Skip Links
Scientific Investigations Report 2011–5193
Prepared in cooperation with the
Massachusetts Department of Conservation and Recreation, the
Massachusetts Department of Environmental Protection, and the
Massachusetts Department of Fish and Game
Factors Influencing Riverine Fish Assemblages in Massachusetts
By David S. Armstrong, Todd A. Richards1, and Sara B. Levin
1Massachusetts Department of Fish and Game, Division of Fisheries and Wildlife
Abstract
The U.S. Geological Survey, in cooperation with the
Massachusetts Department of Conservation and Recreation,
Massachusetts Department of Environmental Protection,
and the Massachusetts Department of Fish and Game,
conducted an investigation of fish assemblages in small- to
medium-sized Massachusetts streams. The objective of this
study was to determine relations between fish-assemblage
characteristics and anthropogenic factors, including
impervious cover and estimated flow alteration, relative to
the effects of environmental factors, including physical-basin
characteristics and land use. The results of this investigation
supersede those of a preliminary analysis published in
2010. Fish data were obtained for 669 fish-sampling sites
from the Massachusetts Division of Fisheries and Wildlife
fish-community database. A review of the literature was
used to select fish metrics—species richness, abundance
of individual species, and abundances of species grouped
on life history traits—responsive to flow alteration. The
contributing areas to the fish-sampling sites were delineated
and used with a geographic information system to determine
a set of environmental and anthropogenic factors that were
tested for use as explanatory variables in regression models.
Reported and estimated withdrawals and return flows were
used together with simulated unaltered streamflows to estimate
altered streamflows and indicators of flow alteration for each
fish-sampling site. Altered streamflows and indicators of flow
alteration were calculated on the basis of methods developed
in a previous U.S. Geological Survey study in which unaltered
daily streamflows were simulated for a 44-year period
(water years 1961–2004), and streamflow alterations were
estimated by use of water-withdrawal and wastewater-return data previously reported to the State for the 2000–04 period and estimated domestic-well withdrawals and septic-system
discharges.
A variable selection process, conducted using principal
components analysis and Spearman rank correlation, was
used to select a set of 15 non-redundant environmental and
anthropogenic factors to test for use as explanatory variables
in the regression analyses. Twenty-one fish species were used
in a multivariate analysis of fish-assemblage patterns. Results
of nonmetric multidimensional scaling and hierarchical cluster
analysis were used to group fish species into fluvial and
macrohabitat generalist habitat-use classes.
Two analytical techniques, quantile regression and
generalized linear modeling, were applied to characterize the
association between fish-response variables and environmental
and anthropogenic explanatory variables. Quantile regression
demonstrated that as percent impervious cover and an
indicator of percent alteration of August median flow from
groundwater withdrawals increase, the relative abundance
and species richness of fluvial fish decrease. The quantile
regression plots indicate that (1) as many as seven fluvial fish
species are expected in streams with little flow alteration or
impervious cover, (2) no more than four fluvial fish species are
expected in streams where flow alterations from groundwater
withdrawals exceed 50 percent of the August median flow
or the percent area of impervious cover exceeds 15 percent,
and (3) few fluvial fish remain at high rates of withdrawal
(approaching 100 percent) or high rates of impervious cover
(between 25 and 30 percent). Three generalized linear models
(GLMs) were developed to quantify the response of fluvial
fish to multiple environmental and anthropogenic variables.
All variables in the GLM equations were demonstrated to
be significant (p less than 0.05, with most less than 0.01).
Variables in the fluvial-fish relative-abundance model were
channel slope, estimated percent alteration of August median
flow from groundwater withdrawals, percent wetland in a
240-meter buffer strip, and percent impervious cover.
Variables in the fluvial-fish species-richness model were
drainage area, channel slope, total undammed reach length,
percent wetland in a 240-meter buffer strip, and percent
impervious cover. Variables in the brook trout relativeabundance
model were drainage area, percent open water,
and percent impervious cover. The variability explained by
the GLM models, as measured by the pseudo R2, ranged from
18.2 to 34.6, and correlations between observed and predicted values ranged from 0.50 to 0.60. Results of GLM models indicated that, keeping all other variables the same, a one-unit
(1 percent) increase in the percent depletion of August median
flow would result in a 0.9-percent decrease in the relative
abundance (in counts per hour) of fluvial fish. The results of
GLM models also indicated that a unit increase in impervious
cover (1 percent) resulted in a 3.7-percent decrease in the
relative abundance of fluvial fish, a 5.4-percent decrease in
fluvial-fish species richness, and an 8.7-percent decrease in
brook trout relative abundance.
|
First posted December 2011
Part or all of this report is presented in Portable Document Format (PDF); the latest version of Adobe Reader or similar software is required to view it. Download the latest version of Adobe Reader, free of charge. |
Suggested citation:
Armstrong, D.S., Richards, T.A., and Levin, S.B., 2011, Factors influencing riverine fish assemblages in Massachusetts: U.S. Geological Survey Scientific-Investigations Report 2011–5193, 58p. (Also available at http://pubs.usgs.gov/sir/2011/5193.)
Contents
Acknowledgments
Abstract
Introduction
Purpose and Scope
Description of Study Area
Previous Studies
Factors Limiting Riverine Fishes
Methods of Determining Factors Affecting Fish Assemblages
Fish-Assemblage Characteristics
Environmental Factors
Simulated Streamflow Data and Measures of Flow Alteration
Selection of Environmental Factors and Fish Assemblage Metrics
Environmental Factor Selection using Principal Component Analysis and
Spearman Rank Correlation
Fish Assemblage Classification using Ordination and Hierarchical Cluster
Analysis
Analysis of Associations between Fish Assemblages and Environmental Factors
Quantile Regression
Generalized Linear Models
Factors Affecting Riverine Fish Assemblages
Environmental Factors
Streamflow-Alteration Indicators
Variable Reduction and Selection
Fish-Assemblage Characteristics
Patterns in Fish-Assemblage Structure
Quantile Regression
Generalized Linear Models
Fluvial-Fish Species-Richness Models
Fluvial-Fish Relative-Abundance Models
Brook Trout Relative-Abundance Model
Model Validation and Limitations
Comparison of Study Results to Preliminary Analysis
Possible Future Modeling Efforts
Summary and Conclusions
References Cited