Evaluating the relative effectiveness and cost of contrasting techniques for determining trout passage over culvert restoration sites
Principal Investigators:
Dr. Helen Neville (Trout Unlimited, Boise, ID, )
Dr. Douglas Peterson (US Fish Wildlife Service, Helena, MT,).
September 15, 2010
Issue and Need:
Thousands of culverts across the western U.S. present passage barriers to inland trout (e.g., Hendrickson et al. 2008), and as a result the U.S. Forest Service and other agencies have dedicated a great deal of money towards removing or restoring culverts to allow passage. Determining whether or not fish are actually passing through restoration sites, however, is difficult, time-consuming and expensiveandin general there is little direct evidence of fish passage to confirm the effectiveness of these restoration efforts. Clearly, there is a need to develop effective methods for determiningsuccessful passage of aquatic organisms,not only to demonstrate “biological success” of restoration sites, but to develop the scientific basis for future decision making and project planning. Here, we outline a study plan to compare and contrast various techniques for measuring inland trout movement, with the aim of providing a protocol useful to agency managers for implementing effective and efficient techniques to determine fish passage on restoration sites of interest. We capitalize on an on-going study of culvert impacts and restoration effectiveness incutthroat trout in Montana, which we describe first before outlining our current study and plan for additional work.
Ongoing Lolo Creek Project: Background and Study Design
In 2008, the US Fish and Wildlife Service and Trout Unlimited, in collaboration with the LoloNational Forest, began a research project to evaluate the demographic and genetic response of cutthroat trout to barrier removal in the Lolo Creek watershed in western Montana (Figure 1). The study uses a combination of presence/absence surveys, estimates of relative abundance and age/size structure, and population-genetic sampling to measure the demographic and genetic response of westslope cutthroat trout to restored connectivity of the stream network. The objective is not simply to measure if fish pass the improved road crossings, but alsoto generalize how quickly, and by which mechanisms trout populations respond to restored connectivity.
The ongoing Lolo Creek study is modeled on a before-after-control-intervention (BACI) design, where changes (detected as changes in genetic diversity, relative abundance, size/age composition, occurrence, life history expression) are compared through time for sites where barriers are removed (intervention or treatment) relative to where they are not (controls). Control populations can either be populations that are expected to remain isolated or that are already “connected” and will remain so. Demographic, occupancy, and genetic sampling was carried out on all sites in 2008 (before culvert replacement for ‘treatment’ sites) and is scheduled for 2010 (after culvert replacement for ‘treatment’ sites) and 2012.
The Lolo Creek watershed was selected for this study based on LoloNational Forest plans to remove or replace numerous culverts that were identified as partial or total passage barriers. This provided an opportunity to concentrate multiple sample sites within a single geographic area, and thus control for some confounding variables (climate, hydrology, fish community) and reduce logistic and cost limitations associated with having sample locations spread over large areas.
Ongoing Lolo Creek Project: Methods and Sampling 2008
The 2008 field season was dedicated to selecting the specific streams and road crossings to include in the study, and conducting baseline (pre-barrier removal) demographic and genetic sampling. Field crews conducted preliminary surveys to determine fish occupancy in 12 tributaries in the upper Lolo Creek drainage, primarily in the East Fork of Lolo Creek (and its tributaries) and Granite Creek (Figure 1). From this collection of sites, we selected two “treatment streams” that are unnamed tributaries of East Fork Lolo Creek and where culverts ranked as “partial barriers” by USFS Region 1 were scheduled to be replaced with passage-friendly culverts in 2008 (Figure 2). These streams were selected because the target species (westslope cutthroat trout) occurs both upstream and downstream of the culvert location. We selected an “isolated control” in Sally Basin Creek, also tributary to East Fork Lolo Creek, where fish barriers exist but where no passage projects are planned (Figure 2).
The basic sampling design in each stream involved demographic and genetic sampling along a longitudinal profile that encompassed identified fish passage barriers. Within each stream, we selected 6-7 sample reaches above the culvert to be replaced (‘treatment’ sites) or the lower-most culvert (isolated ‘control’ site). To determine population structure of westslope cutthroat trout more generally in the connected East Fork Lolo Creek drainage we conducted occupancy sampling (presence/absence) and collected tissue from cutthroat trout at three additional locations (two in East Fork Lolo Creek, one in Lost Park Creek, referred to below as “background populations”). Collectively, the above samples were representative of the area in East Fork Lolo Creek occupied by westslope cutthroat trout.
Single and multi-pass backpack electrofishing was used to collect trout species for tissue samples (genetic analyses); determine occurrence, relative abundance, and size/age structure; and tag fish (Passive Integrated Transponders or PIT) forsubsequent mark-recapture to estimate movement directly and measure individual growth between years. Population genetic analysis of westslope cutthroat trout permits us to contrast genetic diversity and population structure between connected vs. isolated sites, and detect changes in the treatment streams that are associated with improved connectivity For example, we would expect to observe increased genetic diversity within and decreased genetic differentiation among sites after connectivity is restored. Sampling in treatment streams was conducted in July 2008, prior to replacement of the most downstream culverts in treatment streams 1 and 2 (see Figure 1).
Ongoing Lolo Creek Project: Brief summary of findings to date
Occupancy surveys conducted in 2008 indicated that native westslope cutthroat trout and nonnative brook trout were the most common fish species detected at all sites. In treatment streams, the relative abundance of westslope cutthroat trout generally decreased with increasing elevation (i.e., as one moved upstream). Brook trout have apparently not entirely invaded the isolated control stream (Sally Basin Creek; Figure 2). Multiple fish passage barriers (culverts) are present in treatment and control streams (Peterson, unpublished data). Thus trout in these streams may be subject to a serial barrier effect, where migratory fish will encounter multiple partial (or total) barriers to upstream movement. It has been proposed that the increased fecundity associated with migratory life histories can contribute to greater population resilience (Rieman and Dunham 2000; Neville et al. 2006) to environmental and biotic stressors, such as competition with nonnative brook trout (e.g., Peterson et al. 2008). The decreased relative abundance of westslope cutthroat trout in the two treatments streams with increasing distance from the main stem (Figure 2, and and increasing number of barriers, Peterson, unpublished data) is thus consistent with the hypothesis that migratory cutthroat trout only infrequently spawn in upstream stream reaches. The study is thus situated to test whether restoring passage for westslope cutthroat trout will facilitate biotic resistance to effects of brook trout invasion. If the biotic resistance hypothesis is correct, then we predict then that the relative abundance of westslope cutthroat trout in the upstream reaches of treatment streams 1 and 2 will increase through time as improved connectivity provides passage for the migratory life history.
Genetic analyses indicated significant genetic differentiation, lower genetic diversity, and smaller effective population sizes in the westslope cutthroat trout populations in the isolated control and two treatment streams relative to other sample locations within the connected East Fork Lolo Creek drainage (Neville and Peterson, unpublished data). These data are consistent with an “isolation effect” anticipated for trout populations fragmented by passage barriers(e.g., Wofford et al. 2005, Neville et al. 2006b). These results illustrate two points relative to management. First, even partial barriers can significantly limit gene flow in the westslope cutthroat trout populations. Second, efforts to improve connectivity in the treatment streams is clearly warranted, as very low effective populations sizes (Ne <50) indicate that westslope cutthroat trout populations may be at high risk of inbreeding depression.
Ongoing research – expanding the Lolo Creek project to include direct assessment of movement through culverts
Now that the culverts have been removed in our Lolo creek treatment sites, we are alsoevaluating whether resident fish are successfully passing through the replacement culvertsdesigned for passage by continuing the methods described above that focus on population-level responses (i.e., characterization of population genetics, occupancy and demographics). Additionally, we propose to use a capture-mark-recapture-detect (CMRD) design using state-of-the-art PIT tag technology (Figure 3), and implement a new genetic “sibship” method for direct evaluation of movement. Our objective is to develop and implement a hierarchical monitoring design allowing us to partition the data based on technique, cost, and effort, and determine the information gain relative to cost and effort. Our primary target species is native westslope cutthroat trout, but brook trout will be included in all field investigations (occupancy, CMRD, collection of tissue samples).
A capture-mark-recapture-detect (CMRD) design to measure fish movement past improved culverts
We plan to use a CMRD design to assess fish movement at improved road crossings. The design involves capturing and marking trout (with PIT tags) above and below the culvert or road crossing of interest, then using active capture (electrofishing), active detection (mobile PIT antenna) and passive detection (stationary PIT antennas) to measure movement of individuals past the road crossing. The proposed design integrates many of the approaches used to estimate fish passage at road crossings in the western US and elsewhere (e.g., Coffman 2005; Solcz 2007; Cahoon et al. 2007; Burford et al. 2009). A strength of this design is that we will be able to compare and contrast the relative effectiveness, and cost effectiveness, of techniques that collect data at discrete times across space (electrofishing, mobile antenna)versusone that relies on continuous data collection at a fixed location (stationary antenna). In practice the stationary antennas will be operated continuously at each target culvert from spring through fall, and active recapture and detection events will take place at multiple times during that same time period. In addition, the active sampling techniques differ in effort. Three people (one operator, two netters) are typical crew requirements for electrofishing in small streams; and time allowances must be made to handle and release capture fish. In contrast, a mobile back pack sized PIT tag antenna can be operated by a crew of one or two people, and no fish handling is required as the tagged fish will simply be detected (interrogated by the antenna) without physical capture.
Population-level vs. pedigree (sibship) genetic techniques to measure movement
Genetic data can be useful for monitoring a variety of biological questions, and in many cases provide information that is difficult or impossible to attain with more traditional methods (demographic, mark-recapture, telemetry), often at less expense (Neville et al. 2006a, Schwartz et al. 2006). However, while various genetic techniques are possible for evaluating passage over restored culverts, their relative effectiveness and costs have yet to be evaluated across differing field scenarios and species. Results from our above work as well as work on two otherwestern trout species (rainbow trout, Neville et al. 2009, as well as Lahontan cutthroat trout, Neville unpublished data), have demonstrated genetic impacts of culverts using ‘traditional’ population-level genetic metrics (i.e., heterozygosity, allelic richness, and estimates of effective population size). By contrasting isolated with non-isolated sites, this collective work shows that populations above culvert barriers generally have reduced genetic variability and smaller effective sizes (Ne) than populations in connected habitats(see also Wofford et al. 2005, Neville et al. 2006b). While the above population genetic metrics characterize population responses over time, another technique may also be effective in actually measuring movement directly. Hudy et al (2010) recently used genetic sibship analysis and family reconstructionto characterize movement of young-of-year (yoy) brook charr, and the authors of this study first realized the potential application of sibship analyses for addressing culvert passage (Ben Letcher, personal communication). The idea behind this approach is that finding siblings (as determined by genetic ‘fingerprinting’ and pedigree analysis) on both sides of a restored culvert indicates movement of yoy across the site;the direction of movement is inferred by ‘majority rule’, where it is assumed the family originated on the side of the culvert where the most siblings were captured. Initial simulations demonstrate that sibship analyses may have greater power than individual assignment-based methods (see below: Request for Further Funding)to capture movement(Ben Letcher, personal communication), and similarly this method is also likely to have greater power than the traditional population-based evaluations (e.g., using heterozygosity, Ne, differentiation) used in our previous work.
However, while the sibship approach offers great promise for evaluating immediate movement over culverts in some areas and with some species, it may not be logistically feasible or as powerful in certain scenarios we have faced in our study regions (USFS Regions 1 and 4, see Figure 4). For instance, the sib-ship approach would be logistically impossible in the an on-going study of Lahontan cutthroat trout (Neville, unpublished data), where culvert sites are at the confluence with the mainstem river andthe lower reaches of tributaries are dry the majority of the year, making sampling for siblings at the culvert sites physically impossible. In this system, movement through restored culvert sites and among tributaries likely occurs through larger individuals migrating when high flows seasonally connect the tributaries to the mainstem river. Similarly, in the rainbow trout study area (Neville et al. 2009), sib-ship analyses would not be possible because in the steep mountainous terrain of Idaho roads typically follow river channels, and thus culverts are at the tributary confluence with large mainstem rivers. In such instances, sampling yoy below the culvert would be impossible because suitable spawning habitat is minimal and habitat is too large to sample effectively for yoy in the mainstem river.
Furthermore, there may be important differences in the spawning and movement behavior between trout species (and other fishes/organisms) that affect the power of sibship analyses. For instance brook trout, the focus of the Hudy et al (2010) study, spawn in the fall and it was feasible for the authors to sample them 4 months post-emergence. In their study yoy were still ‘highly clumped’ at this time, yet the mean dispersal distance was over 50 meters in the tributary and 100m in the ‘mainstem’ river (assumedly a 2nd order stream). Cutthroat trout spawn in the early to late summer depending on the local thermal and hydrologic regimes, andfor logistical reasons we can only sample cutthroat trout at low flows during the late summer and early fall, when yoy have already emerged and are large enough tobe safely captured by electrofishing. As a result of the timing of sampling for yoy cutthroat trout, it is unlikely that dispersal distances will be as large in our study, which may affect the power of sibship analyses for addressing culvert passage. In some sense, assuming cutthroat trout yoy disperse a minimum distance large enough to span a culvert site, the greater family clumping expected at sampling in our system may actually increase the power to detect movement by increasing the proportion of siblings per family we capture at the culvert site. It may also reduce the spatial extent around the culvert we would need to sample for reconstructing families and detecting movement. At the same time, we may have decreased power to use this approach if siblings do not disperse far enough to cross the culvert site by the time we sample.
Work plans for funded work 2010-2012
Fieldwork 2010:
Field work in 2010 will focus on three tasks (Table 2). First, we will conduct demographic and population-level genetic sampling of trout populations in the primary study streams (treatments and controls) and 5 “background sites” in the Lolo Creek drainage (the original 3 mentioned above, plus 2 additional sites added in 2010, see Figure 2). Genetic samples from the background sites will not be analyzed under the current study, but will be collected to enable individual genetic assignment tests pending further funding (see below: Request for Further Funding.) Second, we will investigate the efficacy of different stationary antenna designs; estimate detection efficiencies with stationary PIT antennas andmobile PIT antennas; and estimate capture efficiency by electrofishing (see Figure 3). Preliminary investigations of the PIT tag antennas will focus on one of the Lolo Creek treatment sites, and electrofishing capture efficiency estimates will be made in five different streams in the Lolo Creek drainage. These data will be used to estimate statistical power to detect passage based on numbers of marked fish. Third, we will implement a new field protocol in the two treatment streams to sample yoy to allow for sibship analyses of movement across the restoration sites (Figure 5). For population-level evaluation of genetic characteristics, it is typically advised to avoid sampling yoy which may represent spatial clusters of siblings and thus over-represent families and bias estimates of genetic parameters in the ‘population’ (Hansen et al. 1997). For sibship analyses, in contrast, we obviously want to sample yoy and our sampling needs to straddle the culvert directly (Figure 5). Furthermore, an important consideration in developing sampling protocols for sibship analysis is the spatial extent of sampling and the number of fish in a collection required above and below the culvert. To evaluate the minimum sampling intensity required, we propose to target sampling 300 m upstream and downstream of the replaced culverts and stratify collections by distance from the culvert (e.g., 100m vs. 200m vs. 300m; see Figure 5) to facilitate statistical power analyses based on samples collected within each stratum vs all strata.