Environmental Engineering Reference
In-Depth Information
TABLE 25.1
Land cover (km
2
and percent of study area) observed (2002) and predicted (2003, 2027), change in percent of the study area
from 2002 to 2003 and 2027, and percent increase by 2027 from 2002 land cover.
Observed
Predicted
2002
2003
2027
km
2
%
2
km
2
% Change in %
% Change in %
% Increase
Heavy Urban
29.2
5.7
30.6
6.0
0.3
103.8
20.4
14.7
255.9
Medium Urban
66.0
13.0
51.0
10.0
-3.0
53.9
10.6
-2.4
-18.4
Light Urban
99.1
19.5
113.5
22.3
2.8
144.8
28.5
9.0
46.1
Grass
33.3
6.6
64.6
12.7
6.2
57.8
11.4
4.8
73.7
Agriculture
49.5
9.7
34.2
6.7
-3.0
0.1
0.0
-9.7
-99.9
Deciduous and Mixed Forest
118.3
23.3
123.1
24.2
0.9
67.2
13.2
-10.1
-43.2
Coniferous Forest
55.7
11.0
43.3
8.5
-2.4
23.1
4.6
−
6
.
4
-58.4
Regenerating Forest
37.3
7.3
28.7
5.7
-1.7
38.3
7.5
0.2
2.5
Other
19.7
3.9
19.1
3.8
19.1
3.8
applied to various spatial partitions of the Central Puget Sound
was reported in Hepinstall, Alberti and Marzluff (2008) and
was calculated by comparing predictions for 2003 with observed
land cover for 2002 (the closest available date of land cover
data); kappa values varied between 0.63 and 0.77. Land cover
can be predicted for as many time steps as desired, but research
by Pontius, Huffaker and Denman (2004) has shown that the
accuracy of these models likely converge on the level of accuracy
derived from random models after multiple time steps. Prior runs
of LCCM in the Central Puget Sound indicated the accuracy of
model output likely approached the accuracy of random models
after 8 - 10 time steps (Hepinstall-Cymerman, unpublished data).
heavy urban), agriculture (converting to urban or grass), and
forest (deciduous and mixed forest and coniferous forest both
converting to urban classes) (Table 25.1). The greatest increase
is predicted to be in heavy urban and this pattern is visible
in Fig. 25.3.
25.2.6
Predicted changes in avian
biodiversity for study area
The predicted changes in land cover obviously have great poten-
tial for changing the avian species able to inhabit this landscape.
Indeed, the total species richness is expected to decline across
94% of the study area (Table 25.2). The losses in total species
richness (Fig. 25.4) clearly follow the spatial pattern of conversion
of forests and agriculture to heavy urban (Fig. 25.3). For native
forest birds, already low in this area because of the dominance
of urban and agriculture, 59% of the area is predicted to harbor
fewer of these species. As expected by the area totals, the overall
pattern for native forest birds is approximately equal between
areas of loss where forest is predicted to convert to urban and
stable where agriculture is predicted to convert to urban since
native forest birds currently do not occupy these agricultural
lands (Fig. 25.5). Interestingly, some of the greatest predicted
losses of species are in large patches; patterns that would not
be easily discerned by looking only at the predicted land cover
maps. This results points to the importance of modeling spe-
cific biodiversity responses rather than relying on land use or
land cover changes alone to tell the story. Other published studies
have found similar results with differential effects being predicted
depending on what aspects of biodiversity were being considered
(Schumaker
et al
., 2004; Gude, Hansen and Jones, 2007)
25.2.4
Avian biodiversitymodel
In this example, avian biodiversity, as indexed by expected
species richness (total and for native forest species) on a site,
was modeled using published models for the Central Puget
Sound (Hepinstall, Marzluff and Alberti, 2009) and are described
in detail below. These models were developed from extensive
field data designed to quantify species richness on sites repre-
senting different locations along an urban-to-wildland gradient
(Donnelly and Marzluff, 2004, 2006; Marzluff, 2005; Marzluff
et al
., 2007). I used linear regression equations that related the
percentage forest, aggregation of residential land use, and mean
development age per km
2
to: (a) total species richness (for
a subset of 57 common species); and (b) species richness of
native forest species (19 species typically found in intact, mature
forests) (Hepinstall, Marzluff and Alberti, 2009). These vari-
ables were selected in the avian community models because they
were correlated with vegetation characteristics to which birds
were responding and were variables output by the UrbanSim
and LCCM models, and therefore available as spatially explicit
predictions of future conditions.
Conclusions
25.2.5
Predicted land cover change
for study area
I have provided a brief overview of one example of integrated
land use, land cover, and biodiversity modeling in an urbaniz-
ing environment. Such integrated modeling efforts provide the
opportunity to simulate the interrelated components of urban
development (Alberti and Waddell, 2000). I have shown how
The study area in 2027 is predicted to have more heavy urban,
light urban, and grass and less medium urban (converting to
Search WWH ::
Custom Search