species-specific edge effects on nest success and breeding bird

0 downloads 0 Views 462KB Size Report
BREEDING BIRD DENSITY IN A FORESTED LANDSCAPE. DAVIDJ. FLASPOHLER ... creases nest density for some species of migratory birds in a zone adjacent to the opening. This pattern supports ...... The Audubon society encyclopedia of.
Ecological Applications, 1 l(1). 2001, pp. 32-46 O 2001 by the Ecological Society of A ~ n e r ~ c a

SPECIES-SPECIFIC EDGE EFFECTS ON NEST SUCCESS AND BREEDING BIRD DENSITY IN A FORESTED LANDSCAPE DAVIDJ. FLASPOHLER,'~~ STANLEYA. TEMPLE,'AND ROBERTN. ROSENFIELD? 'Department of Wildlife Ecology, University of Wisconsin, Madison, Wisconsin 53706 USA ?Department of Biology, Universih of Wisconsin, Stevens Point, Wisconsin 54481 USA

Abstract. Using natural nests of eight bird species, we provide one of the first multispecies tests for edge effects on reproductive success in a forested landscape. Our primary objective was to assess whether distance to the edge of recent clearcuts was related to nesting success in intact northern hardwood forests. Estimated nest success was generally lower for the two ground-nesting species than for the six canopy-nesting species. Brood parasitism was 0.05).

Ecological Applications Vol. I I , No. I

DAVID J. FLASPOHLER ET AL

TABLE3 . Plot-specific nest success for pooled ground nests (Hermit Thrush and Ovenbird) and canopy nests (Least Flycatcher, Black-throated Green Warbler, Red-eyed Vireo, American Robin, Rose-breasted Grosbeak, and Scarlet Tanager) ranked in descending order based on nest success. Category and plot

Nests (no.)

Exposure (d)

Failurest (no.)

Nest success!: (mean i S E )

Ground nesters Kaine Lake Huff Creek Popple River Camp 2 Holt Road Bear Anvil Lake Brule River Pine Sap Canopy nesters Huff Creek Pine Sap Anvil Lake Popple River Kaine Lake Brule River Holt Road Camp 2 Bear Note: All three years of nest data are included.

f Only predation-related failures are included.

Nest success calculations for ground nesters and canopy nesters were done using an average

weighted by exposure days. Rows with same superscript letter are not significantly different ( P > 0.05) based on chi-square analysis using multiple comparisons. Comparisons are made within each nest category (i.e., ground nesters and canopy nesters).

:

For comparisons of nest success at different distances from the edge, we used cu = 0.05. We ranked the nine plots in order of plot-level nest success for each species. We then compared patterns of plot-level nest success for ground nesters and canopy nesters. We also tested for species-specific year and plot effects using logistic regression models. For all logistic regression analyses, we used simple nest fate (success = +. failure = -) in place of Mayfield estimates. Statistical analyses were conducted using SYSTAT (Wilkinson 1997). For calculation of nest success at various distances from the forest edge, each plot was partitioned into distance categories with similar numbers of nests in each category, based on Gates and Gysel (1978). We used two methods to examine the relationship between nest success and proximity to edge. First, we blocked nests and compared nest success in two zones (0-300 m vs. 301-950 m from the forest edge). These two zones contained roughly equal numbers of nests for most species. Using nest success values in the two zones, we used contingency tables to test the hypothesis that nest success was different in the two zones. For species that showed significant differences between nest success in the two zones, and for which we had sufficient sample sizes ( 2 2 0 nests). we then blocked nests into smaller (100 m) zones and tested multiple hypotheses related to effects at 100 m intervals from the forest edge.

Second, we used logistic regression to examine the influence of distance to forest edge on nest fate. This technique does not require that data from the independent variable be normally distributed (Hosmer and Lemeshow 1989). Logistic regression also allowed us to fit a logit model to data from ground-nesting species which appeared to have a nonlinear or threshold distribution based on the pattern generated by blocking nest success values by distance from edge. We generated second-order variables using a log transformation of distance. In all logistic regression models, we tested five alternative models containing a distance to edge parameter against the model containing only a constant. We included year and plot as categorical variables and as a second measure of year and/or plot effects. We used the inflection point of the logit model for each species to test for threshold changes in nest success at different distances from the forest edge. When this modeling exercise generated a significant model containing an edge parameter, we performed the identical logit modeling exercise on the outcome of the incubation and nestling phases to explore the effect of proximity to edge at different stages of nesting (i.e., prehatching and posthatching). In the first set of models, the outcome was the logit transformation of the probability of a nest being depredated during the full nesting period. In the second set of models, the outcome was the logit transformation of the probability of

MULTIPLE EDGE EFFECTS ON A BIRD COMMUNITY

February 2001

TABLE4. Nest success (mean + 1 SE) in two zones (0-300 m and 301-950 m from a forest edge) for each year and for the entire study period.

Species and year Hermit Thrush

1995 1996 1997 All years Ovenbird

1995 1996 1997 All years Least Flycatcher

1995 1996 1997 All years Black-throated Green Warbler 1995 1996 1997 All years Red-eyed Vireo?

All years Rose-breasted Grosbeak?

All years American Robin?

All years Scarlet Tanagert

All years

Nest success at indicated distances from forest edge (no. of nests)

0-300 m

301-950 m

0.23 i 0.14 (10)

0.29 + 0.1 1 (19)

0.31 i 0.18 (7)

0.28 i 0.08 (36)

0.58 i 0.13 (19)

0.34 i 0.15 (13)

0.37 + 0.12 (10) 0.44 + 0.08 (42)

0.87 + 0.12 (9) 0.89 + 0.06 (24) 0.81 + 0.10 (8)

0.86 i 0.05 (41)

1.0 + 0.0 (2) 1.0 + 0.0 (10)

0.68 i 0.15 (10)

0.78 + 0.1 1 (22)

0.56

+ 0.12 (19)

0.69

+ 0.11 (19)

0.37 i 0.12 (21)

0.84 i 0.11 (18)

Notes: Mayfield (1961, 1975) nest success values calculated from exposure.~Apower analysis (for 20% change in nest success at a = 0.05; Motulsky 1995) for the six nonsignificantly different canopy-nesting species yielded the following estimates of power (1 -P): Least Flycatcher, 65.5%; Black-throated Green Warbler, 19.8%;Red-eyed Vireo, 15.9%; Rose-breasted Grosbeak, 18.4%; American Robin, 5.5%; and Scarlet Tanager, 11.5%. For data from all years pooled, only the Ovenbird had significantly different ( P < 0.05) estimates of nest success in the two zones relative to the edge. f Sample sizes of nests were insufficient to allow us to estimate nest success for single years.

the nest being depredated during the prehatching or posthatching stages. In all logistic regression models, we used Akaike's Information Criterion (AIC) (Akaike 1973, 1985) to compare models based on log-likelihood values (Burnham and Anderson 1992). This method uses the Principle of Parsimony (Goodman 1984, McCullagh and Nelder 1989) to balance the competing tendencies of model underfitting and overfitting. We calculated nest-density indices based on the number of nests found at different distances from the forest edge. Nests from all three years were grouped into 50m zones relative to distance from the forest edge, and the number of nests in each zone was expressed as nestslhectare. We also expressed nest density as the annual and cumulative (1995-1997) number of nests found in edge (