Although metapopulation theory is widely used in basic and applied ecology,
there are still few empirical studies that explore the relationships betwe
en dispersal, patch size, and the persistence of natural populations. Here,
we reanalyzed data from a six-year study of a spatially structured field v
ole (Microtus agrestis) metapopulation in the Tvarminne archipelago, Finlan
d. Our goal was to address several issues relevant to metapopulation studie
s: (1) the relationships between within-subpopulation dynamics, dispersal b
ehavior, habitat quality, and metapopulation dynamics; (2) the generality o
f one of the most common conclusions of metapopulation theory-that smaller
and less frequently inhabited islands are less important for metapopulation
dynamics; and (3) the comparison of different methods for understanding an
d predicting dynamics in "metapopulation-like" systems. Our results suggest
that this vole metapopulation is driven by extinctions and colonizations o
f island subpopulations. However, contrary to expectation, colonizations by
voles from tiny, ephemeral skerry subpopulations were about as important f
or metapopulation persistence as were colonizations from the more persisten
t subpopulations on large islands. This pattern resulted from less stable v
ole densities on smaller islands, combined with increased emigration preced
ing subpopulation extinctions. Either spatially implicit Levins models or i
ncidence function models provided reasonable predictions of the structure a
nd function of this metapopulation, but parameters fitted to incidence func
tions varied dramatically among years. Our results suggest that models of r
eal metapopulations need not become highly detail oriented or spatially com
plicated to provide good predictive power. However, applications of metapop
ulation theory require careful consideration of how underlying ecological a
nd behavioral mechanisms will shape metapopulation dynamics of particular s
pecies and situations.