Ecological modeling in urban environments: predicting changes in biodiversity in response to future urban development - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 349

FIGURE 25.4 Predicted total bird diversity for 2003 and 2027 and the difference in species richness with time (2027-2003).

However, many limitations still exist. In the example provided,
there was no attempt to model how the error in the initial
land cover classifications may have influenced the results. Such
errors would have propagated through: (1) the LCCM model
specification, leading to erroneous associations between land
cover change and predictor variables and ultimately erroneous
predictions of future land cover; (2) erroneous avian biodiversity
models since they were derived from the land cover data; and (3)
erroneous predictions of future biodiversity change.
Additionally, each model had its own set of assumptions. For
example, the LCCM assumes that observed change will continue
into the future; specifically that current drivers of land cover
change will be the same and of the same magnitude into the future.
Because of these changes, future land cover change (rates and
types of change) are likely to be different than what was observed
in the past. Other studies have demonstrated how changes in
data used to specify the LCCM and application to different areas
have lead to different future predictions (Hepinstall, Alberti

and Marzluff, 2008). Avian community models represent only
one aspect of the biodiversity of the area and only one way
of modeling avian diversity. Looking at individual species will
likely yield different results with more complicated patterns
(Hepinstall, Marzluff and Alberti in press).
We are now in an era where increased remotely sensed data,
other geospatial data, computer power, decreased storage costs,
and improved training at our undergraduate institutions will
combine to provide many opportunities to model the past,
present, and future effects of urbanization on ecological systems.

Acknowledgments

I thank Marina Alberti and the Urban Ecology Research Labora-
tory staff, and Paul Waddell and the Center for Urban Simulation
and Policy Analysis staff, and John Marzluff. I thank Krista Merry