Relationship between habitat area and the distribution of tidal marsh birds

—To assess the relationship between marsh area and relative abundance of tidal marsh bird species, we surveyed birds on 86 circular plots in 40 salt and brackish tidal marshes in Connecticut. We measured marsh area in two ways: the amount of contiguous marsh vegetation not interrupted by broad barriers (.500 m of open water or .50 m of upland habitat) and by narrow barriers (.30 m of open water or .10 m upland). We determined the relationship between marsh area and the relative abundance of particular species (mean number of individuals per survey plot) with linear or logistic regression. When the broad barrier definition was used, we found that all three species of short grass meadow specialists, Willets (Catoptrophorus semipalmatus), Seaside Sparrows (Ammodramus maritimus), and Saltmarsh Sharp-tailed Sparrows (A. caudacutus), were less abundant or absent in survey plots in smaller marshes. The Seaside Sparrow and Willet also showed a significant tendency to be less frequent in smaller marshes when the narrow barrier definition was used. In contrast, species that used a wider range of wetland types, as in the Virginia Rail (Rallus limicola), Marsh Wren (Cistothorus palustris), and Swamp Sparrow (Melospiza georgiana), were equally frequent on plots in marshes of different areas. Our results are consistent with the hypothesis that fragmentation of marsh systems with artificial habitat causes a decline in the density of short grass meadow specialists in the remaining patches of appropriate habitat. Received 25 July 2001, accepted 20 September 2002. Connecticut lost about 30% of its tidal wetlands between the 1880s and the 1970s (Rozsa 1995), and most of the remaining marshes have been heavily modified by ditching, tidal restriction, and the spread of common reed (Phragmites australis). These habitat changes are associated with population declines in salt marsh birds (Brawley et al. 1998, Benoit and Askins 1999, Clarke et al. 1984, Craig 1990), but the role of habitat fragmentation in these declines remains an open question. Species that are sensitive to the negative effects of habitat fragmentation would decline not only in areas where habitat has been altered, but also in remaining small patches of apparently suitable habitat. Habitat fragmentation is associated with changes in the composition of bird communities in a wide range of habitats, including deciduous forests in Japan and eastern North America (Roberts and Norment 1999, Askins 2000, Askins et al. 2000), shrubsteppe in Idaho (Knick and Rotenberry 1995), temperate rain forests in Chile (Willson et al. 1994), and tropical rain forests in Brazil (Laurance et al. 1 Dept. of Zoology, Connecticut College, New London, CT 06320, USA. 2 Current Address: Dept. of Environmental Protection, Office of Long Island Sound Programs, 79 Elm St., Hartford, CT 06106, USA. 3 Corresponding author; E-mail: [email protected] 2002). Some species (usually habitat specialists) in each of these habitats are area sensitive, with a tendency to decline or disappear in small remnant patches of apparently suitable habitat. However, area sensitivity has not been demonstrated conclusively in North American marsh birds despite the fact that Brown and Dinsmore (1986) and Craig and Beal (1992) showed that there was a positive relationship between the number of species of birds and marsh area, and that some species were missing from smaller marshes. The results of both of these studies were inconclusive because more time was spent surveying birds in large marshes than in small marshes. Consequently, more species may have been detected in larger marshes because of the passive sampling effect (Connor and McCoy 1979, Horn et al. 2000). Because there was less surveying effort in smaller marshes, fewer individuals would be detected, increasing the chance that some species would be missed even if none of the species were area sensitive. Moreover, neither study showed that the density of particular species of marsh birds was lower in smaller marshes than in larger marshes. In both forests (Robbins et al. 1989, Askins et al. 1990) and grasslands (Vickery et al. 1994, Johnson and Igl 2001), area sensitive species tend to have lower densities in small patches of habitat than in large blocks of continuous habitat. This may be due to negative 315 Benoit and Askins • DISTRIBUTION OF TIDAL MARSH BIRDS edge effects such as higher predation rates in smaller patches (Johnson and Temple 1990, Faaborg et al. 1995). We especially focused on two species of sparrows that are salt marsh specialists, the Seaside Sparrow (Ammodramus maritimus) and Saltmarsh Sharp-tailed Sparrow (A. caudacutus), because they are taxonomically and ecologically similar to area sensitive sparrows of dry grasslands. Moreover, in New England the two salt marsh sparrows are largely restricted to short grass meadows, salt and brackish tidal wetlands dominated by low grasses such as Spartina patens, Distichlis spicata, and Juncus gerardi (Greenlaw and Rising 1994, Post and Greenlaw 1994, Benoit and Askins 1999). In many respects, these habitats are structurally similar to upland grasslands. Studies in dry grasslands such as prairie preserves in Illinois (Herkert 1994a), blueberry barrens in Maine (Vickery et al. 1994), fields in western New York (Norment et al. 1999), and restored grasslands in the northern Great Plains (Johnson and Igl 2001) showed that Vesper Sparrows (Pooecetes gramineus), Savannah Sparrows (Passerculus sandwichensis), Grasshopper Sparrows (Ammodramus savannarum), Baird’s Sparrows (A. bairdii), and Henslow’s Sparrows (A. henslowii) were more likely to be detected on standard plots in large grasslands than in plots of the same size in small grasslands. Moreover, these species tend to be missing in survey plots located in the smallest grasslands. Consequently, an important concern in managing or restoring grasslands is to provide large enough areas of contiguous habitat to support populations of these sparrows. Similarly, if the salt marsh sparrows are area sensitive, then it will not be sufficient to consider the total amount of suitable habitat needed to support populations; it also will be important to maintain or create large blocks of uninterrupted short grass meadow. Our goal was to determine whether specialized marsh birds are area sensitive. If they are, then we would expect them to display either of the following patterns: (1) a lower density in smaller marshes, or (2) a tendency to be absent from survey plots in marshes smaller than some minimum area. We completed surveys in a large number of tidal marshes to test these predictions. METHODS Survey plots.—During the summers of 1995 and 1996, we surveyed birds and vegetation on 86 standardized circular plots in 40 brackish and salt marshes along the coast and tidal rivers of Connecticut (see Benoit and Askins 1999 for locations and descriptions of these sites, including the number of survey plots per site). We surveyed 20 marshes during each of the two years. These encompassed nearly all salt and brackish marshes .10 ha in the state as well as some marshes ,10 ha. The 50-m radius plots were located $200 m apart and $75 m from upland habitats. We recorded all birds detected during an observation period during each of two visits, one in June and the other ($2 weeks later) in July. We commenced the study in early June because Seaside and Saltmarsh Sharp-tailed sparrows are still migrating during late May (Saunders 1959). We conducted observations between 05:00 and 10:00 EST, and we surveyed #4 plots per day. The observation period consisted of 10 min of passive observation followed by 7 min of broadcasting, in sequence, the taped calls of the following species: Least Bittern (Ixobrychus exilis), American Bittern (Botaurus lentiginosus), Virginia Rail (Rallus limicola), King Rail (R. elegans), Clapper Rail (R. longirostris), Sora (Porzana carolina), and Black Rail (Laterallus jamaicensis). Playback was not necessary for highly detectable birds such as sparrows, Willets (Catoptrophorus semipalmatus), and Marsh Wrens (Cistothorus palustris). We quantified the relative abundance of each species as the total number of individuals seen or heard during the initial 10 min plus any additional birds that responded to conspecific calls during the playback period. We counted only the adults of each species. Individuals of the same species had to be detected simultaneously to be recorded as different individuals. These survey methods were appropriate for sampling bird distribution across a regional landscape and were not intended to characterize particular marshes. We chose survey plots by stratified random design. We mapped major vegetation types using aerial photographs supplemented with field checking. We initially classified vegetation into three categories (Table 1): (1) short grass meadow (areas dominated by low marsh grasses such as Spartina patens, Juncus gerardi, and Distichlis spicata), (2) cattail (areas dominated by Typha spp.), and (3) Phragmites (areas dominated by Phragmites australis). We used a table of random numbers to select coordinates of survey plots in each sufficiently extensive vegetation type on a grid superimposed on a map of each site. Each marsh had 1–5 survey plots, depending upon its size. We used the line intercept method (Brower and Zar 1977) to estimate percent cover of different species of plants on each plot. Two 50-m perpendicular transects were laid out from the center of each plot. One of the transects was oriented toward the nearest tidal creek. We calculated percent cover from the total distance that the line intercepted the foliage of each plant species. Based on the dominant vegetation indicated by 316 THE WILSON BULLETIN • Vol. 114, No. 3, September 2002 TABLE 1. Mean percent cover for different plant species and water features for six vegetation categories in 40 tidal marshes on the coast of Connecticut, 1995–1996. Vegetation categories Short grass meadow Phragmites Cattail Brackish mixture Short S. alterniflora Other No. of survey plots 36 14 7 14 6 9