Northern Prairie Wildlife Research Center
We also conducted a study of sparrow breeding habatiat at San Diego bay in 1995. San Diego Bay lost >92% of its salt marsh habitat (Macdonald 1990) between 1856 and 1984. The remaining 94 ha of salt marsh consist of isolated fragments surrounded by highly modified upland plant communities and urban landscapes. Remnants of the original San Diego Bay salt marshes were studied within the Sweetwater Marsh National Wildlife Refuge (32°36´N, 117°07´W), which includes F Street Marsh and a small remnant of salt marsh on the western edge of D Street Fill (Fig. 1). All three marshes were in the southern portion of San Diego Bay and were chosen because they had similar tidal flows but different degrees of distribution and isolation.
|Fig. 1. Map of Sweetwater Marsh National Wildlife Refuge wetland complex, San Diego Bay, California. Study plots not shown to scale.|
We established four study plots in the Refuge, two within the mid-littoral zone of salt marsh habitat (SM and Vener Pond) and two within the upper-littoral zone (SH1 and SH2). Sweetwater Marsh (50.6 ha of salt marsh) was separated from Vener Pond (6.84 ha) by a levee road 10 m wide. We also established plots at two isolated marshes (F Street and D Street), both within the mid-littoral zone of salt marsh habitat (Fig. 1). F Street Marsh (2.85 ha) was separated from Sweetwater Marsh by 0.5 km, and D Street Marsh (1 ha) was separated by a distance of 0.1 km. The landscape between F Street Marsh and Sweetwater Marsh is industrial/urban, whereas D Street Marsh was separated by natural tidal channels and salt marsh (Fig. 1).
The study plots at Sweetwater Marsh, Vener Pond, and D Street Marsh measured 100 x 100 m (1 ha). Owing to impassable channels and size constraints, the plot at F Street Marsh measured only 90 x 100 m (0.9 ha). All of the plots were marked at 25-m intervals. We visited all plots once a week from 22 March until 1 September 1995. Sparrow surveys began within one-half hour of dawn and continued until about 1000, when birds typically left their territories to forage in the mid- and lower-littoral zones of the marsh. A total of 209 hours was spent on the six plots in 1995. During each visit, we mapped territory boundaries using a combination of territory-flush and spot-mapping techniques (Wiens 1969). Territory boundaries were recorded on grid maps using landscape features and plot markers as reference points. Boundary disputes, countersinging, and pair-bond associations were recorded. On average, five singing locations were mapped for each territorial male (Wiens 1969). The outermost points of each territory were taken from the grid maps and transformed into X and Y coordinates. These coordinates were entered into program CALHOME to calculate the area of each territory using the minimum convex polygon method (Kie et al. 1996).
We used mist nets to capture sparrows prior to nesting between March 1994 and April 1995. We individually color-banded 277 adult sparrows to determine dispersal among habitat patches and other marshes in the region within seasons and between years. Sex determination of adults was based on presence of a brood patch or cloacal protuberance (Pyle et al. 1987).
We chose a relatively noninvasive method for determining nesting success because of the endangered status of Belding's Savannah Sparrows, and because nests were difficult to find without risking trampling and abandonment. We estimated nesting success using a reproductive index developed for rare sparrows by Vickery et al. (1992). Territories were ranked on a scale from 1 to 7 using behavioral cues to determine stages in the breeding cycle (see Vickery et al. 1992). If a bird was seen with food we considered it direct evidence that a pair had nestlings. Observations of feeding and begging behavior between adults and fledglings determined fledging success. Territories with "high success" (ranks 5 to 7) fledged at least one offspring, whereas territories with "low success" (ranks 1 to 4) fledged no offspring. Territories that produced fledglings from more than one brood within the nesting season received the highest rank of 7. If a nest was found incidentally, we banded the nestlings and recorded the type of vegetation used to build the nest, the direction of the nest opening, and degree of nest concealment (Martin and Geupel 1992). In addition, we measured the vegetation around the nest using a 1-m² quadrat, with the nest placed in the center of the sampling unit.
We sampled vegetation within each plot along 10 100-m transects (except F Street Marsh, n = 9 transects) that were spaced at 5-m intervals along one side of each plot. This ensured the entire length of each plot was sampled. Five randomly placed circular quadrats, each measuring 1 m², were then sampled along each transect, for a total of 591 samples, or 10% of the study area. We measured the following variables within each quadrat: (1) percent cover by species; (2) average vegetation height across the quadrat; (3) maximum height of vegetation; and (4) percent cover of bare ground, channel, and litter (both organic and inorganic). We established six cover classes (< 1%, 1-5%, 6-25%, 26-50%, 51-75%, 76-100%) to rank percent cover, and used the midpoints of these intervals in statistical analyses (Pacific Estuarine Research Laboratory 1990). Maximum height was the highest point of herbaceous vegetation recorded in a quadrat, and mean height was determined by measuring three random plants in each quarter of the quadrat. The coefficient of variation (CV) of plant height was calculated from mean plant heights.
Territories with ≥50% of their area outside of a study plot were excluded in the statistical analyses. We used ANOVA to assess differences in reproduction among study plots and habitat types. Data for high-marsh (SH1 and SH2) and mid-marsh (SM1 and Vener Pond) plots were similar biologically and were combined for this analysis. Study plots were categorized as "large, connected" marsh (SH1, SH2, SM, Vener Pond), "small, connected" marsh (D Street), or "small, isolated" marsh (F Street). Comparisons among groups were made using Sheffé's method (Sokal and Rohlf 1981). We used stepwise discriminant function analysis (DFA) to explore how well the habitat variables distinguished high- from low-success territories, and sparrow territories from non-territories (Wray and Whitmore 1979). We grouped percent cover by plant species into high-marsh (Cordylanthus maritimus, Cressa truxillensis, Distichlis spicata, Juncus acutus, Lasthenia glabrata, Limonium californicum, Monanthochloe littoralis, Salicornia subterminalis), mid-marsh (Cuscuta salina, Frankenia salina, Jaumea carnosa, Salicornia virginica, Suaeda esteroa, Triglochin concinnum), and low-marsh (Batis maritima, Salicornia bigelovii, Spartina foliosa) species. Edge (Amsinckia spectabilis, Baccharis pilularis, Lycium californicum, Mesembryanthemum spp., Sonchus asper, Spergularia spp.), exotic (Bromus diandrus, Bromus mollis, Bromus rubens, Chrysanthemum spp., Hemizonia pungens, Parapholis incurva, Raphanus sativus), and dune (Camissonia cheiranthifolia, Heliotropium curvassavicum, Melilotus indica) species also were combined. Thus, the 14 variables included total percent plant cover; cover by high-marsh, mid-marsh, low-marsh, edge, exotic and dune species; mean, maximum, and CV of plant height; number of plant species; and percent cover by bare ground, litter, and tidal channels.