Geoscience Reference
In-Depth Information
possibly because of confounding effects of weather, human disturbance, meas-
urement error, and other factors (McEwan and Hirth 1979; Hines 1987;
Rumble and Hodoroff 1993; Bruggink et al. 1994; Gilbert et al. 1996; Max-
son and Riggs 1996). Even experimental demographic response studies have
been plagued with unexpected variations in confounding variables (Taitt et al.
1981; Harper et al. 1993). Nevertheless, some carefully designed studies
observed links between habitat and reproduction or survival and through
further investigation discovered the underlying causes (Chasko and Gates
1982; Brown and Litvaitis 1995; Greenwood et al. 1995; Loegering and Fraser
1995).
A final major problem, discussed previously in relation to use-availability
and site attribute designs, is scale. Levin (1992) showed clearly that there is no
single correct scale for studying ecological relationships. Animals view and
react to their environment at various scales. Human perceptions of ecological
systems are inescapably biased or incomplete because they are filtered by the
observational scale chosen for the investigation. Some demographic response
studies have recognized this and have adopted a multiscaled approach. For
example, Orians and Wittenberger (1991) found that densities of yellow-
headed blackbirds (
Xanthocephalus xanthocephalus
) were higher on marshes
with higher food (insect) abundance but that density of blackbird territories
within these marshes was related to vegetational structure, not food. The
authors postulated that these birds may not select food-rich territories because
they often hunt outside their territories. Also, they establish their territories
before the full emergence of insects and hence may not be able to predict
future food abundance on a scale smaller than the marsh level. Pedlar et al.
(1997) developed habitat models on two different scales to explain variation in
raccoon (
Procyon lotor
) density. One model was fit to macrohabitat features
and the other to microhabitat variables, after which the two were combined to
form a more comprehensive model relating density to habitat at both scales.
Morris (1984, 1987, 1992) observed both macrohabitat and microhabitat dif-
ferences among several species of small mammals, suggesting habitat selection
on both scales, but found that variation in density was much more evident at
the macrohabitat level. On a larger scale, Dooley and Bowers (1998) discov-
ered, counter to their expectations, that densities and population growth rates
of voles were higher in patches within a fragmented landscape than in an
unfragmented landscape, whereas total population size was higher in the
unfragmented landscape (because more total habitat was available). Landscape
fragmentation caused overall habitat loss but, on a finer scale, enhanced repro-
duction within individual habitat fragments. Similarly, Brown and Litvaitis
