Predicting population abundance and age structure: testing theory with field data

Abstract

Relatively simple demographic models can be used to predict abundance and age-structure of populations whose spatial boundaries can be delimited, especially fi there is little dispersal of individuals beyond those boundaries. Because these conditions are generally not fulfilled by marine populations, researchers rarely attempt to predict abundance or structure of these populations. In this study, these predictions were made for the bivalve Gemma gemma using finite population growth rates and limiting age distributions associated with Leslie matrices. The field population is enclosed in small bay in Rhode Island, USA. Dispersal to and from the bay is low because the clam's life cycle does not include planktonic stages. The data set analysed consisted of a time series of clam abundance from 1978 to 1983, as well as age-specific survivorship and fecundity for each cohort present. Annual forecasts for June 1979, 1980 and 1981 deviated from observed values by only 12 to 17%. The June 1982 forecast predicted a major population crash which did occur 1 mo later. By multiplying successive matrices, annual forecasts were made up of 5 yr ahead. In 4 of 5 years predictions fell within 95% confidence intervals for mean observed abundance. Empirical age distributions fit expected distributions, derived from the theory of imprimitive or periodic matrices. This study's success at predicting major properties of the field population from the Leslie model suggests that (1) assumptions of the model were satisfied, (2) the population was in equilibrium state with the environment each year, and (3) the model was realistic enough to describe dynamics of this natural population