Df. Doak et W. Morris, Detecting population-level consequences of ongoing environmental change without long-term monitoring, ECOLOGY, 80(5), 1999, pp. 1537-1551
The frequent lack of correspondence between measured population stage struc
tures and those predicted from demographic models has usually been seen as
an embarrassment, resulting from poor data, or a testimony to the failings
of overly simplistic models. However, such mismatches can also arise due to
natural or anthropogenic changes in the: environment, thus providing the d
ata needed to test hypotheses about the ecological effects of local or glob
al environmental change. Here, we present a method that allows this type of
comparison to rigorously test for the population-level effects of past and
ongoing environmental change in situations where no long-term monitoring d
ata exist. Our approach hinges on the fact that changing environmental cond
itions will cause population size structure to lag behind that predicted by
current demographic rates. We first develop the methods needed to calculat
e the likelihood of an observed population structure, given different stoch
astic models of demography responding to environmental changes. We next use
simulated data to explore the method's power in the face of estimation err
ors in current vital rates, environmental noise, and other complications. W
e conclude that this method holds promise when applied to slowly growing, l
ong-lived species and when model structures are used that allow for realist
ic time lags in population structure. Researchers using this approach shoul
d also be careful to assess the importance of other phenomena (rare catastr
ophes, recent founding of populations, genetic changes, and density depende
nce) that may compromise the method's accuracy. Although large data sets ar
e required for the method to be accurate and powerful, the data required wi
ll be readily obtainable for abundant and easily sampled species. While mos
t ecological efforts to detect global environmental changes have focused on
long-term monitoring, or indicators such as tree rings that are unique to
organisms not present in all biomes, this method allows tests for past and
ongoing changes in situations where neither past monitoring data nor unique
indicator species are available.