probably occurred for the follow- foraging species were slightly higher on the ing reasons: ( 1) (Northern Parula, Yellow- islands than in the large forests, these differ- rumped Warbler, and Song Sparrow) Northern ences did not approach significance (P > Parulas on the tiny islands existed in the ab- 0.05). sence of several potential competitors. Gold- en-crowned Kinglets, which appeared to be DENSITY OF INDIVIDUAL SPECIES a major factor responsible for the low density When a single pair of a species occurred on of Northern Parulas in dense spruce forests an island they usually ranged widely over it, (Morse 1967), bred but rarely on the tiny islands. This species nested only on Thief probably using all parts as frequently as they would parts of intensely defended territories Island, the only primarily spruce-covered is- on the adjacent mainland. Some species used land larger than the average Northern Parula much smaller areas on the islands than they did on the mainland, while other species did TABLE 10. Comparisonof the sizes of areas (ha) not markedly change the size of their terri- occupied on small islands and in large populations. tories on the islands, only occupying islands of a size consistent with their mainland terri- ;!$;yzy Size of Size of Mini- tories. By comparing the data on mainland in large smallest next mum popu- island smallest contrac- territory size in Table 5 with the smallest- Species lations occupied island tion sized island occupied, one may roughly esti- Winter Wren 3.64 3.86 1.86 None mate the amount of intraspecific crowding Golden-crowned that will be tolerated by the different species Kinglet 0.72 1.50 0.69 None (Table 10). Five of the eight species for Northern Parula 0.40 0.16 0.11 2.5 Magnolia Warbler 0.72 1.50 0.69 None which the necessary data were available did Yellow-rumped not decrease their areas of occupancy (Winter Warbler 0.81 0.35 0.16 2.3 Wren, Golden-crowned Kinglet, Magnolia Black-throated Green Warbler 0.43 0.49 0.39 None Warbler, Black-throated Green Warbler, and White-throated White-throated Sparrow), while three spe- Sparrow 0.49 0.49 0.39 None cies did (Northern Parula, Yellow-rumped Song Sparrow 0.49 0.10 - 4.9+ 408 DOUGLASS H. MORSE territory in dense mainland spruce forests 1967). This result suggests that the islands (Table 5). In edge situations on islands are almost as available to would-be colonists smaller than the Northern Parulas’ customary as is the mainland. The three smallest islands mainland territory size (which were seldom were no farther from the mainland on the frequented by Golden-crowned Kinglets), average than were the somewhat larger islands only one other warbler (Yellow-rumped) was (Table 4), so that their small number of spe- present, in contrast to four species of warblers cies cannot be attributed to isolation. Since (Yellow-rumped, Black-throated Green, Mag- these islands are considerably smaller than nolia, Blackburnian) and the kinglet in the the average territory sizes of many likely large forests. Further, on the smallest island colonizers, this itself probably accounts for occupied by Northern Parulas, where their the result obtained. Diamond and Mayr territories were only 40% of their average size (1976) noted a similar trend on the smallest in edge situations on the mainland, the birds islands surveyed in their study of the Solomon occurred in the absence of other foliage- Archipelago, though no islands censused were gleaning species. (2) A similar pattern held as small as those considered here. for the Yellow-rumped Warbler, which also Increases in numbers of species with area was present with a reduced number of other are often believed to result, in part, from large warblers on the small islands, a number which areas having greater habitat ranges than small decreased with forest size. (3) The Song Spar- ones (Hamilton et al. 1964). However, be- row coexisted in large forests and deciduous- cause of the uniformity of the islands in this edge situations with two other common study, the actual size of an island probably ground-dwelling sparrows, the Dark-eyed is more important in accounting for the num- Junco and White-throated Sparrow. Juncos ber of species present on it than are more did not nest on these islands during the study, specific measures of the environment. Simber- and White-throated Sparrows only nested loffs’ (1976a) experimental results upon the sporadically. Islands containing the highest fauna of mangrove islets suggest a similar ex- density of Song Sparrows did not have White- planation. throated Sparrows on them. Song Sparrows The absence of a species, even for a single foraged regularly into the midst of the spruce year, may provide the opportunity for another forests. Also, they foraged in the low, live species to occupy the area. The cumulative foliage of the spruces on the tiny islands, a count over the 1967-1975 period thus is the mode of foraging not observed elsewhere in best measurement of the number of species this area except occasionally in juveniles that find the attributes of the islands satis- (unpubl. data). Yeaton and Cody (1974) factory for inhabitation; its magnitude is prob- noted similar patterns of release in this species. ably not primarily a function of interactions between species. If the mean number of spe- DISCUSSION cies per year and the minimum yearly counts differ markedly, as occasionally occurred in Many of my results invite comparison with this study, non-interactive conditions may the predictions of MacArthur and Wilson have been experienced as well. In particular, (1967) and those of other more recent papers certain results obtained from 1973 to 1975 on island biogeography. In contrast to many (Table 6) favor this interpretation. However, of those studies, the islands that I have studied the largest number of species to occupy an are distinctive in at Ieast five ways: no dis- area in a year should produce a situation in tance effect exists; dispersal is probably ade- which interspecific interactions reach a maxi- quate to insure saturation in most instances; mum. the size of these islands is smaller than that Numbers of species on the islands remained apparently required by many species to es- rather constant during the study period. The tablish themselves; these island populations decrease of the spruce-forest component and are not genetically isolated; and the species corresponding increase of the deciduous-edge are migratory. component following two poor breeding seasons provided a strong test of the hypothesis NUMBER OF SPECIES that these islands are basically in equilibrium. The low z values obtained after the three I cannot explain, however, why the spruce- smallest islands (<0.2 ha) were eliminated forest group apparently dropped below its from the calculations approximate the values carrying capacity, while the deciduous-edge usually obtained from mainland censuses of group did not. The spruce-forest species now different-sized areas (MacArthur and Wilson appear to be re-establishing themselves, and BIRDS ON SMALL ISLANDS 409 it will be instructive to determine to what rates of 33%/yr on small islands off the Con- degree they will reattain their pre-1972 con- necticut coast. centrations. PATTERNS OF RESOURCE EXPLOITATION TURNOVER This study and my earlier one (Morse 1971a) While the turnover of species on these islands demonstrate that some species modify their is not identical to that occurring on islands patterns of resource exploitation on these is- with low invasion rates and isolated breeding lands, while others show little, if any, such populations (see Lynch and Johnson 1974, tendency. The most flexible species were the Smith 1975, Simberloff 1976b), comparison most frequent occupants of these small islands, is profitable. The rates of new colonization and in at least some cases, were also the most and extinction on these islands (16.7%/yr) successful reproductively ( Morse 1971a). Sub- are well over an order of magnitude higher tle changes in niche overlap and niche width than those reported from the California Chan- occurred among some of these species (Morse nel Islands (0.3-1.2%/yr); Karkar, East In- 1971a), but the present study indicates that dies (0.32%/yr; Diamond 1969, 1971); and the greatest changes in resource exploitation Mona Island, West Indies (0.29%/yr; Ter- in the presence of other species lay in changes borgh and Faaborg 1973). However, these in the intensity with which any given part of authors ’ turnover rates are minimal, as their a territory was used. In the absence of certain study areas were only censused twice at ap- other species, some species successfully oc- proximately 50-yr intervals, with the result cupied territories less than one-fifth the size that many colonizations and subsequent ex- of those used when the others were present. tinctions may have gone unnoticed between Similar patterns have been noted by other the censuses. When I use only my 1967 and workers (e.g., Beer et al. 1956, Yeaton and 1975 data (first and last years), the mean Cody 1974). turnover rate calculated for my islands drops The tendency of deciduous-edge species to to 3.6%/yr, still much higher than the results make regular use of non-arborescent vegeta- of others, but considerably nearer to them. tion next to the forests probably accounted for Most of the colonizations and extinctions on the disproportionately high success of the de- my islands took place between 1967 and 1975 ciduous-edge species in colonizing these is- and hence were not accounted for in this lands. While the deciduous-edge species oc- abbreviated calculation. These results sup- cupied these islands with a frequency similar port the notion that actual turnover rates for to that of spruce-forest species, deciduous- the other studies were considerably higher edge species occupied areas for which the than the minimum ones that were calculated spruce-forest species, but not the deciduous- with their data base. edge species, were presumably specially Lynch and Johnson ( 1974) have challenged adapted. If new areas of unspecified nature many of Diamonds interpretations of indi- became available, the latter group would vidual colonizations and extinctions, and con- probably make the heavier contribution to cluded that his turnover rates were probably colonization. This result is not surprising, far too high. While several of Lynch and because others (e.g., Wilson 1961, MacArthur Johnsons’ arguments could be resolved only et al. 1972, Diamond 1974) have reported with new data, my results are based upon that species from marginal or disturbed main- year-to-year intensive censusing of demon- land habitats are the most frequent of island strably breeding individuals and, thus, are inhabitants. not subject to alternative explanation. DENSITY, BIOMASS, AND METABOLIC Though my results seem high in the con- DEMAND text that I have presented them, even higher Population densities of individual species were turnover rates than these have been recorded often much greater than those of the adjacent in near-shore situations for a nearly complete mainland; further, the cumulative density, insect fauna (Simberloff and Wilson 1970, biomass, and metabolic demand of birds on Simberloff 1974) and for ants (Goldstein these islands often considerably exceeded 1975). Simberloff and Wilson found turn- those of the mainland. Other studies of popu- over rates of 2%/day on mangrove islets in lation density on islands (reviewed by Sim- the Florida Keys (subsequently considered an berloff 1974) do not as a whole reveal any overestimate of true turnover by Simberloff simple pattern, and presumably data on bio- [1976bl)> and Goldstein reported turnover mass and metabolic demand would yield a 410 DOUGLASS H. MORSE comparable picture. It appears necessary, as concrete information, the competition hypoth- Diamond (1970) suggested, to examine each esis is consistent with the data, and the re- situation individually. The result obtained source-availability hypothesis has little to rec- could be caused by differences in (1) num- ommend it. bers of competitors from taxonomic groups not considered here, (2) simple concentrations SUMMARY of available resources (a non-interactive situation), (3) numbers of predators, or (4) Breeding landbirds on 12 small (0.14.0 ha), a combination of some or all of these factors. forested islands in Muscongus Bay, Maine, Numbers of species from potentially com- were censused from 1967 through 1975. All peting taxonomic groups also decrease mark- 17 species that bred also occupied spruce edly on these small islands (e.g., mammals, forests or adjacent disturbed habitats on the Crowell 1973, Morse 19731,). However, since nearby mainland. Similar numbers of species the possibility of competition with these ani- came from both areas, but since most islands mals has not been explored, little more can were spruce-covered, the contribution from be said about this matter. The food supply the disturbed areas exceeded that predicted should be correlated with the numbers of by chance. Size of forest accounted for the competitors present, but the ease with which greatest amount of variation in numbers of food may be procured could be a factor de- species present, while the total area vege- termining the density and biomass of indi- tated, amount of ground cover, and forest viduals present in a non-interactive situation. height made lesser contributions. Total vege- Since the amount of insects does not differ tated area accounted for the greatest amount markedly in large forests and on tiny islands of variation in numbers of species present (Fig. 2) during the periods most critical to from the disturbed areas, however, even the success of nesting individuals (Morse though this study only treated the species 1968), this factor appears unlikely to account that occupied the forested parts of the islands. for the differences in density observed. The When the smallest islands (<0.2 ha) were volume of foliage per unit area in large forests removed from the calculation, z (slope of the and on tiny islands does not differ, though log-log regression of species against area) trees in the large forests are taller than those fell near 0.20, which approaches a typical on the tiny islands (Morse 1971a, 1976). En- mainland figure. Including the smallest is- ergy flow could be higher on the islands than lands in the calculation, z was about 0.50, on the mainland, but there is no clear reason probably because these islands were too small why this should be the case. to support most species. Some species oc- Known predatory losses on the mainland curred only on islands at least as large as their result primarily from nest-robbing and capture territory sizes on the mainland, while others of fledglings by Blue Jays and red squirrels decreased their territory size considerably. ( Tamiasciurus canadensis; Morse 1976). With Those bird species not predicted on the basis one exception, I have never seen a red squir- of the size of their mainland territories often rel on these tiny islands, and Blue Jays oc- were low in the social hierarchy and demon- curred but infrequently. Therefore, preda- strated ecological release on the islands. Fol- tion almost certainly is less intense on the lowing two poor breeding seasons, spruce- islands than on the mainland. It seems ques- forest species declined markedly, while those tionable, however, whether predatory pres- of disturbed areas increased; consequently, the sur,e would significantly depress the numbers islands remained in equilibrium when taken of breeding adults on the mainland, espe- as a group. Turnover rates of species averaged cially when extra nonbreeding individuals 16.7%/yr. Numbers of vagrants actually ob- were present, as was often the case during served were adequate to saturate the is- some years of the study. The only large for- lands, at least in some years. Overall density, est without red squirrels (Harbor Island) biomass, and metabolic demand generally that I have studied had a substantially higher changed inversely with forest size and ex- overall density of birds than did the four ceeded those on the adjacent mainland. other mainland-type spruce forests censused (highest score plotted for large forests in Fig. ACKNOWLEDGMENTS 3). No one of these factors can unequivocally I thank J. H. Fellers, D. E. Gill, and M. L. Reaka for account for these differences in density, but comments on the manuscript. The National Audu- bon Society kindly permitted me to work on Hog the predation hypothesis is supported by some Island and Hockomock Point. This research was BIRDS ON SMALL ISLANDS 411 supported by the National Science Foundation (GB- MACARTHUR, R. H., AND E. 0. WILSON. 1963. An 6071, GB-31005). equilibrium theory of insular zoogeography. Evo- lution 17:373-387. LITERATURE CITED MACARTHUR, R. H., AND E. 0. WILSON. 1967. The theory of island biogeography. Princeton Univ. BALDWIN, S. P., AND S. C. KENDEIGH. 1938. Varia- Press, Princeton, New Jersey. tions in the weight of birds. Auk 55:416467. MORSE, D. H. 1966. Hog Island (a coastal spruce BEER, J. R., L. D. FRENZEL, AND N. HANSEN. 1956. forest) and its breeding vertebrate fauna. Maine Minimum space requirements of some nesting Field Nat. 22: 127-133. nasserine birds. Wilson Bull. 68:200-209. MORSE, D. H. 1967. Competitive relationships be- CADI&RY, J. M., AND A. D. CRUICKSHANK. 1937-58. tween Parula Warblers and other species during Climax red and white spruce forest (in reports of the breeding season. Auk 84:490-502. the annual breeding bird census). Bird-Lore MORSE, D. H. 1968. A quantitative study of forag- (later Audubon Msg. ) Vols. 39-60. ing of male and female spruce woods warblers. CROWELL. K. L. 1973. Exnerimental zooaeogranhv: Ecology 49:779-784. introductions of mice to small islands- Am. Nat. MORSE, D. H. 1971a. The foraging of warblers 107:535-558. isolated on small islands. Ecology 52:216-228. CRUICKSHANK, A. D. (no date). Summer birds of MORSE, D. H. 1971b. Effects of the arrival of a Lincoln County, Maine. Natl. Audubon Sot., new species upon habitats utilized by two forest New York. thrushes in Maine. Wilson Bull. 83:57-65. DIAMOND, J. M. 1969. Avifaunal equilibria and MORSE, D. H. 1972. Habitat differences of Swain- species turnover rates on the Channel Islands of sons’ and Hermit thrushes. Wilson Bull. 84:206- California. Proc. Natl. Acad. Sci. U.S.A. 64: 208. 57-63. MORSE, D. H. 1973a. The foraging of small populations of Yellow Warblers and American Red- DIAMOND, J. M. 1970. Ecological consequences of island colonization bv southwest Pacific birds. starts. Ecology 54:346-355. 197313. Habitat utilization of meadow II. The effect of species diversity on total pop- MORSE, D. H. ulation density. Proc. Natl. Acad. Sci. U.S.A. voles on small islands. J. Mammal. 54:792-794. 67: 1715-1721. MORSE, D. H. 1974. Niche breadth as a function of social dominance. Am. Nat. 108:818-830. DIAMOND, J. M. 1971. Comparison of fauna1 equi- MORSE, D. H. 1976. Variables determining the librium turnover rates on a tropical island and density and territorial size of breeding spruce- a temperate island. Proc. Natl. Acad. Sci. U.S.A. woods warblers. Ecology 57:290-301. 68 :2742-2745. MORSE, D. H. 1977. Hostile encounters among 1974. Colonization of exploded DIA~LIOND, J. M. spruce-woods warblers ( Dendroica, Parulidae ). volcanic islands by birds: the supertramp strat- Anim. Behav. 24:764-771. egy. Science 184: 803-806. POOLE, E. L. 1938. Weights and wing areas in DIAMOND, J, M. 1975. Assembly of species com- North American birds. Auk 55:511-517. munities, p. 342-444. In M. L. Cody and J. M. SIMBERLOFF, D. S. 1974. Equilibrium theory of Diamond [eds.]. Ecology and evolution of com- island biogeography and ecology. Annu. Rev. munities. Harvard Univ. Press, Cambridge, Mass. Ecol. Syst. 5: 161-182. DIAMOND, J. M., AND E. MAYR. 1976. Species-area SI~~BERLOFF, D. S. 1976a. Experimental zoogeog- relation for birds of the Solomon Archipelago. raphy of islands: effects of island size. Ecology Proc. Natl. Acad. Sci. U.S.A. 73:262-266. 57: 629-648. GOLDSTEIN, E. L. 1975. Island biogeography of SI~~BERLOFF, D. S. 197613. Species turnover and ants. Evolution 293750-762. equilibrium island biogeography. Science 194: HAMILTON, T. H., R. H. BARTH, JR., AND I. RUBINOFF. 572-578. 1964. The environmental control of insular SIMBERLOFF, D. S., AND E. 0. WILSON. 1970. Ex- variation in bird species abundance, Proc. Natl. perimental zoogeography of islands. A two-year Acad. Sci. U.S.A. 52:132-140. record of colonization. Ecology 51:934-937. LASIEWSKI, R. C., AND W. R. DAWSON. 1967. A re- S~~ITH, F. E. 1975. Ecosystems and evolution. Bull. examination of the relation between standard Ecol. Sot. Am. 56( 4) :2-6. metabolic rate and body weight in birds. Con- STEWART, P. A. 1937. A preliminary list of bird dor 69: 13-23. weights. Auk 54:324-332. LYNCH, J. F., AND N. K. JOHNSON. 1974. Turnover TERBORGH, J., AND J. FAABORG. 1973. Turnover and equilibria in insular avifaunas, with special and ecological release in the avifauna of Mona reference to the California Channel Islands. Island, Puerto Rico. Auk 90:759-779. Condor 76:370-384. WILSOX, E. 0. 1961. The nature of the taxon cycle MACARTHUR, R. H. 1972. Geographical ecology. in the Melanesian ant fauna. Am. Nat. 95:169- Harper and Row, New York. 193. MACARTHUR, R. H., J. M. DIAMOND, AND J. R. KARR. YEATON, R. I., AND M. L. CODY. 1974. Competitive 1972. Density compensation in island faunas. release in island Song Sparrow populations. Ecology 53 :330-342. Theor. Popul. Biol. 5:42-58. 412 DOUGLASS H. MORSE

APPENDIX I. Species pool of breeding landbirds OTHER SPECIES COMPONENT”-Black-billed from mainland and large inshore islands.” Cuckoo ( Coccyzus erythropthulmus), Whip-poor-will ( Caprimulgus vociferus), Eastern Kingbird ( Tyrun- nus tyrunnus), Great Crested Flycatcher ( Myiurchus SPRUCE FOREST COMPONENTb-Yellow-bellied crinitus), Alder Flycatcher ( Empidonax alnorum), Flycatcher ( Empidonux fluviventris), Blue Jay ( Cy- Least Flycatcher ( E. minimus), White-breasted Nut- anocitta cristata), Black-capped Chickadee ( Purms hatch ( Sitta curolinensis), Brown Thrasher ( Toxos- utricapiZZus), Red-breasted Nuthatch ( Sitta cunaden- toma rufum), Wood Thrush (Hylocichla mustelinu), sis), Brown Creeper ( Certhia familiaris), Winter Veery ( Cuthurus fuscescens) , Eastern Bluebird Wren (Troglodytes troglodytes)d, Hermit Thrush ( Sialiu siulis), Tennessee Warbler ( Vermivoru pere- (Cutharus guttutus), Swainsons’ Thrush (C. ustulu- grinu), Black-throated Blue Warbler (Dendroicu tus), Golden-crowned Kinglet ( Regulus sutrupu), caerulescens), Northern Waterthrush ( Seiurus nov- Solitary Vireo (Vireo soliturius), Northern Parula eboracensis), Bobolink ( Dolichonyx oryzivorus), (Par& umericunu)d, Magnolia Warbler (Dendroica Eastern Meadowlark (Sturnella magna), Redwinged magnolia), Cape May Warbler (D. tigrinu), Yellow- Blackbird ( Agelaius phoeniceus ), Common Grackle rumped Warbler ( D. coroneta), Black-throated Green ( Quisculus quisculu), Northern Oriole (Icterus gul- Warbler (D. sirens), Blackburnian Warbler (D. hula), Scarlet Tanager (Pirangu olivaceu), Rose- fuscu), Bay-breasted Warbler (D. custuneu), Brown- breasted Grosbeak (Pheucticus ludoviciunus), Indigo headed Cowbird (Molothrus ater), Purple Finch Bunting (Passerinu cyunea), Savannah Sparrow (Pus- ( Curpoducus purpureus), Pine Siskin ( Curduelis serculus sandwichensis), Field Sparrow ( Spizellu pu- pinus), Red Crossbill (Loxiu cumirostru), White- s&z), Swamp Sparrow ( Melospiza georgiana). winged Crossbill ( L. leucopteru), Dark-eyed Junco (Junco hyemulis), White-throated Sparrow (Zono- e Nesting on mainland within 2 km of shore. trichia albicollis). DECIDUOUS-EDGE COMPONENT”-Mourning Dove (Zenuida mucroura), Ruby-throated Humming- bird ( Archilochus colubris), Common Flicker (Col- uptes auratus ) , Hairy Woodpecker ( Picoides villosus ) , APPENDIX II. Weights (g) of species used in cal- Downy Woodpecker (I.’ pubescent), Eastern Wood culations.” Pewee ( Contopus virens), Olive-sided Flycatcher ( Nuttullornis borealis), Catbird ( Dumetella curolinensis), American Robin (Turdus migratorius), Cedar Red-breasted Nuthatch, 10.3; Brown Creeper, 8.4; Waxwing (Bombycilla cedrorum), Red-eyed Vireo Winter Wren, 9.0; Catbird 34.5; American Robin, ( Vireo olivaceus ) , Black-and-white Warbler ( Mnio- 76.2; Hermit Thrush, 29.7; Swainsons’ Thrush, 29.6; tiltu vuriu), Nashville Warbler (Vermivoru rufi- Golden-crowned Kinglet, 6.0; Ruby-crowned Kinglet cupilla), Yellow Warbler ( Dendroicu petechia), ( Regulus calendulu), 6.5; Red-eyed Vireo, 16.9; Chestnut-sided Warbler (D. pensylvanica), Oven- Black-and-white Warbler, 10.2; Northern Parula, 7.0; bird ( Seiurus aurocapillus), Yellowthroat (Geothlypis Yellow Warbler, 9.6; Magnolia Warbler, 9.8; Yellow- trichus ) , Canada Warbler ( Wilsoniu canudensis ) , rumped Warbler, 11.5; Black-throated Green Warbler, American Redstart ( Setophaga ruticillu), American 9.2; Blackburnian Warbler, 10.3; Yellowthroat, 9.5; Goldfinch ( Carduelis tristis), Rufous-sided Towhee Canada Warbler, 9.0; American Redstart, 8.0; Purple ( Pipilo erythrophthulmus), Chipping Sparrow ( Spi- Finch, 24.5; Dark-eyed Junco, 20.5; White-throated zellu pusserina), Song Sparrow ( Melospizu melodiu) . Sparrow, 26.5; Song Sparrow, 20.7.

a Data represent mean weights of males and females com- a From Cruickshank (no date); Morse (1966, unpubl. bined from birds captufed in study ?reas, supplemented by Does not include several species directly dependent data). y1gy8y’ (1937), Baldwm and Kendelgh (1938), and Poole upon human habitation for nest sites. . With rare exceptmns sample wes or data from b Many of these species also nested in the deciduous-edge other sources are inadequate to calculate confidence intervals. habitats. c Deciduous growth, mixed deciduous growth, areas of blown-over trees, and edge of forest. Many of these species also nested in habitats supporting species in last category. Department of Zoology, University of Maryland, Col- d Placed in deciduous-edge component by Morse ( 1966). lege Park, Maryland 20742. Accepted for publication Subsequent observations justify inclusion in spruce forest component. 9 December 1976.