Redband Trout Habitat Assessment: Owyhee

Redband Trout Habitat Assessment:
Owyhee, Bruneau-Jarbidge, and Salmon Falls Creek
Basins
Version 2.1, February 2015
Kurt Fesenmyer and Dan Dauwalter
Trout Unlimited, Arlington, VA
0
1
Executive Summary
The US Bureau of Land Management (BLM) manages a substantial portion of public lands in the western
United States. Many of these public lands contain significant biological and cultural resources that span
not only BLM lands but private, tribal, state, and other federal lands as well. Thus, multiple-use land
management (mining, grazing, recreation) is best done with a landscape-scale approach that places local
biological and cultural values into a larger landscape context for informed land management and
conservation decision making.
The Owyhees are a high desert region in southwestern Idaho, southeastern Oregon, and northern
Nevada that is one of the most remote regions in the United States and where BLM manages a
significant portion of lands. Of the many biological resources, interior redband trout (Oncorhynchus
mykiss gairdneri) inhabit the stream and rivers of the region – the Owyhee, Bruneau, and Salmon Falls
basins. Redband trout are known for their ability to inhabit harsh desert streams, and the species is
listed as a sensitive species by Nevada and Idaho BLM offices and by state fish and wildlife agencies.
This broad-scale assessment of redband trout habitat in the Owyhees is intended to compile existing
spatial data, as well as develop new spatial data sources, that are analyzed and made available in a
framework that can aid in making strategic conservation decisions focused on redband trout habitats. In
short, this assessment briefly reviews the instream and riparian habitat needs of redband trout in desert
streams; develops new measures of stream temperature and riparian conditions for redband trout
streams; develops linkages between these new measures and redband trout distribution and
abundance; and, incorporates these new measures of habitat into Trout Unlimited’s Conservation
Success Index (CSI), as a tool that can help inform redband trout habitat conservation in the Owyhees.
Importantly, key spatial data sources used in developing the CSI, as well as key riparian condition
measures and stream temperature predictions, are made available in a web-mapping application. This
application allows the data to be continually queried based on specific land management and redband
trout conservation questions. The results of such queries facilitate strategic, place-based conservation
of a significant but sensitive aquatic resource in one of the more remote regions of the United States,
and which is managed, to a large degree, by the US Bureau of Land Management.
Recommended Citation: Fesenmyer, K., and D. Dauwalter. 2014. Redband Trout Habitat Assessment:
Owyhee, Bruneau-Jarbidge, and Salmon Falls Creek Basins. Final report to Nevada State Office, U.S.
Bureau of Land Management. Trout Unlimited, Arlington, Virginia.
Cover photos: Clockwise from top left, Trail Creek, NV (K. Fesenmyer); Deep Creek, ID redband (J.
Kellner); Owyhee River Canyon (K. Fesenmyer); Diversion with fish ladder, Cottonwood Creek, NV (D.
Dauwalter); Redband habitat, South Fork Cottonwood Creek, NV (D. Dauwalter).
2
Table of Contents
Executive Summary ................................................................................................................................. 1
Assessment Objectives ............................................................................................................................ 3
Overview of Study Area ........................................................................................................................... 3
1.1 Climate and Topography ................................................................................................................ 3
1.2 Land Ownership ............................................................................................................................. 4
Salmonids and the Fish Community ......................................................................................................... 5
2.1 Fish communities ........................................................................................................................... 5
2.2 Salmonids ...................................................................................................................................... 6
2.3 Redband Trout ............................................................................................................................... 7
Stream Temperature, Riparian Vegetation, and Links to Redband Trout ................................................ 12
3.1 Stream temperature monitoring ................................................................................................. 13
3.2 Riparian Vegetation Monitoring ................................................................................................... 15
3.3 Redband Population and Fish Community Monitoring ................................................................. 17
Conservation Success Index ................................................................................................................... 18
4.1 CSI Overview................................................................................................................................ 18
4.2 CSI Factors and Indicators ............................................................................................................ 19
4.3 CSI Results ................................................................................................................................... 20
4.4 Conservation Strategies ............................................................................................................... 26
Discussion ............................................................................................................................................. 27
Acknowledgements ............................................................................................................................... 33
Reference List........................................................................................................................................ 34
Appendix A. Stream Temperature Monitoring and Validation of the NorWeST Stream Temperature
Model ................................................................................................................................................... 42
Appendix B: Evaluate Classification of NAIP imagery as a Tool for Riparian Vegetation Classification and
Monitoring ............................................................................................................................................ 45
Appendix C. Development of a Solar Radiation Model from Terrain and Riparian Vegetation Attributes49
Appendix D: Redband trout associations with instream and riparian habitat in Salmon Falls Creek,
Nevada .................................................................................................................................................. 54
Appendix E. A model to predict smallmouth bass distribution above Hells Canyon and Below Shoshone
Falls....................................................................................................................................................... 69
Appendix F. Conservation Success Index (CSI) Application ..................................................................... 73
3
Assessment Objectives
The US Bureau of Land Management (BLM) manages over 260 million acres of public land across the
West. At one-eighth of the United States land mass, this includes some of the most significant biological
and cultural resources in the country. Multiple uses of BLM lands requires that management include a
broad-scale view of these resources to ensure that complex decisions regarding use - such as mining,
grazing, and recreation - are made using an informed process that considers the local value of specific
resources relative to those elsewhere across the landscape.
The Owyhee uplands is a general description of the high desert region in southwest Idaho, southeast
Oregon, northern Nevada. This region is one of the most remote in the United States, and it is dissected
by deep canyons, high plateaus, and distant mountain ranges. Also within this region is a significant
biological resource - the Interior redband trout (Oncorhynchus mykiss gairdneri). The species is known
for its ability to reside in harsh desert streams not typically associated with trout habitat, and is listed as
a sensitive species by the Nevada and Idaho BLM. In order to make informed decisions about land
management, especially while considering the long-term conservation of redband trout, information
about the status and trends of redband trout habitat is required. As such, the Nevada State Office of the
BLM contracted Trout Unlimited (TU) to develop relevant spatial and non-spatial data sources that could
inform a broad-scale spatial assessment of redband trout habitats and riparian conditions in the
Owyhee, Bruneau-Jarbidge, and Salmon Falls Creek basins. To do so effectively, four main objectives
were addressed:
•
•
•
•
Review key linkages between redband trout populations and instream and riparian habitat
conditions in the study area,
Develop stream temperature and riparian vegetation measures of redband trout habitat that
can be used for broad-scale assessment and monitoring,
Develop linkages between landscape-scale measures of stream temperature and riparian
vegetation and redband trout distribution and abundance,
Integrate landscape-scale measures of stream temperature and riparian vegetation into TU’s
Conservation Success Index as a tool that can help inform strategic redband trout conservation.
By addressing each of the objectives, this assessment can serve as a resource for making informed land
management and strategic habitat restoration decisions targeting redband trout conservation in the
basin. This assessment updates an older version released in 2012 with newly available redband trout
data and stream temperature forecasts. Technical appendices provide detailed information on the
methods.
Overview of Study Area
1.1 Climate and Topography
The Owyhee River, Bruneau-Jarbidge River, and Salmon Falls Creek basins drain 18,900 square miles of
deep canyons, plateaus, and high mountain ranges in southwestern Idaho, northern Nevada, and
4
southeastern Oregon. This varied topography, ranging in elevation from 10,843 to 2,185 ft, creates a
wide range of climatic conditions, including daily summer temperatures averaging from 10°C (50°F) to
25°C (77°F) and daily winter temperatures averaging from -20°C (-4°F) to 5°C (41°F). Precipitation falls in
the basins primarily in the winter and annually totals 7 to 45 inches, supporting sagebrush steppe at low
elevations and stands of conifers and aspens at high elevations.
The main Nevada tributaries of the Owyhee in Nevada are the Little Owyhee, East Fork Owyhee, and
South Fork Owyhee Rivers (Figure 1). The upper East Fork is impounded at Wild Horse Dam, a Bureau of
Reclamation facility completed in 1937; flows downstream are managed for irrigation in the Duck Valley
Indian Reservation. The main tributaries of the East Fork in Idaho are Battle, Deep, and Red Canyon
Creek. The Middle Fork Owyhee, North Fork Owyhee, and Jordan Creek flow out of Idaho to join the
mainstem river in Oregon. The lower river is impounded by Owyhee Dam, a Bureau of Reclamation
facility completed in 1932. Tributaries to the Snake River on the northeastern edge of the Owyhee
Basin, including Reynolds, Sinker, Castle, Big and Little Jacks Creeks, are included in this assessment.
The Jarbidge and Bruneau rivers originate from the north side of the Jarbidge Mountains in Nevada
(Figure 1). The main tributaries include Sheep and Mary’s Creeks. The system is undammed. The North
and South Forks of Salmon Falls Creek drain from the east side of the Jarbidge Mountains and come
together to flow into Salmon Falls Reservoir in Idaho. Salmon Falls Dam is a Bureau of Reclamation
facility completed in 1910.
Discharge in the basins is driven by snow melt and typically peaks between mid-March and mid-May.
Flows can exceed 50,000 cfs at the USGS gage at Rome, OR, approximately 100 miles upstream of the
confluence of the Owyhee with the Snake River near Nyssa, OR; 7,000 cfs at the Bruneau River gage at
Bruneau, ID; and 3,500 cfs at the Salmon Falls Creek gage near Jackpot, NV above the reservoir. Low
flows typically occur during late August and mid-September and average 150 cfs at Rome (Owyhee), 75
cfs at Bruneau (Bruneau), and 25 cfs at Jackpot (Salmon Falls Creek).
1.2 Land Ownership
Land ownership in the basins is primarily public, with management by BLM (72%), USFS (5.5%), Idaho
Department of Lands (2.8%), and Oregon Department of State Lands (1.8%) (Figure 1). Private lands
comprise 15% of the basins and include irrigated pasture and rangelands, as well as mining districts in
vicinity of Silver City (ID), South Mountain (ID), Tuscarora (NV), and Jarbidge (NV). The Duck Valley
Indian Reservation of the Shoshone and Paiute tribes occupies 2.3% of the basin. Federally designated
wilderness areas, roadless areas, or wilderness study areas cover just under 14% of the basins. These
include the Jarbidge, Big and Little Jacks Creek, Pole Creek, North Fork Owyhee, Bruneau-Jarbidge, and
Owyhee River Wilderness areas and several BLM wilderness study areas along the South Fork Owyhee in
Nevada and throughout the Owyhee plateau of Oregon; several USFS roadless areas exist in Nevada.
5
Figure 1. Study area, including Owyhee, Bruneau, and Salmon Falls Creek basins.
Salmonids and the Fish Community
2.1 Fish communities
The Owyhee, Bruneau, and Salmon Falls river basins are inhabited by a suite of native coldwater and
warmwater fishes (Meyer et al. 2013; Wallace and Zaroban 2013). Across the basins, native fishes
include: three salmonid species (Salmonidae), three sucker species (Catostomidae), seven minnow
species (Cyprinidae), and three sculpin species (Cottidae) (Table 1); only the leopard dace was not
collected during recent extensive surveys by Idaho Department of Fish and Game (IDFG; Meyer et al.
2013).
Although there are non-native salmonids (brook trout and hatchery rainbow trout) present, there are
few non-native non-game species inhabiting redband streams (Meyer et al. 2013). Most notably,
smallmouth bass (Micropterus dolomieu) have been stocked into several reservoirs in the Snake River
Basin, including Owyhee Reservoir, and the species has invaded into the main forks of the Owyhee River.
Currently, barriers to fish passage prohibit smallmouth bass invasion in the Bruneau River and in Salmon
Falls Creek above Salmon Falls Reservoir. Non-native common carp (Cyprinus carpio) and grass carp
6
(Ctenopharyngodon idella) are common in large rivers not typically inhabited by redband trout in
southern Idaho (Meyer et al. 2013).
Table 1. Native and notable non-native fish species present in the Owyhee, Bruneau, and Salmon Falls basins (Meyer et al.
2013; Wallace and Zaroban 2013). BLM status as a Type 1 (Threatened, endangered, proposed, and candidate species) or
Type 2 (Rangewide/globally imperiled) sensitive species, and state agency status as a Species of Greatest Conservation Need
(SGCN) are noted.
Common Name
Native
Interior Redband trout
Bull trout
Mountain whitefish
Bridgelip sucker
Largescale sucker
Mountain sucker
Chiselmouth
Leopard dace
Longnose dace
Speckled dace
Redside shiner
Peamouth
Northern pikeminnow
Mottled sculpin
Paiute sculpin
Shorthead sculpin
Non-native
Coastal rainbow trout
Brook trout
Brown trout
Smallmouth bass
Scientific Name
Oncorhynchus mykiss gairdneri1,2
1,2
Salvelinus confluentus
Prosopium williamsoni2
Catostomus columbianus
Catostomus macrocheilus
Catostomus platyrhynchus
Acrocheilus alutaceus
Rhinichthys falcatus2*
Rhinichthys cataractae
Rhinichthys osculus
Richardsonius balteatus
Mylocheilus caurinus
Ptychocheilus oregonensis
Cottus bairdii
Cottus beldingii
Cottus confusus
Oncorhynchus mykiss
Salvelinus fontinalis
Salmo trutta
Micropterus dolomieu
1 BLM Special Status Species
2 Species listed as SGCN in ID, OR, or NV
*Species listed as SGCN but status not yet assessed.
2.2 Salmonids
The Snake River below Shoshone Falls once had a diverse salmonid fish community. Fall and spring runs
of anadromous chinook salmon (Oncorhynchus tshawytscha) and summer run steelhead (Oncorhynchus
mykiss) once occupied the basins. Historical accounts exist for runs in the Owyhee (Jordan Creek, along
the East Fork Owyhee at the Duck Valley Indian Reservation, and as far inland as the South Fork Owyhee
near Tuscarora, NV), in the Bruneau-Jarbidge, and Salmon Falls Creek (near Contact, NV, below the
confluence of the North and South forks) (Praggastis and Williams in press; Shock et al. 2007). The
completion of Swan Falls Dam on the Snake River in 1901 blocked access of anadromous salmonids to
the Bruneau-Jarbidge and Salmon Falls Creek drainages, while Owyhee Dam eliminated the runs from
the upper Owyhee basin in 1932. Lower reaches of the Owyhee near Ontario, Oregon may have
supported runs of fall chinook and coho salmon until the completion of the first Hell’s Canyon complex
dam in 1958 (Shock et al. 2007). In addition to anadromous salmon and steelhead, the bull trout
(Salvelinus confluentus) inhabits only the Jarbidge River among all the southern tributaries to the Snake
7
River. The Jarbidge bull trout population is a Distinct Population Segment (DPS) that was listed as
Threatened under the Endangered Species Act in 1999 (USFWS 2004).
2.3 Redband Trout
The interior redband trout is an iconic species native to the high desert basins of the western US, as the
species is renowned for its persistence in harsh desert conditions (Benke 1992). The species is
designated as a sensitive species (Table 1) by Idaho and Nevada BLM, and is a Species of Greatest
Conservation Need in Idaho and Oregon (e.g., IDFG 2005). The redband trout in desert environments
have twice been petitioned for listing under the Endangered Species Act but listing was found to be not
warranted in either case (USFWS 1995; 2000).
2.3.1 Redband Trout Ecology
Native, resident redband trout reside primarily small to medium-sized, coldwater streams, where they
have been shown to mature in their first or second year of life (Schill et al. 2010). Because of the desert
environment they inhabit, these redband trout have long been thought to have an inherited genetic
ability to withstand the higher temperatures commonly observed in desert streams (Benke 1992).
Redband trout have been observed in streams with maximum daily temperatures as high as 28-29°C
(Benke 1992; Zoellick 1999). Recent research has suggested that redband trout in desert streams are
genetically distinct from montane populations, and they have genes likely to have been selected for
adaptation to local thermal environments (Narum et al. 2010). Despite genetic evidence of local
adaptation, laboratory studies have not been able to demonstrate that desert populations of redband
trout are more adapted to the thermal regime of desert streams than are montane populations
(Cassinelli and Moffitt 2010).
Redband trout abundance has been observed to be as high as 130 redband trout / 100 m2 in desert
streams (Zoellick et al. 2005), and it is typically higher in colder streams that are more shaded (Zoellick
2004; Zoellick and Cade 2006). Despite high abundance in some places only a few streams have been
shown to have any exploitation from angling, and where exploitation has been observed it has been a
only 3.4 to 5.7% of the population (Schill et al. 2007). Abundance has also been shown to be lower in
stream with higher concentrations of silty substrate and in the presence of piscivorous fish, particularly
northern pikeminnow (Ptychocheilus oregonensis) and smallmouth bass (Meyer et al. 2010).
The distribution of redband trout is typically defined to streams with mean summer stream
temperatures less than 16°C (Meyer et al. 2010). The species distribution is generally perceived to have
constricted out of higher order mainstem rivers in the Owyhee basin since the 1970’s, resulting in
decreased population connectivity (Allen et al. 1998). However, the extent of habitat occupied by
individual populations has been shown to expand and contract over time in conjunction with annual
changes in precipitation and streamflow (Zoellick 1999). Additionally, redband abundance in the Idaho
Owyhees decreased from the late 1970’s to the late 1990’s only in streams below 1,500 meters
elevation; abundances were stable or increasing at higher elevation sites (Zoellick et al. 2005). In a more
recent survey in Salmon Falls Creek (NV), redband trout densities have decreased since the early 2000’s
(see Section 3.3).
8
2.3.2 Redband Rangewide Status Assessment
In 2012, state wildlife management agencies, federal land management agencies, tribes, and other
interested agencies and organizations across the northwest produced a rangewide status assessment for
redband trout. This effort compiled various data sources, including expert opinion, to map and attribute
stream reaches with information related to redband distribution, abundance, life history, genetic status,
habitat quality, and the presence of non-native species and barriers to population connectivity (May et
al. 2012). In all, the assessment identified 2,090 stream miles of redband trout habitat within the
Owyhee, Bruneau-Jarbidge, and Salmon Falls Creek basins (Figures 2 and 3). The status assessment
highlights 18 populations or sub-populations within larger connected habitat patches as conservation
populations, designations assigned to populations with <10% introgression or that have unique genetic,
ecological, or life history traits, such as a migratory behaviors (UDWR 2000; May et al. 2012).
Figure 2. Populations of redband trout in the northern portion of the study area. For population identities, consult Table 2.
9
Across the basin, nearly 60% of populations persist at the lowest densities, 19 % have some indications
of introgression (>1%), and 34% co-occur and compete with non-native trout or bass (Table 2). All
populations exhibiting an anadromous life history no longer express it due to fish passage barriers,
mainly large dams. Large redband are known to occur in some mainstem river systems like the Owyhee
below the confluence of the East Fork and South Fork, but it is unknown if these fish express fluvial life
histories. Similarly, some redband populations occur in artificial reservoirs, but none have a known
adfluvial life history.
Figure 3. Populations of redband trout in the northern portion of the study area. For population identities, consult Table 2.
10
Owyhee basin
Table 2. Populations of redband trout in the Owyhee, Bruneau-Jarbidge, and Salmon Falls Creek basins and summary of
attributes from the Redband Rangewide Status Assessment (May et al. 20112). Populations are identified by habitat
contiguity and values are length-weighted by stream miles within populations. Metapopulations with sufficient available
habitat are identified with bolded population names; conservation populations are identified by italicized population names;
# denotes populations with subpopulations identified as conservation populations; * denotes inferred genetic status; ^
denotes populations or portions of populations with unknown densities inferred here as 100-250 fish/km.
Map ID
Population
Mi.
stream
habitat
1
Mainstem Owyhee R. # ^
317.6
Average
population
density
(fish/km)
35- 100
2
Upper Jordan Cr.^
131.0
35- 100
3
Snake R. below Swan Falls #
89.0
4
Snake R. above Swan Falls #
5
Minimum
genetic
status
Non-native
species present
Average
habitat
quality
1 - 10%
Bass (modeled)
Fair
1 - 10%
Trout
Fair
0 - 35
1 - 10%
Bass (modeled)
Poor
84.8
0 - 35
1 - 10%
Trout
Poor-fair
East Fork Owyhee R.
75.0
35- 100
1 - 10%
Trout
Fair
6
Big Jacks Cr.
51.9
35- 100
0 - 1% *
None
Fair
7
Middle Jordan Cr.^
47.1
100 - 250
0 - 1% *
Trout
Poor
8
Lower Succor Cr.^
43.0
100 - 250
0 - 1% *
None
Poor-fair
9
Dry Cr. (Oregon) ^
40.6
100 - 250
0 - 1%
None
Poor
10
South Fork Owyhee R.
37.7
0 - 35
1 - 10%
Trout
Poor
11
Red Canyon
27.3
0 - 35
1 - 10%
None
Good
12
Upper Deep Cr. (ID) ^
26.8
35- 100
0 - 1% *
None
Fair
13
Little Jacks Cr.
21.9
250 - 625
0 - 1%
None
Good
14
Sinker Cr.
20.8
0 - 35
0 - 1% *
None
Fair
15
Thomas Cr.
16.0
0 - 35
0 - 1% *
None
Fair
16
Jacks Cr. (NV)
15.5
35- 100
0 - 1% *
Trout
Fair-good
17
Columbia Cr.
15.1
0 - 35
> 20%
Trout
Fair
18
Martin Cr.
12.6
0 - 35
1 - 10%
Trout
Fair-good
19
Upper Reynolds Cr.
11.8
0 - 35
1 - 10%
Trout
Fair
20
Fawn Cr.^
9.8
35- 100
0 - 1% *
None
Fair
21
Rock Cr.
9.8
0 - 35
0 - 1% *
None
Fair
22
Mill Cr.
9.3
35- 100
0 - 1% *
None
Poor-fair
23
California Cr.
8.1
0 - 35
0 - 1% *
None
Fair-good
24
Deep Cr. (NV)
7.7
100 - 250
0 - 1% *
None
Fair
25
Jerritt Canyon
7.5
35- 100
0 - 1% *
None
Fair
26
Shoofly Cr.
6.8
0 - 35
0 - 1%
None
Fair
27
Middle Succor Cr.
6.2
0 - 35
1 - 10%
None
Poor
28
Water Pipe Canyon
4.9
0 - 35
0 - 1% *
Trout
Poor-fair
29
Upper Succor Cr.^
4.3
100 - 250
1 - 10%
None
Fair
30
Breakneck Cr.
4.2
0 - 35
0 - 1% *
Trout
Fair
31
Fourmile Cr.
4.1
0 - 35
0 - 1%
None
Fair
32
Riffle Cr.
3.7
0 - 35
0 - 1% *
None
Fair
33
Bull Run below Res.^
3.5
100 - 250
1 - 10%
Trout
Good
34
Indian Cr.
3.4
35- 100
0 - 1%
None
Poor
Salmon Falls Cr. basin
Jarbidge – Bruneau basin
Owyhee basin, continued
11
Average
population
density
(fish/km)
Minimum
genetic
status
Non-native
species present
Average
habitat
quality
Map ID
Population
Mi.
stream
habitat
35
Jump Cr.
3.3
0 - 35
10 - 20%
Trout
Fair
36
McCann Cr.
3.2
0 - 35
0 - 1% *
None
Fair
37
Mitchell Cr
2.9
0 - 35
0 - 1%
None
Fair
38
Capp Winn Cr.
2.8
0 - 35
0 - 1% *
None
Poor-fair
39
Winters Cr.
2.5
0 - 35
0 - 1%
None
Good
40
Red Cow Cr.
2.4
35- 100
0 - 1%
None
Poor
41
Upper Cow Cr.
2.2
100 - 250
0 - 1% *
None
Fair
42
Wall Cr
2.1
0 - 35
0 - 1%
None
Good
43
Wickahoney Cr.
2.1
0 - 35
0 - 1% *
None
Fair
44
Marsh Cr.
2.1
35- 100
0 - 1% *
None
Fair
45
Dip Cr.
2.1
0 - 35
0 - 1% *
None
Fair
46
Wood Gulch Cr.
2.0
0 - 35
0 - 1% *
None
Poor
47
Burns Cr.
1.6
0 - 35
0 - 1%
Trout
Poor
48
Clear Cr.
1.0
0 - 35
0 - 1%
None
Fair
49
Schmidt Cr.
0.7
100 - 250
0 - 1%
None
Fair
50
Doby George Cr.
0.4
0 - 35
0 - 1%
None
Fair
51
Frost Cr.
0.1
0 - 35
0 - 1% *
None
Poor
52
Bruneau R.# ^
234.2
35- 100
> 20%
None
Fair-good
53
Clover Cr.#
97.8
0 - 35
0 - 1% *
Trout
Fair
54
Upper Bruneau R.
97.7
35- 100
0 - 1% *
None
Fair-good
55
Sheep Cr.
53.1
0 - 35
0 - 1% *
Trout
Fair-good
56
Mary's Cr.^
20.0
0 - 35
0 - 1% *
None
Fair
57
Deadwood Cr.
9.7
0 - 35
0 - 1% *
Trout
Fair
58
Cat Cr.
4.5
0 - 35
0 - 1% *
None
Fair
59
Crab Cr.
3.5
35- 100
0 - 1%
None
Fair
60
Log Cr.
3.4
0 - 35
0 - 1% *
None
Fair
61
Columbet Cr.
3.1
0 - 35
0 - 1% *
None
Poor
62
Corral Cr.
2.5
0 - 35
0 - 1% *
None
Fair
63
Bear Cr.
2.1
0 - 35
0 - 1% *
Trout
Fair-good
64
Louse Cr.
1.8
35- 100
0 - 1% *
None
Fair
65
Mason Cr.
0.3
0 - 35
0 - 1%
None
Fair
66
Upper Salmon Falls Cr.# ^
129.0
35- 100
0 - 1% *
Trout
Fair-good
67
Shoshone Cr.# ^
51.9
35- 100
1 - 10%
Trout
Poor-fair
68
Salmon Falls Cr. below Res.
50.1
100 - 250
> 20%
Bass (observed)
Good
69
Mainstem Salmon Falls Cr.^
48.9
0 - 35
> 20%
Trout
Poor-fair
70
Upper Cottonwood Cr.
24.6
0 - 35
0 - 1% *
None
Good
71
House Cr.^
19.7
35- 100
> 20%
Trout
Fair-good
72
Lower mainstem Salmon Falls Cr.^
10.6
100 - 250
> 20%
Trout
Poor
Salmon Falls Cr. basin
12
Average
population
density
(fish/km)
Minimum
genetic
status
Non-native
species present
Average
habitat
quality
Map ID
Population
Mi.
stream
habitat
73
Cedar Cr. and Res.
9.3
35- 100
1 - 10%
Trout
Good
74
Sun Cr.
7.2
35- 100
0 - 1%
None
Poor
75
South Fork Jakes Cr.
6.4
0 - 35
0 - 1% *
Trout
Fair
76
Cottonwood Cr - lower SFC
5.0
35- 100
1 - 10%
None
Poor
77
China Cr. (SFC)^
4.9
100 - 250
> 20%
Trout
Poor
78
West Fork Deer Cr.
3.8
0 - 35
0 - 1%
None
Fair
79
Middle Fork Deer Cr.
3.8
0 - 35
0 - 1% *
None
Good
80
Dry Cr. (SFC)
3.6
35- 100
0 - 1% *
None
Poor
81
Shack Cr.
3.5
35- 100
0 - 1% *
None
Fair
82
Trout Cr. (SFC)
3.4
0 - 35
0 - 1%
None
Poor
83
East Fork Deer Cr.
2.4
0 - 35
0 - 1% *
None
Fair
84
Bull Camp Cr. (SFC)
2.3
0 - 35
0 - 1% *
None
Poor
Several researchers have identified criteria for identifying metapopulations of fish likely to persist over
time due to the increased resiliency afforded by habitat diversity and connectivity of large habitat
patches (Rieman and Dunham 2000; Hildebrand and Kershner 2000; Haak and Williams 2012).
Metapopulations are defined as populations that are sufficiently large to accommodate local extinctions
through migration and recolonization (Hanski and Simberloff 1996). When applied to this assessment,
metapopulations include redband populations with large, connected habitats (> 9 miles) occurring at
any density or moderate-sized connected habitats (6 – 9 miles) occurring at moderate to high densities
(>35 fish/km) (Haak and Williams 2012); there are 39 metapopulations in the basin. 54% of populations
occur in isolated habitats or at densities too low to warrant this designation.
Stream Temperature, Riparian Vegetation, and Links to Redband Trout
Strategic conservation decisions require information on key habitat attributes across broad landscapes.
A main objective of this assessment was to develop methods to characterize stream temperatures and
riparian vegetation conditions associated with redband trout habitats across the study area. To do this
we deployed a stream temperature monitoring network to establish baseline stream temperature data
in un-monitored streams and to validate a newly developed stream temperature model developed for
the region. As stream temperature and riparian vegetation have been shown to be important to
redband trout distribution and abundance previously (Zoellick et al. 2006; Meyer et al. 2010), we
conducted a pilot study to link these landscape-scale measures of temperature and riparian vegetation
to the distribution and abundance of redband trout, and the entire fish assemblage, to ensure they
provide meaningful measures of habitat across the landscape.
13
3.1 Stream temperature monitoring
The development of small underwater data loggers with sufficient battery and memory resources to
record hourly water temperatures has greatly facilitated stream temperature monitoring during the last
decade (Sowder and Steel 2012). The US Forest Service, Rocky Mountain Research Station (RMRS)
recently assembled a large dataset - the NorWeST Regional Database and Modeled Stream
Temperatures – of hourly August stream temperatures collected from 1993– 2011 throughout the
northwest US (Isaak et al. 2011). They used these data to develop a spatial statistical model that
predicts average August stream temperatures for every stream reach in the Columbia River Basin as a
function of climate, topography, and land cover factors. These predictions of stream temperature have
recently been released for the middle Snake River and its tributaries (Figure 4).
From 2010 to 2012, we deployed a network of 103 data loggers set to record hourly stream
temperatures year-round for up to 4 years (Figure 4). The temperature monitoring network was
designed to fill data gaps in the NorWeST database, as well as focus on BLM lands in the Owyhee and
Salmon Falls Creek basins (Appendix A). The primary intent of the temperature monitoring network in
Owyhee and Salmon Falls Creek basin monitoring locations is to provide baseline data for tracking
trends and change in stream temperatures over time. Year-round monitoring captures important
monthly and seasonal patterns in stream temperature beyond the peak annual temperatures in summer
that are typically the focus of temperature studies.
14
Figure 4. Location of temperature loggers installed in streams of the Owyhee River and Salmon Falls Creek basins. TU
monitoring locations represent loggers deployed for this assessment. NorWeST mean August temperature predictions also
shown.
The secondary intent of our temperature monitoring has been to serve as an independent dataset for
validating the NorWeST stream temperature models. This was important because the model was
developed for both montane and desert streams in the middle Snake Basin, but stream temperature
data were sparse in the Owyhee and Salmon Falls Creek basins (Appendix A). To date, stream
temperature data were downloaded from 23 loggers representing 42 summers of data. To validate the
NorWeST model, we compared observed mean August stream temperatures in 2012 and 2013 to mean
August temperatures predicted by NorWeST in 1994 (the year that closely corresponds to air
temperature and precipitation in 2012 and 2013). Predicted temperature was typically with +/- 1°C of
observed temperatures, indicating that the model accurately predicted mean August stream
temperatures. The exceptions were on mainstem rivers like the East Fork Owyhee below Wildhorse
Reservoir and the mainstem Owyhee below the confluence of the East and South Forks, where the
NorWeST models under predicted stream temperatures by 3°C (Appendix A). This error is likely induced
by a lack of data for large streams for use in model development, the alteration of expected
temperature regimes by Wildhorse Dam operations, and extensive irrigation practices on the Duck
Valley Reservation. Despite these anomalies, a majority of temperature predictions were accurate for
15
small streams representing redband trout habitat, and the NorWeST model provides an excellent
measure of stream temperatures, and suitability for redband trout, across the basin that can be used for
broad-scale assessments of stream temperatures.
3.2 Riparian Vegetation Monitoring
We developed two new spatial datasets that have immediate application for assessment of riparian
vegetation and its influence on stream shading and contribution to redband habitat quality in the
Owyhee and Salmon Falls Creek basins – a riparian vegetation classification derived from high-resolution
aerial photography and estimates of stream shading and exposure.
3.2.1 Riparian land cover classification
To assess the condition, gross composition, and
structure of riparian vegetation, we performed a
supervised classification of land cover using high
resolution (1-m2) USDA National Agriculture
Imagery Program (NAIP) photographs. High
resolution aerial imagery interpretation and
classification is increasingly used as a tool for
monitoring riparian vegetation (Booth et al.
2007). Our classification used available NAIP
imagery from 2006 (entire Owyhee basin,
Nevada), 2009 (redband streams in Owyhee
basin, Idaho), and 2010 (redband streams in
Salmon Falls Creek basin, Nevada and portions of
the Owyhee Front, Idaho) to classify vegetation
within the riparian zone as woody riparian,
herbaceous/emergent riparian, upland shrubland
or grassland, bare ground, agriculture, or open
water types (Figure 5). We did not produce
riparian mapping for portions of the study area in
Oregon, the Bruneau-Jarbidge basin, Idaho
portions of Salmon Falls Creek, or for Little and
Big Jacks Creeks, Idaho because of limited
Figure 5. Example of supervised classification of NAIP imagery
riparian restoration opportunity along mainstem (top panel) into land cover categories (bottom panel).
rivers (e.g. Owyhee River in OR) and in wilderness
areas and due to existing riparian monitoring within the range of bull trout (Bruneau-Jarbidge).
Appendix B provides additional details pertaining to the riparian classification and its interpretation.
3.2.2 Shade modeling
We characterized how our classification of riparian vegetation across the landscape contributes to
stream shading by modeling field-measured percent August solar radiation (percent of total direct solar
radiation for the month of August) as a function of woody riparian vegetation cover and local shading
16
from terrain features (e.g., canyon walls) and azimuth using a regression model (Appendix C). Because
riparian vegetation and shading can be measured across a large portion of the landscape, we were then
able to use the model to predict mean August solar radiation at all stream reaches with available data in
the study area based on percent woody vegetation (derived from NAIP imagery) and terrain (Figure 6).
Figure 6. Proportion of August solar radiation reaching the stream surface due to shading from riparian vegetation and
terrain features (e.g., canyons).
These two mapped products – riparian vegetation and percent solar radiation - provide a simple means
to monitor riparian vegetation condition and structural composition and its influence on stream shading
across a large landscape. Monitoring is possible because NAIP imagery are collected periodically (~5
years). These data will provide baseline measures for tracking changes in riparian condition over time,
and provide and indicator of stream habitat quality, that can then be used to identify potential
restoration strategies (see Section 4.4).
17
3.3 Redband Population and Fish Community Monitoring
0.6
0.0
20
40
60
80
1.0
Woody Vegetation (%)
0.0
0.2
0.4
0.6
0.8
Redband >100mm
Probability
3.3.2 Stream temperature and woody riparian vegetation links
to redband trout populations
Redband trout distribution and abundance has been linked to
stream temperature (or surrogates) and woody riparian
vegetation measured in the field (Zoellick et al. 2006; Meyer et
al. 2013). In 2013, we sampled 21 sites in the Salmon Falls
Creek (NV) to develop associations between redband trout
occurrence and abundance, and fish assemblage structure, and
models of stream temperature (NorWeST model) and remotely
sensed measures of woody riparian vegetation – two
landscape-scale measures of redband trout habtiat (see Section
3.2.1). In fact, percent woody vegetation in a 5-m stream
buffer, as classified using NAIP imagery, was the best predictor
of the presence of both small (<100mm TL) and large
(>100mm) redband trout when compared against field
measures of riparian vegetation, bank conditions, and instream
physical habitat (Figure 7; Appendix D). Likewise, mean August
stream temperature, as modeled for the NorWeST project, and
percent woody vegetation are most likely the main factors
limiting abundance of large redband trout in Salmon Falls Creek
basin streams (Figure 8; Appendix D). As expected, they are also
main drivers of fish community structure as well (Appendix D).
0.2
0.4
Probability
0.8
1.0
3.3.1 Redband population monitoring
Redband trout population trend monitoring has been ongoing in southwest Idaho jointly between IDFG
and Idaho BLM since the 1970’s. As noted earlier, Zoellick et al. (2005) reported decreasing population
trends at lower elevations, but stable to increasing population trends at higher elevations. Resampling
some of these sites in 2010, IDFG found no significant trend in abundance or change in distribution
(Butts et al. 2011; Kozfkay et al. 2011; Butts et al. 2013). Our redband trout trend monitoring in Salmon
Falls Creek (NV) showed a 5% annual decline, on average, across eight sites that were sampled in 2003
by IDFG; two of these sites were dry. An additional two sites were inundated by beaver ponds, and an
additional six sites were not sampled due to a lack of land access (Appendix D). While trends at
individual sites are hard to detect and trend estimates may not
accurately index trends in local populations as a whole,
Redband <100mm
population monitoring across a suite of sites allows for
generalizations to be made regarding regional trends
(subbasin) or trends in specific parts of the study basins
pertaining to large-scale changes in land use or broad-scale
restoration programs (Dauwalter et al. 2010).
20
40
60
80
Woody Vegetation (%)
Figure 7. Influence of woody riparian
vegetation, measured from classified NAIP
imagery, on probability of redband trout
occurrence in streams. Circles represent
observed presence (1) or absence (0) at 21
sites.
18
Figure 8. Redband trout abundance relationships with woody riparian vegetation and stream temperature (mean August) in
Salmon Falls Creek, NV. The pseudo R-squared (R1) is a measure of model fit ranging from 0 to 100.
3.3.3 Smallmouth bass distribution modeling
Non-native smallmouth bass have invaded some streams in the Owyhee, Bruneau, and Salmon Falls
streams, and their presence has been shown to reduce the likelihood of redband trout presence in
streams (Meyer et al. 2010). Although some records of smallmouth bass exist where fish surveys have
been completed, their current distribution and invasion potential is not completely understood. To
address this uncertainty and better understand the species’ invasion potential, we developed a model
that predicts smallmouth bass occurrence as a function of landscape-scale environmental variables in
the Lower Snake Basin (Appendix E). Stream temperature, stream gradient, stream size, reservoir, and
land conversion all influenced smallmouth bass distribution, and when the model was applied to the
study area smallmouth bass were most likely to occur, or capable of invading habitats, in the mainstem
and main forks of the Owyhee basin and lower Salmon Falls Creek. Although the model suggest there to
be some invasion potential in the lower Bruneau River and upper Salmon Falls Creek, barriers to fish
passage currently limit the species’ invasion in those systems.
Conservation Success Index
4.1 CSI Overview
Trout Unlimited developed the Conservation Success Index (CSI) to provide a strategic, landscape-scale
planning tool for cold-water conservation that is focused on watersheds (see Williams et al. 2007). The
CSI is a compilation and assessment of spatial information related to a species’ distribution, populations,
habitats, and future threats. It summarizes spatial (GIS) data within watersheds related to a broad suite
of population metrics, anthropogenic stressors, and environmental conditions and assigns the
summaries a categorical score (5 through 1, reflecting exceptional through poor condition) based on the
best scientific understanding of the significance of the particular data on aquatic species persistence and
effects on habitat quality. The data considered are not intended to comprise a comprehensive list of
factors affecting instream habitat or aquatic species; rather, they include factors that exist as broadly
19
available, mapped data. This watershed data “summary and scoring” approach is a standard
conservation planning tool and is similar to products developed by other land management agencies
and conservation partners, including the Watershed Condition Framework developed by the US Forest
Service and the NFHAP Data System created by the National Fish Habitat Partnership. The watershed
units for this CSI analysis are the subwatershed or 12 digit hydrologic unit code watershed (HUC12) and
subbasin or 8 digit hydrologic unit code watershed; these units average 15,000 acres and 1.2 million
acres, respectively.
4.2 CSI Factors and Indicators
The CSI has a hierarchical structure in which each spatial data source is summarized and interpreted as
individual factors, suites of similar factors are rolled up into thematic indicators, and indicators are
combined into four simple groups – Rangewide Conditions, Population Integrity, Habitat Integrity, and
Future Security (Figure 9). Rangewide Conditions indicators compare current and historical distribution
patterns at multiple spatial scales. Population Integrity indicators are based on data directly from the
Redband Rangewide Status Assessment (May et al. 2012), including population density, habitat patch
size, genetic status, competition with non-native species, and life history diversity. Habitat Integrity
indicators account for factors related to instream habitat (such as road densities), temperature
(including NorWeST stream temperature models and shade models from this study), watershed
connectivity (as affected by barriers data from the status assessment and other sources), water quality
(such as land use), and flow regime (such as water storage and conveyance infrastructure). Threats to
subwatersheds are accounted for in the Future Security indicators, which address conversion risk (exurban development), resource extraction (development of renewable and non-renewable resources),
climate change, sedimentation (landslide and erosion risk), and land stewardship (public ownership and
protected status). All indicators can be further organized to identify conservation strategies that may be
appropriate in watersheds given the pattern of species occurrence, habitat condition, and likely future
threats, providing a landscape-scale blueprint for management efforts on public and private lands.
20
Figure 9. A schematic of CSI structure. Data are summarized and scored within factors and then organized into Indicators
and Groups. Each Indicator may have multiple factors – for example, the Climate Change Indicator is comprised of factors
related to winter precipitation regime change, stream temperature change, annual precipitation volume change, base flow
index, drought risk, and fire regime change risk.
The Climate Change indicator includes several factors which assess the vulnerability of aquatic habitats
to climate change based on a composite analysis of six risk factors: changes in precipitation and flow
regime based on winter precipitation type (snow vs. rain); increasing August stream temperatures based
on temperature models for 2040 within the NorWeST dataset; changes in flow volume based on total
annual precipitation volume models for 2050; ability of watersheds to buffer changes in flow through
base flow condition (groundwater vs. surface flows); heat-related moisture loss measured through the
Palmer Drought Severity Index; and changes in fire regime associated with earlier spring warming and
anticipated lengthening of fire seasons.
CSI methods, including factors and interpretive scoring rules, are outlined in detail in Appendix F.
4.3 CSI Results
The following brief summaries describe the broad patterns of the data summary and scoring for the CSI
as applied to the Owyhee, Bruneau-Jarbidge, and Salmon Falls Creek basins. While the maps provide
21
one way to compile, summarize, and interpret the data across the landscape, the underlying data are
available in a web-mapping application available online. This application allows one to view specific data
sources and overlay them in specific ways to answer specific questions regarding the populations,
habitats, and future threats, and, thus, facilitating informed conservation decision-making over time.
Rangewide Conditions scores are generally high because redband trout are persisting within their
historical range at the subwatershed and subbasin scale (Figure 10). The majority of redband trout
range contraction in the basins has occurred in the tributaries to the East Fork and South Fork Owyhee.
There, and in the southern-most tributaries to Salmon Falls Creek, populations persist in small,
disconnected habitat patches. Most populations persist in higher order streams outside of the highest
elevation headwater reaches of the Bull Run and Jarbidge Mountains, where many populations occur in
first order streams.
Figure 10. CSI Rangewide Condition scores representing distribution across subbasins (HUC8), distributions within subbasins,
distribution across subwatersheds (HUC12), and distribution within first order streams.
Population integrity indicator scores reflect the data within the redband trout rangewide status
assessment and are high to moderate across much of the Owyhee basin (Figure 11). Redband
22
population densities generally correspond to elevation, with highest densities in the highest elevations.
The habitat extent of connected populations is greatest for populations that include mainstem river
distributions, including the Owyhee River and Salmon Falls Creek above the reservoirs. However, many
of those same mainstem rivers, including the Owyhee below Rome, South Fork Owyhee in Nevada,
mainstem Snake, and lower reaches of the mainstem Bruneau River, have lower genetic status scores,
reflecting a history of stocking. Similarly, mainstem rivers, except for the Bruneau River, also have nonnative species present which compete with or prey upon redband trout. Life History Diversity scores are
uniformly moderate, reflecting the loss of an anadromous steelhead life history across all three basins.
Additional data related to redband trout density, genetic status, occurrences of smallmouth bass and
brook trout, and presence of fluvial life histories across the basins will further refine these results as that
information becomes available. Much of the population density and genetic status attributes are
inferred for redband across the basins. No data for characterizing populations are currently available
from the Duck Valley Indian Reservation.
Figure 11. CSI Population Integrity scores representing population density, habitat extent occupied, genetic status,
competition, and life history diversity.
Habitat integrity scores are generally moderate to high across the Owyhee basin (Figure 12). Moderately
low and low watershed condition scores are associated with high ratios of miles of road within riparian
23
and floodplain zones to miles of stream, reflecting that human development in the basin tends to occur
along watercourses and that the topography confines roads to level terrain near streams.
Subwatershed-scale road densities are generally low. Temperature scores are highest in high elevation
subwatersheds with intact riparian vegetation; low to moderate scores occur on many of the largest
systems, with substantial portions of each subbasin 303d listed for temperature. Watershed
connectivity scores are generally moderately high, but likely inflated due to the lack of a comprehensive
barrier survey across the basins. Agricultural areas along the mainstem Snake, along Jordan Creek, in
Duck Valley Indian Reservation on the East Fork Owyhee, along the upper South Fork Owyhee, and along
Salmon Falls Creek upstream of Jackpot and widespread 303d listing for toxins (arsenic, mercury, zinc,
and manganese), dissolved oxygen, and E. coli cause low-scoring exceptions to overall high water quality
scores. Flow regime scores are low in association with extensive canal networks and diversions for flood
irrigation of pastures in low and mid-elevation portions of the basin and associated with reservoirs such
as Wild Horse, Owyhee, and Salmon Falls dams, and numerous smaller structures.
Figure 12. CSI Habitat Integrity scores representing watershed conditions, temperature, watershed connectivity, water
quality, and flow regime.
24
The CSI analysis of future security suggests that redband in the Owyhee are at moderate risk (Figure 13).
The basins have a high future security for land conversion, with minimal threat of urban or suburban
development. The threat of resource extraction is moderate. Renewable energy development areas
exist in the upper South Fork Owyhee and Shoofly Creek areas (geothermal), in the western portion of
the basin in the headwaters of Crooked and Antelope Creeks and along the Owyhee Front (wind), and
across plateau areas (solar). Additionally, substantial mining claims exist in the vicinity of the Tuscarora,
Jarbidge, and Silver City mining districts and along Salmon Falls Creek above Jackpot. The risk of
increased sedimentation from riparian soil erosion and severe wildfire events is high across the basins
due to inherently erosive soils and altered fire regimes. The Owyhee basin has a range of land
stewardship scores, with high scores related to wilderness and wild and scenic river designations along
several canyonlands and in the Jarbidge Mountains, but moderate scores elsewhere reflecting
substantial public lands in general management status.
Figure 13. CSI Future Security scores representing land conversion, resource extraction, climate change, sedimentation, and
land stewardship.
The CSI considers the effects of climate change on redband in the basins (Figure 14). Precipitation
regime change risk is high across the basins as mid-elevation portions of the basin outside of the highest
25
elevations of the War Eagle, Bull Run, Independence, and Jarbidge Mountains are forecast to transition
from snow to rain or mixed winter precipitation regimes by 2050 (Figure 14 A). The volume of annual
precipitation forecast to fall in the basins in2050 is equivalent and not forecast to exceed 113% of
current annual precipitation and base flows are generally moderate to high (Figure 14 B and D),
indicative of some contributions from groundwater sources and snowmelt. Redband are at moderate
risk to range constriction due to forecast summer temperatures for 2040, except at the highest
elevations in the basin (Figure 14 C). Potential range contraction is most pronounced in the mainstem
Owyhee below the confluence of the East Fork and South Fork, in lower Jordan Creek, and in the
mainstem Snake. Drought scores associated with heat-related moisture loss are moderate to low
(Figure 14 E). Fire regime change risk is highest in the forested portions of the basin, where longer fire
seasons and fine fuels contribute to increase risk (Figure 14 F).
Figure 14. CSI Climate Change scores: A) Winter precipitation regime change, B) Annual total precipitation change, C) August
stream temperature change, D) Base flow index, E) Drought risk, F) Fire regime change.
26
4.4 Conservation Strategies
The CSI provides an assessment of species, habitat, and threat data from multiple data sources
summarized at a consistent scale and interpreted using transparent scoring rules. By comparing factors
from the combined indicators, we can categorize watersheds according to generalized conservation
strategies (Figure 15).
•
Protection strategies occur in subwatersheds with best habitat conditions, as indicated by the
highest CSI Habitat Integrity scores, and with the highest CSI Population Integrity scores.
•
Habitat restoration strategies are appropriate in watersheds with a lower relative habitat
condition, as reflected in CSI Habitat Integrity scores, but highest CSI Population Integrity scores
indicating healthy populations. Restoration strategies may need to address single factors that
lower the CSI scores (e.g. addressing water quality impairment caused by heavy metals from a
mining legacy) or a broader suite of factors to secure populations into the future.
•
Population restoration strategies occur in subwatersheds with the best habitat conditions, as
reflected in CSI Habitat Integrity scores, but low CSI Population Integrity scores. Restoration
strategies may be feasible for specific causes of low Population Integrity (e.g. habitat
connectivity) but may require large-scale population restoration for other causes (e.g.
introgression or non-native species).
•
Habitat and population restoration strategies may both be necessary in subwatersheds with
impaired population and habitat conditions. In some cases , addressing habitat restoration
needs may result in a population level response (e.g. addressing flow regime modification
caused by irrigation diversions and canals may allow for expansion of redband-occupied habitat
downstream).
High CSI Future Security scores and the presence of agency special status species (e.g., Greater sage
grouse, Columbia spotted frogs) provide useful overlays for further prioritization within these areas,
reflecting the likelihood of success of conservation actions or additional priority driven by special status
species or agency habitat objectives.
27
Figure 15. Conservation Strategies for subwatersheds based on arraying Population Integrity and Habitat Integrity.
Discussion
This redband trout population and habitat assessment and the associated webmap for the Owyhee,
Bruneau-Jarbidge, and Salmon Falls Creek basins is a compilation and interpretation of various data
sources to describe broad patterns of redband trout distribution, populations, habitats, and the threats
those habitats and redband trout are likely to face in the future. Taken together, the Redband
Rangewide Status Assessment data, riparian vegetation and shade maps developed specifically for this
assessment, redband stream temperature predictions, and CSI provide species, habitat, and threat data
from multiple sources at multiple spatial scales.
The results outlined in this document are one of a multitude of interpretations of the original data based
on a suggested set of scoring rules and the organizing structure of the indicators and factors. However,
the primary utility of this effort and other watershed conservation planning tools is to provide a single
source for filtering and querying a large set of disparate but important data with user-defined questions
28
about landscape scale patterns (see Game et al. 2013). These questions start with a specific
conservation strategy in mind. For example, for identifying where riparian restoration projects can have
the most “bang for the buck,” the following questions are relevant: Where are the highest quality
redband trout populations in the basins? Which stream reaches within those populations have the
warmest predicted stream temperatures? Finally, which of those reaches have degraded riparian
vegetation cover that could be restored to mitigate those projected increases in temperature? These
questions can be asked at a variety of scales, including within state, basins, and for individual
populations (Box 1).
Some examples of conservation strategies relevant to the Owyhee, Bruneau-Jarbidge, and Salmon Falls
Creek basins include:
•
Protection
These strategies can occur in watersheds or for populations with best habitat conditions, as
indicated by the highest categories within the Redband Rangewide Status Assessment, high CSI
Habitat Integrity scores, most intact riparian zones offering shade and physical structure, and
the coldest stream temperatures (e.g. Big and Little Jacks Creeks, Upper Cottonwood Creek).
Examples of protection strategies include land status designations (e.g. BLM Areas of Critical
Environmental Concern) or limiting and mitigating development.
•
Connecting and expanding existing populations
The resiliency and likelihood of persistence of existing metapopulations can be bolstered by
increasing their connectivity. While some populations in the basin are so isolated that
increasing connectivity is not possible (e.g. McCann Creek), many are separated by relatively
short reaches of habitats that are currently unoccupied for a variety of reasons, including
seasonally passable barriers (e.g. the diversion isolating Cottonwood Creek from Upper Salmon
Falls Creek), permanent barriers (e.g. the ARS research dam isolating Upper Reynolds Creek
from Lower Reynolds Creek and the mainstem Snake), degraded instream habitat (e.g. the
headcut isolating the Upper Bruneau and mainstem Bruneau populations), or potential thermal
barriers (e.g. Combination Creek and Upper Jordan Creek). Increasing connectivity may have
negative consequences if restoration opens up new habitat to non-native species (or
introgressed populations) and should be carefully evaluated with agency partners.
•
Riparian restoration
In many locations, riparian plantings or grazing management may be a means to restore the key
functions establishing riparian vegetation, including mitigating highest summer temperatures,
where it is currently absent. These activities will have the greatest immediate impact to trout
habitat in places currently at the species’ thermal margins (i.e. > 17°C August stream
temperatures) and where shade and solar exposure are mapped at levels observed to decrease
trout abundance (i.e. < 60% shaded). Upper Reynolds Creek below Dobson Creek is one such
location.
29
•
Climate change mitigation
The primary climate change threat identified for redband trout in the basins is the shift from
snow to rain or mixed winter precipitation regimes associated with warmer winter
temperatures. The primary impacts to stream habitat from these shifts will likely be decreasing
summer base flows, higher winter stream flow, and increasing frequencies of rain-on-snow
flood events (Williams et al. 2009; Mantua et al. 2010). Mitigation strategies for these include
protecting the pools and groundwater upwelling zones that provide persistent habitat at low
flows, increasing floodplain connectivity to attenuate flows, and promoting beaver. Beaver
impoundments can provide persistent perennial water and bolster downstream flows in many
circumstances (Pollock et al. 2003). Strategies with immediate redband benefits that also lessen
impacts from climate change include riparian vegetation restoration to increase stream shading
in thermally marginal habitats and increasing habitat connectivity to facilitate fish movement
and recolonization following disturbance. Fish passage improvements with larger culverts has
the additional benefit of decreasing road and infrastructure damage during flooding; NorWeST
temperature predictions provide a means to identify areas forecast as thermally marginal.
An equally useful approach is to use the information synthesized in this assessment to evaluate projects.
As potential projects arise, the Redband Rangewide Status Assessment and CSI results become two
landscape-scale criteria in the project evaluation phase. Additional considerations can be gained from a
limiting factors analysis using the stream temperature predictions, or local data sources such as agency
field data or surveys of on-the-ground habitat conditions. Figure 16 provides a conceptual model of this
project evaluation process, in which different tools are used to identify priorities.
30
Box 1. Using Web-Mapping to Identify Strategic Restoration Opportunities for Redband Trout Populations Vulnerable
to Climate Change.
Strategic conservation decisions use science and data to focus limited resources. Scientific assessments are commonly completed to
provide strategic, science-based guidance for largescale conservation initiatives. Often these assessments are done in the early
stages of an initiative, and they necessarily need to anticipate many different types of conservation actions that might be
implemented. Because of this, tools are needed that allow conservation practitioners to continue to use science and data to answer
new questions that arise as initiatives mature over time. Web-based mapping applications provide a means to help answer
unanticipated questions not addressed in initial scientific assessments.
Here we show a simple example of how the data assembled and developed for this assessment – and served using a web-mapping
interface – can be used to identify strategic opportunities to restore redband trout habitat and provide resilience to climate change
across the large Owyhees landscape. Woody riparian vegetation is known to reduce solar insulation to streams through shading and,
as a result, can help to reduce stream temperatures. It also helps to stabilize stream banks and provide quality instream habitat for
trout. Thus, one habitat restoration strategy that can help buffer climate change impacts to redband trout is to restore riparian
vegetation with a goal of reducing stream temperatures and improving habitat quality. However, to be most strategic in restoration
implementation, one needs to have an answer to the following question: Where can riparian restoration best benefit redband trout
populations that are predicted to be impacted by climate change?
This question can easily be answered by using the web-mapping application developed for this assessment. The application contains
data on redband trout populations (Section 2.3.2), stream temperature predictions under climate change (Section 3.1), and percent
solar radiation reaching the stream (Section 3.2.2) that were assembled or developed as part of this assessment. Within the
application you can filter the genetically pure redband trout populations (‘Population Integrity Indicator 3 (PI_3)’ score > 3), where
stream temperatures are predicted to be marginal for trout (‘NorWeST modeled avg. Aug temp C (2040)’ > 17°C) and where stream
shading is sparse (‘Total ave. August direct solar radiation’ > 0.40 [40%]). Applying this filter highlights strategic restoration
opportunities in several tributaries to Jordan Creek, Salmon Creek (Reynolds Creek), upper Sinker Creek, North Fork Owyhee River,
Red Canyon Creek, Nip and Tuck Creek, Salmon Falls Creek in Nevada, and more (see Figure below). Additional filters, such as land
ownership, may help to further refine strategic restoration opportunities with specific conservation partners.
31
Figure 16. Example of how landscape assessment information can be used with conservation needs and local knowledge
(agency partners, or other data sources) to evaluate the efficacy of potential projects.
When interpreting assessment results, there are two important considerations related to the input data
– data quality and missing data. The data we use represent the best available datasets for representing
a particular feature. Most data are from the period from 2000-2010, and may not be the most up to
date: the assessment and analyses presented here provides a snapshot – and potential baseline – for
features and conditions for that period. Additionally, there may be variability within a particular factor
not captured by the coarse resolution of some of the GIS data used in the assessment. For example, we
use road densities to approximate sedimentation effects from road networks, but roads will vary greatly
in their delivery of sediment to streams based on their quality of construction, position in the
watershed, and bedrock geology (Black et al. 2010). Likewise, there may be local spatial datasets
overlooked during the data gathering of broader, more general datasets that may provide additional
resolution for considering conditions or resources on the ground. Thus, this assessment and CSI results
32
should not be considered the final says in habitat conditions or conservation strategies, but should be
considered a starting point for making strategic conservation decisions across a large landscape.
A second consideration is what important factors are missing. For CSI Habitat Integrity results, several
factors lack any spatial data for approximating or measuring impact, including direct or indirect
measures of grazing and accurate maps depicting seasonality of stream flows or dewatering. In
particular, the lack of measures of grazing impacts will likely overestimate instream habitat quality in
much of the study area, where grazing is often the only land use or anthropogenic disturbance on large
blocks of public land. For Future Security, we lack overlays on the resilience of aquatic systems or the
interactive effects of natural disturbances like fire and floods that may compound existing or future
threats.
Restoration opportunities may be best pursued at the watershed scale, given recent evidence of the
importance of concentrating restoration efforts in limited areas in order to produce measurable changes
in aquatic species abundance (Roni et al. 2010). Agency management plans (e.g., Johnson 2005) and
local knowledge will provide important information on fine-scale condition and opportunities within
watersheds identified based on their relative condition across the analysis landscape. For example,
restoration opportunities likely exist on local scales within watersheds with a protection priority.
The presence of agency special status species within redband trout habitat may increase the availability
of funding for restoration, so long as restoration activities for redband are compatible with other
species’ needs. Two candidate species for listing under the Endangered Species Act- Columbia spotted
frog (Rana luteiventris) and Greater sage grouse (Centrocercus urophasianus) – both utilize riparian
areas during portions of their life histories.
Despite the abundance of newly available information assembled within this assessment, many data
gaps and monitoring needs remain in the basin, including:
•
Monitoring flows and fish passage
Surveying for diversions, impassible road culverts, and flow limited reaches would fill a critical
information gap across the basin. Throughout the basin, diversions irrigating pastures with
surface water both block upstream passage of fish and limit downstream flows. Modifying the
largest barriers provides an opportunity to increase the connectivity of populations, while
improving the efficiency of diversions provides an opportunity to increase downstream flows,
which is particularly important during the late summer low flow period.
•
Monitoring non-native fishes
Non-native fishes like brook trout and smallmouth bass pose threats to some redband
populations. Although recent surveys have been completed (Meyer et al. 2010; Meyer et al.
2013; Butts et al. 2011; Butts et al. 2013), the exact distribution of non-natives in the basin is not
known. Modeled predictions of smallmouth bass occurrence, can, for example, help target
33
efforts to define and monitor the potential spread of this non-native species, and how it might
interaction with specific populations of redband trout.
•
Fill data gaps in Rangewide Status Assessment
For many stream reaches in the assessment area, the Redband Rangewide Status Assessment
assumes abundance or genetic status. Clarification of those population characteristics will help
refine conservation strategies for the species. Additionally, population connectivity or potential
connectivity with downstream sources of non-native species is undetermined for many
watersheds (e.g. Red Canyon or North Fork Owyhee with Mainstem Owyhee).
•
Continued temperature monitoring
Continued monitoring with our network of 104 sensors provides key baseline temperature data
for the basins and allows for further evaluation of NorWeST models. With the development of
riparian vegetation and shade maps specific to the basin, the opportunity exists to develop
higher-resolution, basin-specific models to predict stream temperatures with higher accuracies.
Additional variables worth exploring are beaver dam and impoundment locations and
topographic predictors of snow drift formation.
Broad-scale assessments of species and habitats represent a specific view of conditions across a
landscape that can then be used as a starting-point for strategic conservation decisions that are based
on a blend of broad-scale data, local information, and informed discussion with agency and other
potential project partners. Delivery of these various data sources allows for exploration of various data
sources viewed in ways done explicitly for this report.
Acknowledgements
This project was supported by grant/cooperative agreement BLM L10AC20195 from the BLM Nevada
State Office. We are grateful for data provided by Elko and Owyhee Field Offices – BLM, Nevada
Department of Wildlife, Idaho Department of Fish and Game, and Oregon Department of Fish and
Wildlife and for discussions on redband habitat with C. Evans, P. Coffin, G. Johnson, J. Elliott, J. Kozfkay
and M. Koenig. R. Pahl provided helpful overview of stream temperature and shade assessment
techniques as used in the Dixie and Hanks Creeks Temperature TMDLs. Field work and GIS support were
provided by: R. Bjork, J. Kellner, M. Mayfield, P. Gardner, I. Shives, and J. Barney.
34
Reference List
Allen, D. B., B. J. Flatter, J. Nelson, and C. Medrow. 1998. Redband trout population and stream habitat
surveys in northern Owyhee County and the Owyhee River and its tributaries, 1997.Technical
Bulletin No. 98-14.
Benke, A. C. 1990. A perspective on America's vanishing streams. Journal of the North American
Benthological Society 9(1): 77-88.
Benke, R. J. 1992. Native trout of western North America. Monograph 6, American Fisheries Society,
Bethesda, Maryland.
Black, T. 2010. Road inventory and monitoring with GRAIP. CDM Water Resources Discipline webinar.
Booth, D. T., S. E. Cox, and G. Simonds. 2007. Riparian monitoring using 2-cm GSD aerial photography.
Ecological Indicators 7: 636-648.
Bozek, M. A., and F. J. Rahel. 1991. Assessing habitat requirements of young Colorado River cutthroat
trout by use of macrohabitat and microhabitat analysis. Transactions of the American Fisheries
Society 120:571-581.
Breiman, L. 2001. Random forests. Machine Learning 45:5-32.
Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference. Springer-Verlag,
New York.
Burton, T. A., S. J. Smith, and E. R. Cowley. 2011. Riparian area management: Multiple indicator
monitoring (MIM) of stream channels and streamside vegetation. Technical reference 1737-23.
BLM/OC/ST-10/003+1737. Bureau of Land Management, Denver, Colorado. 155pp.
Butts, A. E., J. R. Kozfkay, C. Sullivan, and J. C. Dillon. 2011. 2009 Southwest Region (Nampa) Fisheries
Management Report. IDFG #11-108. Idaho Department of Fish and Game, Boise, Idaho. (pdf)
Butts, A. E., M. Koenig, J. R. Kozfkay, P. Gardner, and R. Gillingham. 2013. 2012 Southwest Region
(Nampa) Fisheries Management Report. IDFG #13-122. Idaho Department of Fish and Game,
Boise, Idaho. (pdf)
Cade, B. S., B. R. Noon, and C. H. Flather. 2005. Quantile regression reveals hidden bias and uncertainty
in habitat models. Ecology 86:786-800.
Cade, B. S., and B. R. Noon. 2003. A gentle introduction to quantile regression for ecologists. Frontiers
in Ecology and Management 1:412-420.
Cassinelli, J. D., and C. M. Moffitt. 2010. Comparison of growth and stress in resident redband trout
held in laboratory simulations of montane and desert summer temperature cycles. Transactions
of the American Fisheries Society 139:339-352.
35
Coffin, P. and C. Evans. Elko BLM District fish distribution. 2006. Elko, NV, Bureau of Land Management.
Cummins, K. W. 1962. An evaluation of some techniques for the collection and analysis of benthic
samples with special emphasis on lotic waters. American Midland Naturalist 67:477-504.
Dauwalter, D. C., F. J. Rahel, and K. G. Gerow. 2010. Power of revisit monitoring designs to detect
forest-wide declines in trout populations. North American Journal of Fisheries Management
30:1462-1468.
Davies-Colley, R. J. and D. G. Smith. 2001. Turbidity, suspended sediment, and water clarity: a review.
Journal of the American Water Resources Association 37(5): 1085-1101.
Evans, J. S., M. A. Murphy, Z. A. Holden, and S. A. Cushman. 2011. Modeling species distribution and
change using random forest. Pages 139-159 in C. A. Drew, Y. F. Wiersma, and F. Huettmann,
editors. Predictive species and habitat modeling in landscape ecology. Springer, New York.
Fausch, K. D., B. E. Rieman, M. K. Young, and J. B. Dunham. 2006b. Stategies for conserving native
salmonid populations at risk from nonnative fish invasions: tradeoffs in using barriers to
upstream movement.General Technical Report RMRS-GTR-174,Fort Collins, Colorado.
Game, E. T., P. Kareiva, and H. P. Possingham. 2013. Six common mistakes in conservation priority
setting. Conservation Biology 27(3): 480-485.
Haak, A. L., and J. E. Williams. 2012. Spreading the risk: native trout management in a warmer and lesscertain future. North American Journal of Fisheries Management 32:387-401.
Hanski, I.A. and D. Simberloff. 1996. The metapopulation approach, its history, conceptual domain, and
application to conservation. In: Hanski, I.A. & Gilpin, M.E., ed. Metapopulation biology: ecology,
genetics, and evolution. New York: Academic Press, pp.5-26.
Hawkins, C. P., J. L. Kershner, P. A. Bisson, M. D. Bryant, L. M. Decker, S. V. Gregory, D. A. McCullough, C.
K. Overton, G. H. Reeves, R. J. Steedman, and M. K. Young. 1993. A hierarchical approach to
classifying stream habitat features. Fisheries 18:3-12.
Hilderbrand, R. H. and J. L. Kershner. 2000. Conserving inland cutthroat trout in small streams: how
much stream is enough? North American Journal of Fisheries Management 20: 513-520.
Hyndman, P. C. and H. W. Campbell. BLM mining claim recordation system: mining claim density. 1996.
Fort Collins, Colorado, Open-File Report 99-325. Natural Resource Ecology Labaoratory, U.S.
Geological Survey.
Idaho Department of Water Resources. Idaho Water Right Point of Diversion. (August 2002 website
addition, accessed 2008). 2002.
IDEQ. Streams of Idaho: 305(b) & 303(d) integrated report - water quality - 2010. 2012. Boise, Idaho,
Idaho Department of Environmental Quality.
36
IDFG. 2005. Idaho comprehensive wildlife conservation strategy. Idaho Department of Fish and Game
report to U.S. Fish and Wildlife Service. Boise, Idaho.
IGDC. Integrated Road Transportation of Idaho. (2008-06-22). 2008. Moscow, Idaho, Idaho Geospatial
Data Clearinghouse (INSIDE Idaho).
Idaho National Laboratory. Hydropower Resource Assessment. 2004. Idaho Falls, Idaho, Idaho National
Laboratory.
Isaak, D. J., and D. L. Horan. 2011. An evaluation of underwater epoxies to permanently install
temperature sensors in mountain streams. North American Journal of Fisheries Management
31:134-137.
Isaak, D.J., S.J. Wenger, E.E. Peterson, J. M. Ver Hoef, S. Hostetler, C.H. Luce, J.B. Dunham, J. Kershner,
B.B. Roper, D. Nagel, D. Horan, G. Chandler, S. Parkes, and S. Wollrab. 2011. NorWeST: An
interagency stream temperature database and model for the Northwest United States. U.S. Fish
and Wildlife Service, Great Northern Landscape Conservation Cooperative Grant. Project
website:www.fs.fed.us/rm/boise/AWAE/projects/NorWeST.html
Johnson, G. L. 2005. Redband trout species management plan for Nevada.
Klausmeyer, K. R. and M. R. Shaw. 2009. Climate Change, Habitat Loss, Protected Areas and the Climate
Adaptation Potential fo Species in Mediterranean Ecosystems Worldwide. PLoS ONE 4(7).
Kozfkay, J. R., A. E. Butts, J. C. Dillon. 2011. 2010 Southwest Region (Nampa) Fisheries Management
Report. IDFG #11-110. Idaho Department of Fish and Game, Boise, Idaho. (pdf)
Lee, D. C., J. R. Sedell, B. E. Rieman, R. F. Thurow, and J. E. Williams. 1997. Broadscale assessment of
aquatic species and habitats. Pages 1057-1496 in T. M. Quigley and S. J. Arbelbide, editors. An
assessment of ecosystem components in the Interior Columbia Basin and portions of the
Klamath and Great Basins: Volume III. USDA Forest Service, General Technical Report PNW-GTR405, Portland, Oregon.
Lloyd, D. S. 1987. Turbidity as a water quality standard for salmonid habitats in Alaska. North American
Journal of Fisheries Management 7(1): 34-45.
Lukacs, P. M., K. P. Burnham, and D. R. Anderson. 2010. Model selection bias and Freedman's paradox.
Annals of the Institute of Statistical Mathematics 62:117-125.
Manel, S., H. C. Williams, and S. J. Ormerod. 2001. Evaluating presence-absence models in ecology: the
need to account for prevalence. Journal of Applied Ecology 38:921-931.
Mantua, N. J., I. Tohver, and A. F. Hamlet. 2010. Climate change impacts on streamflow extremes and
summertime stream temperature and their possible consequences for freshwater salmon
habitat in Washington State. Climatic Change 102(1-2): 187-223.
37
Maurer, E., L. Brekke, T. Pruitt, and P. Duffy. 2007. Fine-resolution climate projections enhance regional
climate change impact studies. Eos Trans.AGU 88(47).
May, B. E. and S. Albeke. 2005. Range-wide status of Bonneville cutthroat trout (Oncorhynchus clarki
utah): 2004.Bonneville Cutthroat Trout Status Update Interagency Coordination Workgroup,
Publication Number 05-02,Salt Lake City, Utah.
May, B.E, B.J. Writer, and S. Albeke. 2012. Redband Trout Status Update Summary. Wild Trout
Enterprises. Bozeman, MT.
McElhany, P., M. H. Ruckelshaus, M. J. Ford, T. C. Wainwright, and E. P. Bjorkstedt. 2000. Viable
Salmonid Populations and the Recovery of Evolutionarily Significant Units.NMFS-NWFSC-42.
Meyer, K. A., J. A. Lamansky, Jr., and D. J. Schill. 2010. Biotic and abiotic factors related to redband trout
occurrence and abundance in desert and montane streams. Western North American Naturalist
70(1): 77-91.
Meyer, K. A., J. A. Lamansky, Jr., D. J. Schill, and D. W. Zaroban. 2013. Nongame fish species distribution
and habitat associations in the Snake River Basin of southern Idaho. Western North American
Naturalist 73:20-34.
Murphy, M. A., J. S. Evans, and A. Storfer. 2010. Quantifying Bufo boreas connectivity in Yellowstone
National Park with landscape genetics. Ecology 91:252-261.
Narum, S. R., N. R. Campbell, C. C. Kozfkay, and K. A. Meyer. 2010. Adaptation of redband trout in
desert and montane environments. Molecular Ecology 19:4622-4637.
National Renewable Energy Laboratory. 2008. Wind energy resource atlas of the United States: High
resolution wind data. NREL, Alliance for Sustainable Energy, LLC and U.S. Department of Energy,
Golden, CO.
Natural Resources Conservation Service. 2012. US General Soil Map (STATSGO). NRCS, Washington, DC.
NV DEP. Nevada's 2006 303(d) Impaired Waters List. 2006. Carson City, NV, Nevada Division of
Environmental Protection.
OR Department of Land Conservation and Development. Oregon Urban Growth Boundaries. 2009.
Salem, Oregon Department of Land Conservation and Development.
Oregon Department of Environmental Quality, W. Q. D. Oregon 2004/2006 Integrated Report on Water
Quality. 2007. Portland, OR.
Oregon Water Resources Department. Oregon Water Rights. 2010. Salem, OR.
Pahl, R. 2010. Dixie and Hanks Creeks temperature TMDLs. Nevada Department of Environmental
Protection. Carson City, NV.
38
Perez, R., P. Ineichen, Moore K, M. Kmiecikm, C. Chain, R. George, and F. Vignola. 2002. A new
operational satellite-to-irradiance model. Solar Energy 73(5): 307-317.
Pierce, K.B., T. Lookingbill, and D.L. Urban. 2005. A simple method for estimating potential relative
radiation (PRR) for landscape-scale vegetation analysis. Landscape Ecology 20:137-17.
Platts, W. S., W. F. Megahan, G. W. Minshall. 1983. Methods for evaluating stream, riparian, and biotic
conditions. Intermountain Forest and Range Experiment Station, US Forest Service. Ogden,
Utah.
Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks, and J. C.
Stromberg. 1997. The natural flow regime. BioScience 47(11): 769-784.
Pollock, M.M., M. Heim, and D. Werner. 2003. Hydrologic and geomorphic effects of beaver dams and
their influence on fishes. In S.V. Gregory, K. Boyer, and A. Gurnell (eds.), The ecology and
management of wood in world rivers, p. 213- 233. Symposium 37. American Fisheries Society,
Bethesda, MD.
Praggastis, A. and J. E. Williams. Salmon's presence in Nevada's past. Nevada Historical Society Quarterly
In Press.
PRISM Group. PRISM 800m Normals (1971 - 2000). 2008. Corvallis, Oregon, Oregon State University.
R Development Core Team. 2013. Foundation for statistical computing. Version 3.0.2. Vienna, Austria
http://www.R-project.org/
Rahel, F. J. 2004. Unauthorized fish introductions: fisheries management of the people, for the people,
or by the people. Pages 431-443 in M. J. Nickum, P. M. Mazik, J. G. Nickum, and D. D. MacKinlay,
editors. Propagated fish in resource management. American Fisheries Society Symposium 44,
Bethesda, Maryland.
Ramirez, C, J.H. Viers, J.F. Quinn, J.H. Johnson, and N.B. Kalman. 2003. Geospatial risk assessment of
North Coast Watersheds in California. Information Center for the Environment, UC-Davis, Davis,
CA.
Rieman, B. E. and J. B. Dunham. 2000. Metapopulations and salmonids: a synthesis of life history
patterns and empirical observations. Ecology of Freshwater Fish 9(1-2): 51-64.
Roni, P. and T. P. Quinn. 2001. Density and size of juvenile salmonids in response to placement of large
woody debris in western Oregon and Washington streams. Canadian Journal of Fisheries and
Aquatic Sciences 58(2): 282-292.
Schill, D. J., G. W. LaBar, F. S. Elle, E. R. J. M. Mamer. 2007. Angler exploitation of redband trout in eight
Idaho desert streams. North American Journal of Fisheries Management 27:665-669.
39
Schill, D. J., G. W. LaBar, E. R. J. M. Mamer, and K. A. Meyer. 2010. Sex ratio, fecundity, and models
predicting length at sexual maturity of redband trout in Idaho desert streams. North American
Journal of Fisheries Management 30:1352-1363.
Shaw SC and D. Johnson. 1995. Slope morphology model derived from digital elevation data: SMORPH.
Proceedings of Northwest ArcInfo Users Conference.
Shock, C. B., M. P. Shock, and C. C. Shock. 2007. Lower Owyhee Watershed Assessment.
Soule, M. E. 1987. Where do we go from here? Viable populations for conservation. Cambridge
University Press, Cambridge, England.
Sowder, C., and E. A. Steel. 2012. A note on the collection and cleaning of water temperature data.
Water 4:597-606.
Stephens, S. E., J. A. Walker, D. R. Blunck, A. Jayaraman, D. E. Naugle, J. K. Ringelman, and A. J. Smith.
2008. Predicting risk of habitat conversion in native temperate grasslands. Conservation Biology
22(5): 1320-1330.
Theobald, D. 2005. Landscape patterns of exurban growth in the USA from 1980 to 2020. Ecology and
Society 10 (online).
UDWR. 2000. Cutthroat trout management, a position paper: genetic considerations associated with
cutthroat trout management. Publication Number 00-26. Utah Division of Wildlife Resources,
Salt Lake City, Utah.
US Bureau of Land Management. Ground Transportation Roads - Idaho. 2004. Boise, ID, Idaho BLM
State Office.
US Bureau of Land Management. Ground Transportation Roads - Oregon and Washington. 2009.
Oregon BLM State Office.
USACE (U.S. Army Corps of Engineers). 2008. National Inventory of Dams. U.S. Army Corps of Engineers.
http://crunch.tec.army.mil/nidpublic/webpages/nid.cfm.
USBLM. Geocommunicator. 2008. USBLM and USFS.
USDA. National Cropland Data Layer. 2009. USDA National Agricultural Statistics Service.
USEPA and USGS. National Hydrography Dataset Plus - NHDPlus (1:100,000 scale). 2005. Sioux Falls,
South Dakota, U.S. Environmental Protection Agency and U.S. Geological Survey.
USFWS. 2004. Draft recovery plan for the Jarbidge River Distinct Population Segment of Bull Trout
(Salvelinus confluentus). U.S. Fish and Wildlife Service. Portland, Oregon.
40
USFWS. 1995. Endangered and threatened wildlife and plants: 12-month finding for a petition to list
desert redband trout in the Snake River Basin above Brownlee Reservoir and below Shoshone
Falls as threatened or endangered. Federal Register 60:187:49819-49821.
USFWS. 2000. Endangered and threatened wildlife and plants: 12-month finding for a petition to list
the Great Basin redband trout as threatened or endangered. Federal Register 65:54:1493214936.
US Fish and Wildlife Service. 2013. Nevada Passage Assessment Database. USFWS, Reno, NV.
USGS. National Land Cover Database. 2006. Sioux Falls, South Dakota, U.S. Geological Survey.
USGS. Land Ownership in Western North America, 180 m. (1986-2003). 2004a. Boise, Idaho, Sage-grouse
rangewide conservation assessment, Snake River Field Station, U.S. Geological Survey.
USGS. 2004b. Oil and natural gas wells, western U.S.Conservation Assessment of Greater Sage-grouse
and Sagebrush Habitats, Western Association of Fish and Wildlife Agencies,Cheyenne, Wyoming.
[online] http:\\sagemap.wr.usgs.gov.
USGS. All Roads in the Western United States (2000 TIGER) (1:100,000). 2008a.
USGS. Mineral Resources Data System (MRDS) (Active). (2005). 2008b. Reston, Virginia, U.S. Geological
Survey.
USGS. National Elevation Dataset (30m) (1:24,000). 2008c. Sioux Falls, SD, USGS EROS Data Center.
US Geological Survey. 2009. LANDFIRE: LANDFIRE Fire Regime Condition Class layer. USGS, Reston, VA.
USGS. Protected areas database of the United States (PAD-US) Version 1.3. 2013. Moscow, ID, USGS
NBII.
Wallace, R. L. and D. W. Zaroban. 2013. Native fishes of Idaho. American Fisheries Society, Bethesda,
Maryland.
Westerling, A. L., H. G. Hidalso, D. R. Cayan, and T. W. Swetnam. 2006. Warming and earlier spring
increases western U.S. forest wildfire activity. Science 313: 940-943.
Williams, J. E., A. L. Haak, N. G. Gillespie, and W. T. Colyer. 2007. The Conservation Success Index:
synthesizing and communicating salmonid condition and management needs. Fisheries 32(10):
477-492.
Wind Powering America and National Renewable Energy Laboratory. Wind Resource Potential. 2003.
National Renewable Energy Laboratory, USDOE.
Wolock, D. M. Base-flow index grid for the conterminous United States. Open-File Report 03-263. 2003.
Reston, VA, US Geological Survey.
41
Zippin, C. 1958. The removal method of population estimation. Journal of Wildlife Management 22:8290.
Zoellick, B. W. 1999. Stream temperatures and the elevational distribution of redband trout in
southwestern Idaho. Great Basin Naturalist 59:136-143.
Zoellick, B. W. 2004. Density and biomass of redband trout relative to stream shading and temperature
in southwestern Idaho. Western North American Naturalist 64:18-26.
Zoellick, B. W., D. B. Allen, and B. J. Flatter. 2005. A long-term comparison of redband trout
distribution, density, and size structure in southwestern Idaho. North American Journal of
Fisheries Management 25:1179-1190.
Zoellick, B. W., and B. S. Cade. 2006. Evaluating redband trout habitat in sagebrush desert basins in
southwestern Idaho. North American Journal of Fisheries Management 26:268-281.
42
Appendix A. Stream Temperature Monitoring and Validation of the NorWeST Stream
Temperature Model
Objectives
•
•
Establish a network of temperature data loggers in the Owyhee and Salmon Falls Creek basins to
monitor stream temperatures year-round.
Validate US Forest Service Rocky Mountain Research Station stream temperature predictions
using preliminary data from loggers.
Methods
We selected preliminary monitoring sites from existing Idaho Department of Fish and Game fish
sampling locations and strata related to elevation and stream order. We deployed Onset Hobo TidBit
dataloggers within PVC housing epoxied to boulders or anchored in the stream channel to rebar (Isaak
and Horan 2011). These loggers were deployed in 2010 – 2013 and have the capacity to record hourly
temperatures for over 4 years. At each location, we also measure the solar exposure at the monitoring
location using a Solar Pathfinder. We used these data (August average direct solar radiation) to predict
shading on all stream reaches within the current distribution of redband trout based on the results of
the riparian vegetation classification and a measure of shading provided by terrain features (Appendix
C).
We evaluated the stream temperature predictions produced by the NorWeST Stream Temp Regional
Database and Modeled Stream Temperatures (Isaak et al. 2011) using the subset of our stream
temperature monitoring data that has been downloaded. Because the NorWeST dataset provides
predictions for mean August temperatures for the years 1991 – 2011 and our field measurements date
from 2012 and 2013, we first had to identify which year from the 1991 – 2011 time period was most
similar to 2012 and 2013 climatically. We downloaded monthly average temperature and precipitation
from weather stations within the study area via the NOAA National Climatic Data Center. 23 weather
stations provided a continuous temperature dataset and 16 provided a continuous precipitation record.
We compared the average difference in mean August temperature and total precipitation from the
preceding water year across sites for each year relative to 2012 and 2013 and selected the year that was
most similar to both criteria. We assigned each field observed temperature to a NorWeST stream
segment and compared the difference in observed temperature to predicted temperature.
Results
We have established a stream temperature monitoring network of 103 data loggers in the Owyhee and
Salmon Falls Creek basins (Figure A-1). These locations complement existing networks of dataloggers,
including existing US Forest Service monitoring in the Bruneau-Jarbidge (PIBO and bull trout), a currently
inactive network of sites maintained by the Idaho BLM Owyhee Field Office, and annual, summer-only
monitoring by Nevada BLM Elko Field Office and Idaho Department of Water Quality. Our network is
unique in the basins as the only source of year-round temperature monitoring, which will prove
43
invaluable for tracking changes in stream temperatures outside of the summer period (i.e. spring and fall
temperature trends).
Based on average August air temperatures and preceding water year total precipitation measured at 16
locations throughout the basin, 1994 is most similar to both 2012 and 2013 (for observed average
August air temperature and preceding water year total precipitation from the 1993-2011 period within
NorWeST). Data from 29 sites representing 42 unique summer-observations were available for
comparison with NorWeST predictions. Of those, 50% have observed temperatures less than 1 C from
predicted and 71% fall within the standard error values reported by NorWeST (mean standard error for
basins = 1.9°C). Of the 28% of sites with the greatest disparity in observed vs. predicted temperatures
(i.e. those with an absolute difference > 2 C), half are located on the mainstem Owyhee or below
Wildhorse Reservoir on the East Fork Owyhee and are warmer than predicted. Systems colder than
predicted include the Middle Fork Owyhee above Three Forks, Flint Creek, and North Boulder Creek
(Figure A-2).
Figure A-1. Stream temperature monitoring network.
44
Figure A-2. NorWeST stream temperature model validation sites.
45
Appendix B: Evaluate Classification of NAIP imagery as a Tool for Riparian Vegetation
Classification and Monitoring
Objectives:
•
•
Use widely available, free, high-resolution National Agriculture Imagery Program (NAIP)
photographs to classify riparian vegetation composition
Assess NAIP image classification as a tool for long-term monitoring of riparian vegetation.
Methods
We acquired National Agriculture Imagery Program (NAIP) photographs from 2006 (entire Owyhee
basin, Nevada), 2009 (redband streams in Owyhee basin, Idaho), and 2010 (redband streams in Salmon
Falls Creek basin, Nevada and portions of the Owyhee Front, Idaho). NAIP images are color aerial
photographs with a horizontal resolution of 1 meter and include spectral information for 4 bands – red,
green, blue, and infrared. Imagery for Nevada date from the summer of 2006; Idaho and Oregon images
are from 2009.
We conducted a supervised classification of the NAIP imagery using two main steps – delineating
potential riparian zones and characterizing the vegetation within that area. For identifying potential
riparian areas, we two used variables derived from a 10 meter resolution digital elevation model (USGS
2008c) - slope and distance from NHD Plus streams (USEPA and USGS 2005) - to identify a threshold
value at which the product of a slope x distance raster encompassed the zone of productivity within the
riparian zone (Ramirez et al. 2003). For classifying vegetation, we used the feature extraction tools
within the Feature Analyst GIS software package and the “Land cover feature” and “Manhattan input
representation” algorithms. These tools take the spectral information contained within multiple
example polygons delineated by GIS technicians for each vegetation type and computes statistics from
all four bands of the NAIP imagery to cluster pixels within the entire image into land cover classes based
on common signatures. Our classification is limited to the floodplain area and includes the following
classes: woody riparian vegetation, herbaceous riparian vegetation, upland sagebrush
shrubland/grassland, bare ground/sand/rock, agriculture, roads, and water.
Several classes proved problematic to correctly classify and we used a series of steps to improve the
quality of the final classification. We cross-checked our mapping of agricultural areas against the
National Land Cover Dataset (USEPA 2001) and 2009 cropland data layer (USDA 2009) to improve the
agriculture mapping. To discriminate between bare ground/rock and roads, we overlaid existing roads
data (IGDC 2008; USGS 2008a) onto the final vegetation map after buffering primary, paved roads by 25
meters and secondary road data by 12 meters. Water mapping proved difficult due to the similarity in
spectral statistics in the NAIP images between open water, shadows, and dark rock flows (basalts). We
improved the water class though a separate mapping effort focused solely on water and limited to low
slope areas within the floodplain area and in patches at least 120 m². We conducted an accuracy
assessment to determine the quality of our riparian mapping effort after completing the classification by
46
comparing the mapped vegetation categories against the vegetation type directly interpreted from the
imagery at randomly placed points throughout the basin.
As mapped, the woody riparian vegetation class includes willows (Salix sp.), cottonwoods and aspens
(Populus sp.), dogwood (Cornus sp.), Douglas fir (Pseudotsuga menziesii), western juniper (Juniperus
occidentalis) and other woody tree and large shrub species that serve a functional role of providing
shade within the riparian zone. The herbaceous riparian vegetation class includes sedges (Carex sp.),
grasses, and forbs that occupy the wetted area adjacent to streams and exhibit robust annual leafy
growth. Sagebrush (Artemisia sp.), bitterbrush (Purshia tridentata), native grasses, and introduced
annual grasses (especially cheatgrass (Bromus tectorum)) characterize the upland sagebrush
shurbland/grassland vegetation class. Agriculture in the basins is comprised solely of hay pasture
typically flood-irrigated with diverted surface water.
Results
Figure B-1 provides an example of an original false-color NAIP image and the final land cover
classification of the potential riparian zone. The error matrix in Table B-1 compares the mapped
vegetation categories against the vegetation type directly interpreted from the imagery at randomly
placed points throughout the basin. Although a rigorous accuracy assessment would compare the
mapped vegetation type to vegetation identified on the ground in field surveys, direct image
interpretation still provides a useful set of data for understanding the quality of the classification
product. These data are summarized to report producer’s accuracies and user’s accuracies by
vegetation type. Producer’s accuracies reflect how well a vegetation type on the ground is mapped
within the classification product, while user’s accuracies reflect the likelihood that a pixel mapped as a
particular vegetation type actually represents that type. Overall accuracy for the riparian vegetation
mapping effort is high – 77%. Similarly, the accuracy of the key vegetation type – woody riparian shrubs
and trees that provide shade – is high. Most of the classification errors are associated with the adjacent
type of vegetation within the moisture-driven productivity gradient of woody riparian --- herbaceous
riparian --- sagebrush/grass --- bare ground/rock. Thus, woody riparian vegetation patches on the
ground are most likely to be misclassified in the map product as herbaceous riparian vegetation,
sagebrush and grassland patches are most likely to be misclassified as bare ground or as herbaceous
riparian vegetation, and so forth. Road mapping errors are associated with our method of using roads
data to overlay on the vegetation mapping; due to inaccuracies in the data, dirt and two-track roads are
misplaced into the adjacent upland vegetation (typically sagebrush). Other errors are understandable
given the similarities in vegetation type – hay meadows mapped as agriculture are often similar in
structure and function to wet meadows mapped as herbaceous riparian vegetation.
47
Classification
Table B-1. NAIP image classification error matrix resulting from classifying randomly sampled test pixels.
Woody Rip
Emerg Rip
Sage/grass
Bare
Ag
Roads
Water
Total
Training Set Data
Woody Rip. Emerg. Rip. Sage/grass
Bare
Ag
21
4
4
1
4
18
1
2
1
29
11
19
2
1
25
5
2
25
25
51
22 27
Woody Riparian
Emergent Riparian
Sagebrush/grassland
Bare ground/rock
Agriculture
Roads
Water
Overall Accuracy
Producer's Accuracy
84.0%
72.0%
56.9%
86.4%
92.6%
100.0%
100.0%
Roads
Water
6
6
12
12
User's Accuracy
70.0%
72.0%
96.7%
63.3%
89.3%
54.5%
85.7%
77.4%
Total
30
25
30
30
28
11
14
168
48
Figure B-1. Example of the supervised classification of NAIP imagery (top panel) into vegetation classes (bottom panel).
49
Appendix C. Development of a Solar Radiation Model from Terrain and Riparian
Vegetation Attributes
Objective:
•
•
Develop a model that predicts percent August solar radiation as a function of percent woody
vegetation and terrain exposure
Predict Percent August Solar Radiation for all streams segments in the study area.
Methods:
A model of percent total August solar radiation was developed to link field measures of total solar
radiation in the month of August with GIS-derived measures of woody riparian vegetation and
topography. Similar methods have been used by the Nevada Department of Environmental Quality for
development of TMDL documents for temperature impaired streams (Pahl 2010). To develop the
model, we measured the percent of total solar radiation that reaches the stream surface, hereafter
referred to as Percent Solar Radiation, using Solar Pathfinder™. Solar Pathfinder identifies the amount
of solar insulation intercepted by shade-producing objects such as riparian vegetation and topographic
features such as valley walls, and estimates the average daily thermal input reaching the stream surface
for each month of the year by integrating the effects of azimuth, topographic altitude, vegetation
height, latitude, hour angle, and time of year. The location of Solar Pathfinder measurement was
recorded with a GPS unit in order to link it with the NAIP image classification.
At each Solar Pathfinder location, percent woody vegetation and potential relative radiation were
measured as predictors of percent solar radiation, and they are known to influence the amount of solar
radiation reaching stream surfaces (Zoellick 2004) and are inherently accounted for by Solar Pathfinder
measurements. Percent woody vegetation was measured from the NAIP imagery classification (see
Section 3.2.1 above) as the percent of woody vegetation within a 5-m buffer of the Solar Pathfinder
location. Terrain exposure was evaluated by generating an August potential relative radiation layer for
the basin from a 10-meter resolution digital elevation model. Potential relative radiation reflects the
shading contributions of terrain features, including canyon walls and northerly aspects, for a specific
time period (Pierce et al. 2005). Potential relative radiation values are scaled around 1 (flat ground) and
have values < 1 when terrain shelters a site and > 1 when the site is oriented to maximize exposure (e.g.,
southerly aspects).
Percent woody vegetation and terrain exposure were used to predict percent total August solar
radiation using beta regression in a model selection framework. Beta regression is a commonly used
form of regression that assumes the dependent variable is beta-distributed with values ranging from 0
to 1. Beta-regression also has a precision parameter that can be used to account for heteroskedastisity
and skewness. A global model included percent woody vegetation, terrain exposure, and their
interaction as predictors. Other candidate models included both woody vegetation and terrain
50
exposure terms but excluding an interaction term, as well as two single variable models with percent
woody vegetation and terrain exposure terms only. Candidate models were evaluated using Akaike’s
Information Criterion adjusted for small sample size (AICc). The model with the lowest AICc was
considered the most plausible model, and all other models within 4 ΔAICc unis of the best model were
also considered plausible. Akaike weights (wi) were used to assess the probability of each model being
the most plausible. Pseudo R2 was used to evaluate model fit.
Since percent woody vegetation and terrain exposure variables were measured in a GIS and are
available for redband streams in the Owyhee (Idaho, Nevada) and Salmon Falls Creek (Nevada) basins,
the best model was used to predict percent August solar radiation for all stream segments in those
basins.
Results:
Solar Pathfinder was used to measure the percent of total August solar radiation estimated to reach the
stream surface at 79 sites. At these sites, percent solar radiation ranged from 1 to 99%, indicating that a
wide range of conditions were measured. Likewise, the percent woody vegetation at the sites ranged
from 0 to 100%, and terrain exposure values ranged from 0.658 (low exposure due to terrain and
azimuth) to 1.06 (high exposure).
When candidate beta-regression models were fit to these data, the model with both percent woody
vegetation and terrain exposure terms, including their interaction term, was suggested to be the most
plausible(Table C-1; C-2). This model predicted percent solar radiation to be higher in highly exposed
terrain when woody vegetation was absent, but to be lower in highly exposed terrain when woody
vegetation was near 100% (Figure C-1, bottom panel). This most plausible model explained 56% of the
variance in percent August solar radiation (pseudo-R2 = 0.557).
After the model was applied to redband streams in the Owyhee and Salmon Falls Creek basin, predicted
percent solar radiation was higher in lower elevation streams with sparse woody vegetation and along
larger streams and rivers with wide valleys (Figure C-2). For example, the South Fork Owyhee River was
predicted to have a high percentage of August solar radiation reaching surface water because of low
percentages of woody vegetation, flat terrain, and a north-south oriented river course.
51
2
Table C-1. Pseudo-R , AICc, ΔAICc, and Akaike weights (wi) for candidate beta-regression models predicting percent August
solar radiation as a function of percent woody riparian vegetation and terrain exposure.
Pseudo-R2
0.557
0.502
0.515
0.001
Candidate Models
Percent Woody Vegetation x Terrain Exposure
Percent Woody Vegetation
Percent Woody Vegetation + Terrain Exposure
Terrain Exposure
AICc
-57.91
-55.15
-54.19
-1.81
ΔAICc
0.00
2.76
3.72
56.09
wi
0.71
0.18
0.11
0.00
Table C-2. Parameter estimates and standard errors for a beta-regression model predicting percent solar radiation as a
function of percent woody vegetation, terrain exposure, and their interaction.
Parameter
Intercept
Percent Woody Vegetation
Terrain Exposure
Percent Woody Vegetation x Terrain Exposure
bi
-2.594
4.444
4.236
-7.139
se(bi)
1.583
2.760
1.674
2.908
1.0
0.8
0.6
0.0
0.2
0.4
August Solar Radiati
52
0.2
0.4
0.6
0.8
1.0
0.6
0.8
1.0
Woody Vegetation (%)
0.0
0.2
0.4
August Solar Radiati
0.0
0.8
0.9
Terrain Exposure
(High)
1.0
(Low)
1.0
0.4
0.6
0.8
Low Exposure
High Exposure
0.0
0.2
August Solar Radiati
0.7
0.0
0.2
0.4
0.6
0.8
1.0
Woody Vegetation (%)
Figure C-1. Plots of Percent Woody Vegetation versus observed Percent August Solar Radiation (top panel), Terrain Exposure
versus observed Percent August Solar Radiation (middle panel), and predicted Percent Solar Radiation versus Percent Woody
Vegetation at low and high levels of Terrain Exposure from a beta regression model (bottom panel). Dotted lines are
standard errors of predictions.
53
Figure C-2. Percent August Solar Radiation predicted for redband streams in the Owyhee basin in Idaho and Nevada, and the
Salmon Falls Creek basin in Nevada. Percent solar radiation was predicted from percent woody vegetation and terrain
exposure using a beta-regression model.
54
Appendix D: Redband trout associations with instream and riparian habitat in Salmon
Falls Creek, Nevada
Objectives
•
•
•
•
•
Determine the interrelations between riparian vegetation, stream temperature, and other
stream habitat characteristics.
Evaluate how redband trout distribution is associated with instream and riparian habitat, and
non-native salmonids, with an emphasis on woody riparian vegetation and stream temperature.
Evaluate woody riparian vegetation and stream temperature as limiting factors to redband trout
abundance.
Assess trends in redband trout abundance over the last 10 years.
Determine how the fish assemblage is structured by instream and riparian habitat.
Methods
Fish Sampling - Fishes were sampled by electrofishing at 21 sites in the Salmon Falls Creek basin from
June to August, 2013. At each site, a stream reach 50 – 165m in length was established and isolated
with 6.35-mm bar mesh block nets. Fishes were sampled with multiple-pass electrofishing using a
Smith-Root LR-24 backpack electrofisher with one netter for 13 sites, 2 LR-24 backpack electrofishers
and four netters at seven sites, and by visual observation (1-person) at one site. Electrofishing was done
using pulsed (30-40 Hz) direct current and 200-450 V. All species were collected during the first
electrofishing pass. If one or more salmonids were collected during the first pass, then subsequent
passes were made until a decreasing number of salmonids were collected during subsequent passes or
until no salmonids were captured. All fishes were identified to species, and all salmonids were
measured to the nearest mm total length and weighed to the nearest g. Abundance of salmonids in
each reach was estimated using the Zippin removal method (Zippin 1958) as implemented in the FSA
package in Program R (R Development Core Team 2013). The one site where visual observation was
used was in an intermittent reach where fish were congregated into two small pools. Fish were easily
identified and the total length of counted individuals was estimated. The visual count was divided by a
0.8 detection probability as an estimate of abundance (Bozek and Rahel 1991).
Instream Habitat Survey – In association with fish surveys, instream habitat, streambank conditions, and
riparian vegetation were assessed using transect-based sampling. At each site, 10 transects were
established every 10-m along the reach beginning at the downstream reach boundary. Transects were
placed across the stream channel at bankfull height. Bankfull height was identified as the greenline
(Burton et al. 2011). Channel depth, water depth, stream substrate, and cover were recorded at 10
equidistant points along each transect (Platts et al. 1983). Channel depth (m) was measured as the
depth of the channel from bankfull elevation to the stream bottom. If the point fell within the wetted
portion of the channel, then water depth (m) was also measured. Stream substratum at each point was
classified according to the modified Wentworth scale as: bedrock, silt/clay (<0.064-mm diameter on baxis), sand (0.064-2-mm), gravel (2-15mm), pebble (15-64mm), cobble (64-256mm), and boulder (>256-
55
mm) (Cummins 1962). Cover was classified as: boulder, large wood (>10-cm diameter, >4-m in length),
small wood, rootwad, submergent vegetation, emergent vegetation, and undercut bank (>10-cm depth).
Channel unit type was classified as riffle, run, or pool based on water depth and velocity (Hawkins et al.
1993). Streambank alteration, woody vegetation height, and streambank stability and cover were
assessed within 25-cm of the transect between the greenline and the water’s edge using the US Bureau
of Land Management’s Multiple Indicators Monitoring protocol (Burton et al. 2011). Streambank
alteration was classified as none, cattle hoof prints, or cattle trail. Woody species height was classified
above each transect at the greenline as 0.0-0.5-m, 0.5-1.0-m, 1.0-2.0-m, 2.0-4.0-m, 4.0-8.0-m, and >8.0m. Streambank cover and stability was classified at each transect between the water’s edge and the
greenline. Streambank cover was classified as covered or uncovered. Streambank stability for erosional
banks was classified as fracture, slump, slough, eroding, or absent. Mean August temperature (°C) was
based on a spatially-explicit stream temperature model developed for streams in the Salmon Falls Creek
watershed (NorWeST project). Percent Woody Riparian vegetation was based on a supervised
classification of National Agricultural Imagery Program (NAIP) imagery from 2010.
Associations Among Instream and Riparian Habitat – The inter-relationships among instream habitat
variables at the 21 sites were assessed using a principal components analysis (PCA). The variables
included in the analysis were: watershed size (km2); Residual pool depth (m); Channel width:depth ratio;
Channel slope (%); Mean woody vegetation height (m); Percent woody vegetation within a 5-m stream
buffer; Percent bank sloughed or slumped (%); Percent submergent aquatic vegetation; Percent cobble
substrate; Percent fine substrate (silt, clay, or sand); Mean August temperature (°C); Density of brown
trout >100mm TL (#/m2). A scree plot was used to determine the number of meaningful principal
components for interpretation. The PCA was fit using scaled and centered data (mean = 0, SD = 1) and
the correlation matrix in princomp in Program R (R Core Development Team 2013). To visualize how
redband trout densities varied with habitat variables, densities (#/m2) <100 mm and >100 mm TL were
overlain on PCA biplots.
Redband Trout Associations with Instream Habitat – Instream and riparian habitat were evaluated for
their effect on redband trout occurrence and abundance (both <100 mm and >100 mm TL). Their effect
on occurrence was assessed using logistic regression models within a model selection framework. All
combinations of one and two-variable models were fit with the same variables used in the PCA;
candidate models were limited to two predictors variables to keep the ratio of sample size to number of
predictor variables to 10:1 or greater. PCA axis scores were also evaluated as predictors of occurrence.
Akaike’s information criterion adjusted for small sample size (AICc) was used to evaluate the plausibility
of all candidate models; the model with the lowest AICc value was considered the most plausible, but all
models within 4 AICc units of the best model were also considered plausible (Burnham and Anderson
2002). If multiple candidate models were plausible, then parameter estimates (and variances) were
averaged using Akaike weights and shrinkage (Burnham and Anderson 2002; Lukacs et al. 2010). In
addition to unstandardized parameter estimates, we computed standardized parameter estimates (from
models fit with data that were standardized with mean = 0, and standard deviation = 1) for comparison
of relative effect sizes for the variables included in plausible models. Variable importance was evaluated
using the sum of Akaike weights for models containing each variable. Fit of the most plausible model,
56
for each size class, was evaluated with a Hosmer Lemeshow test, and predictive performance was
evaluated using a 10-fold cross-validated Area Under the Curve (AUC) of a receiver operating
characteristic plot (Hosmer and Lemeshow 2000), where values of 0.5 indicate no discrimination
(predictive ability) and values of 1 indicate complete model discrimination. Logistic regression models
were fit using the glm function with a logit link in Program R (R Core Team 2013).
We evaluated instream and riparian habitat variables as limiting factors on redband trout density using
quantile regression models, again in a model selection framework (Cade and Noon 2003). Quantile
regression conditionally models the quantiles of a response variable distribution that may be more
informative in understanding ecological process – compared to a mean response - when not all variables
affecting the response are measured and included in a model. When upper quantiles are modeled they
effectively represent the potential maximum response to a variable at different levels of that variable,
and thus can be viewed as an evaluation of limiting factors (Cade and Noon 2003). Because of the small
number of sites sampled, we only evaluated single-variable models for their potential as limiting factors
to redband trout density. Thus, single variable models for the 90th percentile were fit using all instream
and riparian habitat variables, ln-transformed redband trout density (<100mm, >100mm, and all sizes)
was the response variable, and models were evaluated using the rqAICc statistic, which is AICc adjusted
for quantile regression (Cade et al. 2005). As before, the model with the minimum AICc was considered
the most plausible, and models within <4 rqAICc units were considered plausible as well. Model fit was
evaluated using quantile coefficient of determination (R1)(Cade et al. 2005).
Trends in Redband Trout Density – We resampled 10 sites in the Salmon Falls Creek basin that were
sampled by Idaho Department of Fish and Game in 2003 (IDFG 2007) to estimate trends in redband
trout densities. Previously sampled sites were located with a global positioning system coordinate at
the downstream reach boundary. We navigated to this reach boundary, which was used as the
beginning (or downstream) of the sampling reach during electrofishing surveys. Trends in redband trout
densities between 2003 and 2013 were estimated for two size classes (<100mm and ≥100mm TL) as
percent annual change in redband density (#/100m2; intrinsic rate of population change, r):
% 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑐𝑐ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 = 100 ∗ (𝑒𝑒𝑒𝑒𝑒𝑒
�ln�𝑁𝑁𝑦𝑦𝑦𝑦2 + 0.5� − ln�𝑁𝑁𝑦𝑦𝑦𝑦1 + 0.5��
− 1)
(𝑦𝑦𝑦𝑦2 − 𝑦𝑦𝑦𝑦1)
where Nyr2 is the estimated density of fish during the second sampling year (yr2 or 2013), and Nyr1 is the
estimated density of fish in the first sampling year (yr1 or 2003).
Fish Assemblage Associations with Instream Habitat - Associations between fish assemblage structure
and instream and riparian habitat was evaluated using non-metric multidimensional scaling (NMDS). A
Wisconsin-standardization on square root-transformed species abundances was used to compute NMDS
axis scores. Significance of instream and habitat variables was assessed by permuting the variables 1000
times to assess the likelihood of random associations with NMDS axes. Fit of variables was evaluated
using a squared correlation coefficient, and significance was assessed at α = 0.10. A thinplate spline was
fitted to ordination plots for selected significant variables to view non-linear relations between species
scores and habitat variables.
57
Results
Fish Sampling – Across the 21 sites sampled by backpack electrofishing (Figure D-1), redband trout were
collected at 11 sites, brook trout were collected at one site, and brown trout were collected at seven
sites (Table D-1). In addition to trout, the non-game species collected were: mottled sculpin, bridgelip
sucker, speckled dace, longnose dace, mountain sucker, northern pikeminnow, redside shiner,
chiselmouth, and largescale sucker (Table D-2). Speckled dace were collected at the most sites, whereas
the brook trout was only collected at one site. Likewise, speckled dace represented the highest
percentage of all fish caught (45%) and brook trout represented the lowest percentage (0.6%).
Associations Among Instream and Riparian Habitat – The PCA showed that most variation in instream
and riparian habitat was associated with a stream-size gradient, where larger streams had higher mean
August stream temperatures, deeper residual pools, lower channel slopes, and less woody vegetation
(PCA axis 1, 31% of variance; Figure D-2). A secondary gradient was one of stream disturbance, where
sites with more bank sloughing and slumping had more fine substrates (silts and sands), more aquatic
vegetation, and little woody riparian vegetation (PCA axis 2, 20% of variance; top panels of Figure D-2).
Brown trout were more abundant at these sites. A third gradient represented a substrate size gradient
that distinguished site with more cobble substrates from those with more fine substrates (PCA axis 3,
14% of variance; bottom panels of Figure D-2). A scree plot suggested the first three axes explained
most of the variance among instream and riparian habitats; therefore, PCA axes 4 and higher were not
interpreted. When redband trout density (#/m2) was plotted on the PCA biplots, it showed densities to
be higher in smaller streams, with no relation with the stream disturbance gradient, with a similar
pattern between small (<100mm; right panels of Figure D-2) and large (>100mm; left panels of Figure D2) redband trout. When redband trout densities were plotted against individual habitat and riparian
variables, these relationships were also apparent (Figure D-3).
Redband Trout Associations with Instream Habitat – Logistic regression models suggested that percent
woody riparian vegetation within a 5-m stream buffer was the strongest predictor of redband trout
occurrence, both for small (<100mm) and large (>100mm) trout. Specifically, the most plausible logistic
regression model for small redband trout included two measures of riparian vegetation: percent woody
vegetation and woody vegetation height (Table D-3; top panels in Figure D-4). When interpreted
together, the parameter estimates for those two variables suggested that small redband trout occur
more frequently when woody riparian vegetation is dense (covers a high percent of stream corridor) but
short (e.g., willows versus mature cottonwoods) (Table D-4). This most plausible model fit the data and
had good predictive ability (Hosmer Lemeshow: χ2= 3.45, P = 0.90; AUCxval = 0.80). Small redband trout
were also more likely to occur in higher gradient streams with few or no large brown trout (Tables D-3
and D-4; Figure D-4). Large redband trout were also more likely to occur at sites with more woody
riparian vegetation – the major driver of occurrence. The most plausible model with percent woody
vegetation and percent cobble substrate variables fit the data and had good predictive ability (Table D-3;
Hosmer Lemeshow: χ2= 5.01, P = 0.76; AUCxval = 0.78). Large redband trout were more likely to occur
when there was a higher percentage of woody vegetation and more cobble substrate at a site (Table D4; bottom panels of Figure D-4). However, model selection uncertainty suggested that other habitat
variables influenced occurrence as well (Table D-3). Parameter estimates suggested large redband trout
58
were more likely occur at sites with more cobble substrates, taller woody riparian vegetation, and fewer
brown trout (Table D-4). Other variables had lesser and more uncertain influences in occurrence (Tables
D-3 and D-4). Models with individual habitat variables were better predictors of redband trout
occurrence than models predicting occurrence from PCA axes scores that represented synthetic stream
size and stream disturbance variables (Table D-4).
Quantile regression models evaluating the potential of instream and riparian habitat variables as limiting
factors suggested that channel slope was most likely limiting abundance of small (<100mm) redband
trout; although not nearly as plausible, percent woody vegetation was the next most likely limiting
factor (Table D-5; Figure D-5). Percent woody vegetation and mean August stream temperature were
first and second-most plausible models for large (>100mm) and all redband trout abundance;
abundance could potentially be higher with more woody vegetation along the stream corridor and in
streams with cooler mean August temperatures (Table D-5; Figure D-5). Despite these most likely
models, parameter estimates for the 90th percentile were highly uncertain and included zero for every
model. Pseudo-R2 were low, ranging from 18.6 to 26.2%.
Trends in Redband Trout Density – Of the 10 sites that were able to be relocated with landowner
permission, two were dry and two were colonized and ponded by beaver. Thus, of the eight that were
revisited, including dry sites, all sites that had fish in 2003 decreased in the density of small and large
redband trout. The mean percent annual change in density was -5.6% (SE = 2.5) for small redband trout,
and -5.3% (SE = 2.4) for large redband trout (Table D-6).
Fish Assemblage Associations with Instream Habitat – Fish assemblage structure was associated with
several instream and riparian habitat variables, as revealed by the NMDS ordination. Stress was 0.067,
indicating a good ordination fit. Of the 10 habitat variables evaluated, seven were significantly
correlated with NMDS axes (Table D-7): temperature, slope, width:depth ratio, watershed size, percent
woody vegetation, percent fines, and percent cobble substrate. Redband trout were most abundant in
high gradient, cold streams, whereas longnose dace, largescale sucker, and chiselmouth were in the
largest, low-gradient streams that were warm (Figure D-6). In contrast, brook trout and the two sculpin
species were most abundant in streams with more cobble substrate and narrower channels, whereas
speckled dace, northern pikeminnow, and mountain sucker were more abundant at sites with wide
channels with more fines and less cobble substrate. Redband trout were most abundant when there
was a higher percentage of woody riparian vegetation.
59
Table D-1. Trout densities (#/100m2) for two size classes (<100mm and ≥100mm TL) at sites survey using
electrofishing in Salmon Falls Creek, NV.
Site ID
BULLCAMP1a*
CANCK01
CC01
CC02
CW00
CW01
CW02
JAKE01
LIME01
MFCAN01
NFSALM01
SF02
SF03
SF04
SF05
SF06
SFJAKE01
SFSF00
SHACK01
WILS01
Stream name
Bull Camp Creek
Canyon Creek
Camp Creek
Camp Creek
Cottonwood Creek
Cottonwood Creek
Cottonwood Creek
Jake Creek
Lime Creek
M Fk Canyon Creek
N Fk Salmon Falls Cr
Salmon Falls Creek
Salmon Falls Creek
Salmon Falls Creek
Salmon Falls Creek
Salmon Falls Creek
S Fk Jakes Creek
S Fk Salmon Falls Cr
Shack Creek
Wilson Creek
Date
8/5/2013
6/27/2013
7/26/2013
8/08/2013
7/10/2013
7/25/2013
7/24/2013
8/07/2013
7/15/2013
6/27/2013
7/16/2013
7/25/2013
7/22/2013
7/22/2013
7/23/2013
7/23/2013
8/06/2013
7/24/2013
7/9/2013
7/10/2013
Length
(m)
50
100
80
100
100
80
90
100
95
90
100
100
135
132
90
100
100
165
100
100
Area
2
(m )
39
401
294
325
322
563
295
284
226
154
590
1150
1000
1284
833
849
96
1168
195
303
Redband trout
<100mm ≥100mm
20.51
20.51
-2.24
-0.34
0.31
0.92
-1.86
----0.70
0.70
7.52
14.15
1.30
3.25
------------1.04
--0.26
--0.66
5.94
Brook trout
<100mm ≥100mm
--0.50
5.49
-------------------------------------
Brown trout
<100mm ≥100mm
---------------0.35
----0.85
1.86
--0.70
-0.39
-0.36
1.08
0.82
0.12
--0.26
1.03
-----
*Fish were counted using visual observation, and abundance was estimated using a visual observation efficiency of 80% for
small streams (Bozek and Rahel 1991).
Table D-2. Counts of non-game fish species collected during electrofishing surveys in Salmon Falls
Creek, NV.
Site ID
BULLCAMP1a*
CANCK01
CC01
CC02
CW00
CW01
CW02
JAKE01
LIME01
MFCAN01
NFSALM01
SF02
SF03
SF04
SF05
SF06
SFJAKE01
SFSF00
SHACK01
WILS01
Chiselmouth
----------1
14
16
-4
-1
---
Bridgelip
sucker
--3
---2
28
---2
6
1
3
1
-33
---
Largescale
sucker
----------2
5
5
2
-------
*Fish were counted using visual observation.
Mountain
sucker
-----15
4
--------------
Mottled
sculpin
-2
--33
---25
-----------
Paiute
sculpin
-6
-83
----4
----------37
Northern
pikeminnow
-----12
-----3
---------
Longnose
dace
---------3
4
1
2
-8
-2
---
Speckled
dace
--12
1
1
47
138
74
4
-95
12
79
228
48
143
-95
5
21
Redside
shiner
-----28
25
---87
45
92
74
81
79
-149
---
60
Table D-3. Candidate logistic regression models predicting redband trout occurrence (<100mm and
>100mm) as a function of instream and riparian habitat variables in Salmon Falls Creek, NV.
Candidate models
K
LogLikelihood
AICc
ΔAICc
wi
Redband trout <100mm
Percent Woody Vegetation + Woody Vegetation Height
Percent Woody Vegetation + Slope
Percent Woody Vegetation
Percent Woody Vegetation + Brown trout >100mm
3
3
2
3
-3.63
-4.04
-6.22
-5.29
14.7
15.5
17.1
18.0
0.00
0.83
2.45
3.32
0.466
0.308
0.137
0.089
PC1
PC1 + PC2
PC1 + PC3
2
3
3
-9.633
-9.079
-9.613
23.9
25.6
26.6
0.00
1.64
2.70
0.588
0.259
0.152
Redband trout >100mm
Percent Woody Vegetation + Percent Cobble
Percent Woody Vegetation
Percent Woody Vegetation + Woody Vegetation Height
Percent Woody Vegetation + Percent Aquatic Vegetation
2
Percent Woody Vegetation + Brown Trout Density (# ≥100mm / m )
Percent Woody Vegetation + Watershed size
Percent Woody Vegetation + Slope
Percent Woody Vegetation + Temperature
Percent Woody Vegetation + Percent Fines
Percent Woody Vegetation + Percent Slough Slump
Percent Woody Vegetation + Channel Width:Depth Ratio
3
2
3
3
3
3
3
3
3
3
3
-6.468
-7.888
-7.038
-7.465
-7.476
-7.534
-7.612
-7.709
-7.770
-7.860
-7.881
20.348
20.442
21.488
22.342
22.363
22.480
22.635
22.829
22.952
23.133
23.174
0.000
0.094
1.140
1.994
2.016
2.133
2.287
2.481
2.605
2.785
2.826
0.201
0.192
0.114
0.074
0.073
0.069
0.064
0.058
0.055
0.050
0.049
PC1 + PC3
PC1
PC1 + PC2 + PC3
PC1 + PC2
3
2
4
3
-9.180
-11.217
-9.153
-11.177
25.8
27.1
28.8
29.78
0.00
1.33
3.03
3.99
0.535
0.275
0.117
0.073
Table D-4. Unstandardized and standardized unconditional parameter estimates (model averaged using
shrinkage) and standard errors for logistic regression models predicting redband trout occurrence in
Salmon Falls Creek, NV. Akaike weights (wi) were summed across all models as a measure of variable
importance.
Variable
𝛽𝛽�𝚤𝚤̅
Unstandardized
𝑆𝑆𝑆𝑆�
�
𝛽𝛽
𝚤𝚤
Redband trout <100mm
Intercept
Percent Woody Vegetation
Woody Vegetation Height
Slope (%)
2
Brown Trout Density (# ≥100mm / m )
-21.880
0.327
-0.814
0.695
-0.198
17.104
0.221
1.151
1.743
1.021
Redband trout >100mm
Intercept
Percent Cobble Substrate
Percent Woody Vegetation
Woody Vegetation Height
Percent Aquatic Vegetation
2
Brown Trout Density (#≥100mm / m )
2
Watershed Size (km )
Slope (%)
Temperature (C)
Percent Fine Substrate
Percent Bank Sough/Slump
Channel Width:Depth Ratio
-8.806
0.016
0.135
0.159
-0.002
-0.075
-0.00007
-0.019
-0.012
-0.0008
0.0004
-0.0004
3.419
0.039
0.047
0.642
0.013
0.057
0.00005
0.015
0.009
0.0006
0.0003
0.0003
𝛽𝛽�𝚤𝚤̅
Standardized
Sum(wi)
-3.980
7.007
-0.848
1.011
-0.198
𝑆𝑆𝑆𝑆�
�
𝛽𝛽
2.904
4.747
1.199
2.537
1.021
1.000
1.000
0.308
0.137
0.089
-0.474
0.267
2.890
0.166
-0.042
-0.075
-0.046
-0.028
-0.021
-0.018
0.007
-0.003
0.632
0.650
1.014
0.669
0.220
0.057
0.035
0.021
0.016
0.014
0.006
0.003
1.000
0.201
1.000
0.114
0.074
0.073
0.069
0.064
0.058
0.055
0.050
0.049
𝚤𝚤
61
Table D-5. Parameter estimates with lower and upper confidence intervals (rank inversion method), and
model selection statistics for single-variable quantile regression models predicting the 90th percentile of
ln(redband trout abundance+1) as a function of instream and riparian habitat variables in Salmon Falls
Creek, NV.
Variable
𝛽𝛽�𝚤𝚤̅
Parameter (τ = 0.90)
𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 �
�
𝛽𝛽
Redband trout <100mm
Slope (%)
0.690
Redband trout >100mm
Temperature (C)
Percent Woody Vegetation
-0.843
0.044
-0.957
-0.064
Redband trout (all sizes)
Temperature (C)
Percent Woody Vegetation
-0.981
0.053
-1.133
-0.062
-0.067
𝚤𝚤
𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈𝑈 �
�
𝛽𝛽
rqAICc
ΔrqAICc
Sum(wi)
65.59
0.00
1.000
0.543
0.047
68.31
70.75
0.00
2.44
0.772
0.228
0.069
0.057
75.58
77.53
0.00
1.95
0.727
0.273
+∞
𝚤𝚤
Table D-6. Trends in redband trout densities (#/100m2) for two size classes (<100mm and ≥100mm TL)
from 2003 to 2013 in the Salmon Falls Creek basin. Trends are represented as a percent annual change.
Site ID
BULLCAMP2
CANCK01
CC01
CW00
CW01
CW02
MFCAN00
MFKDEER
NFSALM01
NFSALM02
NFSALM03
NFSALM04
SF00
SF02
SF03
SF04
SHACK01
SHOCR01
WILS01
MEAN (SE)
Stream name
Bull Camp Creek
Canyon Creek
Camp Creek
Cottonwood Creek
Cottonwood Creek
Cottonwood Creek
MF Canyon Creek
MF Deer Creek
NF Salmon Falls Creek
NF Salmon Falls Creek
NF Salmon Falls Creek
NF Salmon Falls Creek
Salmon Falls Creek
Salmon Falls Creek
Salmon Falls Creek
Salmon Falls Creek
Shack Creek
Shoshone Creek
Wilson Creek
2003 (IDFG)
<100mm ≥100mm
0.00
0.00
0.00
0.00
1.39
0.23
1.23
3.47
3.28
0.35
0.00
0.31
8.79
2.20
11.40
15.71
0.00
0.38
0.00
0.36
0.00
4.18
0.00
0.70
0.00
0.00
0.00
0.27
0.00
0.00
0.00
0.75
51.02
51.02
0.00
0.00
6.74
2.95
2013
<100mm
DRY
No access
Beaver pond
0.00
0.00
Beaver pond
Not sampled, difficult access
DRY
No access
No access
No access
Not sampled, difficult access
No access
0.00
0.00
0.00
0.00
No access
Not sampled, difficult access
≥100mm
% change
<100mm ≥100mm
0.0
0.0
1.86
0.00
-5.2
-8.4
-2.2
-2.3
-12.9
-14.0
0.0
0.0
0.0
-18.2
-1.9
0.0
-3.9
-18.2
-5.6 (2.5)
-5.3 (2.4)
0.00
0.00
0.00
0.00
62
Table D-7. Axis scores, squared Pearson correlation coefficient (R2), and significance (P) from a nonmetric multidimensional scaling (NMDS) ordination for instream and riparian habitat variables on fish
assemblage structure in Salmon Falls Creek, NV.
Variable
Temperature (C)
Slope (%)
Width:depth ratio
2
Watershed size (km )
Woody Vegetation Height (m)
Percent Woody Vegetation
Percent Bank Slough/Slump
Percent Aquatic Vegetation
Percent Fines
Percent Cobble
NMDS1
0.732
-0.875
0.811
0.630
-0.994
-0.999
-0.104
0.993
0.441
-0.256
NMDS2
-0.681
0.484
0.585
-0.777
0.109
0.036
-0.995
-0.114
0.897
-0.967
R2
0.521
0.468
0.252
0.493
0.143
0.426
0.053
0.089
0.330
0.509
P
0.003
0.006
0.080
0.004
0.257
0.010
0.639
0.481
0.032
0.004
63
Figure D-1. Locations of fish and temperature monitoring in Upper Salmon Falls Creek. Redband trout habitat from redband
trout rangewide database (May et al. 2012).
64
JAKE01
JAKE01
4
3
PFines
BNTgt_den
SFJAKE01
SF0
PctAqVeg
PctSloSlu
1
PctSloSlu
SFSF00
NFSALM01
NFDRY01
2
2
PctAqVeg
PC2 ( 20.2 %)
NFSALM01
NFDRY01
1
PC2 ( 20.2 %)
RBT
Present
Absent
3
4
RBT
Present
Absent
PFines
BNTgt_den
SFJAKE01
CW02
CW02
-1
CW01
BULLCAMP1a
PctWoody
MFCAN01
WILS01
Temperature
LIME01
PCob
ResPoolDepth
WoodVegHt
-1
0
1
MFCAN01
WILS01
SF04
SF03
SF02
WD_Ratio
2
-2
PCob
ResPoolDepth
SF06
WatSize_km2
SHACK01CC01
WoodVegHt
-1
0
PC1 ( 30.8 %)
ResPoolDepth
Temperature
3
RBT
Present
Absent
BNTgt_den
SFSF00
CC02
WILS01
PctAqVeg
CC01
SF0
SF06
ResPoolDepth
Temperature
PctWoody
0
CANCK01
MFCAN01
LIME01
SF04
PctSloSlu
-1
NFSALM01
2
SF0
SF06
LIME01
WatSize_km2
JAKE01
WD_Ratio
Slope_pct
SFJAKE01
SF04
PctSloSlu
SF03
BULLCAMP1a
SF02
WD_Ratio
Slope_pct
SFJAKE01
SF02
CW02
CW02
PFines
CW01
SHACK01
SHACK01
-2
NFDRY01
-2
-1
WatSize_km2
JAKE01
SF03
BULLCAMP1a
PFines
CW01
-2
3
WoodVegHt
-1
1
SFSF00
PctAqVeg
PctWoody
2
CW00
PC3 ( 14.1 %)
BNTgt_den
WoodVegHt
0
PC3 ( 14.1 %)
2
CW00
CANCK01
MFCAN01
1
PCob
1
3
RBT
Present
Absent
PCob
CC01
SF04
SF03
SF02
WD_Ratio
PC1 ( 30.8 %)
NFSALM01
CC02
WILS01
Temperature
LIME01
CW00
3
CW01
BULLCAMP1a
PctWoody
CC02
CANCK01
SF06
WatSize_km2
SHACK01CC01
CW00
-2
0
Slope_pct
-1
0
Slope_pct
CC02
CANCK01
SF0
SFSF00
0
1
PC1 ( 30.8 %)
2
3
NFDRY01
-2
-1
0
1
2
3
PC1 ( 30.8 %)
Figure D-2. Biplots of principle components analysis axis scores (top panels: axes 1 vs 2; bottom panels:
axes 1 vs 3) showing interrelationships among stream habitat characteristics (arrows) and stream sites in
Salmon Falls Creek, Nevada. Variables are: WatSize_km2 = watershed size (km2); ResPoolDepth =
Residual pool depth (m); WD_Ratio = channel width:depth ratio; Slope_pct = channel slope (%);
WoodyVegHt = Mean woody vegetation height (m); PctWoody = Percent woody vegetation within a 5-m
stream buffer; PctSloSlu = Percent bank sloughed or slumped (%); PctAqVeg = Percent submergent
aquatic vegetation; PCob = Percent cobble substrate; PFines = Percent fine substrate (silt, clay, or sand);
Temperature = Mean August temperature (°C); BNTgt_den = Density of brown trout >100mm (#/m2).
Site names are shown on Figure D-1. Site symbols are scaled by the density (#/m2) of redband trout
<100mm TL (left panels) and >100mm TL (right panels).
PFines
100
2
3
2
1
0
log(RBT / m +1)
3
2
2
3
2
1
0
log(RBT / m +1)
3
2
1
2
1
0
2
4
3
PctWoody
log(RBT / m +1)
3
2
1
2
6
0
Temperature
3
2
1
40
0
40
Conductivity
3
2
1
2
500
0
log(RBT / m +1)
3
300
20
PPeb
2
100
40
0
2
20
1
2
17
20
PctAqVeg
log(RBT / m +1)
3
2
1
0
0
log(RBT / m +1)
3
2
15
20 40 60 80
PCob
1
13
5
0
2
2
10 20 30 40
0
11
4
PctUndBnk
log(RBT / m +1)
3
2
1
2
2
60
0
PBldr
log(RBT / m +1)
3
2
1
2
log(RBT / m +1)
0
20
40 60 80
0
log(RBT / m +1)
3
2
1
2
log(RBT / m +1)
0
0
PctSmWd
0
3
PctSloSlu
8
ResPoolDept
WoodVegHt
log(RBT / m +1)
2
3
2
1
log(RBT / m +1)
3
2
1
0
2
log(RBT / m +1)
0
20
PctBnkImp
4 6
2
Slope_pct
0
0.0 0.4 0.8 1.2
0
2
0 1 2 3 4 5
20 40 60 80
0 2
1
2
log(RBT / m +1)
3
2
1
30
WD_Ratio
0
0.3
WDepth
0
2
20
0.1
WWidth
log(RBT / m +1)
3
2
1
0
2
log(RBT / m +1)
WatSize_km
10
0
log(RBT / m +1)
3
2
0 2 4 6 8
1500
0
500
1
2
0
0
log(RBT / m +1)
3
2
1
0
2
log(RBT / m +1)
65
0.0
1.0
BNTgt_den
Figure D-3. Biplots of stream characteristics versus redband trout density (all sizes) in 21 stream sites in
Salmon Falls Creek, NV. Variables are: WatSize_km2 = watershed size (km2); WWidth = mean wetted
width (m); WDepth = Mean water depth (m); ResPoolDepth = Residual pool depth (m); WD_Ratio =
channel width:depth ratio; Slope_pct = channel slope (%); WoodyVegHt = Mean woody vegetation
height (m); PctWoody = Percent woody vegetation within a 5-m stream buffer; PctBnkImp = Percent
streambank impacted; PctSloSlu = Percent bank sloughed or slumped (%); PctUndBnk = Percent
undercut bank; PctAqVeg = Percent submergent aquatic vegetation; PctSmWd = Percent small wood;
PBldr = Percent boulder substrate; PCob = Percent cobble substrate; PPeb = Percent pebble substrate;
PFines = Percent fine substrate (silt, clay, or sand); Temperature = Mean August temperature (C);
Conductivity = conductivity (µS/cm); BNTgt_den = Density of brown trout >100mm (#/m2).
1.0
1.0
66
0.0
0.2
0.4
0.6
Probability
0.6
0.0
0.2
0.4
Probability
0.8
Redband <100mm
0.8
Redband <100mm
20
40
60
80
0
1
2
5
1.0
Redband >100mm
0.8
0.6
0.2
0.0
0.0
0.2
0.4
0.6
Probability
0.8
Redband >100mm
0.4
Probability
4
Woody Veg. Height (m)
1.0
Woody Vegetation (%)
3
20
40
60
80
Woody Vegetation (%)
0
10
20
30
40
50
Cobble (%)
Figure D-4. Predicted probability of redband trout occurrence (<100mm and >100mm) versus percent
woody vegetation, wooyd vegetation height, and percent cobble substrate in Salmon Falls Creek, NV.
Open points represent observed presences (1) or absences (0).
20
20
67
Redband <100mm
Redband <100mm
R1 = 20.3
15
0
5
10
2
RBT / m
10
0
5
RBT / m
2
15
R1 = 18.6
0
20
40
60
80
100
0
1
2
5
6
15
Redband >100mm
R1 = 22.5
0
5
10
2
RBT / m
10
0
5
RBT / m
2
15
Redband >100mm
R1 = 23.2
0
20
40
60
80
11
100
12
13
14
15
16
17
Temperature (C)
40
40
Woody Vegetation (%)
30
Redband all
R1 = 26.2
20
10
0
0
10
20
RBT / m
2
30
Redband all
R1 = 25.5
2
RBT / m
4
Slope (%)
20
20
Woody Vegetation (%)
3
0
20
40
60
80
Woody Vegetation (%)
100
11
12
13
14
15
16
17
Temperature (C)
Figure D-5. Redband trout density versus Percent Woody Vegetation (5-m stream buffer), Percent
channel slope, and Mean August Stream Temperature (°C) at 21 sites in Salmon Falls Creek, NV. Lines
represent 50th (solid line) 75th (long dash), and 90th (dotted) percentiles predicted from quantile
regression models. Pseudo R-squared (R1) is noted for each 90th percentile quantile regression model.
-2
-1
0
1
16
2
LND
CSM
LSS
PSC
-1
0
65
1
2
1.6
2.2
-1
0
NPM
-0.5
BLS
MSC
2
RSS
BNT
LND
CSM
LSS
PSC
-1.5
-1.5
0.5 1.0 1.5
NMDS2
10
0.5 1.0 1.5
-0.5
NMDS2
21
NMDS1
1
20
0
17
-1
RBT
14
PSC
BKT
-2
MTS
SPD
15
RSS
BNT
LND
CSM
LSS
16
15
40 35
MSC
19
MTS
NPM
30
2
Width:Depth ratio
18
5
BLS
1
NMDS1
SPD
25
RSS
BNT
LND
CSM
LSS
BKT
-2
Cobble (%)
RBT20
NPM
6
0.
BLS
NMDS1
10
1
1.8
0.5 1.0 1.5
0.5 1.0 1.5
2
8
0.
PSC
-1.5
0
2
MTS
SPD
MSC
LSS
PSC
-1.5
RBT
-0.5
RSS
BNT
50 CSM
LND
BKT
-1
1
Slope (%)
2.4
45
40
BLS
55
-0.5
NMDS2
NPM
60
NMDS2
70
SPD
RBT
-2
RSS
16.5 BNT
NMDS1
MTS
70
NPM
BLS
BKT
-2
Woody Vegetation (%)
MSC75
MTS
SPD
15.5
MSC
NMDS1
65
5
13.
1.
2
-1.5
PCob
15
14.5
RBT
-1.5
Temperature
WatSize_km2
PSC
BKT
14
13
MSC
12
13
1.
4
RBT
PctWoody
Temperature (C)
-0.5
WD_Ratio
SPD
NPM
BLS
RSS
BNT
LND
CSM
LSS
0.5 1.0 1.5
Slope_pct
MTS
NMDS2
0.5 1.0 1.5
PFines
-0.5
NMDS2
68
BKT
-2
-1
0
1
2
NMDS1
Figure D-6. Non-metric multidimensional scaling plots (axes 1 vs 2) showing relations between fish
species and instream and riparian habitat variables. Surfaces for inidivdual varaibles are a thinplate
spline. Variable abbrebiations are: PctWoody = Percent Woody Vegetation; Slope_pct = Percent slope;
PCob = Percent cobble; PFines = Percent fines; WD_Ratio = channel width:depth ratio; WatSize_km2 =
Watershed size (km2); Temperature = Mean August Stream Temperature (C). Species abbreviations are:
RBT = redband trout; MSC = mottled sculpin; BLS = bridgelip sucker; SPD = speckled dace; LND =
longnose dace; MTS = mountain sucker; NPM = northern pikeminnow; RSS = redside shiner; BNT =
brown trout; BKT = brook trout; CSM = chiselmouth; LSS = largescale sucker.
69
Appendix E. A model to predict smallmouth bass distribution above Hells Canyon and
Below Shoshone Falls.
Objective
•
•
Develop a model to predict the probability of occurrence for smallmouth bass in streams and
rivers as a function of landscape-scale habitat variables.
Use the model to predict smallmouth bass occurrence probability for all stream segments in the
Snake River Basin above Hells Canyon and below Shoshone Falls.
Methods
Presence-absence data for smallmouth bass (Micropterus dolomieu) were used in conjunction with
landscape-scale environmental variables to fit a random forest model that can predict the probability of
occurrence for smallmouth bass in streams and rivers in the Lower Snake River basin above Hells Canyon
and below Shoshone Falls. Random forests represent a flexible modeling approach that can model high
order interactions and non-linear relationships between environmental variables and species occurrence
without a priori model specification (i.e., predefined variable interactions or non-linear responses
between predictor and response variables) that is needed in traditional modeling approaches such as
generalized linear models (Olden et al. 2008; Evans et al. 2011). Fish data were from a database
compiled for the Upper Snake River basin above Hells Canyon (e.g., Idaho Department of Fish and Game;
Oregon Department of Fish and Wildlife; US EPA; Chris Walser, College of Idaho; Idaho Department of
Environmental Quality); data from reservoirs were excluded. Landscape-scale variables evaluated were:
Mean Annual Streamflow (m3/s; USEPA and USGS 2005); Stream Segment Slope (%; USEPA and USGS
2005); Mean Annual Precipitation (cm; USEPA and USGS 2005); Mean August Stream Temperature (°C;
NorWeST); Cumulative Reservoir Storage (m3/km2; USACE 2008); Percent 100-m Stream Buffer
Converted (USEPA 2001); Percent Watershed Converted (USEPA 2001); and road density in the
watershed (km/km2; US Census Bureau 2001). The approach of Murphy et al. (2010) was used to screen
variables based on their importance (via the Gini Index) and contribution to explaining smallmouth bass
occurrence; final models were refit with the remaining variables that best explained species
occurrences. Final species-specific models were then used to predict probability of occurrence for all
species for perennial stream segments in the NHDPlus data set. Model predictive ability was assessed
by using 10-fold cross-validated Area Under the Curve (AUC) of a plot of a receiver operating
characteristic plot (i.e., a plot of 1-specificity vs. sensitivity across a range of probability values), which is
a measure of model performance ranging for 0.5 (no discrimination ability) to 1.0 (complete
discrimination) that is unaffected by species prevalence (proportion of sites where species was
observed)(Hosmer and Lemeshow 2000; Manel et al. 2001).
Results
Of the 661 records in the fish database, smallmouth bass were present at 45 sites (6.8%). Seven
variables predicted smallmouth bass occurrence. In order of predictive importance, the variables were:
Mean August Stream Temperature, Stream Segment Slope; Mean Annual Streamflow, Mean Annual
70
Precipitation, Percent of 100-m Stream Buffer Converted Land Use; Percent Watershed Converted Land
Use; and Reservoir Storage (Table E-1). Smallmouth bass were more likely to occur in larger, warmer
streams of higher gradient with lower mean annual precipitation. They were also more likely to occur in
streams and rivers with watersheds were land cover has been converted for agriculture or other uses
(Figure E-1). Spatial predictions of occurrence probability shows the Owyhee River basin to have the
highest probability of occurrence, as did the lower portions of Salmon Falls Creek, Boise River, Weiser
River, and Malheur River (Figure E-2). The 10-fold cross-validated AUC was 0.92 (sensitivity = 0.87;
specificity = 0.85), indicating the model had excellent predictive performance.
Table 1. Mean decrease in Gini Index values as a measure of variable importance in a random forest
model predicting smallmouth bass occurrence.
Variable
Mean August Stream Temperature (C)
Slope (%)
Mean Annual Streamflow (m3·s-1)
Mean Annual Precipitation (cm)
Percent Watershed Converted
Percent of 100-m Stream Buffer Converted
reservoir storage (m3·km-2)
Mean Decrease Gini Index
16.8
13.2
13.1
12.0
8.8
7.6
7.0
71
10
15
20
0.00
0.05
0.10
0.15
0.20
0.25
maugstrtemp
slope
Partial Dependence on m
Partial Dependence on m
-5
-1.00
-4
-3
-0.90
-2
-0.80
-1
5
Partial Dependence on s
-6 -5 -4 -3 -2 -1
-4
-3
-2
-1
Partial Dependence on m
1000
2000
3000
4000
5000
6000
mannflow
400
600
800
1000
1200
1400
maprecip
Partial Dependence on b
-4.0
-3.0
-2.0
-4.0 -3.0 -2.0 -1.0
Partial Dependence on c
200
-1.0
0
0
20
40
60
80
cpconv
0
20
40
60
buf_pconv
-4.0
-3.0
-2.0
Partial Dependence on d
0
50000
150000
250000
damstor
Figure E-1. Partial dependence plots from a random forest model predicting smallmouth bass
occurrence as a function of: maugstrtemp = Mean August Stream Temperature (°C), slope = Slope (%),
mannflow = Mean Annual Streamflow (m3·s-1), maprecip = Mean Annual Precipitation (cm), buf_pconv =
Percent of 100-m Stream Buffer Converted, cpconv = Percent Watershed Converted, and damstor =
reservoir storage (m3·km-2).
72
Figure E-2. Probability of occurrence predictions for smallmouth bass in the Snake River Basin above
Hells Canyon.
73
Appendix F. Conservation Success Index (CSI) Application
Objective:
•
•
Assemble and summarize relevant spatial data within subwatersheds related to redband trout
distribution, populations, habitat quality, and future threats.
Interpret the subwatershed summaries based on current scientific understanding of the
significance of the particular data on aquatic species persistence and effects on habitat quality.
Methods:
Trout Unlimited developed the Conservation Success Index (CSI) to provide a strategic, landscape-scale
planning tool for cold-water conservation that is focused on watersheds (see Williams et al. 2007). The
CSI summarizes spatial (GIS) data at the subwatershed scale (12-digit hydrologic unit (NRCS USGS),
equivalent to approximately 10,000 acres) related to a broad suite of population metrics, anthropogenic
stressors, and environmental conditions and assigns the summaries a categorical score (5 through 1,
reflecting exceptional through poor condition) based on the best scientific understanding of the
significance of the particular data on aquatic species persistence and effects on habitat quality. The data
considered are not intended to comprise a comprehensive list of factors affecting instream habitat or
aquatic species, rather they include factors that exist as broadly available, mapped data.
The CSI is a species-specific assessment comprised of four groups of “indicators.” Each indicator is a
summary of several factors or metrics grouped thematically. For example, the Watershed Condition
indicator includes summaries of data related to factors which affect instream habitat condition,
especially through sedimentation: the footprint of road networks in watersheds, status of streams on
EPA’s 303d list for sediment impairment, and the presence of active sand and gravel mining operations
in the riparian zone. Data summaries in some cases are normalized by watershed area or stream
mileage within watersheds (e.g. percent agricultural land or diversions per stream mile) and in other
cases summarized just for the riparian zone. Each indicator receives a score and indicators are
organized into groups that can be summed for overall scores related to Range-wide Conditions,
Population Integrity, Habitat Integrity, and Future Security (main document,Figure 9). Scores can be
further organized to identify conservation strategies that may be appropriate in watersheds, given the
pattern of species occurrence, habitat condition, and likely future threats, providing a landscape-scale
blueprint for management efforts on public and private lands.
The following summaries provide an overview of the indicators and factors in the CSI. Table F-1 outlines
the scoring rules and data sources used for each indicator and metric.
Rangewide Conditions
The Rangewide Conditions indicator compares the observed distribution of redband trout to the
inferred historical distribution at multiple scales – across subbasins (how many historically occupied
HUC8s remain occupied across the study area), within subbasins (how many historically occupied
74
HUC12s remain occupied within each HUC8 in the study area), and within subwatersheds (how many
historically occupied stream miles remain occupied within HUC12s). These factors assess range
contraction as regional and watershed scales. The indicator also summarizes the percent of current
distribution that occurs in at least 2nd order streams, which indicates the degree to which the species is
not relegated to a distribution limited to headwaters streams.
Population Integrity
Population Integrity indicators take and interpret data directly from the Redband Rangewide Status
Assessment (May et al. 2012), including population density, habitat patch size, genetic status,
competition with non-native species, and life history diversity. Population density and habitat patch
size factor scores reflect that small, non-sustaining populations are more vulnerable to extirpation
(Soule, 1987). Genetic status scores, which reflect the degree of introgression with stocked rainbow
trout, address the reduced fitness that can be conferred by non-localized brood stocks. Non-native
species considered for the competition factor include brook and brown trout and smallmouth bass.
Distribution for these species is highlighted in the Rangewide Status Assessment and further clarified
using the smallmouth bass distribution models produced for this study (Appendix E). The variety of life
histories present in a population contributes to genetic variation essential for responding to
environmental changes and facilitates the ability to occupy a greater variety of habitats, mitigating risks
across space and time (McElhany et al 2000).
Overview: Habitat Integrity
The current condition of aquatic habitats is analyzed in the CSI through five Habitat Integrity indicators:
Watershed Condition, Temperature, Watershed Connectivity, Water Quality, and Flow Regime. We
summarized and scored individual metrics within each of these indicators, calculated the average and
minimum score by indicator, and summed indicator scores for an overall Habitat Integrity score.
Habitat Integrity – Watershed Conditions
Sedimentation is addressed through the Watershed Conditions indicator, which summarizes the miles of
303(d)-listed streams for sediment, the overall road density, and the ratio of road miles within the
riparian zone to stream miles in each subwatershed. These factors reflect the presence of sediment in
streams or the footprint of roads in watersheds, a source of fine sediments (Lee et al. 1997), which
smother benthic invertebrates, embed spawning substrates, and increase turbidity (Lloyd 1987; DaviesColley and Smith 2001).
Habitat Integrity – Temperature
The Temperature indicator assesses instream water temperatures by looking at the miles of stream
303(d)-listed for temperature, the predicted average August stream temperature from the NorWeST
database, and average August total direct solar radiation received by the stream surface, as modeled in
this study (Appendix C). 303(d) impairment for temperature reflects a departure from anticipated
natural water temperatures required to sustain aquatic biota. Stream temperatures and exposure are
interpreted in consideration of the observed relationship between redband abundance and average
summer stream temperatures below 16°C and ample shading (Meyer et al. 2010).
75
Habitat Integrity – Watershed Connectivity
The Watershed Connectivity factors compares the amount of currently connected habitat to the amount
of historically connected habitat within the entire connected stream network within subwatersheds and
interpret the overall count of stream x road intersections, the likely locations of culverts that limit
habitat connectivity. Increased hydrologic connectivity provides more habitat area and better supports
multiple life stages of aquatic species, an important viability criterion which increases their likelihood of
persistence (McElhany et al. 2000).
Habitat Integrity – Water Quality
The Water Quality indicator incorporates information on 303(d)-listed streams for toxicity and nutrients,
the amount of agricultural land, number of active mines, and number of oil and gas wells. Impaired
water quality, including reduced dissolved oxygen, increased turbidity, toxins, and nutrients associated
with land uses, reduces aquatic habitat suitability.
Habitat Integrity – Flow Regime
The Flow Regime indicator represents the count of dams and their storage capacity in each
subwatershed, the miles of canals that divert water from streams, the count of diversions per stream
mile, the amount of dense, early successional forest habitat, and the amount of private land in rural
residential land use. Natural flow regimes are critical to proper aquatic ecosystem function (Poff et al
1997) and dams, reservoirs, diversions, and canals alter flow regimes (Benke 1990).
Overview: Future Security
Threats to aquatic habitats are addressed in the CSI through five Future Security indicators - Land
Conversion, Resource Extraction and Development, Climate Change, Water Quality, and Land
Stewardship – which are evaluated at the subwatershed scale only (Table F-1).
Future Security – Land Conversion
The Land Conversion indicator evaluates the risk of unconverted private land being developed for
residential purposes. Such changes will likely reduce aquatic habitat quality and availability through
land disturbances and changes in water use (Stephens et al. 2008).
Future Security – Resource Extraction
The Resource Extraction indicator includes information on the amount of oil and gas leases, hard rock
mineral claims, renewable energy development resources, and potential dam sites. Increased resource
development will increase road densities, modify natural hydrology, increase water uses associated with
development, and increase the likelihood of pollution to aquatic systems. Dam construction is likely to
be associated with habitat loss, changes in flow regimes and habitat suitability, and increased likelihood
of invasion by non-native species.
Future Security – Climate Change
76
The Climate Change indicator includes several factors assesses the vulnerability of aquatic habitats
climate change based on a composite analysis of six risk factors: changes in precipitation and flow
regime based on winter precipitation type (snow vs. rain); predicted average August stream
temperatures for 2040; changes in flow volume based on precipitation models for 2050; ability of
watersheds to buffer changes in flow through base flow condition (groundwater vs. surface flows); heatrelated moisture loss measured through the Palmer Drought Severity Index; and changes in fire regime
associated with earlier spring warming.
The CSI identifies areas vulnerable to changes in flow timing and magnitude related to climate change.
Transitions in winter precipitation regimes throughout the western United States – especially from snow
to rain - may be associated with changes in spring peak flow timing and magnitude, summer low flow
magnitude, and increased likelihood of rain-on-snow events (Williams et al. 2009; Mantua et al. 2010).
For each watershed, we predict the transition in precipitation regime, where regimes include snowdominated (December through February mean temperature < - 1°C), mixed (December through
February mean temperature between – 1°C and 1°C), and rain-dominated (December through February
mean temperature > 1°C), based on current climate and forecasts for 2050.
Increasing water temperatures will displace species from portions of their current distribution (Williams
et al. 2009; Mantua et al. 2010). The CSI anticipates stream temperature increase using the forecast
average August temperature values reported in the NorWeST database for 2040 under the A1B warming
trajectory for 2040s (2030-2059) with accounting for differential warming of streams by using historical
temperatures to scale temperature increases so that cold streams warm less than warm streams (Isaak
et al. 2011). These values are interpreted based on the observed relationship between redband
abundance and summer water temperature less than 17°C
The CSI summarizes total annual precipitation forecasts for 2050 (Maurer et al. 2007) and characterizes
watersheds with a 10% or greater forecast increase in precipitation volumes as low risk, stable
precipitation volumes as moderate risk, and any forecast decrease in precipitation greater than 10% as
high risk.
Base Flow Index measures the ratio of base flows to total stream flows expressed as a percentage
(Wolock 2003). High base flow watersheds have groundwater or snow melt dominated flows, while low
base flow watersheds have surface run-off dominated flows. Watersheds with large components of
their annual flow provided by stable sources such as groundwater or snow are likely to have lower
fluctuations in flow in response to climate variability.
Heat-related moisture loss is forecast to overwhelm any increase in precipitation in much of the interior
western United States anticipated with changing climate causing a perpetual state of drought (Hoerling
and Eischeid 2007). Areas with low total annual precipitation volumes and high temperatures will be
especially at risk and likely to have less water available for instream flows (Haak et al. 2010).
77
Earlier spring snowmelt coupled with warmer spring temperatures are forecast to increase the duration,
extent, and severity of wildfire seasons as climates change, affecting instream habitats directly through
burning and indirectly through post-fire flooding and debris flow (Williams et al. 2009). Fire regime
changes in the western United States are likely to be particularly amplified in mid-elevation watersheds
currently dominated by fine fuels (Westerling et al. 2006).
Future Security – Sedimentation
The potential for new sedimentation through shallow landslides on unstable slopes (Shaw and Johnson
1995), particularly in habitats vulnerable to severe, stand replacing wildfire, is addressed through the
Sedimentation indicator. The Sedimentation indicator also looks at the susceptibility of riparian soils to
erosion and the rate of runoff, as measured by the soil K factor, and the fire regime condition class, a
measure of the departure of vegetation from expected historical fire regime and fuel characteristics.
Sediment smothers benthic invertebrates, embed spawning substrates, and increase turbidity (Lloyd
1987; Davies-Colley and Smith 2001); landslides and post-fire erosion can be significant sources of
sediment.
Future Security – Land Stewardship
Finally, the Land Stewardship indictor represents the fraction of each subwatershed with lands that have
a protected status. Protected lands have some mandate for conservation via federal, state, or other
conservation ownership with some additional regulatory or congressionally-established protections
(e.g., Wilderness Areas, Research Natural Areas, Areas of Critical Environmental Concern). Stream
habitats and watersheds with higher portions of protected lands are likely to experience less
anthropogenic disturbance than other lands.
Results:
CSI results are described in the main assessment document and can be explored in an online webmap.
78
Table F-1: Indicators and factors within the Redband Trout CSI and their scoring rules and datasources. Rangewide Conditions and
Population Integrity indicators are only scored for subwatersheds with redband trout present.
Range-wide Conditions
Group
Indicator
Factor
Persistence
across
subbasins
Percent of
historical HUC8s
currently
occupied
Percent of
historical
HUC12s
currently
occupied w/in
HUC8
Percent of
historical stream
habitat occupied
within HUC12
Percent of
habitats
occupied >= 2nd
order
Population
density (Fish/km)
Persistence
within
subbasins
Persistence
within
subwatersheds
Population Integrity
Persistence in
large streams
within
subwatersheds
Population
density
Habitat extent
Genetic status
Miles of
occupied,
connected
habitat
Genetic status
Score = 1
0 – 50%
0 – 20%
0 – 10%
0 – 10%
Score = 2
50 – 70%
20 – 40%
10 – 20%
Score = 3
70 – 80%
40 – 60%
20 – 35%
Score = 4
80 – 90%
60 – 80%
35 – 50%
90 – 100%
80 – 100%
50 – 100%
10 – 15%
15 – 20%
0 – 35
35 – 100
100 – 250
or
“Unknown”
250 – 625
> 625
<6
6 – 12
12 – 18
18 – 30
> 30
10 – 20%
1 – 10% or
“suspected
hybridized”
“Suspected
unaltered”
0 – 1%
> 20%
20 – 25%
Score = 5
25 – 100%
Data and Scoring Sources
Redband Rangewide Status
Assessment 2012 (May et al.
2012)
Redband Rangewide Status
Assessment 2012 (May et al.
2012)
Redband Rangewide Status
Assessment 2012 (May et al.
2012)
Redband Rangewide Status
Assessment 2012 (May et al.
2012); NHD Plus
Redband Rangewide Status
Assessment 2012 (May et al.
2012)
Redband Rangewide Status
Assessment 2012 (May et al.
2012); Haak and Williams
2012
Redband Rangewide Status
Assessment 2012 (May et al.
2012)
79
Population Integrity,
continued
Group
Indicator
Factor
Competition
Non-native trout
or bass present
Life history
diversity
Life history
diversity
Instream
habitat
Miles 303d listed
for sediment
Habitat Integrity
Road density
(miles/miles²)
Temperature
Roads in riparian
zone (miles road
within 200m of
streams/miles
stream)
Miles 303d listed
for temperature
Watershed
average
predicted
summer stream
temperature (C°)
Score = 1
-
-
Score = 2
Score = 3
Bass
observed
Bass
modeled at
probability
≥ 0.65
Trout
-
All forms
present
except
anadromous
-
Score = 4
> 0.1% of
streams
>= 4.7
1 - 0.5
4.7 - 3
0.5 - 0.25
3 - 2.5
0.25 - 0.1
23 - 26
19 - 23
None
All historical
forms present
0%
2.5 - 1.6
0.1 - 0.05
> 0.1% of
streams
> 26
Score = 5
< 1.6
0.05 - 0
0%
17 - 19
< 17
Data and Scoring Sources
Redband Rangewide Status
Assessment 2012 (May et al.
2012); Appendix E
Redband Rangewide Status
Assessment 2012 (May et al.
2012)
ID Dept. of Env. Quality,
2010; OR Dept. of Env.
Quality, 2004/06; NV Dept. of
Env. Protection 2006
US Census Bureau TIGER
2000; BLM ID Ground Transp.
2004; BLM OR Ground
Transp. 2009
EPA NHD Plus (1:100K); US
Census Bureau TIGER 2000;
BLM ID Ground Transp. 2004;
BLM OR Ground Transp. 2009
ID Dept. of Env. Quality,
2010; OR Dept. of Env.
Quality, 2004/06; NV Dept. of
Env. Protection 2006
Isaak et al 2011 (1993 – 2011
Average); Meyer et al. 2012
80
Group
Indicator
Watershed
connectivity
Factor
Watershed
average
predicted shade
(redband
streams only)
Ratio of current
maximum
stream network
connectivity (mi)
to historical
Score = 1
Score = 2
Score = 3
Score = 4
Score = 5
Appendix C
< 20%
20 – 35%
35 – 50%
50 – 75%
> 75%
< 50%
50 - 75%
75 - 90%
90 - 95%
> 95%
>= 12
8 - 11
5-7
1-4
0
> 175
100 - 175
50 – 100
20 – 50
0 - 20
Habitat Integrity,
continued
Barrier count
Road – stream
intersection
counts (likely
passage barriers)
Water quality
Miles 303d listed
for toxins or
nutrients
% urban or
agricultural land
use
Active mine
count
Data and Scoring Sources
> 0.1% of
streams
0%
58 - 100%
28 - 58%
15 - 28%
5 - 15%
0 - 5%
>= 10
7-9
4-6
1-3
0
Redband Rangewide Status
Assessment 2012 (May et al.
2012); USFWS NV Passage
Assessment Database 2013 ;
USACE National Inventory of
Dams 2008
Redband Rangewide Status
Assessment 2012 (May et al.
2012); USFWS NV Passage
Assessment Database 2013 ;
EPA National Hydrography
Dataset Plus (1:100K); US
Census Bureau TIGER 2000;
BLM ID Ground Transp. 2004;
BLM OR Ground Transp. 2009
ID Dept. of Env. Quality,
2010; OR Dept. of Env.
Quality, 2004/06; NV Dept. of
Env. Protection 2006
USGS National Landcover
Dataset 2006
USGS Minerals Resources
Data System (Active) 2005
81
Group
Indicator
Factor
Active oil/gas
well count
Habitat Integrity,
continued
Flow regime
Dam count
Ratio of dam
storage (ac-ft) to
stream miles
Miles canal
Future Security
Diversions per
stream mile
Score = 1
Score = 2
>= 400
300 - 400
200 - 300
50 - 200
0 - 50
>= 5
3-4
2
1
0
>= 2500
1000 2499
250 -999
1 - 249
0
>= 20
10 - 20
5 - 10
1-5
0-1
>1
0.6 - 1
Score = 3
0.4 - 0.6
Score = 4
0.2 - 0.4
Score = 5
0 - 0.2
Data and Scoring Sources
USGS Western Oil and
Natural Gas Wells 2004
USACE National Inventory of
Dams 2008
EPA NHD Plus (1:100K);
USACE National Inventory of
Dams 2008
EPA Nat'l Hydrography
Dataset Plus (1:100K)
EPANat'l Hydrography
Dataset Plus (1:100K); Idaho
DWR 2002
Conversion
risk
% vulnerable to
urban/ex-urban
conversion
80 - 100%
60 - 80%
40 - 60%
20 -40%
0 - 20%
D. Theobald - US Forests on
the Edge Spatially Explicit
Growth Model v3
Resource
extraction risk
% suitable for
solar
development
50 - 100%
25 - 50%
10 - 25%
1 - 10%
0 - 1%
% suitable for
geothermal
development
50 - 100%
25 - 50%
10 - 25%
1 - 10%
0 - 1%
A.L. Haak, unpublished data.
Suitability a function of
annual direct normal
irradiance >= 500
KwH/m2/day, slopes < 5%,
and within 50 mi. of 115-345
kV transmission lines.
BLM Geothermal potential
areas
82
Future Security, continued
Group
Indicator
Climate
change risk
Factor
Score = 1
Score = 2
Score = 3
Score = 4
Score = 5
Data and Scoring Sources
% suitable for
wind
development
50 - 100%
25 - 50%
10 - 25%
1 - 10%
0 - 1%
Count hydro
development
sites
>= 1 in
local subwatershed
>5 in
subbasin
3- 5 in
subbasin
1 - 2 in
subbasin
0 in subbasin
WPA and NREL, Wind
Resource Potential 2003;
excludes USGS Protected
Areas Database 1.3 2012
Idaho National Laboratory,
Hydropower Resource
Assessment 2004
% acreage of
mining claims
50 - 100%
25 - 50%
10 - 25%
1 - 10%
0 - 1%
BLM LR2000 2003
% in oil and gas
lease areas
50 - 100%
25 - 50%
10 - 25%
1 - 10%
0 - 1%
BLM Geocommunicator 2008;
excludes USGS Protected
Areas Database 1.3 2012
Winter
precipitation
regime change
risk (2050
current and
forecast average
winter temp °C)
2040 forecast
average summer
stream
temperature (°C)
Current
snow (< 1°C) and
future rain
(>1°C) or
mixed (-1
to 1°C)
Current and
future snow
(<-1°C) or rain
(> 1°C)
E. Maurer et al. "Fineresolution climate projections
enhance regional climate
change impact studies", Eos
Trans.AGU 88, (2007).
Flow volume
change risk I
(precipitation
forecast)
> 26
<90% of
current
levels
Current
mixed (-1
to 1°C) and
future rain
(>1°C) or
mixed (-1
to 1°C)
23 - 26
19 - 23
90 - 110%
of current
levels
17 - 19
< 17
>110% of
current levels
Isaak et al 2011 (Scenario
S30_2040D); Meyer et al.
2012
E. Maurer et al. "Fineresolution climate projections
enhance regional climate
change impact studies", Eos
Trans.AGU 88, (2007).
83
Future Security, continued
Group
Indicator
Factor
Score = 1
Flow volume
change risk II
(Base Flow)
0 - 33
(Surface
flow
regime)
Flow volume
change risk III
(Heat-related
moisture loss)
Upper 1/3
of temp or
below
average to
average
temps
W/in 1680
- 2690 m
elevation
range and
fine fuels
majority
Altered fire
regime risk
Sedimentation
% shallow
landslide risk
area
Average soil
erodibility (k
factor) within
riparian zone
Average Fire
Regime
Condition Class –
departure from
expected fire
regime
58 - 100%
0.3 – 0.5
70 – 100
(Highly
departed)
Score = 2
28 - 58%
0.25 – 0.3
50 – 70
Score = 3
Score = 5
Data and Scoring Sources
33 - 66
66 - 100
(Groundwater
or snowmelt
flow regime)
USGS Base Flow Index 2003
1 - 2 st. dev
above ave
precip or
middle 1/3
of temp
> 3 st dev
above ave
precip or
lower 1/3 of
temp
E. Maurer et al. "Fineresolution climate projections
enhance regional climate
change impact studies", Eos
Trans.AGU 88, (2007).
W/in 1680
- 2690 m
elevation
range and
fine fuels in
minority
Outside of
1680 - 2690
m elevation
range
USGS National Elevation
Dataset 30m; USGS/USFS
LANDFIRE Anderson 13 Fuel
Models
15 - 28%
0.2 – 0.25
30 – 50
Score = 4
5 - 15%
0.15 – 0.2
10 – 30
0 - 5%
0 – 0.15
0 – 10 (Least
departed)
USGS National Elevation
Dataset 30m; SMORPH
shallow landslide models
NRCS SSURGO soil survey;
EPA NHD Plus (1:100K) 200m
buffer
USGS LANDFIRE: Fire Regime
Condition Class 2009
84
Future Security,
continued
Group
Indicator
Factor
Land
stewardship
% public
ownership
% public
ownership with
protected status
Score = 1
0%
0%
Score = 2
0 - 30%
0 - 30%
Score = 3
30 - 50%
30 - 50%
Score = 4
50 - 90%
50 - 90%
Score = 5
Data and Scoring Sources
> 90%
USGS Protected Areas
Database 1.3 2012- GAP
status code 1,2,3
> 90%
USGS Protected Areas
Database 1.3 2012- GAP
status code 1,2