Environmental Engineering Reference
In-Depth Information
However, measures of urbanization tend to be highly redun-
dant. Therefore, Hahs and McDonnell (2006) created an index
that explained variability along the urbanization gradient. It
was based on two measures: density of people working in non-
agricultural employment (obtained from census information),
and the proportion of the landscape covered by impervious sur-
faces (determined from satellite imagery). They suggested that
this index could constitute a broadmeasure of urbanization, suit-
able for comparative studies. Remote sensor data plays a key role
here for capturing urbanization measures and their changes over
time, especially when urban growth is rapid and/or unplanned,
causing topographic maps to be misleading with respect, for
example, to the extent of a city (Angel, Sheppard and Civco,
2005). Thus, in order to allow comparisons, future studies try-
ing to explain species richness in a local urban habitat need to
combine both field studies and remote sensing techniques and
examine different hierarchical levels.
valuable information on how landscape fragmentation affects
birds in local urban woodlands could be provided if different
geographical scales were examined.
As pointed out, landscape fragmentation due to urbaniza-
tion is considered to be a major cause of the decline in species
richness. Andren (1994) suggested a theoretical landscape level
threshold of 20-30% of remaining habitat below which the
loss of species, or populations of species, is higher than the
combined effects of habitat loss and fragmentation. Few empir-
ical studies of landscape thresholds exist (but see Betts
et al
.,
2007). In Sweden, 1-34% (average 20%) of the area of cities
consists of urban woodlands in the form of remnant forests
(Hedblom and Soderstrom, 2008). In comparison, the propor-
tion of urban woodland averages 7% in the 22 largest cities in
Holland. Thus, with this range in extent of urban green species,
it is possible to undertake studies of thresholds in Swedish urban
landscapes.
Hedblom and Soderstrom (2008, 2010) used remote sensing
to select 34 cities out of 100 in which the proportion woodland
varied from 1-34%. The cities were also chosen on the basis
of other criteria such as: being in different regions of Sweden
(regional differences with respect to climate, etc.), cities sur-
rounded predominantly by agricultural land or forests (the two
most common landscape characteristics in Sweden) and whether
the city was mainly suburban, an industrial city, a major city, etc.
(see Hedblom and Soderstrom, 2008, for details). The woodlands
were chosen using remote sensing and thereafter habitat struc-
ture (whether deciduous or coniferous woodland dominated and
the age structure of the trees) and bird species were inventoried.
Along a gradient fromcenter to the surrounding, woodlands were
selected in triplets of equal size (i.e., in three size intervals:
>
1
to
20.4
Using remote sensing
to interpret effects of
urbanization on species
distribution
In order to explain species abundances in urban areas it is nec-
essary to select field data at different hierarchical levels (multiple
geographical scales) from the local habitat of urban green spaces
to differences between cities, nations and regions (Clergeau,
Jokimaki and Snep, 2006; McDonnell and Hahs, 2008). Thus, it
is costly and time consuming to have field personnel working in
many green spaces and cities at the same time (sometimes large
sampling efforts are required over short time periods, for example
when conducting breeding bird surveys). Moreover, it is difficult
to use only remote sensing without complementary field surveys
to explain species distribution. For example, Clergeau
et al
. 2006
compared cities by collecting data from inventories compiled
in 19 European cities. The results supported previous findings
that indicated a decline in species richness from rural areas to
urban centers. However, the species data had been collected using
different methods without detailed knowledge of the habitats.
Therefore it was difficult to draw conclusions about whether it
was the distance to the center, the surrounding landscape, or
the quality of the habitat that had the main effect on species
distribution. In a detailed study conducted in one city by Blair
(1996) there were clear differences in species richness depending
on whether the habitat studied was a golf course, park, lawn
or natural area. The author described the dominant landscape
character surrounding the study areas (e.g., residential area) but
did not quantify them making it difficult to interpret possible
fragmentation effects.
In northern Europe remnant forests are under severe threat
from development infilling the remaining natural spaces and
as a result of weak legal protection (Carlborg, 1991). These
urban woodlands also have high potential for biodiversity and
are popular for recreation (Hedblom and Soderstrom, 2008).
Studies by Hedblom and Soderstrom (2008, 2010) revealed that
8haor
>
8 ha, see Hedblom and Soderstrom,
2010). After using remote sensing to find woodlands fulfilling
these criteria, field assistants were despatched primarily to con-
firm that the woodlands had the characteristics indicated by the
remote sensor data (namely, the tree species configuration, age
structure and tree size).
Bird surveys were conducted by skilled ornithologists in 474
urban woodlands in 34 cities (see Hedblom and Soderstrom,
2010, for details). Since the digitalized remote sensor data ade-
quately represented the habitat characteristics (as confirmed by
the ground truth collected by the field assistants), digital maps
were used to quantify the proportion of woodland in cities and
in the surrounding area. Thereafter, species abundance in local
urban woodlands was compared with the proportions of wood-
land at the city level and in surrounding areas. The results revealed
that some bird species breeding in local urban woodlands were
affected by the total proportion of woodland at the city level,
indicating possible thresholds. Moreover, some species breeding
in local urban woodlands were affected by the proportion of
woodland in the surrounding landscape; this was especially the
case in cities surrounded by agricultural land (see Hedblom and
Soderstrom, 2010). Thus, urban planners and decision mak-
ers could take these results into consideration when examining
the effects of urban sprawl; in addition, they should carefully
consider applications to exploit local urban woodlands in cities
where they constitute 20-30% of the original landscape cover.
However, to understand fully species distribution and landscape
effects it is also important to include temporal studies of species
fluctuations.
≤
3ha,
>
3to
≤
Search WWH ::
Custom Search