The Influence of Area and Habitat on the Avian Community in Red Maple Swamps of Southern Rhode Island

Red maple swamps are common throughout the glaciated Northeast and, along with other wetland types, are protected for their wildlife habitat and other functions. Yet there are few descriptions of red maple swamp wildlife communities, and little research on how the wildlife are influenced by habitat features. Several states -Gefine jurisdictional wetlands on the basis of wetland size, but the influence of area on wetland wildlife communities is largely unknown. Birds were cen~used in 12 mature, ve~y poorly drained red maple swamps in southern Rhode Island. Swamps ranged from 0.49 to 19.24 ha and were placed in four size categories. Avian community composition was described and the influence of area and habitat on the avian community were e xamined. Five species made up the majority (66%) of singing bird observations: Canada warbler, Gray catbird, Black-and-white Warbler, Veery, and Northern Waterthrush. The avian association was similar in composition to that observed by other researchers in red maple swamps in west-central Massachuse tts. Species richness at individual sites ranged from 3 to 15 singing bird species and from 7 to 24 total species. Richness was strongly (P<0.0001) related to swamp area: for

Birds were cen~used in 12 mature, ve~y poorly drained red maple swamps in southern Rhode Island. Swamps ranged from 0.49 to 19.24 ha and were placed in four size categories. Avian community composition was described and the influence of area and habitat on the avian community were e xamined.
Five species made up the majority (66%) of singing bird observations: Canada warbler, Gray catbird, Black-and-white Warbler, Veery, and Northern Waterthrush. The avian association was similar in composition to that observed by other researchers in red maple swamps in west-central Massachuse tts.
Species richness at individual sites ranged from 3 to 15 singing bird species and from 7 to 24 total species. Richness was strongly (P<0.0001) related to swamp area: for ii singing birds, R 2 =D.81, and for all species observed, Rz~o.84. Wetlands in either of the two larger size categories supported signif icant~y more species than wetlands in either of the two smaller categories. Area did not relate significantly to avian relative abundance.
The smallest s wamps studied, down to -0.5 ha, supported several breeding species, including the Northern waterthrush, an obligate wetland species. Thus red maple swamps down to at least o._5 ha have significant wildlife habitat value and support "wetland species." There was a rapid increase in the number of species in swamps up to about 6-8 ha in size, and a slower increase in species richness beyond this size.
In stepwise regression models, swamp area and measures of shrub structure were significantly related to species richness.
Avian relative abundance was significantly related only to the thickness of the organic soil layer; the nature of the relationship between these variables is unknown.
iii  The avian community of red maple swamps has been described at only a few sites in the Northeast. Breeding

LIST OF TABLES
Bird census (BBC) results have been reported for three red maple swamps in New Jersey Seeley 19 54 1 1955Seeley 19 54 1 , 1956Seeley 19 54 1 , 1957Seeley 19 54 1 , 1966Taylor l984) and one in western New York ). Golet et al. (in prep.) have summarized these results.  used mist nets to census birds 2 in a red maple-Atlantic white cedar (Chamaecyparis thyoides) swamp in southeastern Massachusetts as part of an encephalitis research project. Mist-netting samples a relatively small segment of the bird community and the results are comparable only to studies using similar netting techniques and sampling effort  .  censused the breeding birds in eight large (30-45 ha) west-central Massachusetts red maple swamps. Their study sites were heterogeneous, including both forested swamp and shrub swamp, and both poorly drained and very poorly drained soils (see  for drainage class definitions) . Some of the sites also included features such as upland islands and powerline corridors.
Clearly, more studies are needed before the avian community of Northeastern red maple swamps can be characterized definitively.
Further, the effect of ·ations in geography, plant community structure and vari floristics, water regime, and other factors should be examined.
Many aspects of habitat influence the density or 3 species richness of forest-dwelling breeding birds. Among these features are the relative length of habitat edge (Kroodsma 1984, Gotfryd and; the types or diversity of surrounding habitat (Whitcomb et al. 1981, Gotfryd and; the degree of isolation of the habitat patch from similar habitats Whigham 1984, Opdam et al. 1985); vegetation structure MacArthur 1961, James andWarner 1982); and the extent of surface water .
Only swift et al. (1984) have investigated the effect of habitat on the avifauna of red maple swamps. Using . multipl e regression models, they found that the abundance of breeding birds was positively related to the stem density of shrubs 1-3 m tall, the percent cover of surface water, and the depth (thickness) of the organic soil layer. Bird spe~ies richness within census plots was positively related to stem density of shrubs 1-3 m tall and organic soil depth; it was negatively related to tall (3-5 m) shrub stem density and to lowest overstory branch height. As swift et a l.
pointed out, however, there was extreme collinearity among their independent variables; this makes it difficult to identify the most important variables in their regression models. The area of a habitat patch strongly affects wildlife community composition , King 1987. Larger blocks of habitat tend to support more species in greater numbers , provide a buffer against the influence of external factors such as parasitic edge species , and reduce the rate of species extinction within a given patch  imP 5 significant avian species-area relationships have been found in forested habitats (e.g., ) and in freshwater marshes Dinsmore 1986, Tyser 1983). However, no speciesarea research has been done in forested wetlands.
Breeding bird density has been found to decrease as the area of an island or habit.at patch increases , tynch and Whigham 1984.  proposed four rules on breeding bird density in woodland habitat, including a rule that density decreases as the area of uniform habitat increases.  reviewed European Breeding Bird Census data and found that "small and moist" areas had higher breeding bird density than larger, drier ones.  studied 1-to 14-ha forest patches in Delaware and, although he found no clear relationship between area and density, he asserted that the densities found were higher than those from "interior" habitats (i.e., areas removed from habitat edges).  found a highly significant (P<0.001) decrease in density with increasing area of shelterbelts in the U.S.
Midwest.  studied a wide range of forest patch si zes in the U.S. Middle Atlantic States, and found patch area to be inversely related to the relative abundance of birds (the total number of pairs of birds Su sed at a single point within each forest). cen 6 The influence of wetland size on the abundance and species richness of the avifauna of red maple swamps is unknown; investigation of that topic should provide some basis for judging whether the current size minima of wetland regulations are warranted.
The research reported on here addresses several of the above topics.
Specific objectives of this research were: 1. To describe the breeding bird community of mature, very poorly drained red maple swamps, which predominate in much of southern New England.
2. To determine the influence of wetland size on the breeding bird community of these swamps.
3. To further elucidate the influence of habitat on breeding bird abundance and richness in red maple swamps. Island. Sites meeting the above criteria were examined on large-scale aerial photographs and then field-checked. The sites selected (Figure 1 and Appendix A) ranged from 0.49 ha to 19.24 ha and were grouped in 4 size categories (Table 1).

Bird Censuses
Censusing birds in wooded swamps presents two major problems. First, the vegetation is dense, so that birds are difficult to see and must be censused primarily by vocalizations. This makes spot-mapping, which depends largely on visual observations, difficult. Second, in small swamps, the observer is often close to a habitat edge, so that unlimited-distance methods such as Emlen's variable-width transect  and the Indice  ponctuel D'Abondance (IPA) point count  are inappropriate.
For these reasons, the fixed-radius circular plot technique was selected , DeGraaf 1987, Morrison et al. 1987 Birds were censused six times at each plot center between 25 May and 2 July 1988. Each plot census consisted of a 1-min "settling" period followed by a 5-min observation period during which all bird observations within the plot were recorded. The bird species and type of observation--singing, calling, or visual--were noted. Only clearly identifiable territorial or mate-attraction vocalizations that are frequently repeated by birds on territories were considered "songs." All censuses were carried out within 4 hours after sunrise.
Each morning one group of sites was censused. Each group consisted of two or three sites. Groups were fixed throughout the census season and were based on site size categories and proximity: sites within a group were from different size categories and were as far apart geographically as feasible.
The order of censusing of groups, sites within a group, and plots within a site was varied in the following systematic way in an effort to minimize the effects oi time of day and season.
1. The order in which groups were c e nsused was rota ted after each complete round of sites, so that the group that was censused first in a given round was censused last in the following round.
2. The order in which sites within a group were censused was rotated each successive time, as with groups.
3 . The order of censusing of plots within a site was reversed for each successive census.
species richness (the total number of species observed during all six censuses) and relative abundance (the a v erage number of birds per plot per census) were calculated for both singing bird observations and all bird observations at each site.

Measurement of Independent Variables
Measurements of study-site area and surrounding habitat diversity were based on 1:4,800-scale panchromatic aerial photographs. Area measurements were made with a digital planimeter, and the length of wetland edge corresponding to each surrounding habitat type was determined with a map measurer. Thes e edge lengths were then entered into the Shannon diversity formula  to obtain a measure of surrounding habitat diversity for each study site.
Patches of surrounding habitat had to be within 50 m of the wetland edge and at least 50 m long and wide to b e measured. 3. Tree diversity.
Live trees were identified to species, and the Shannon index  was used  This method adopts the categorical approach found in two-dimensional cover estimators (Daubenmire 1959, and is similar in concept to other recent efforts to quantify foliage volume in dense habitats   (11), Veery (11), Canada Warbler (10), and Black-and-white Warbler (9).
The most abundant species generally were found in the greatest number of study sites. Although most species observed were represented by only a few individuals, many were found at a relatively large number of study sites.
Rufous-sided Towhees, Great Crested Flycatchers, and Redeyed vireos each accounted for 3% or less of all observations, yet were censused at seven sites. Brown creepers were observed only seven times overall (1% of observations), but were found at five study sites.
In eight west-central Massachusetts red maple swamps studied by , the most abundant species were largely the same as those encountered in southern Rhode 22 Island (Table 4). The minor differences may be attributable to differences in habitat complexity or geographic variation in species abundances or habitat use. It is apparent that the avian community of red maple swamps in southern New England is dominated by fewer than 10 common species.
Relative abundance.--Relative abundance (birds/plot/census) of singing birds ranged from 0.58 at the Narragansett site to 2.00 at Tuckertown (Table 5), with an average for all study sites of 1.05 ± 0.11 (SE). Relative abundance at most study sites (9 of 12) fell within the relatively narrow range of 0.80 to 1.08. The relative abundance of all birds observed ranged from 1.58 at Carolina ble 4. Comparison of the most abundant birds in red maple Ta mps in southern Rhode Island (this study) and westswatral Massachusetts    Analysis of variance revealed significant differences (P<0.05) among the mean richness values of wetland size categories for both singing birds and all birds observed (Table 6). Species richness was not significantly different between the two smallest size categories or between the two largest size categories; however, wetlands in either of the two largest categories supported significantly more species than wetlands in either of the two smallest categories.
Although area explained the great majority of the variation in the singing bird species richness among study sites, the variation in richness can also be explained simply by the variation in sample area (i.e., number of census plots) among sites (R 2 =0.77). However, because sample area and wetland area are very closely correlated (r=0.99), the results are believed to accurately reflect the influence of wetland area on species richness.
The species-area analysis showed that (1) there were several species singing--and presumably breeding--in swamps less than 1 ha in size, (2) there was a rapid increase in the number of species in swamps up to about 6-8 ha in size, and (3) there was a slower increase in species richness beyond 6-8 ha.
The smallest sites, down to 0.5 ha, all supported several breeding species. One of these species, the Northern Waterthrush, was an obligate wetland species.  and  also found (1) these species are responding to the larger forested landsca pe in which the swamps are located, or (2) they are not actually area-sensitive.
In this study, species richness increased at a relatively rapid rate until the wetlands exceeded about 6-8 ha in size. The same trend was apparent for singing b i rds and all birds observed (Figure 2). Richness continued to increase, but at a lower rate, in sites larger than 8 ha.   four forested, very poorly drained study sites. As noted above, there were some methodological differences between the two studies, bu~ the results are consistent (Figure 3), and suggest that species richness may continue to increase gradually as swamp size increases beyond the size range sampled in the present study.
A frequently discussed aspect of the species-area relationship is the slope or z value of the regression equation (Connor and McCoy 1979). This parameter expresses the rate of increase in species richness, and may be  (Table 6).
Previous research (Oelke 1966, Lynch andWhigham l984, Martin 1980) has found bird density · to decrease with increasing area; the higher density of animal populations near habitat edges   richness, also may be important in determining habitat "value," for example, the preservation of rare species, forest-interior species, species restricted to certain habitats, or species of special interest. In this study, the important conclusion is that all sites, regardless of size, display breeding bird habitat value, and that "wetland species" breed in even very small sites.

The Influence of Habitat
Habitat   Table 8. Stepwise regression models, including significant (P<0.05) ~ndependent varia~les explaining variation in bird species richness and relative abundance. The influence of shrub layer structure on avian species richness in red maple swamps has been documented-by . In a multiple regression analysis using 80 census plots (rather than study sites) as samples, they found significant relationships between avian richness and depth of peat (+), density of shrubs 1-3 m high (+), density of shrubs 3-5 m high (-), and lowest overstory branch height (-). Both this study and that of  suggest that dense shrubs within 2-3 m of the ground, combined with a more open shrub layer above that level, support more avian species. Dense, low shrubs may provide ~scape cover and foraging and nesting substrates for a wide range of species. Other studies also have documented the importance of shrub layer structure, and foliage density in particular, to both bird species richness (Blake andKarr 1987, Martin 1988) and abundance .
However, none of these studies distinguished between shrub height categories.
Habitat and relative abundance.--Peat depth was the only variable significantly related (P<0.05) to avian relative abundance in stepwise multiple regression models. When peat depth was excluded from the analyses, no other variables were significant at P<0.05, and only tree basal area was significant at the P<0.15 level (and only for singing birds). Simple correlations (Appendix E) between relative abundance and most habitat variables also were very weak. More research is needed on the possible relationships between the avian community, soil characteristics, vegetation, and perhaps also invertebrate prey abundance. Unidentif ied spec i es a re not included in species richness t otal s , except for unidenti f i ed woodpeckers (a ll Pi coides s peci es ) in s it es wh ere thes e s pecies did not oth e rwi se occu r . Appendix C.