Density-dependent Reproductive Success in Dabbling Ducks: Using a 20-year Dataset to Identify Patterns Josh T. Ackerman, United States Geological Survey, Western Ecological Research Center, Davis Field Station, One Shields Ave., University of California, Davis, CA 95616, Email:
[email protected] John M. Eadie, Wildlife, Fish, and Conservation Biology, One Shields Ave., University of California, Davis, CA 95616, Email:
[email protected] Greg Yarris, California Waterfowl Association, 4630 Northgate Blvd., Suite 150, Sacramento, CA 95834, Email:
[email protected] Dan Loughman, California Waterfowl Association, 4630 Northgate Blvd., Suite 150, Sacramento, CA 95834, Email:
[email protected] Bob McLandress, California Waterfowl Association, 4630 Northgate Blvd., Suite 150, Sacramento, CA 95834, Email:
[email protected] Julie Yee, United States Geological Survey, Western Ecological Research Center, Modoc Hall, 3020 State University Drive East, Sacramento, CA 95819, Email:
[email protected] Extended Abstract: Although much research has focused on testing density dependence in waterfowl populations, it remains unclear whether reproductive success is generally densitydependent or independent. A complicating factor is that the scale of study will often influence the results, and interpretation can be different among ecological scales. Density dependence has been detected in several studies on duck population dynamics at the continental scale, but results at local scales are equivocal. Furthermore, nest predation often is the focus of density-dependent studies at local scales, yet other reproductive parameters, such as clutch size, may also be influenced by density.
may not be density-dependent at these smaller scales, it is still possible that nest predation may depend on the overall abundance of duck nests in any given year.
During 1985 to 2003, we monitored the natural variation in nest success and nest densities within a large block of uplands at the Grizzly Island Wildlife Area in Suisun Marsh. In total, we monitored 11,856 duck nests (9,178 mallards). We calculated Mayfield nest success for each year; total duck nest success was calculated using an average clutch age at hatching of 35 days (mainly mallard and gadwall, with small numbers of cinnamon teal and northern pintail). We used Mayfield nest-success rates to estimate nesting densities (Miller and Johnson 1978). Briefly, we Herein, we use our 20-year dataset on nesting divided the number of successful nests (≥1 egg ducks in Suisun Marsh, California to examine density-dependent reproduction among years. We hatched) by Mayfield nest success to estimate the consider this to be the largest scale of analysis for total number of nests initiated each year. We then nesting ducks, in contrast to the majority of studies divided the number of nests initiated by the total area (ha) searched each year to determine the which have assessed density-dependent nest success among study sites or small habitat patches. annual density of initiated nests. We logePreviously, we assessed whether nest predation transformed Mayfield nest success, nest densities, was density-dependent using both natural and and clutch size. artificial nests at three spatial scales, and we found little evidence for density-dependent nest Annual nest success was negatively correlated predation among fields 5-33 ha in size, plots 1 ha with nest densities for mallards (N=19 years, in size, or among nearest neighboring nests R2=0.24, P=0.0317) and all dabbling ducks (Ackerman et al. 2004). Although nest predation combined (N=19 years, R2=0.32, P=0.0123), and
there was no evidence of autocorrelation among years using a first-order autoregressive time series process. Mallard nest success
0.8 0.6 0.4 0.2 0.0 1 10 Mallard nest density per ha
Mallard clutch size also was negatively correlated with mallard nest density (N=19 years, R2=0.30, F1,1=4.78, P=0.04), while accounting for median nest initiation date (F1,1=1.86, P=0.19). Mallard clutch size
Duck nest success
0.8
The adjusted P-values for the negative correlations between Mayfield nest success and Mayfield nest density in Figure 1 was P=0.080 for mallards and P=0.048 for all ducks. These results confirm the pattern of negative density-dependence, but the statistical significance was reduced relative to our initial results.
0.6 0.4 0.2 0.0 1 10 Duck nest density per ha
Figure 1. Annual Mayfield nest success was negatively related to Mayfield nest densities for mallards (top panel) and all dabbling ducks combined (bottom panel) from 1985-2003 in Suisun Marsh, CA.
However, there may be an inherent negative correlation between these variables because Mayfield nest density (X variable) is calculated by dividing the density of successful nests by Mayfield nest success (Y variable).
9.2 8.8 8.4 8.0 1 10 Mallard nest density per ha
Figure 2. Annual mean mallard clutch size was negatively related to Mayfield nest densities from 19852003 in Suisun Marsh, CA.
Mallard egg success was not correlated with mallard nest densities (F1,1=2.00, P=0.18) after statistically accounting for the positive relationship between egg success and nest success using partial regression (F1,1=14.21, P=0.002). Our results indicate that nest success and clutch size may be density-dependent among years, in contrast to our prior findings of no density dependence at three smaller spatial and temporal scales. Further, tests for density dependence in breeding waterfowl should account for the inherent correlation between estimates of nest success and nest density.
We examined this possible bias using simulated data from independently selected values of Mayfield nest success and Mayfield nest density based on the distribution of these values in the long-term dataset. As expected, simulated Mayfield nest success was negatively correlated with simulated Mayfield nest density at an inflated Literature Cited: rate (11.7% by chance rather than the prescribed Ackerman, J. T., A. L. Blackmer, and J. M. Eadie. 5% type I error); indicating that there is an 2004. Is predation on waterfowl nests density inherent bias towards finding a negative dependent? Tests at three spatial scales. Oikos relationship between Mayfield nest success and 107:128-140. Mayfield nest density by chance alone. Therefore, we used simulation analysis to correct for this bias Miller, H. W. and Johnson, D. H. 1978. Interpreting the results of nesting studies. by calculating an adjusted P-value using 10,000 Journal of Wildlife Management 42:471-476. simulations.