Data to replicate: Forecasting community reassembly using climate-linked spatio-temporal ecosystem models

Ecosystems are increasingly impacted by human activities, altering
linkages among physical and biological components. Spatial community
reassembly occurs when these human impacts modify the spatial overlap
between system components, and there is need for practical tools to
forecast spatial community reassembly at landscape scales using monitoring
data. To illustrate a new approach, we extend a generalization of
empirical orthogonal function (EOF) analysis, which involves a
spatio-temporal ecosystem model that approximates coupled physical,
biological, and human dynamics. We then demonstrate its application to
five trophic levels for the eastern Bering Sea by fitting to multiple,
spatially unbalanced datasets measuring physical characteristics
(temperature measurements and climate-linked forecasts), primary producers
(spring and fall size-fractionated chlorophyll-a), secondary producers
(copepods), juveniles (age-0 walleye pollock), adult consumers (five
commercially important fishes), human activities (seasonal fishing
effort), and mobile predators (seabirds). We identify the spatial niche
for each ecosystem component, as well as dominant modes of variability
that are highly correlated with a known bottom-up driver of dynamics. We
then measure spatial overlap between interacting variables (using
Schoener’s-D) and identify that age-0 pollock have decreased spatial
overlap with copepods and increased overlap with adult pollock during warm
years, and also that adult pollock have increased overlap with arrowtooth
flounder and decreased overlap with catcher-processor fishing effort
during these warm years. Given the warming conditions that are projected
for the coming decade, the model forecasts increased prey and competitor
overlap involving adult pollock (between age-0 pollock, adult pollock and
arrowtooth flounder) and decreased overlap with the copepod forage base
and with the catcher-processor fishery during future warming. We recommend
that joint species distribution models be extended to incorporate
“ecological teleconnections” (correlations between distant locations
arising from known mechanisms) arising from behavioral adaptation by
mobile animals as well as passive advection of nutrients and planktonic
juvenile stages.