Use of Invertebrates by Birds in Red Maple Forested Wetlands and Contiguous Forested Uplands in Southern Rhode Island

Successful management of wetland wildlife populations requires a basic understanding of invertebrate ecology and their availability as food to higher life forms. Community structure, abundance, and seasonal dynamics of litter invertebrates in red maple forested wetlands are unknown . Differences in these parameters may influence where wildlife species forage along a wetland-upland gradient. I studied invertebrate use by ground-foraging birds along moisture gradients from upland forests to red maple (Acer rubrum) forested wetlands at three sites in Washington County, Rhode Island from April through August 1991. I examined diets of ground-foraging birds by stomach-flushing birds with saline solution and immediately preserving stomach contents. The invertebrates within the stomach samples were identified at least to order. I collected invertebrates along the upland-wetland gradient at each site to determine the mean biomass of invertebrates available to ground-foraging birds . Additionally, I monitored water tables, sampled shrub density and identified microhabitat types along the gradient at each site to correlate with invertebrate biomass . The most common invertebrates found in ground litter were larval Diptera, larval Coleoptera, adult Hymenoptera (Formicidae), adult Coleoptera and Araneae. The mean biomass of the litter invertebrates was greater in the wetland habitats at all three sites (f < 0.05) . The mean biomass of litter invertebrates differed significantly from month to month along the gradient at two sites

x

I. INTRODUCTION
Food availability commonly limits wildlife populations and affects how wildlife select habitats (Martin 1987). Food type, food abundance, and the locations from which food is taken are of critical importance to the survival of any animal (Morrison et al. 1992). Undisturbed upland habitat surrounding wetlands may be necessary for survival of many wetland species because of the food resources it provides to these species.
Managing wetland wildlife populations requires a basic understanding of invertebrate ecology and availability of invertebrates for higher life forms. Invertebrates in forest litter are of particular interest because several vertebrates (e.g., birds, small mammals, reptiles, and amphibians) depend on invertebrates for food (Martin 1987). Community structure, abundance, and seasonal dynamics of litter invertebrates in red maple forested wetlands and adjacent upland forests are unknown. Differences in these parameters may influence where upland or wetland wildlife species forage along this wetland-upland gradient. Knowledge of wildlife foraging patterns will aid wetland managers in determining biologically significant widths of protected upland habitat (buffer zones) around red maple forested wetlands.
Birds are an appropriate taxon for studying the importance of invertebrate communities along the wetland-upland gradient because they are conspicuous consumers of invertebrates. Birds can choose foraging sites along the moisture gradient because they are highly mobile.
Also, there are reliable methods for determining invertebrate use by birds (Rosenberg and Cooper 1990).
-1-Ground-foraging insectivorous birds tend to track temporal variation in resource availability more precisely than do other bird species; positive correlations between bird capture rates and abundances of litter arthropods were stronger for ground insectivores (Karr and Brawn 1990). Therefore, measuring abundances of arthropods in litter is a good way of determining the relative value of potential foraging areas for ground-foraging birds along a moisture gradient.
In Rhode Island, red maple forested wetlands comprise 77% of all palustrine wetlands (Golet et al., In press.). Several species of birds are found in red maple forested wetlands during the breeding season (Swift 1980 (Bent 1958). Veeries and Ovenbirds direct >50% of their foraging attacks toward prey in the forest litter, on ground layer herbs, ferns, and low seedling foliage (Holmes and Robinson 1988).
Other species that are mainly ground f?ragers include Northern Waterthrush (Seiurus novaboracensis) (Bent 1953, Eaton 1957, Craig 1984 and Canada Warbler (Wilsonia canadensis) (Bent 1953). In red maple forested wetlands of southern New England, these latter two species tend to be wetland-dependent (Merrow 1990).
Insect availability is the abundance of potential prey in microhabitats used by an insectivore when searching for food (Wolda 1990). All the arthropods detected in abundance estimates are not potential prey items for birds; some may be unpalatable or require excessive time or energy for capture (Martin 1986, Wolda 1990. Use is -2-a demonstrated presence of a particular prey item in an animal's diet (Morrison et al. 1992). Selection is use coupled with evidence that the frequency of occurrence in the diet (by number of prey items, biomass, etc.) is significantly greater statistically than the frequency in the animal's environment (Morrison et al. 1992).
Scientists studying ecological relationships of animals seldom attempt to quantify the occurrence of prey items in the diet. Rather, studies have concentrated on indirect measures of food use such as bird foraging locations (Hutto 1985;Rosenberg and Cooper 1990).
There are several benefits to measuring food use directly by obtaining diet samples from birds.
In the field, researchers often do not have time to identify the prey item in the bird's mouth before the item is swallowed or the bird flies away .
When diet samples are brought back to the lab, researchers can more accurately identify the prey items because the invertebrates (whole or fragmented) may be placed under the microscope for detailed analysis and expert taxonomists can be consulted if necessary.
Comparatively few field experiments have tested the role of specific factors in regulating wetland invertebrate populations (Neckles et al. 1990). Shelford (1951) and Kendeigh (1979) emphasized that environmental factors regulated invertebrate populations. Two significant factors influencing avian community abundance in studies of red maple forested wetlands of southern Rhode Island were water regime and vegetation structure (Golet et al., In press.). Perhaps these factors also influence the insectivorous prey base of birds in red maple forested wetlands and contiguous forested uplands. -3- The extent of flooding, and its timing and duration determine the composition and productivity of plants and associated animal within wetland systems (Fredrickson and Reid 1986). In 8 Massachusetts shrub and forested wetlands, spanning a wide range of hydrologic, edaphic and structural conditions, wetter sites also had greater peat depths, denser shrub layers, and a larger, more diverse breeding bird community (Swift et al. 1984) .
Shrub layer structure appeared to be most closely related to avian richness and abundance (Swift et al. 1984, Merrow 1990  Lastly, the type of litter and litter ~oisture may influence types of invertebrates and their abundances found along a gradient from wetland to upland. Litter is a key element in the productivity of wetlands and eventually determines the value of a site for animal life (de la Cruz 1979, Nelson and Kadlec 1984, Batema et al. 1985, White 1985, Wylie 1985.
In this study, I obtained baseline data on the invertebrate food of birds associated with red maple forested wetland ecosystems . I identified invertebrates along the gradient from forested upland through forested wetland; determined differences in biomass of -4-invertebrates among moisture zones and throughout the breeding season; determined use of prey by 5 target birds species; assessed food selection by comparing abundance of invertebrate taxa in bird diets to abundance of invertebrate taxa along the gradient, and determined the relationship among environmental variables and invertebrate biomass.
-5-  containing invertebrates by placing a 30 cm diameter plastic ring on 2 the ground and scraping to the root mass. A total of 0.64 m of litter was collected in each zone during each sample period. To identify the type of microhabitat for each sample, we categorized the collected litter by percent cover type (e.g., 100% loose dry leaves; 90% sphagnum moss, 10% wet compact leaves). I extracted invertebrates from litter using Tungren funnels with 1-mm mesh size. I placed them under 100-watt lights for 48 hours -8- (Steyskal et al. 1986). I determined 48 hours was the minimum amount of time needed to extract the maximum number of invertebrates in an earlier pilot study. I categorized invertebrates by taxon (at least to order) and total body length (mm). Length categories included: 1-5 mm, 6 -10 mm, 11-15 mm, and ~16 mm. The number of individuals in each taxon was counted for each sample .
I determined the total number of individuals for each taxon in each zone by order and length class. I determined mean biomass of invertebrates for each zone using previously established models of length-weight relationships (Rogers et al. 1977) . I compared mean biomass of invertebrates among zones using Wilcoxon signed-rank tests (SAS Institute, Inc. 1989). I compared mean biomass of invertebrates among microhabitat types zones using Wilcoxon signed-rank tests (SAS Institute, Inc. 1989).
I divided the summer season by months 13 April-30 April; 1 May-30 May; 31 May-30 June; 1 July-30 July; 31 July-28 August. I compared mean biomass of invertebrates through the summer season using Wilcoxon signed-rank tests (SAS Institute, Inc. 1989).

Diet Sampling
To sample bird diets, I mist-netted 5 bird species, including Ovenbird, Northern Waterthrush, Veery, Gray Catbird, and Canada Warbler 2 times/week at each site and zone from May through August, 1991. I stomach-flushed the birds with lukewarm water (Forde et al. 1982 Mean biomass of invertebrates throughout the summer at each plot was correlated with mean shrub density at each plot using Spearman's rank-order correlation (SAS Institute, Inc. 1989) .
Differences in mean biomass of invertebrate taxa selected as prey were noted between upland and wetland zones only at Arrow Swamp. Mean biomass of prey taxa was greater in the wetland zones than in the upland zones for Canada Warblers and Northern Waterthrush (f < 0.01) (Table 4). No significant differences in mean biomass of prey taxa among zones were noted for any target bird species at Burlingame State Park or Great Swamp (Table 5). However, the biomass of prey taxa for Veeries, Northern Waterthrush, Canada Warblers, and Gray Catbirds was consistently higher in the wetland at Great Swamp (Table 6).

Monthly Differences in Biomass
The mean biomass of litter invertebrates . at Arrow Swamp was significantly different from month to month in zones A, C, and D (f < 0.04) although there was no clear pattern in how the biomass changed (Table 7). The mean biomass of litter invertebrates at Burlingame State Park decreased significantly from May to June and June to July in zone C (f < 0.0001) (Table 3) .. The mean biomass of invertebrates decreased significantly from June to July in zone D (f < 0.002) ( Table   8). There were no differences in mean biomass of invertebrates among the months in any zone at Great Swamp (Table 9).      bMeans within a column followed by the same letter do not differ (Wilcoxon signed-rank test) (£ < 0.05).  aZones were based on soil moisture and distance from the wetland-upland edge. Zone A (furthest upland zone from wetland-upland edge), Zone B (closest upland zone to upland-wetland edge), Zone C (closest wetland zone to upland-wetland edge), Zone D (furthest wetland zone from wetland-upland edge).
bMeans within a column followed by the same letter do not differ (Wilcoxon signed-rank test) (f < 0.05).       (Table 13).

Invertebrate Community
Thirty-eight percent of all the invertebrates I found in zone C and 23% of all the invertebrates I found in zone D were larval Diptera. It is not surprising that larval Diptera were the most common individuals found in the litter layer of these zones because many larval Diptera live in a variety of microhabitats; e.g., water, soil, under bark or stones, or on vegetation (Borror et al. 1989) typically found in red maple forested wetlands. Studies of invertebrates from seasonally flooded freshwater wetlands reveal remarkable similarities in community structure (Neckles 1990). Depressions which are flooded for only short periods during the year are characterized by very high densities of aquatic invertebrates with low taxonomic diversities (Wiggins et al. 1980).
The order Araneae is a large and widespread group. They occur in many types of habitats and are often very abundant (Borror et al. 1989 These results may reflect the bias of finding hard-bodied invertebrates , which are more difficult for birds to digest, more frequently in diet samples. Bent (1953) found Veeries principally ate ground beetles, ants, caterpillar and grasshoppers. Holmes and Robinson (1988) found Veeries foraged more on the ground than other -43-There are biases associated with analyzing diet samples (Rosenberg and Cooper 1990). Coleoptera, most of which were adults, were the most frequently found item in all diet samples. However, these results may be biased because Coleoptera body parts, especially elytra, probably persist longer in the stomachs than those of other types of prey (Robinson and Holmes 1982) .
Like other researchers, I found it difficult to identify and quantify small and/or fragmented food items. I used whole invertebrates that I had collected in litter and keys from previous diet studies to help me identify these fragments or small parts.
Although I usually obtained more than one prey item per sample, In general, the biomass of invertebrates eaten by the target birds was not significantly different among zones. However, the biomass of invertebrates eaten by Canada Warblers and Northern Waterthrush at Arrow Swamp was greater in the wetland zones . There was also a tendency for the biomass of invertebrates eaten by Canada Warblers and Northern Waterthrush to be greater in the wet~and zones at Great Swamp .

In red maple forested wetlands, Canada Warblers and Northern
Waterthrush are wetland dependent species (Merrow 1990) . This suggests food may be a factor involved in habitat selection for Canada Warblers and Northern Waterthrush. Food may restrict bird species such as the Canada Warbler and Northern Waterthrush moving between upland and wetland habitats or it may attract birds from the uplands to forage where there is higher invertebrate biomass . This finding is consistent with Robinson and Holmes's (1982) hypothesis that food influences the pattern of bird habitat selection and community structure. Similarly, Craig (1984) studied seasonal changes in invertebrate biomass in Waterthrush territories along a deciduous forested riparian system in northeastern Connecticut. He sampled three times between mid-May and late June. He found biomass was highest early in the season and declined afterwards.
When I analyzed prey taxa and non-prey taxa together, no pattern of biomass increase or decrease was noted over time or within zones, although significant differences in inverteb~ate biomass throughout the breeding season were found at all three sites. Perhaps I masked what was actually happening to the target bird's food resources during the breeding season by analyzing changes in invertebrate biomass using all the extracted invertebrates from our samples .
If invertebrate food resources are more limited late in the breeding season, birds that arrive on site early may have greater nest success. Evolution would favor the "early bird".
However, invertebrate taxa and abundance may vary from year to year (Stenger 1958). Therefore, invertebrate communities and -46-abundances along wetland-upland gradients would have to be sampled for several years to be certain that food is a limited resource during the breeding season at these sites.

Biases in Biomass Estimates
Most techniques which sample food availability are biased because researchers lack the birds' perception and do not know their feeding constraints (Robinson and Holmes 1982, Heinrich and Collins 1983, Sherry 1984. What is present in the field may not be what is actually available to the bird. We do not know which prey items a bird ignores because of the prey's inaccessibility (Kantak 1979, Avery andKrebs 1984), difficulty of capture (Hespenheide 1973), mechanical defenses (Sherry and McDade 1982) or chemical defenses (Eisner 1970, Janzen 1980. I had no previous data on food habits of birds in red maple forested wetlands and contiguous uplands. I felt collecting litter down to the root mass at each sample point was a comprehensive way of sampling all potential food items which bird~ feeding on the surface of the litter and in the litter would encounter. Many types of arthropods have clumped distributions which can greatly inflate variance estimates (Southwood 1966, Cooper andWhitmore 1990). I suspect many of the invertebrate taxa I collected were patchily distributed because I found high variances among samples.
This made it difficult to detect differences in abundances of invertebrates between zones and prevented me from pooling sites.
More detailed work with species identification of invertebrates needs to be done to help understand differences in abundance among -47-moisture zones in red maple forested wetlands and contiguous uplands.
I identified most individuals of the invertebrate community only to order or family level because of time constraints.
Changes in biomass over time or space may be masked when data for invertebrates are pooled at these ordinal levels.
Species often have different life cycles and are associated with different habitats (Borror et al. 1989).
Therefore, changes in abundance in species are not necessarily representative of those in another (Hutto 1985).
Invertebrate communities and abundances along these upland/wetland gradients would have to be sampled for several years in order to be certain that food is a limited resource during the breeding season at our sites because invertebrate types and abundance may change from year to year (Stenger 1958     aZones were based on soil moisture and distance from the wetland-upland edge. Zone A (furthest upland zone from wetland-upland edge), Zone B (closest upland zone to upland-wetland edge), Zone C (closest wetland zone to upland-wetland edge), Zone D (furthest wetland zone from wetland-upland edge).