398
proportion of their native biota may be of exceptionally high value for biodiversity conservation
and restoration (Sedell et al., 1990; Moyle and Sato,
1991; Doppelt et aI., 1993; Frissell and Bayles,
1996). In ecological assessments for which protection of biological diversity is a goal, locations of
sensitive species must be explicitly identified in a
drainage network context to formulate plans that
ensure their continued function as biodiversity refugia (Doppelt et aI., 1993; Frissell and Bayles, 1996;
Stanford et aI., 1996). Moyle and Randall (1998)
provide an example of such a spatially explicit assessment for watersheds of the Sierra Nevada area
of California. Their multimetric approach incorporates data on distribution and status of native fish
and amphibians, disruption of ecosystem integrity
by introduced fishes (e.g., those that prey on native
amphibians), and alteration of aquatic environments by dams, roads, and other human developments.
The global and regional assessments of the status of aquatic animals cited previously universally
point to habitat alteration and loss as the most widespread cause of aquatic biodiversity decline. Yet,
for both streams and lakes, there are few (if any)
universally accepted protocols for general assessment of habitat condition or change in habitat quality over time. Thus few monitoring networks or
comprehensive surveys provide comparable data
for either spatial or temporal assessment of habitat
trends. Many kinds of physical data can be useful
in attempting to reconstruct such trends (see extensive discussions and citations earlier in this
chapter), but in no case can this be expected to be
a cookbook exercise. Any method that pretends to
offer a simple or certain answer to this problem
should not be considered reliable and most certainly has not met with widespread professional or
scientific acceptance. For example, measures of
channel unit composition of streams are often
adopted as the survey and monitoring method of
choice for federal land management agency assessments, yet this widespread approach has serious limitations for this purpose (Poole et aI., 1997).
Structural and dynamic changes in physical habitat that frequently impair biota are often not predictably associated with changes in water chemistry or violation of chemical water quality
standards (Karr, 1991; Schlosser, 1991). Therefore,
routine monitoring of chemical water quality conditions to meet water quality standards is not sufficient for ecological assessments.
Nevertheless, several studies offer useful models for comparative analysis of habitat data and selected biotic information at the scale of tributaries
Assessment of Biotic Pattems in Freshwater Ecosystems
within a region or large river basin. For example,
Roth et aI. (1996), Minns et al. (1996), Richards et
aI. (1996), and Allan et aI. (1997) provide evidence
through such comparative studies indicating correspondence of various habitat measurements to metrics of biotic integrity and correlation of both habitat and biotic integrity to land use or land cover. In
Europe, using somewhat more explicit but in many
ways similar methodological framework, Bravard
et al. (1986) and Amoros et aI. (1987a, b) also developed a synthetic understanding of landscapelevel patterns in stream habitat changes and biotic
responses to them. On the other hand, Hughes
(1985) found that, although invertebrate assemblages corresponded well to the principal gradient
of environmental stress (mining wastes) among
study streams in Montana, trout biomass did not
show correspondence (perhaps partly because the
trout populations included a mix of native and introduced species whose varying tolerances and interactions were not taken into account). Working
at a similar scale with a comparative design across
a montane landscape, Baxter et al. (1999) demonstrated strong correlations of bull trout spawning
populations with land-use pattern (road density)
and with valley geomorphology (reflecting groundwater efflux and other channel conditions). Following Frissell et aI. (1986), Frissell and Bayles
(1996) suggest that comparative studies of this
type, incorporating a selected mosaic of catchments
with similar geological and biophysical templates
(including shared biogeographical history), but
which have been differently altered by human activities and recent natural events, are likely to provide the most useful information for ecological assessments. Such a design plows the middle ground
in scale between ecoregionally comprehensive assessments and local or site-focused assessments.
Frissell and Bayles (1996) argue that this approach
has already proved more fruitful than singlewatershed analyses, which are presently fashionable among federal and state agencies and industrial landowners and some scientists (e.g., Montgomery et aI., 1995). In a typical single-watershed
analysis, the ability to address critical questions
about physical and biological linkages rests on the
end to end coupling of simplified, mechanistic simulation models. Individually and collectively, these
models lack empirical validation, precision, and
any means of appropriate scaling (among other
problems).
Some conservation assessments of aquatic
species have employed population viability analyses (PV A) to understand the long-term probability
of persistence or extinction of a populations under
proportion of their native biota may be of exceptionally high value for biodiversity conservation
and restoration (Sedell et al., 1990; Moyle and Sato,
1991; Doppelt et aI., 1993; Frissell and Bayles,
1996). In ecological assessments for which protection of biological diversity is a goal, locations of
sensitive species must be explicitly identified in a
drainage network context to formulate plans that
ensure their continued function as biodiversity refugia (Doppelt et aI., 1993; Frissell and Bayles, 1996;
Stanford et aI., 1996). Moyle and Randall (1998)
provide an example of such a spatially explicit assessment for watersheds of the Sierra Nevada area
of California. Their multimetric approach incorporates data on distribution and status of native fish
and amphibians, disruption of ecosystem integrity
by introduced fishes (e.g., those that prey on native
amphibians), and alteration of aquatic environments by dams, roads, and other human developments.
The global and regional assessments of the status of aquatic animals cited previously universally
point to habitat alteration and loss as the most widespread cause of aquatic biodiversity decline. Yet,
for both streams and lakes, there are few (if any)
universally accepted protocols for general assessment of habitat condition or change in habitat quality over time. Thus few monitoring networks or
comprehensive surveys provide comparable data
for either spatial or temporal assessment of habitat
trends. Many kinds of physical data can be useful
in attempting to reconstruct such trends (see extensive discussions and citations earlier in this
chapter), but in no case can this be expected to be
a cookbook exercise. Any method that pretends to
offer a simple or certain answer to this problem
should not be considered reliable and most certainly has not met with widespread professional or
scientific acceptance. For example, measures of
channel unit composition of streams are often
adopted as the survey and monitoring method of
choice for federal land management agency assessments, yet this widespread approach has serious limitations for this purpose (Poole et aI., 1997).
Structural and dynamic changes in physical habitat that frequently impair biota are often not predictably associated with changes in water chemistry or violation of chemical water quality
standards (Karr, 1991; Schlosser, 1991). Therefore,
routine monitoring of chemical water quality conditions to meet water quality standards is not sufficient for ecological assessments.
Nevertheless, several studies offer useful models for comparative analysis of habitat data and selected biotic information at the scale of tributaries
Assessment of Biotic Pattems in Freshwater Ecosystems
within a region or large river basin. For example,
Roth et aI. (1996), Minns et al. (1996), Richards et
aI. (1996), and Allan et aI. (1997) provide evidence
through such comparative studies indicating correspondence of various habitat measurements to metrics of biotic integrity and correlation of both habitat and biotic integrity to land use or land cover. In
Europe, using somewhat more explicit but in many
ways similar methodological framework, Bravard
et al. (1986) and Amoros et aI. (1987a, b) also developed a synthetic understanding of landscapelevel patterns in stream habitat changes and biotic
responses to them. On the other hand, Hughes
(1985) found that, although invertebrate assemblages corresponded well to the principal gradient
of environmental stress (mining wastes) among
study streams in Montana, trout biomass did not
show correspondence (perhaps partly because the
trout populations included a mix of native and introduced species whose varying tolerances and interactions were not taken into account). Working
at a similar scale with a comparative design across
a montane landscape, Baxter et al. (1999) demonstrated strong correlations of bull trout spawning
populations with land-use pattern (road density)
and with valley geomorphology (reflecting groundwater efflux and other channel conditions). Following Frissell et aI. (1986), Frissell and Bayles
(1996) suggest that comparative studies of this
type, incorporating a selected mosaic of catchments
with similar geological and biophysical templates
(including shared biogeographical history), but
which have been differently altered by human activities and recent natural events, are likely to provide the most useful information for ecological assessments. Such a design plows the middle ground
in scale between ecoregionally comprehensive assessments and local or site-focused assessments.
Frissell and Bayles (1996) argue that this approach
has already proved more fruitful than singlewatershed analyses, which are presently fashionable among federal and state agencies and industrial landowners and some scientists (e.g., Montgomery et aI., 1995). In a typical single-watershed
analysis, the ability to address critical questions
about physical and biological linkages rests on the
end to end coupling of simplified, mechanistic simulation models. Individually and collectively, these
models lack empirical validation, precision, and
any means of appropriate scaling (among other
problems).
Some conservation assessments of aquatic
species have employed population viability analyses (PV A) to understand the long-term probability
of persistence or extinction of a populations under