You are not currently logged in.
Access JSTOR through your library or other institution:
Simple Assumptions on Age Composition Lead to Erroneous Conclusions on the Nature of Density Dependence in Age-Structured Populations
Richard W. Zabel and Phillip S. Levin
Vol. 133, No. 3 (Nov., 2002), pp. 349-355
Stable URL: http://www.jstor.org/stable/4223427
Page Count: 7
Preview not available
Ecologists have debated the nature of density dependence in natural populations for decades, and efforts to detect density dependence from time series of abundance data have paralleled these debates. Yet due to the correlative nature of time series data, these undertakings have been statistically problematic. Most analyses of density dependence have focused on simple population models (i.e., non-overlapping generations), but in reality most vertebrates exhibit more complex life histories, and this complexity has been incorporated into population models in a variety of ways. Unfortunately, adding complexity to population models can further exacerbate efforts to detect density dependence. We examined the effect of adding age structure when inadequate data exist in support; to demonstrate this effect, we adopted Pacific salmon (Oncorhynchus spp.) as our study organism. Most salmon populations are semelparous and have variable age at maturity. Salmon populations (and many other fish species populations) are typically modeled in terms of numbers of recruits arising from spawners in a given brood year. Recruits are enumerated as they return as adults to spawn, and proper assignment of recruits to brood year requires age information. Unfortunately, while adult counts are common, detailed age information is not. A common practice is to apply long-term averages of age composition to returning adults to "reconstruct" time series of recruits. Here, by conducting simulations and analyzing data from natural populations, we demonstrated that this practice leads to a biased portrayal of density dependence by overestimating recruits from small spawning classes and underestimating recruits from large spawning classes. Also, productivity was overestimated and variance was underestimated, which could lead to overly optimistic predictions of extinction risk or overharvesting.
Oecologia © 2002 Springer