ArticlePDF Available

Evaluating habitat associations of a fish assemblage at multiple spatial scales in a minimally disturbed stream using low-cost remote sensing

  • Metro Wastewater Reclamation District

Abstract and Figures

Habitat heterogeneity at multiple scales is a major factor affecting fish assemblage structure. However, assessments that examine these relationships at multiple scales concurrently are lacking. The lack of assessments at these scales is a critical gap in understanding as conservation and restoration efforts typically work at these levels.A combination of low-cost side-scan sonar surveys, aerial imagery using an unmanned aerial vehicle, and fish collections were used to evaluate the relationship between physicochemical and landscape variables at various spatial scales (e.g. micro-mesohabitat, mesohabitat, channel unit, stream reach) and stream–fish assemblage structure and habitat associations in the South Llano River, a spring-fed second-order stream on the Edwards Plateau in central Texas during 2012–2013.Low-cost side-scan sonar surveys have not typically been used to generate data for riverscape assessments of assemblage structure, thus the secondary objective was to assess the efficacy of this approach.The finest spatial scale (micro-mesohabitat) and the intermediate scale (channel unit) had the greatest explanatory power for variation in fish assemblage structure.Many of the fish endemic to the Edwards Plateau showed similar associations with physicochemical and landscape variables suggesting that conservation and restoration actions targeting a single endemic species may provide benefits to a large proportion of the endemic species in this system.Low-cost side-scan sonar proved to be a cost-effective means of acquiring information on the habitat availability of the entire river length and allowed the assessment of how a full suite of riverscape-level variables influenced local fish assemblage structure. Copyright © 2015 John Wiley & Sons, Ltd.
Content may be subject to copyright.
Evaluating habitat associations of a fish assemblage at multiple
spatial scales in a minimally disturbed stream using low-cost remote
Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, Lubbock, TX, USA
US Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, Lubbock, TX, USA
Texas Parks and Wildlife Department, Inland Fisheries Division, San Marcos, TX, USA
1. Habitat heterogeneity at multiple scales is a major factor affecting sh assemblage structure. However,
assessments that examine these relationships at multiple scales concurrently are lacking. The lack of assessments
at these scales is a critical gap in understanding as conservation and restoration efforts typically work at these levels.
2. A combination of low-cost side-scan sonar surveys, aerial imagery using an unmanned aerial vehicle, and sh
collections were used to evaluate the relationship between physicochemical and landscape variables at various
spatial scales (e.g. micro-mesohabitat, mesohabitat, channel unit, stream reach) and streamsh assemblage
structure and habitat associations in the South Llano River, a spring-fed second-order stream on the Edwards
Plateau in central Texas during 20122013.
3. Low-cost side-scan sonar surveys have not typically been used to generate data for riverscape assessments of
assemblage structure, thus the secondary objective was to assess the efcacy of this approach.
4. The nest spatial scale (micro-mesohabitat) and the intermediate scale (channel unit) had the greatest
explanatory power for variation in sh assemblage structure.
5. Many of the sh endemic to the Edwards Plateau showed similar associations with physicochemical and
landscape variables suggesting that conservation and restoration actions targeting a single endemic species may
provide benets to a large proportion of the endemic species in this system.
6. Low-cost side-scan sonar proved to be a cost-effective means of acquiring information on the habitat
availability of the entire river length and allowed the assessment of how a full suite of riverscape-level variables
inuenced local sh assemblage structure.
Copyright #2015 John Wiley & Sons, Ltd.
Received 10 September 2014; Revised 22 February 2015; Accepted 06 April 2015
KEY WORDS: stream; river; habitat mapping; landscape; sh
*Correspondence to: Tim Grabowski, US Geological Survey, Texas Cooperative Fish and Wildlife Research Unit, Texas Tech University, PO
Box 42120, Lubbock, Texas 79409-2120, USA. Email:
Present address:Texas Parks and Wildlife Department, Inland Fisheries Division, Heartof the Hills Fisheries Science Center, Mountain Home, TX, USA
Present address: US Bureau of Land Management, Cañon City, CO, USA
Copyright #2015 John Wiley & Sons, Ltd.
Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
Published online in Wiley Online Library
( DOI: 10.1002/aqc.2569
Applying landscape ecology principles to river
systems has spawned a holistic perspective (Wiens,
2002; Palmer et al., 2010), where researchers are
recognizing the inuences that habitat variation at
multiple spatial scales have on sh assemblage
structure (Wang et al., 1997, 2003; Fitzpatrick
et al., 2001; Benda et al., 2004; Gido et al., 2006;
Wehrly et al., 2006). Understanding the
complexity of a river ecosystem at multiple scales
is essential for accurately quantifying shhabitat
associations since the distribution and abundance
of riverine sh populations is ultimately
determined by both ne and coarse scale spatial
and temporal phenomena (Poff, 1997; Allan,
2004). Modelling the factors inuencing sh
populations at different scales has produced vastly
different outcomes depending on whether focus
was placed on ne scales (Gorman, 1988; Gido
and Propst, 1999; Lammert and Allan, 1999;
Wang et al., 2003; Bouchard and Boisclair, 2008)
or coarser scales (Wang et al., 1997; Fitzpatrick
et al., 2001; Benda et al., 2004; Gido et al., 2006;
Wehrly et al., 2006). Furthermore, data resolution
and degree of disturbance have been implicated in
determining the relative importance of one scale
over another (Allan, 2004; Brewer et al., 2007).
Regardless, there is agreement that sh
assemblages are affected by factors at multiple
scales (Poff, 1997), including human disturbances
such as agriculture, urbanization, and road
development (Naiman et al., 1995; Allan, 2004;
Walsh et al., 2005). Therefore, understanding the
relationships between scale and the dynamic
nature of lotic systems and their effects on
subsequent habitat associations is a critical step
towards effectively managing, conserving, and
restoring sh populations. Attempts to address
disturbances at the wrong scale might, at best,
result in temporary improvements and, at worst,
result in wasted resources that might have been
better applied at different scales.
The importance of approaching riverscape and
catchment conservation from the appropriate scale
is illustrated in the rivers and streams of central
Texas. The disturbance of river systems is a
primary threat for the native sh populations
located throughout the state, and while the
proximate mechanisms of disturbance vary, the
ultimate factor is a growing human population.
The state of Texas is projected to experience a
2025% increase in human populations over the
next 1015 years (Murdock et al., 2002) resulting
in changes in land use, water demand, and water
quality. For the Edwards Plateau ecoregion of
central Texas, which is characterized by high
biodiversity and high regional endemism (Bowles
and Arsuf, 1993), an increase in urban growth is
expected to lead to a higher demand on surface
and groundwater resources resulting in decreased
ows, decreased water levels, and physicochemical
changes to regional streams. These disturbances
have the potential to threaten the integrity of
Edwards Plateau sh assemblages (Garrett et al.,
1992; Hubbs, 1995; Edwards et al., 2004) and are
driving the development of conservation and
restoration plans for some of the more vulnerable
and visible regional endemics, such as Guadalupe
bass Micropterus treculii.
Establishing a target or guiding image(Hughes
et al., 1986; Palmer et al., 2005; Raven et al., 2010)
to measure the success of conservation and
restoration actions is critical, as the effectiveness of
these efforts is too often limited by a lack of
criteria for dening and assessing success (Kondolf,
1995; Palmer et al., 2005). Developing targets to
dene success requires quantifying shhabitat
associations in a catchment that not only supports
the regional sh assemblage, but also represents
the typical habitats that would be found in a
particular region. Preferably, targets would be
developed from near-natural lotic systems within
that region. However, very few systems remain
untouched within the Edwards Plateau ecoregion
owing to human pressures (Benke, 1990), and,
therefore, minimally disturbed systems may be the
next available option. Furthermore, the thorough
habitat inventory across multiple spatial scales
needed to develop a guiding image for
conservation or restoration actions can be
prohibitively time-consuming or costly. The use of
low-cost remote sensing technologies, such as
side-scan sonar (Kaeser and Litts, 2010) and
unmanned aerial vehicles (Birdsong et al., in press),
offers the promise of quickly and cost-effectively
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
mapping instream habitats and, when combined
with other available datasets, provides a thorough
inventory of habitat within a catchment. However,
integration of these data to develop a multi-scale
perspective of stream sh assemblage structure has
not been attempted. The objectives for this study
were to demonstrate the effectiveness of this
approach both to develop a statistically robust
sampling design and to use it to determine the
degree to which landscape and riverscape variables
at different spatial scales inuence the structure
and composition of stream sh assemblages.
Study area
The study area is located within the Colorado River
Basin in west-central Texas on the Edwards Plateau
(Linam et al., 2002). A majority of the 10 million
ha of this karst plateau (Edwards et al., 2004;
Heilman et al., 2009) overlays the Edwards
Aquifer, which supplies water to approximately
half of the springs in Texas (Brune, 1981) and
serves as a signicant water supply for adjacent
fast-growing urban areas, such as Austin and San
Antonio. Research was conducted on the South
Llano River, a small, spring-fed, second-order
tributary of the Colorado River Basin, located
approximately 175km north west of San Antonio.
It is approximately 88km in length from its
headwaters in Edwards County, Texas to its
conuence with the North Llano River in Kimble
County near Junction, Texas. The South Llano
River is composed of the typical habitats that are
found in Hill Country streams, and supports many
of the endemic sh species found throughout this
region (Hubbs et al., 1991, 2008). Compared with
most catchments on the Edwards Plateau the South
Llano River has relatively low levels of human
impact largely because of low population densities
within the catchment and comparatively little water
withdrawn for agricultural or municipal purposes
(Linam et al., 2002). The study area consisted of a
39 km reach beginning approximately 1.5 km
upstream of the conuence of the South Llano
River and Paint Creek and ended at Lake Junction
Dam in Junction (Figure 1). Owing to drought
conditions, a decreasing trend in mean daily
discharge was observed throughout 20122013,
with an overall mean (± SD) discharge of only
1.4 ± 0.3 m
compared with a historical median
of 2.0 m
(mean ± SD = 3.6 ± 25.7 m
Data collection
Habitat characteristics were assessed at ve spatial
scales: micro-mesohabitat, mesohabitat, channel
unit, stream reach, and landscape (riparian buffer).
The micro-mesohabitat habitat was the nest spatial
scale and consisted of a patch of substrate within a
mesohabitat (mean area ± SD: 3152 ± 4115 m
range: 2021 602 m
). The mesohabitat scale was
dened as a single pool, rife, or run (mean area
± SD: 8601 ± 9205 m
; range: 86656 010 m
), while
the channel unit scale consisted of the mesohabitat
located at each sh sampling site and all habitat
between it and the nearest upstream and
downstream neighbour of the same mesohabitat
type (mean area ± SD: 22 997 ± 19 310 m
4211117 381 m
). The reach scale was dened as
sections of river uninterrupted by any type of
articial or natural barriers, e.g. road crossings
or a waterfall (Figure 1). Seven non-overlapping
reaches were identied within the study area
(mean area ± SD: 170 123 ± 119 542 m
52 512337 016 m
). For the largest spatial scale,
land-cover data from Phase 4 of the Texas
Ecological Systems Classication Project were
acquired from the Texas Parks and Wildlife
Department and used to quantify the various
oodplain habitats within the South Llano River
catchment. For each sampling site, any upstream
portion of the landscape that fell within a 50m
buffer from the river bank was exported into a
separate shapele for further analysis.
Side-scan sonar surveys, following the methods
of Kaeser and Litts (2010), were conducted in
October 2011 and June 2012 using a Humminbird
988c SI side-scan sonar unit set to a frequency of
455 kHz and a maximum range of 37.8 m on either
side of the transducer. Depth of the thalweg
during the survey was typically 0.62.5 m.
Habitats that were too shallow to be mapped
using side-scan sonar were mapped using a
handheld GPS unit. To characterize instream
microhabitat availability, substrate classes (Kaeser
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
and Litts, 2010) and subclasses (Barnhardt et al.,
1998) were delineated from resulting sonar
imagery in ArcMap 10 (ESRI, Redlands,
California) based on dominant (>50%) and
subordinate (<50%) substrate types within a
given area. Instream structures 100 cm in length
(e.g. boulders and large woody debris) were
identied also and assigned to separate classes.
Mesohabitat types (i.e. runs, rifes, and pools)
were delineated from a combination of aerial
imagery using unmanned aerial vehicle yovers
and depth proles collected during the side-scan
sonar survey. Aerial imagery was collected by
a remote-controlled airplane modied for
autonomous data collection through the addition
of an autopilot, GPS system, inertial measurement
unit, onboard computer, and red-blue-green and
near infrared digital cameras (Chao et al., 2009).
Photographs of the South Llano River were taken
at approximately 400600 m above ground
resulting in imagery with 1116 cm resolution
that was georeferenced and mosaicked using
EnsoMOSAIC (Mosaic Mill OY, Vantaa,
Finland). This imagery was also used to aid in
the classication of habitat in locations where
the side-scan sonar did not perform well, such as
shallow rifes or extensive beds of submerged
aquatic vegetation.
To assess the accuracy of substrate classications
from side-scan sonar imagery, ground truthing was
conducted by revisiting the study area and
recording the observed substrates and depth
approximately every 130 m. In addition, a subset
(25%) of the instream structures, such as boulders
and large woody debris, were selected for ground
Proportions of habitat types within each spatial
scale, the area and perimeter of habitat types at
each spatial scale, and other landscape metrics,
such as contiguity, were calculated for each spatial
scale using FragStats 4.1 (McGarigal et al. 2012;
Table 1). Contiguity was calculated as the average
Figure 1. Map of the South Llano River study area in Kimble County, Texas with stream reaches as demarcated by road crossings or natural barriers
indicated. Inset maps illustrate the location of the South Llano River within the Colorado River Basin and the continental US.
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
spatial connectedness of habitat types within a
given spatial scale following the method of LaGro
(1991) and McGarigal et al. (2012), where
increasing contiguity index values indicate larger,
more connected patches of a given habitat type.
Additional habitat variables, such as reach
sinuosity, distance to nearest micro-mesohabitat
patch, lengthwidth ratios of channel units and
reaches, total length of upstream eroded banks,
sum of eroded upstream banks within each reach,
distance to nearest eroded upstream bank, and
distance to nearest barrier, were manually
calculated in ArcMap 10.
ArcMap 10 was used to randomly select 70 sample
sites representing all available micro-mesohabitat
combinations. Fishes were surveyed seasonally
using a 3.96 × 1.22 × 2.00 m bag seine with a
0.5 cm mesh pulled along a 25 m transect. Transect
length was calculated from GPS waypoints taken at
each end. Six of the 70 sites were too deep to be
sampled effectively with a seine. Thus, six
alternative sites with comparable habitat
composition and depth were chosen where access
allowed for sampling with a boat electrosher
along 50 m transects. Electroshing was conducted
with pulsed DC power at 120 pulses s
(Hz) with
voltage and pulse width adjusted to maintain an
output of approximately 4 A. Captured shes were
released after being identied to species except for
a small number of voucher specimens that were
euthanized in a 0.06% MS-222 solution and xed
in 10% formalin. Habitat parameters including
water temperature, current velocity, dissolved
oxygen (DO), conductivity, turbidity, canopy
cover, stream width and distance from the nearest
bank were recorded after each sampling event at
each sampling site.
Data analysis
Canonical correspondence analysis (CCA; Ter
Braak, 1986) was performed using CANOCO v. 5
(Microcomputer Power, Ithaca, New York) to
evaluate the shhabitat associations. Before
analysis, the habitat data and environmental
measurements were evaluated for normality using
the SAS 9.2 software package (SAS Institute Inc,
2008). Appropriate transformations were applied
to improve homogeneity of variances and to help
dampen the effects of outliers in the dataset
(McCune and Grace, 2002) and all variables were
standardized to a standard deviation of 1 and a
mean of 0. Correlation between variables was
assessed using Pearsons correlation analysis and a
KruskalWallis test was performed on highly
correlated pairs of variables (r>0.70, P<0.01;
McGarigal et al., 2000) to select the variable with
the greatest among-group variance (Noon, 1981;
McGarigal et al., 2000). Principal component
Table 1. Variables at each spatial scale identied by principal
component analysis as being informative for describing the structure
of stream sh assemblages in the South Llano River, Texas. Variables
were selected through the cumulative proportion of total variance
Scale Variable Abbreviation
Micro-mesohabitat Contiguity mm_Contig
Nearest neighbour mm_nn
Temperature Temperature
Current velocity Current velocity
Turbidity Turbidity
% Canopy cover Canopy cover
Width of channel Width
Distance to nearest bank Distance
Conductivity Conductivity
Depth Depth
Mesohabitat Contiguity m_Contig
Proportion of boulder m_BO
Proportion of woody debris m_LWD
Proportion of bedrock m_BR
Proportion of aquatic
Proportion of cobblegravel m_CoGr
Channel unit Area c_Area
Contiguity c_Contig
Proportion of rife
Proportion run boulder c_RnBO
Proportion rife boulder c_RiBO
Proportion pool gravelsand c_PoGs
Proportion run aquatic
Proportion rife aquatic
Proportion pool aquatic
Proportion pool woody debris c_PoLWD
Proportion run gravelcobble P_RnGc
Reach Perimeterarea ratio r_PA
Proportion pool r_Pool
Proportion rife r_Rife
Proportion run cobblegravel r_RnCg
Proportion pool cobblegravel r_PoCg
Proportion pool aquatic
(riparian buffer)
Proportion of forest Forest
Proportion of barren land Barren
Proportion of vegetation Vegetation
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
analysis (PCA) was used to select the most
informative remaining variables (Jolliffe, 1973,
1986; Khattree and Naik, 2000; Table 1). Fish
relative abundance was used as the response
variable in each partial and full CCA. Only
species that occurred in 5.0% of the samples were
used in the analysis (McCune and Grace, 2002),
and the relative abundance data were square-root
transformed to fulll the unimodal assumption of
CCA (McCune and Grace, 2002).
Variance partitioning, following Borcard et al.
(1992), was performed using partial canonical
ordinations to quantify independent effects of the
explanatory variables at each spatial scale and
identify the residual variation by using
explanatory variables from the other scales as
covariables. Only the three spatial scales that
explained the highest amount of variation in the
species data were selected and used in the analysis
owing to software limitations.
Polythetic agglomerative hierarchical clustering
(PAHC) was used to evaluate species associations
at each spatial scale using the Wards
minimum-variance linkage fusion method (Ward,
1963) based on a Euclidean distance matrix
(McCune and Grace, 2002; McGarigal et al.,
2000). The number of signicant clusters for the
analysis was determined based on a combination
of criteria that included observing peaks of the
cubic clustering criterion, pseudo F-statistic
(Khattree and Naik, 2000; McGarigal et al.,
2000), and visually identifying the major inection
point in scree plots (Khattree and Naik, 2000).
For comparison, the environmental and habitat
variables selected for the CCA were used in the
hierarchical cluster analysis along with the relative
abundances of all the species. The null hypothesis
that environmental/habitat variables did not
differ among clusters was evaluated using a
KruskalWallis test (McGarigal et al., 2000;
Ott and Longnecker, 2008). A signicance level of
α= 0.05 was used for all tests.
Substrate was accurately classied from side-scan
sonar imagery for 315 of the 349 (90.3%) ground
truthing sites. Each misidentied substrate was
corrected in the nal instream habitat map for
subsequent analyses. Habitat availability varied
longitudinally along the length of the South Llano
River. The upstream reaches of the river were
primarily rife and run habitats with coarse
substrates, particularly karst bedrock. Proportion
of pool mesohabitat, ner substrates, and
submerged aquatic vegetation increased in the
middle and downstream portions of the river.
Boulders were more common in the upstream
reaches than in those further downstream.
However, large woody debris was relatively rare
throughout the river and only comprised a
meaningful component of instream habitat in the
lowest reach of the study area.
In total, 3397 individual shes encompassing
25 species were captured over the duration of the
study (Table 2), with the largest number of species
captured in the spring sampling period followed by
the summer, autumn and winter. Blacktail shiner
Cyprinella venusta and Texas shiner Notropis
amabilis were the most abundant species and
accounted for more than 55% of the total catch.
Seasonally, blacktail shiner was the most abundant
species encountered, except during spring when
Texas shiner dominated the catch. Species diversity
was highest during the summer and spring
sampling periods.
Canonical correspondence analysis indicated a
division of species into two large clusters generally
associated with pool and rife/run habitat
characteristics regardless of the spatial scale
examined (Figure 2). The relative abundances of
species associated with rife or run habitats
were strongly correlated with higher current
velocity at the micro-mesohabitat scale, while
species associated with pool habitats were
correlated with the occurrence of deeper and wider
portions of the river with greater canopy cover
and micro-mesohabitat contiguity (Figure 2(A)).
Similarly, separation of species along the rst
axis at the mesohabitat scale was primarily
associated with pool and rife/run habitat
characteristics (Figure 2(B)). The relative
abundance of species associated with rife/run
habitats was correlated with higher proportions
of coarse substrates, such as bedrock and
cobblegravel. Habitat contiguity remained a
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
major inuence on the relative abundance of
pool-associated species, but the proportion of
structured habitat within a pool, such as large
woody debris, was also an important factor
affecting relative abundance for a majority of
these species. Guadalupe roundnose minnow stood
out as being the sole species for which submerged
aquatic vegetation exhibited a strong inuence on
relative abundance (Figure 2(B)).
At the channel unit scale, the total area of the
channel unit and the availability of structured
habitat and coarse substrates were the factors most
Table 2. Number of individuals of each sh species captured per season and in total from the South Llano River during 2012. Per cent total is the
relative contribution of each individual species to the total abundance captured. Rank indicates the relative position of each species according to
their total abundance
Species Summer Autumn Winter Spring Total abundance % Total Rank
Blacktail shiner (BTS) 309 222 187 286 1004 29.56 1
Cyprinella venusta
Texas shiner (TXS) 49 127 172 544 892 26.26 2
Notropis amabilis
Mimic shiner (MS) 184 7 2 54 247 7.27 3
Notropis volucellus
Longear sunsh (LS) 75 37 22 53 187 5.50 4
Lepomis megalotis
Guadalupe roundnose minnow (DI) 0 113 25 21 159 4.68 5
Dionda nigrotaeniata
Redbreast sunsh (RBS) 23 76 15 40 154 4.53 6
Lepomis auritus
Guadalupe bass (GB) 70 47 5 24 146 4.30 7
Micropterus treculii
Central stoneroller (CSR) 58 31 6 10 105 3.09 8
Campostoma anomalum
Western mosquitosh (GAM) 33 27 6 12 78 2.30 9
Gambusia afnis
Rio Grande cichlid (RGC) 37 27 10 1 75 2.21 10
Herichthys cyanoguttatus
Bluegill (BG) 23 19 8 24 74 2.18 11
Lepomis macrochirus
Gray redhorse (GRH) 20 19 12 13 64 1.88 12
Moxostoma congestum
Orangethroat darter (OTD) 2 11 29 14 56 1.65 13
Etheostoma spectabile
Texas logperch (TLP) 6 13 13 14 46 1.35 14*
Percina carbonaria
Largemouth bass (LMB) 22 12 6 6 46 1.35 14*
Micropterus salmoides
Redear sunsh 9 5 0 0 14 0.41 16
Lepomis microlophus
Channel catsh 7 4 2 0 13 0.38 17
Ictalurus punctatus
Gizzard shad 3 2 6 1 12 0.35 18
Dorosoma cepedianum
Greenthroat darter 3 1 4 2 10 0.29 19
Etheostoma lepidum
Green sunsh
Lepomis cyanellus 0 0 4 1 5 0.15 20
Lepomis gulosus 0 1 3 0 4 0.12 21
Common carp
Cyprinus carpio 0 2 0 0 2 0.06 22*
Flathead catsh
Pylodictis olivaris 2 0 0 0 2 0.06 22*
Longnose gar
Lepisosteus osseus 0 0 0 1 1 0.03 24*
River carpsucker
Carpiodes carpio 0 1 0 0 1 0.03 24*
Total 935 804 537 1121 3,397
*Denotes a tie in rank
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
highly correlated with the composition of the
relative abundance of most species (Figure 2(C)).
Although there were still two primary groups of
species, there was no clear grouping of
pool-associated species. Species associated with
channel units containing higher proportions of coarse
substrate rifes and runs with a greater number of
boulders were those captured almost exclusively
from rife and run habitats. Species associated with
larger channel units with a higher proportion of
submerged aquatic vegetation were captured from all
three mesohabitat types (Figure 2(C)).
The relative abundance of species associated with
rife/run habitat was more strongly correlated with
variables from ner spatial scales, while factors at
coarser spatial scales had greater inuence on the
relative abundance of species associated with pool
habitats (Figure 3). Overall, the 28 variables
across three spatial scales were only able to
account for 22.6% of the total variation in the
data (Table 3). The micro-mesohabitat scale
accounted for 35% of this explained variation in
sh assemblage data, followed by the channel unit
scale which accounted for 29%. Only 12.9% of the
explained variation in sh assemblage data was
accounted for by the mesohabitat scale variables.
There was consistent separation between species
that were associated with pools and those
associated with rifes/runs at every spatial scale,
except at the landscape (riparian buffer) level
(Figure 4). Although the number of clusters
ranged from three to four, the composition of
species within clusters tended to be consistent
across most of the spatial scales and followed the
same general mesohabitat type division, i.e. pool
vs. rife-run, observed in the CCA. However, the
Figure 2. Canonical correspondence analysis biplots of stream sh species scores at the micro-mesohabitat scale (A), mesohabitat scale (B), and the
channel unit scale (C) in the South Llano River, Texas during 2012. For all panels, environmental variables at each scale are indicated by arrows
and abbreviations as listed in Table 1 and the species scores are represented by the abbreviations listed in Table 2. The percentages of the explained
variance together with the eigenvalues are as follows: Panel A, Axis 1: 37.8%, 0.25; Axis 2: 23.8%, 16. Panel B, Axis 1: 44.4%, 0.18; Axis 2: 25.3%,
0.10. Panel C, Axis 1: 34.5%, 0.19; Axis 2: 21.7%, 0.12.
Figure 3. Canonical correspondence analysis biplot of stream sh
species scores including a composite of variables from the micro-
mesohabitat, mesohabitat, and channel unit scales in the South Llano
River, Texas during 2012. Axis 1 represents 24.9% of the explained
variance with an eigenvalue of 0.32. Axis 2 describes 17.6% of the
explained variance with an eigenvalue of 0.22. Descriptions of the
abbreviations for the environmental variables are given in Table 1
and species names in Table 2.
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
usual grouping of sh species associated with pool
or rife/run habitat characteristics was not
observed with the landscape (riparian buffer) scale.
Rather the species were classied into mixed
groups determined by their afnities for different
dominant land-cover types in the riparian buffer
(Table 4).
Estimates of the substrate composition and the
spatial distribution of substrate types in the South
Llano River generated from side-scan sonar
surveys were consistent with those from more
traditional, on-the-ground assessments conducted
by Heitmueller (2009). However, the combination
of low-cost side-scan sonar surveys and aerial
imagery produced an inventory of instream
habitats and substrates far more comprehensive
than could be generated at the same cost using
more traditional approaches. The entire process of
surveying the 39 km study reach, including initial
groundtruthing efforts, required approximately
ve full days in the eld, mainly owing to a lack
of access points for motorized watercraft on the
South Llano River. There was a substantial
amount of time involved with processing and
interpreting the sonar and aerial imagery as it
took time to become procient with the image
processing and interpretation, but a complete
instream habitat inventory with sub-metre
resolution was produced for this study in
approximately 45 days. This length of time seemed
to be directly related to the experience level of the
authors with the software and procedures used for
processing and interpreting the sonar imagery, as
subsequent surveys required less time to complete
(T.B. Grabowski, personal observation).
Furthermore, it should be noted that since the
production of the habitat inventory used in the
present survey, commercial software has become
available that can substantially reduce the time
required for georeferencing and mosaicking sonar
Substrate composition at the sampling sites
remained consistent during this study, but ow
variation, particularly large ood pulses, have the
potential to move substrate. Follow-up studies in
the South Llano River should re-survey substrates
to ensure that the spatial distribution of substrate
patches has not shifted. While the instream habitat
map generated by side-scan sonar surveys
remained accurate throughout the duration of the
study, the same was not true for submerged
aquatic vegetation. Follow-up side-scan sonar
surveys of selected river segments indicated that
percentage cover of the submerged aquatic
vegetation substrate class uctuated seasonally
because of the spring and summer growth and
winter dieback of American water-willow Justicia
americana and other species. The inuence of
submerged aquatic vegetation on sh assemblage
structure is likely to be underestimated in the
summer and overestimated in the winter.
However, it is not clear how the seasonal
uctuations in submerged aquatic vegetation cover
affected the combined-season estimates of its
inuence on sh assemblage structure.
Habitat variables at the micro-mesohabitat scale
were identied as the most important in structuring
the sh assemblage of the South Llano River when
considering instream and riparian habitats at
Table 3. Independent (13) and confounded (47) components of explained variation in stream sh assemblage data across three spatial scales in the
South Llano River, Texas from collections made in 2012. Components 13 represent variance explained by micro-mesohabitat, mesohabitat, and
channel unit scale variables alone. Components 47 represent the amount of variation in sh assemblage data explained by different combinations
of the three spatial scales
number Component Variation
Percentage species
variance explained
Percentage of total
explained variance
1 Micro-mesohabitat 0.45 35.0 7.9
2 Mesohabitat 0.17 12.9 2.9
3 Channel unit 0.37 29.0 6.5
4 Micro-mesohabitat × mesohabitat 0.09 7.0 1.6
5 Mesohabitat × channel unit 0.06 4.6 1.0
6 Micro-mesohabitat × channel unit 0.06 4.8 1.1
7 Micro-mesohabitat × mesohabitat × channel unit 0.09 6.7 1.5
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
multiple spatial scales. Specically, the variables at
the micro-mesohabitat scale, such as current
velocity, depth, canopy cover, and spatial
connectedness, i.e. contiguity (LaGro, 1991;
McGarigal et al., 2012), tended to be the most
informative for explaining the variance in the sh
assemblage data. This was not unexpected given
that the majority of the samples comprised relatively
small-bodied cyprinids, juvenile centrarchids, and
percid darters and that sampling was connedtoa
single river. These species tend to be relatively
sedentary and may not exhibit large amounts of
movement (Gerking, 1959). Thus, the composition
and characteristics of their habitats at this ne
determining their occupation and abundance in
Figure 4. Hierarchical dendrogram of sh species associations in the South Llano River, Texas at the micro-mesohabitat scale (A), mesohabitat scale
(B), channel unit scale (C), the reach scale (D), and landscape (riparian buffer) scale (E) using Wards method and Euclidean distance. Distances are
measured in semi-partial R-squared with smaller distances between linkages representing higher associations between each species. Descriptions of the
abbreviations for species names can be found in Table 2.
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
this habitat. However, the micro-mesohabitat
variables would also exhibit a similar, but
potentially less pronounced, inuence on more
mobile, larger-bodied shes as these shes would
probably move between similar habitats or use a
range of habitat types over their larger home
ranges. Furthermore, coarser-scale landscape
factors, such as geology, that may inuence
habitat are not accounted for in the analysis of a
single river.
The channel unit scale accounted for
approximately 30% of the total explained variance
in the sh assemblage data, suggesting that many
of the shes are also inuenced by habitat
characteristics at the channel unit scale potentially
through ontogenetic habitat shifts or large home
ranges (Schlosser, 1991; Schlosser and Angermeier,
1995; Fausch et al., 2002). For example, many
catostomids, such as the grey redhorse, are known
to travel relatively long distances within rivers
(Lucas and Baras, 2001). In an Ozark stream
similar in size to the South Llano River, northern
hogsuckers were observed travelling 500 to 3000 m
upstream in the spring before spawning (Matheney
and Rabeni, 1995). In addition, some ndings
have suggested that home ranges of riverine sh
are much larger than previously anticipated, even
for some smaller-bodied species (Skalski and
Gilliam, 2000). This further supports the idea that
some species in South Llano River could be
exploiting relatively large sections of the river,
presumably at the channel unit or higher
spatial scale. Furthermore, there are likely to be
linkages between the variables measured at the
channel unit scale and the response of specic
micro-mesohabitat characteristics, such that the
values recorded at ner spatial scales are heavily
inuenced or determined largely by factors at
larger spatial scales.
Mesohabitat scale variables, intermediate to the
micro-mesohabitat and channel unit scales, were
not particularly useful in explaining variation in
sh assemblage structure. Furthermore, the two
largest scales, reach and landscape (riparian
buffer), were both inconclusive. Ultimately, a
CCA or variance partitioning for both the reach
Table 4. Association of the variables with each multi-species cluster (based on the results from the hierarchical cluster analysis) using the Kruskal
Wallis test. Any test that resulted with P>0.05 was not considered signicant
Scale Cluster Variable Mean rank Mean rank χ
Micro-mesohabitat Cluster 1 (BG-GRH) vs.
Cluster 2 (BTS-MS)
Width of bank* 8.9 3.2 7.2 0.007
Depth* 9.0 3.0 8.1 0.005
Contiguity* 8.9 3.2 7.2 0.007
Pool-gravel-sand* 8.7 3.4 6.3 0.012
% canopy cover* 8.6 3.6 5.5 0.019
4.3 9.6 6.3 0.012
4.6 9.2 4.8 0.028
Mesohabitat Cluster 1 (BG - RBS) vs.
Cluster 2 (BTS - GAM)
Contiguity* 10.0 4.0 8.1 0.005
Proportion of woody debris* 9.8 4.1 7.2 0.007
Pool* 9.6 4.3 6.3 0.012
3.8 8.4 4.8 0.028
Cluster 1 (BG - RBS) vs.
Cluster 3 (DI - LMB)
Proportion of boulders
3.0 6.5 3.8 0.053
Proportion of SAV
3.0 6.5 3.8 0.053
Cluster 2 (BTS - GAM) vs.
Cluster 3 (DI - LMB)
4.0 8.5 4.2 0.040
Proportion of SAV
4.0 8.5 4.2 0.040
Proportion of cobble-gravel
4.0 8.5 4.2 0.040
Channel unit Cluster 1 (BG - TLP) vs.
Cluster 2 (BTS - OTD)
Channel unit area* 2.8 8.4 6.5 0.012
Proportion of rife-cobble-gravel
10.5 4.5 7.4 0.007
Proportion of run-boulder
10.3 4.6 6.5 0.011
Proportion of run-gravel-cobble
9.8 4.9 4.8 0.027
Proportion of pool-woody-debris
9.5 5.0 4.2 0.042
Stream reach Cluster 1 (BG - RBS) vs.
Cluster 2 (BTS - OTD)
Proportion of pool-aquatic vegetation* 9.8 4.5 5.4 0.020
Proportion of rife habitat
4.8 11.2 8.1 0.005
Perimeter-area ratio
4.8 11.2 8.1 0.005
Landscape (riparian buffer) Cluster 1 (BG - RGC) vs.
Cluster 2 (LS - RBS)
Proportion of forest
5.4 10.5 4.7 0.031
Proportion of barren land* 8.7 3.3 5.4 0.021
*= Variables associated with Cluster 1.
= Variables associated with Cluster 2.
= Variables associated with Cluster 3.
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
and landscape (riparian buffer) scale could not be
performed owing to the lack of variation explained
by the CCA axes. One of the key underlying
assumptions of the hierarchical nature of river
systems is that processes at larger spatial scales
will inuence sh assemblages at smaller spatial
scales (Frissell et al., 1986). However, the variables
at the landscape (riparian buffer) scale explained
little of the observed variance in the sh samples
compared with the other scales. This is a
surprising observation considering the important
roles that riparian buffers can play in determining
the quality and composition of instream habitats,
including natural ltering mechanisms that reduce
sedimentation (Schlosser and Karr, 1981),
contribution of allochthonous matter for biota
nourishment and habitat enhancement (Hauer
et al., 2003), and reduction of bank erosion
(Beeson and Doyle, 1995). The low variance
associated with the riparian buffer scale may be a
function of the relatively low human disturbance
found within the 50 m riparian buffer. Past studies
examining sh assemblages in lotic systems at
multiple scales have shown that ner spatial scales
tend to have greater explanatory power when
the surrounding landscape is either minimally
disturbed (Debano and Schmidt, 1989; Wang
et al., 2003, 2006) or highly disturbed (Stauffer
et al., 2000; Diana et al., 2006; Gido et al., 2006;
Heitke et al., 2006). This trend may result from a
lack of variation in the predictor variables at the
landscape (riparian buffer) level caused by low
heterogeneity of land use and cover types across
the landscape mosaic (Lammert and Allan, 1999;
Johnson and Host, 2010), which was apparent in
the South Llano River catchment.
The literature provides conicting viewpoints on
which scale exerts the greatest inuence on biotic
assemblages both in minimally disturbed (Debano
and Schmidt, 1989; Wang et al., 2003, 2006) and
heavily disturbed catchments (Stauffer et al., 2000;
Diana et al., 2006; Gido et al., 2006; Heitke et al.,
2006). For instance, Esselman and Allan (2010)
found that catchments associated with low levels
of human disturbance in Belize showed more
explanatory power with landscape-scale variables
versus local-scale factors. Their ndings suggest
that some undisturbed catchments may have
strong enough natural gradients to overcome
local-scale inuences on sh assemblages. In
contrast, Heitke et al. (2006) found a correlation
between disturbance in a land-use buffer and index
of biotic integrity (IBI) scores in agriculturally
dominated catchments in Iowa. These ndings
illustrate the lack of understanding of the impacts
of landscape heterogeneity and the roles they play
in different regions of the world (Johnson and
Host, 2010), and may indicate a high degree of
system specicity. Moreover, results can vary
between studies in the same system with slight
alterations to the data resolution and study design
(Allan, 2004; Brewer et al., 2007). For example, an
assessment of the biotic integrity of sh and
macroinvertebrate assemblages of the River Raisin
in south-eastern Michigan found that instream
habitat variables were more informative than
those from scales at the subcatchment level
(Lammert and Allan, 1999). This was in contrast
to a previous study on the same catchment which
found that larger-scale land-use variables were
more informative (Roth et al., 1996). The major
differences between these two studies stemmed
from sampling design and spatial scope. The latter
study included a larger area of subcatchment
and more widely spaced sampling sites, which
potentially introduced greater amounts of variance
or contrast among the variables at the subcatchment
scale (Lammert and Allan, 1999).
The ability to survey available instream habitat
rapidly has produced several tangible benets
beyond identifying the scales at which habitat
variables act to structure local sh assemblages.
The habitat inventory generated in this study is
allowing the identication and assessment of
release locations for hatchery-reared Guadalupe
bass. It is also being used to augment efforts to
estimate the population size and demographic
patterns of Guadalupe bass and other species of
conservation concern in the South Llano River.
The approach is currently being applied in
continuing studies in other Texas rivers to assess
instream habitat availability and evaluate its effect
on recruitment and population size for a variety of
species of conservation concern, such as alligator
gar Atractosteus spatula and blue sucker Cycleptus
elongatus. The side-scan sonar and aerial imagery,
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
augmented with on-the-ground surveys, have also
proved invaluable for planning a large-scale bank
and channel restoration project and public
demonstration area at the state park located within
the South Llano River catchment. The surveys also
serve as a pre-restoration assessment of both the
habitat and the sh assemblages associated with the
river in the impaired portions of the stream within
the state park and those near-natural portions of the
stream that can serve as potential endpoints for
dening success of restoration activities.
The effectiveness of stream restoration efforts
is too often limited by a lack of criteria for dening
and assessing success (Kondolf, 1995; Palmer
et al., 2005), and restoration activities are often
undertaken without a full understanding of the
desired targets for community composition and
ecological functionality (Palmer et al., 2005).
Furthermore, many restoration projects are
designed to restore relatively small portions of river
and riparian habitat without understanding the
landscape-scale factors that inuence species
occupation or use of such habitats, much less the
impact to the river system as a whole (Palmer et al.,
2005). Other projects are targeted towards a single
species of conservation concern with the assumption
that benets will be seen throughout the system.
However, the assumption that the condition of a
single species in a stream system can serve as an
indicator of the health of the entire system is rarely
evaluated. Without ecologically based criteria by
which to measure success, restoration activities are
unlikely to address meaningfully the underlying
causes of habitat degradation and result in minimal
or temporary improvements in the health of the
stream. The results of this study provide a snapshot
of how the sh assemblage of a minimally
disturbed, spring-fed stream on the Edwards
Plateau is structured in relation to instream habitat
at various scales and characteristics of its
catchment, and serves as a guide for developing
similar reference states in other systems.
The authors thank P. Borsdorf, B. Grisham, K.
Linner, D. Logue, J. Mueller, for their assistance
with eld collections and R. Stubbleeld, K.
Lopez, and T. Arsuffor their logistical assistance
in the eld. A. Kaeser provided guidance in the
collection and interpretation of side-scan sonar
data. S. K. Brewer, G. Garrett, A. Pease, and D.
Rogowski provided comments and suggestions
that greatly improved an earlier draft of this
manuscript. This research was supported by Texas
Parks and Wildlife Department through US Fish
and Wildlife Service State Wildlife Grant T-60 and
the US Geological Survey (cooperative agreement
number G11AC20436) and was conducted under
the auspices of the Texas Tech University Animal
Care and Use Committee (AUP 1106208).
Cooperating agencies for the Texas Cooperative
Fish and Wildlife Research Unit are the US
Geological Survey, Texas Tech University, Texas
Parks and Wildlife Department, US Fish and
Wildlife Service, and the Wildlife Management
Institute. Use of trade, product, or rm names is
for descriptive purposes only and does not imply
endorsement by the US Government.
Allan JD. 2004. Landscapes and riverscapes: the inuence of
land use on stream ecosystems. Annual Review of Ecology,
Evolution, and Systematics 35: 257284.
Barnhardt WA, Kelley JT, Dickson SM, Belknap DF. 1998.
Mapping the Gulf of Maine with side-scan sonar: a new
bottom-type classication for complex seaoors. Journal of
Coastal Research 14: 646659.
Beeson CE, Doyle PF. 1995. Comparison of bank erosion at
vegetated and non-vegetated channel bends. Journal of the
American Water Resources Association 31: 983990.
Benda L, Poff NL, Miller D, Dunne T, Reeves G, Pess G,
Pollock M. 2004. The network dynamics hypothesis: how
channel networks structure riverine habitats. Bioscience 54:
Benke AC. 1990. A perspective on America vanishing streams.
Journal of the North American Benthological Society 9:
Borcard D, Legendre P, Drapeau P. 1992. Partialling out the
spatial component of ecological variation. Ecology 73:
Bouchard J, Boisclair D. 2008. The relative importance of
local, lateral, and longitudinal variables on the development
of habitat quality models for a river. Canadian Journal of
Fisheries and Aquatic Sciences 65:6173.
Bowles DE, ArsufTL. 1993. Karst aquatic ecosystems of the
Edwards Plateau region of central Texas, USA a
consideration of their importance, threats to their existence
and efforts for their conservation. Aquatic Conservation:
Marine and Freshwater Ecosystems 3: 317329.
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
Brewer SK, Rabeni CF, Sowa SP, Annis G. 2007.
Natural-occurring landscape and in-channel factors
affecting the distribution and relative abundance of riverine
smallmouth bass in Missouri. North American Journal of
Fisheries Management 27: 326341.
Brune GM. 1981. Springs of Texas, Texas A and M University
Press: College Station, TX.
Chao H, Jensen AM, Han Y, Chen Y, McKee M. 2009.
AggieAir: towards low-cost cooperative multispectral
remote sensing using small unmanned aircraft systems. In
Advances in Geoscience and Remote Sensing, Jedlovec G
(ed). InTech: Rijeka, Croatia; 463489.
Debano LF, Schmidt LJ. 1989. Interrelationship between
watershed condition and health of riparian areas in
southwestern United States. In Practical Approaches to
Riparian Resource Management, Gresswell RE, Barton BA,
Hashage KA (eds). US Bureau of Land Management:
Billings, MT; 4551.
Diana M, Allan JD, Infante D. 2006. The inuences of physical
habitat and land use on stream sh assemblages in
southeastern Michigan. In Landscape Inuences on Stream
Habitats and Biological Assemblages, Hughes RM, Wang L,
Seelbach PW (eds). American Fisheries Society: Bethesda,
MD; 359374.
Edwards RJ, Garrett GP, Allen NL. 2004. Aquifer dependent
shes of the Edwards Plateau region. In Aquifers of the
Edwards Plateau, Mace RE, Angle ES, Mullican III WF
(eds). Texas Water Development Board: Austin, TX;
Esselman PC, Allan JD. 2010. Relative inuences of catchment
and reach-scale abiotic factors on freshwater sh
communities in rivers of northeastern Mesoamerica.
Ecology of Freshwater Fish 19: 439545.
Fausch KD, Torgersen CE, Baxter CV, Li HW. 2002.
Landscapes to riverscapes: bridging the gap between
research and conservation of stream shes. Bioscience 52:
Fitzpatrick FA, Scudder BC, Lenz BN, Sullivan DJ. 2001.
Effects of multi-scale environmental characteristics on
agricultural stream biota in eastern Wisconsin. Journal of
the American Water Resources Association 37: 14891507.
Frissell CA, Liss WJ, Warren CE, Hurley MD. 1986. A
hierarchical framework for stream habitat classication:
viewing streams in a watershed context. Environmental
Management 10: 199214.
Garrett GP, Edwards RJ, Price AH. 1992. Distribution and
status of the Devils River Minnow, Dionda diaboli.
Southwestern Naturalist 37: 259267.
Gerking SD. 1959. The restricted movement of sh
populations. Biological Reviews 34: 221242.
Gido KB, Propst DL. 1999. Habitat use and association of
native and nonnative shes in the San Juan River, New
Mexico and Utah. Copeia 1999: 321332.
Gido KB, Falke JA, Oakes RM, Hase KJ. 2006. Fish-habitat
relations across spatial scales in prairie streams. In
Landscape Inuences on Stream Habitats and Biological
Assemblages, Hughes RM, Wang L, Seelbach PW (eds).
American Fisheries Society: Bethesda, MD; 265286.
Gorman OT. 1988. An experimental study of habitat use in an
assemblage of Ozark minnows. Ecology 69: 12391250.
Hauer FR, Dahm CN, Lamberti GA, Standford JA. 2003.
Landscapes and Ecological Variability of Rivers in North
America: Factors Affecting Restoration Strategies, Bethesda,
MD: American Fisheries Society.
Heilman JL, McInnes KJ, Kjelgaard JF, Owens MK,
Schwinning S. 2009. Energy balance and water use in a
subtropical karst woodland on the Edwards Plateau, Texas.
Journal of Hydrology 373: 426435.
Heitke JD, Pierce CL, Gelwicks GT, Simmons GA, Siegwarth
GL. 2006. Habitat, land use, and sh assemblage
relationships in Iowa streams: preliminary assessment in an
agricultural landscape. In Landscape Inuences on Stream
Habitats and Biological Assemblages, Hughes RM, Wang L,
Seelbach PW (eds). American Fisheries Society: Bethesda,
MD; 287303.
Heitmueller FT. 2009. Downstream trends of alluvial
sediment composition and channel adjustment in the
Llano River Watershed, central Texas, USA: the roles of
a highly variable ow regime and a complex lithology.
PhD dissertation, University of Texas, United States of
Hubbs C. 1995. Springs and spring runs as unique aquatic
systems. Copeia 1995: 989991.
Hubbs C, Edwards RJ, Garrett GP. 1991. An annotated
checklist of the fresh-water shes of Texas, with keys to
identication of species. Texas Journal of Science 43:156.
Hubbs C, Edwards RJ, Garrett GP. 2008. An annotated
checklist of the freshwater shes of Texas, with keys to
identication of species, 2nd edn. Texas Academy of
Science. Available online http://texasacademyofscience.
org/ [accessed 20 September 2011].
Hughes RM, Larsen DP, Omernik JM. 1986. Regional
reference sites a method for assessing stream potentials.
Environmental Management 10: 629635.
Johnson LB, Host GE. 2010. Recent developments in
landscape approaches for the study of aquatic ecosystems.
Journal of North American Benthological Society 29:4166.
Jolliffe IT. 1973. Discarding variables in a principal component
analysis II: real data. Applied Statistics 22:2131.
Jolliffe IT. 1986. Principal Component Analysis,
Springer-Verlag: New York, NY.
Kaeser AJ, Litts TL. 2010. A novel technique for mapping
habitat in navigable streams using low-cost side scan sonar.
Fisheries 35: 163174.
Khattree R, Naik DN. 2000. Multivariate Data Reduction and
Discrimination with SAS Software, SAS Institute Inc.: Cary,
Kondolf GM. 1995. Five elements for effective evaluation of
stream restoration. Restoration Ecology 3: 133136.
LaGro J. 1991. Assessing patch shape in landscape mosaics.
Photogrammetric Engineering and Remote Sensing 57:
Lammert M, Allan JD. 1999. Assessing biotic integrity of
streams: effects of scale in measuring the inuence of
land use/cover and habitat structure on sh and
macroinvertebrates. Environmental Management 23:257270.
Linam G, Kleinsasser L, Mayes B. 2002. Regionalization of the
Index of Biotic Integrity for Texas Streams. Texas Parks and
Wildlife Department, River Studies Program, Austin, TX.
Lucas MC, Baras E. 2001. Migration of Freshwater Fishes,
Blackwell Science: Oxford.
Matheney IV, Rabeni CF. 1995. Patterns of movement and
habitat use by Northern Hogsuckers in an Ozark stream.
Transactions of the American Fisheries Society 124: 886897.
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
McCune B, Grace JB. 2002. Analysis of Ecological
Communities. MjM Software Design: Glendeden Beach, OR.
McGarigal K, Cushman SA, Stafford S. 2000. Multivariate
Statistics for Wildlife and Ecology Research. Springer: New
York, NY.
McGarigal K, Cushman SA, Ene E. 2012. FRAGSTATS v4:
Spatial pattern analysis program for categorical and
continuous maps. University of Massachusetts, Amherst, MA.
Murdock SH, White S, Hoque MN, Pecotte B, You X, Balkan
J. 2002. The Texas challenge in the 21st century: implications
of population change for the future of Texas. Department of
Rural Sociology Technical Report 20021, Texas A and M
University, College Station, TX.
Naiman RJ, Magnuson JJ, McKnight DM, Stanford JA. 1995.
The Freshwater Imperative: A Research Agenda. Island Press:
Washington, DC.
Noon BR. 1981. The distribution of an avian guild along a
temperate elevational gradient: the importance and expression
of competition. Ecological Monographs 51:105124.
Ott RL, Longnecker M. 2008. An Introduction to Statistical
Methods and Data Analysis. Brooks-Cole: Belmont, CA.
Palmer MA, Bernhardt ES, Allan JD, Lake PS, Alexander G,
Brooks S, Carr J, Clayton S, Dahm CN, Follstad Shah J,
et al. 2005. Standards for ecologically successful river
restoration. Journal of Applied Ecology 42: 208217.
Palmer MA, Menninger HL, Bernhardt E. 2010. River
restoration, habitat heterogeneity and biodiversity: a failure
of theory or practice? Freshwater Biology 55: 205222.
Poff NL. 1997. Landscape lters and species traits: towards
mechanistic understanding and prediction in stream
ecology. Journal of the North American Benthological
Society 16: 391409.
Raven PJ, Holmes NT, Vaughan IP, Dawson FH, Scarlett P.
2010. Benchmarking habitat quality: observations using
River Habitat Survey on near-natural streams and rivers in
northern and western Europe. Aquatic Conservation:
Marine and Freshwater Ecosystems 20:1330.
Roth NE, Allan JD, Erickson DE. 1996. Landscape inuences
on stream biotic integrity assessed at multiple spatial scales.
Landscape Ecology 11: 141156.
SAS Institute Inc. 2008. SAS/STAT 9.2 Users Guide. SAS
Institute: Cary, NC.
Schlosser IJ. 1991. Stream sh ecology: a landscape perspective.
American Institute of Biological Science 41: 704712.
Schlosser IJ, Angermeier PL. 1995. Spatial variation in
demographic processes of lotic shes: conceptual models,
empirical evidence, and implications for conservation.
American Fisheries Society Symposium 17: 392401.
Schlosser IJ, Karr JR. 1981. Water quality in agricultural
watersheds: impact of riparian vegetation during base ow.
Journal of the American Water Resources Association 17:
Skalski GT, Gilliam JF. 2000. Modeling diffusive spread in a
heterogeneous population: a movement study with stream
sh. Ecology 81: 16851700.
Stauffer JC, Goldstein RM, Newman RM. 2000. Relationship
of wooded riparian zones and runoff potential to sh
community composition in agricultural streams. Canadian
Journal of Fisheries and Aquatic Sciences 57: 307316.
Ter Braak CJF. 1986. Canonical correspondence analysis: a
new eigenvector technique for multivariate direct gradient
analysis. Ecology 67: 11671179.
Walsh CJ, Roy AH, Feminella JW, Cottingham PD, Groffman
PM, Morgan RP II. 2005. The urban stream syndrome:
current knowledge and the search for a cure. Journal of the
North American Benthological Society 24: 706723.
Ward JH. 1963. Hierarchical grouping to optimize an objective
function. Journal of the American Statistical Association 58:
Wang LZ, Lyons J, Kanehl P, Gatti R. 1997. Inuences of
watershed land use on habitat quality and biotic integrity in
Wisconsin streams. Fisheries 22:612.
Wang L, Lyons J, Rasmussen P, Seelbach P, Simon T, Wiley
M, Kanehl P, Baker E, Niemela S, Stewart PM. 2003.
Watershed, reach, and riparian inuences on stream sh
assemblages in the Northern Lakes and Forest Ecoregion,
USA. Canadian Journal of Fisheries and Aquatic Sciences
60: 491505.
Wang L, Seelbach PW, Hughes RM. 2006. Introduction to
landscape inuences on stream habitats and biological
assemblages. In Landscape Inuences on Stream Habitats
and Biological Assemblages, Hughes RM, Wang L,
Seelbach PW (eds). American Fisheries Society: Bethesda,
MD; 124.
Wehrly KE, Wiley MJ, Seelbach PW. 2006. Inuence of
landscape features on summer water temperatures in lower
Michigan streams. In Landscape Inuences on Stream
Habitats and Biological Assemblages, Hughes RM, Wang L,
Seelbach PW (eds). American Fisheries Society: Bethesda,
MD; 113127.
Wiens JA. 2002. Riverine landscapes: taking landscape ecology
into the water. Freshwater Biology 47: 501515.
Copyright #2015 John Wiley & Sons, Ltd. Aquatic Conserv: Mar. Freshw. Ecosyst. (2015)
... In this study, we used side-scan sonar to classify substrate data and explain variability in fish-catch rates at multiple spatial scales. Though many studies have used side-scan sonar to support fisheries research, few have directly examined relationships among the presence of fish species and habitat at different spatial scales using sonar data (Cheek et al., 2016). ...
... Our current study supports side-scan sonar's use in the management, restoration, and conservation of fish species (Cheek et al., 2016;Flowers & Hightower, 2013;Goclowski et al., 2013). Relatively rapid and cost-effective side-scan sonar data can be used in various management applications, such as: identifying areas in which stocked fish are most likely to be successful; determining what river segments would benefit most from habitat-improvement projects; determining the most effective scale for habitat-improvement projects; and generating predictive models for surveying areas expected to harbour target species. ...
Physical habitat is crucial for structuring local fish assemblages. Understanding habitat structure is important for fish management and conservation. Herein, we assessed whether sonar‐derived substrate data could explain spatial variation in species‐catch rates. Using a side‐scan sonar, we mapped the substrates of two non‐wadeable rivers in Illinois, USA and conducted standardized fish surveys at 40 sites over a 3‐year period. We used four fish species from lentic and lotic guilds, with each guild represented by a large piscivore and small insectivore. For each of the 40 sites, we characterized substrate composition at five spatial scales (0.1, 0.5, 1, 2, and 5 km) and used linear regression to explain site variations in species abundance or biomass. We hypothesized that larger spatial scales would better explain the catch rates of large species, and that biomass would be better explained than abundance. The proportion of variance in fish‐catch rates, explained by substrate composition, varied greatly (0.02 ≤ adjR2 ≤ 0.74, mean adjR2 = 0.38) with respect to species, spatial scales, and predictors used. However, we did not observe a consistent relationship between body size and the most relevant scale. Species biomass was more closely related to substrate composition than was species abundance and the best models selected based on AICc reached an average adjR2 of 0.49 (0.25–0.74) across the four species, compared with that of 0.43 (0.20–0.64) obtained via the best abundance‐based models. We conclude that substrate data obtained using side‐scan sonar are useful for improving our understanding of river fish ecology.
... Although in this study we decided to focus on density or presence/absence of a single species of interest as the response variables, the same type of study can be carried out with a broader focus on fish communities as a multivariate response variable, or using fish biomass or species richness as response variables of interest. These other response variables have been used to good effect in recent studies (e.g., Cheek et al., 2016;Baldigo et al., 2019) and their use in future studies in Krycklan and other boreal catchments could shed further light on the factors controlling boreal fish communities. ...
... In order to explore the mechanisms responsible for the observed correlations, an approach utilizing a more continuous monitoring of the entire riverscape including fish movements through the system, is a key next step. Continuous map-derived landscape-scale data is already available, and can be used with some success to predict variation in chemistry Ågren et al., 2014), but linkages to biota still require a combination of on-the-ground surveys and remote-sensing techniques-excitingly, technologies now available are opening up a new world of possibilities for working within stream networks at multiple spatial scales (e.g., Johnson and Host, 2010;Cheek et al., 2016). ...
Full-text available
We used the distribution of stream-dwelling brown trout (Salmo trutta) in a 67 km2 boreal catchment to explore the importance of environmental organizing factors at a range of spatial scales, including whole-catchment characteristics derived from map data, and stream reach chemical and physical characteristics. Brown trout were not observed at any sites characterized by pH < 5.0 during the spring snowmelt episode, matching published toxicity thresholds. Brown trout distributions were patchy even in less acidic regions of the stream network, positively associated with glaciofluvial substrate and negatively associated with fine sand/silty sediments. A multivariate model including only whole-catchment characteristics explained 43% of the variation in brown trout densities, while models with local site physical habitat characteristics or local stream chemistry explained 33 and 25%, respectively. At the stream reach scale, physical habitat apparently played a primary role in organizing brown trout distributions in this stream network, with acidity placing an additional restriction by excluding brown trout from acidic headwater streams. Much of the strength of the catchment characteristics-fish association could be explained by the correlation of catchment-scale landscape characteristics with local stream chemistry and site physical characteristics. These results, consistent with the concept of multiple hierarchical environmental filters regulating the distribution of this fish species, underline the importance of considering a range of spatial scales and both physical and chemical environments when attempting to manage or restore streams for brown trout.
... More recently, stationary multi-beam sonars have also been used to monitor fish movements in narrow waterbodies, such as rivers [19]. As another method, side-scan sonars produce detailed image from both sides of a vessel and are, therefore, becoming common for habitat and mussel-bed mapping [16,[20][21][22][23]. Sturgeon population estimation methods typically include netting, mark-recapture, genetic population structure analysis, and occasionally video surveys [8,9,24,25] but side-scan sonar methods have also been used with success (e.g., [26][27][28]). ...
... Side-scan imaging using consumer-grade fish finders is becoming a popular tool for various ecological research applications such as fish habitat or mussel bed mapping [13,16,[20][21][22][23]. Here we show that a consumer-grade side-scan sonar can be used to produce a mosaicked image of largebodied fish, such as Sturgeon, close to the bottom. ...
Full-text available
In 1979, the Shortnose Sturgeon (Acipenser brevirostrum) population of the Saint John River, New Brunswick, was estimated at 18,000 ± 5400 individuals. More recently, an estimate of 4836 ± 69 individuals in 2005, and between 3852 and 5222 individuals in 2009 and 2011, was made based on a single Shortnose Sturgeon winter aggregation in the Kennebecasis Bay of the Saint John River, a location thought to contain a large proportion of the population. These data, in combination with the Saint John River serving as the sole spawning location for Shortnose Sturgeon in Canada prompted a species designation of "Special Concern" in 2015 under Canada's Species at Risk Act (SARA). A three-decade span of scientific observations amplified by the traditional knowledge and concerns of local indigenous groups have pointed to a declining population. However, the endemic Shortnose Sturgeon population of the Saint John River has not been comprehensively assessed in recent years. To help update the population estimate, we tested a rapid, low-cost side-scan sonar mapping method coupled with supervised image classification to enumerate individual Sturgeon in a previously undescribed critical winter location in the Saint John River. We then conducted an underwater video camera survey of the area, in which we did not identify any fish species other than Shortnose Sturgeon. These data were then synchronized with four years of continuous acoustic tracking of 18 Shortnose Sturgeon to produce a population estimate in each of the five identified winter habitats and the Saint John River as a whole. Using a side-scan sonar, we identified > 12,000 Shortnose Sturgeon in a single key winter location and estimated the full river population as > 20,000 individuals > ~40 cm fork length. We conclude that the combined sonar/image processing method presented herein provides an effective and rapid assessment of large fish such as Sturgeon when occurring in winter aggregation. Our results also indicate that the Shortnose Sturgeon population of the Saint John River could be similar to the last survey estimate conducted in the late 1970s, but more comprehensive and regular surveys are needed to more accurately assess the state of the population.
... Nevertheless, an approach to understanding the longitudinal use of habitats is to correlate fish traits with local-habitat variables and verify if the EM patterns repeat at broader scales from headwaters to mouth reaches. Since streams are locally structured in mesohabitats with influences on fish distributions (Cheek et al., 2016;Wolff & Hahn, 2017), this approach contributes to associating habitat categories to supposed morphological archetypes that most reside in those habitats. ...
Full-text available
Fish ecomorphological analyses often deal with several traits simultaneously, making it challenging to interpret general patterns, and have not addressed the longitudinal component of streams in morphology‐habitat relationships. We identified ecomorphotypes of fishes and correlated their morphological variation with food, structural/hydrological variables, and abundances to detect their links with diet and local‐scale habitat categories. Eighteen morphometric traits were obtained from specimens collected by electrofishing in a coastal Atlantic stream in Brazil. To test for morphology‐habitat associations, headwaters, middle, and mouth reaches were divided into shorter sampling sites, with the middle reach being classified into riffle, run, and pool mesohabitats. Multivariate analyses highlighted the morphological variations and associated the form categories of 18 fish species with food and habitat categories. It resulted in four combined trophic and habitat ecomorphotypes: (1) benthic AB/lithophilic, composed of smaller to longer loricariids, detritivores, of fast‐water, and pebbly habitats; (2) benthic C/lithophilic of very fast‐waters, algal and detritus scrapers, with longer intestines than the other benthic species, wider suctorial mouths, inhabitants of headwaters, and riffles habitats; (3) nektobenthic/lithophilic with streamlined bodies, invertebrate feeders, and dwellers of faster‐water, cobbly habitats; and (4) nektonic/limnophilic composed mainly by characins, which had strong correlations with terrestrial insect consumption, and lentic/pool, sandy habitats. The morphology–environment correlations linked these ecomorphotypes to the broad habitat gradient arrayed in longitudinal and local scales. These results permit inferring the larger form patterns expected for local fish assemblages and emphasise the fish trait categorisation as a possible surrogate to reveal broader ecomorphological associations.
... Aquatic scientists are increasingly using SSS to inform a range of research efforts. This includes mapping essential habitats (Cheek et al., 2016;Holcomb et al., 2020;Kaeser & Litts, 2008;Walker & Alford, 2016), informing invasive species competition for discrete habitat patches ( habitat modeling for aquatic species (Kaeser et al., 2019;Smit & Kaeser, 2016), and mapping aquatic vegetation (Gumusay et al., 2019). Studies like these depend on the ability to easily convert output from recreation-grade sonar systems into reproducible data sets with minimal expertise in data processing. ...
Full-text available
The characterization of benthic habitats is essential for aquatic ecosystem science and management, but is frequently limited by waterbody visibility and depth. Recreation‐grade side‐scan sonar systems are increasingly used to aid scientific inquiries in aquatic environments due to their relative low‐cost, ease of operation, low‐weight, and ease of mounting on a variety of vessels. However, existing procedures and software for post‐processing these data are either limited, closed source, or fail on data from new sonar models, limiting the development of reproducible workflows. Here, we present PING‐Mapper, an open‐source and freely available side‐scan sonar post‐processing toolset for processing and mapping sonar recordings from popular Humminbird instruments. The modular software automatically: (a) decodes sonar recordings from any Humminbird system, (b) exports ping attributes from every sonar channel, (c) uses sonar sensor depth for water column removal, and (d) exports sonogram tiles and georectified mosaics. Sonar channels are processed in parallel for quick decoding and metadata extraction. Major processing wokflows, including georectification and image export, are optimized to scale with computing resources. The software has been extensively tested using data from several rivers of varying character and distribution of depths, but could also be used in estuarine and lacustrine environments. Usage of PING‐Mapper is illustrated in three case studies focused on mapping large woody debris, bathymetric mapping, and visual interpretation and mapping of substrates for selected reaches of the Pearl and Pascagoula river systems in Mississippi.
... Prior to this analysis, differences in environmental variables between midstream and downstream were tested by Mann-Whitney U tests using SPSS statistical programs (Version 20.0). In addition, correlations between different environmental variables were examined by Pearson correlation analysis (Cheek et al., 2016). Variables that showed no significant spatial differences (P > 0.05) and had high correlation with other variables (correlation coefficients > 0.7) were excluded from further analysis. ...
Full-text available
The Yarlung Zangbo River is the largest river on the Tibetan Plateau, which is considered a hotspot of biodiversity conservation efforts. In this study, the fish fauna of the middle and lower Yarlung Zangbo River was investigated, to explore the species composition and spatial patterns of fish assemblages as well as the associated environmental factors. Thirty-four sites across an elevation of 589 to 3410 m were sampled in 2015 and 2016. A total of 35 species—28 native and seven exotic—was collected. Of these species, four were described for the first time, 14 were endemic to the Yarlung Zangbo-Brahmaputra River drainage and five are considered endangered. Cluster analysis revealed significant spatial changes in fish assemblages along the longitudinal gradient. Two site-groups separated by the Yarlung Zangbo Grand Canyon were identified, corresponding to midstream and downstream. In the midstream, fish assemblages were characterized by typical plateau fish species (e.g., Schizopygopsis younghusbandi, Schizothorax oconnori, Triplophysa stenura, Triplophysa brevicauda, Schizothorax macropogon, Triplophysa tibetana and Ptychobarbus dipogon), while in the downstream fish assemblages were characterized jointly by plateau species (e.g., Schizothorax curilabiatus and Schizothorax molesworthi) and oriental species (e.g., Garra tibetana, Psilorhynchus homaloptera, Pseudecheneis sulcata, Exostoma tenuicaudata, Parachiloglanis hodgarti, Neolissochilus hexagonolepis and Garra yajiangensis). Canonical correspondence analysis (CCA) showed that altitude, channel width and the density of macroinvertebrates were the main environmental factors determining the spatial distribution of fish assemblages. This study is an important step towards understanding fish assemblages in the middle and lower Yarlung Zangbo River and will help with future efforts to protect aquatic communities and their habitats.
... Riffles provide many unique niches that intolerant species can occupy, so expectedly, sites with more riffle macrohabitat had less fish assemblage change. Cheek et al. (2015) examined the influence of habitat at different scales (micro-meso, meso, channel unit, reach) on reach-scale fish assemblages in South Llano River, TX. They found habitat at the micro-meso and channel unit scale explained the most fish assemblage variation in the data. ...
... Uno de los indicadores de la morfología del paisaje que nos otorga una visión sintética de la complejidad estructural del paisaje es el Índice de Contigüidad (Forman 1995;Cheek et al. 2016). Este índice (LaGro 1991) evalúa la forma del paisaje con base en la conectividad y cercanía entre fragmentos del mismo tipo cobertura de suelo. ...
In an era dominated by strong anthropogenic transformations, the search for new approaches to reconcile human activities and natural ecosystems is becoming increasingly important. In this sense, the concepts of "Ecosystem Services" and "Socio-Ecological Systems" are increasingly used in different scientific disciplines and are taken into account in political spheres to draw attention to the benefits that humans derive from ecosystems. These concepts lead to the study of complex links between society and nature, where the spatial dimension and landscape characteristics have a strong influence. However, few studies have been applied in a spatial approach.This thesis presents a research on the spatial dimension of Ecosystem Services, in particulary those of the temperate forests of the mountains on the periphery of Mexico City. Ecosystems have an important role to the human well-being through many ecosystem services. These Ecosystem Services are the result of complex interactions between nature and society.The Central Valley (central zone) of Mexico is a priority area for biodiversity conservation due to the high degree of endemism of the fauna and flora species present. In this zone where several protected natural areas are located, a public-private initiative is seeking to create a new conservation category that includes the management of all these areas. This initiative is known as the "Bosque de Agua" (Water Forest). However, in the "Bosque de Agua" the spatial stakes between nature and society are subject to strong anthropic pressures which are due to mining and agricultural activities and to the urban spread of large conurbations. In this space, these pressures lead to the degradation of natural ecosystems. Research is needed to analyse the spatial dimension of interactions between nature and society, by studying the synergy between Socio-Ecological Systems and Ecosystem Services.The thesis introduces the conceptual framework used to study the links between Ecosystem Services and Socio-Ecological Systems. It is considered in this research that Ecosystem Services are at the centre of the "Bosque de Agua" Socio-Ecological Ecological System.In the spatial modelling section, four Ecosystem Services were evaluated (water supply, wood supply, food supply and local climate regulation).The results of this section include a map of Ecosystem Services by identifying hotspots of Ecosystem Services. These results make explicit in space, through geographically weighted regressions, the impact of landscape characteristics on Ecosystem Services.In addition, the spatial coherence of protected natural areas and the spatial distribution of Ecosystem Services were assessed in terms of area. The objective is to create a typology of natural areas according to the surface area of protected and unprotected Ecosystem Services hotspots.Finally, a third result relates to the perception of local inhabitants of the intrinsic capacity of the "Bosque de Agua" to provide Ecosystem Services. This perceived capacity was assessed through field photo questionnaires.
Fisheries professionals frequently measure habitat type and amount, but less often measure the importance of where those habitats are located and in what combinations. We address this challenge by testing whether the individual and combined type, quantity, and location of habitat affects fish diversity in the upper Neosho River basin, Kansas, as a different approach to measuring habitat heterogeneity. Habitat type mattered in that species richness increased in areas of higher riffle density. Furthermore, variation within habitat type also influenced fish diversity; specifically, slower, shallower riffles had more species of fish. The spatial arrangement (i.e., impact of neighbor habitats) influenced fish diversity patterns in that riffle–run and riffle–glide pairings altered riffle habitat characteristics. The study illustrates a useful approach by measuring the type, amount, and arrangement of habitats to assess fish populations and could be adapted to other stream ecosystems.
Full-text available
Predicting how stream fishes may respond to habitat restoration efforts is difficult due, in part, to an incomplete understanding of how basic biological parameters such as growth and ontogenetic habitat shifts interact with flow regime and riverscape ecology. We assessed age-specific Guadalupe Bass Micropterus treculii habitat associations at three different spatial scales in the South Llano River, a spring-fed stream on the Edwards Plateau of central Texas, USA, and the influence of habitat and flow regime on growth. Substrates were classified using a low-cost side-scan sonar system. Scale microstructure was used to determine age and to back-calculate size at age. Over 65% of captured Guadalupe Bass were age-2 or age-3, but individuals ranged from 0-7 years of age. Habitat associations overlapped considerably among age classes 1-3+, but age-0 Guadalupe Bass tended to associate with greater proportions of pool and run mesohabitats with submerged aquatic vegetation. While habitat metrics across multiple scales did not have a large effect on growth, river discharge was negatively correlated with growth rates. Understanding age-specific Guadalupe Bass habitat associations at multiple scales will increase the effectiveness of restoration efforts directed at the species by assisting in determining appropriate ecological requirements of each life-history stage and spatial scales for conservation actions.
Full-text available
Relatively little information is available regarding the environmental factors influencing water temperature in streams draining low-elevation, glaciated landscapes in the upper Midwest. We used multiple regression analysis and covariance structure analysis (CSA) to identify the landscape features that influence spatial variation in mean July water temperature in 282 lower Michigan stream sites and to determine the spatial scales over which these features operate. Both modeling approaches explained from 63% to 65% of the spatial variation in stream temperatures and suggested that thermal regimes in lower Michigan are influenced by a suite of landscape factors operating at catchment and local scales. However, CSA, because it incorporated both direct and indirect effects, provided a more robust approach for identifying the relative influence of landscape features on stream temperature. Our CSA model suggested that catchment area, latitude, local groundwater inputs, local forest cover, air temperature, percent catchment agriculture, percent catchment lakes and wetlands, and percent catchment coarse-textured geology were important factors structuring spatial variation in stream temperatures. Our analysis also suggested that impacts on stream temperature from land cover/land use changes are of similar or greater magnitude as those resulting from increases in air temperature associated with global climate warming.
Full-text available
The shapes of patches in classified landscape digital images can be characterized, for each land-cover class of interest, by quantifying the spatial contiguity and clustering of the pixels within each patch. In a study of the Finger Lakes National Forest in central New York State, patch-level contiguity and clustering indices were used in conjunction with the fractal dimension to assess changes, between 1938 and 1988, in forest patch morphology. Contiguity and clustering data may prove most useful, for landcsape planning and management decision-making, however, as layers in raster geographic information systems (GISs). -Author
Although the northern hog sucker Hypentelium nigricans is widely distributed throughout the Mississippi and Ohio river basins and is both ecologically and recreationally important, much of its basic ecology is not known. We determined movement and habitat use for 25 fish in the Current River, Missouri, for 1 year using radio telemetry. Seasonal movements were recorded two or three times each week during daylight hours from January to November 1988. Diet movement and habitat use were recorded once each hour for 17 d in winter and 12 d in summer. Mean daily distance traveled was greater in summer (425 m) than in winter (276 m). Home range was greater in winter and spring (812 m) than in summer and fall (426 m). Habitat use changed seasonally from slower, deeper water and smaller substrates during winter to increasing use of taster, shallower water and larger substrates through warmer-water periods. In both seasons, fish had a consistent daily pattern, moving more during the day than at night. Diet patterns of use were distinct. In winter, fish used pool habitat with moderate flow during the day and riffle or edge habitat at night. In summer, fish used run habitat during the day and riffle or edge habitat at night. Patterns of habitat use indicated fish used one area of the river during the day to feed and another at night to rest. Fish remained in their home area during high-flow events but used flooded riparian areas where current velocities were lower. Fish moved up- or downstream short distances (mean = 497 m, N = 7) into spawning areas during late February and early March. This study emphasizes the importance of habitat diversity to accommodate this species' diel and seasonal preferences and the necessity of a connected floodplain for the fish to survive catastrophic events.
In much of statistical inference, it is assumed that a data set consists of n independent observations on one or more random variables, x, and this assumption is often implicit when a PCA is done. Another assumption which also may be made implicitly is that x consists of continuous variables, with perhaps the stronger assumption of multivariate normality if we require to make some formal inference for the PCs.
Principal component analysis has often been dealt with in textbooks as a special case of factor analysis, and this tendency has been continued by many computer packages which treat PCA as one option in a program for factor analysis—see Appendix A2. This view is misguided since PCA and factor analysis, as usually defined, are really quite distinct techniques. The confusion may have arisen, in part, because of Hotelling’s (1933) original paper, in which principal components were introduced in the context of providing a small number of ‘more fundamental’ variables which determine the values of the p original variables. This is very much in the spirit of the factor model introduced in Section 7.1, although Girschick (1936) indicates that there were soon criticisms of Hotelling’s method of PCs, as being inappropriate for factor analysis. Further confusion results from the fact that practitioners of ‘factor analysis’ do not always have the same definition of the technique (see Jackson, 1981). The definition adopted in this chapter is, however, fairly standard.