Salmo salar - NRC Research Press

Color profile: Disabled Composite Default screen

279

Interactions of Atlantic salmon (Salmo salar) and trout (Salvelinus fontinalis and Oncorhynchus mykiss) in Vermont tributaries of the Connecticut River Matthew J. Raffenberg and Donna L. Parrish

Abstract: Competitive interactions among stream salmonids in resource-limited environments have been linked to reduced success for many species. Few studies have focused on interactions at scales larger than individual fish or stream reach. We chose to focus our study on these larger scales to provide information for managing species that have complex life histories transcending multiple scales. Our objective was to explore age-0 Atlantic salmon (Salmo salar) growth and survival in relation to trout abundance (introduced rainbow (Oncorhynchus mykiss) and native brook (Salvelinus fontinalis) trout) and prey resources at 24 stream reaches across two Vermont watersheds that flow into the Connecticut River. Simple linear and multilinear regressions were conducted on response and predictor variables related to fish and invertebrate prey. Age-0 Atlantic salmon survival was greatest at the site with highest trout abundance; however, no linear relationships to trout abundance were detected possibly because Atlantic salmon growth and survival were highly variable across sites. In contrast, a positive significant multivariate relationship was identified among age-0 Atlantic salmon survival, the abundance of age-1+ brook trout (i.e., 100–130 mm), and benthic prey abundance. These results suggest that stocking streams based on trout abundance may not increase Atlantic salmon growth and survival during the first summer of life. Résumé : Les interactions compétitives entre les salmonidés d’eau courante dans des environnements aux ressources limitées expliquent, croit-on, le succès réduit de plusieurs des espèces. Peu d’études se sont concentrées sur ces interactions à des échelles qui dépassent l’individu ou la section de cours d’eau. Nous ciblons dans notre étude des échelles plus grandes pour pouvoir fournir des renseignements utiles à l’aménagement d’espèces à cycles biologiques complexes qui englobent des échelles multiples. Notre objectif est d’étudier la croissance et la survie de saumons de l’Atlantique (Salmo salar) d’âge 0 en fonction de l’abondance de truites (truites arc-en-ciel (Oncorhynchus mykiss) introduites et d’ombles de fontaine (Salvelinus fontinalis) indigènes), ainsi que de l’abondance des ressources de proies dans 24 sections de cours d’eau de deux bassins hydrographiques du Vermont qui se déversent dans le fleuve Connecticut. Des régressions simples linéaires et multilinéaires ont permis de décrire les réactions et les variables prédictives concernant les poissons et leurs proies invertébrées. La survie des saumons de l’Atlantique d’âge 0 est maximale au site où la densité des truites et des ombles est la plus forte; cependant, leur croissance et leur survie sont si variables d’un site à un autre qu’il n’y a pas de relation linéaire avec l’abondance des truites et des ombles. En revanche, il y a une relation multidimensionnelle positive entre la survie des saumons de l’Atlantique d’âge 0 et l’abondance des ombles de fontaine d’âge 1+ (i.e. de 100–300 mm) et l’abondance des proies benthiques. Nos résultats laissent croire que l’empoissonnement des cours d’eau en fonction de l’abondance des truites et des ombles peut ne pas favoriser la croissance et la survie des saumons de l’Atlantique durant le premier été de leur vie. [Traduit par la Rédaction]

Raffenberg and Parrish

Introduction Stream salmonids typically inhabit environments possessing limited food and habitat resources, which may promote territoriality and agonistic interactions between species (Keeley and Grant 1997) resulting in reduced success (e.g., feeding,

285

growth, and survival) for one or more species (Kocik and Taylor 1995; Nakano 1995). Interactions can be magnified in streams in which salmonids are introduced outside of their native range, because species that had not coevolved may be unable to efficiently partition available resources (Connell 1975). Because native salmonids occur with natu-

Received 19 December 2001. Accepted 20 February 2003. Published on the NRC Research Press Web site at http://cjfas.nrc.ca on 16 April 2003. J16673 M.J. Raffenberg1,2 and D.L. Parrish. Vermont Cooperative Fish and Wildlife Research Unit, School of Natural Resources, University of Vermont, Burlington, VT 05405, U.S.A. 1 2

Corresponding author (e-mail: [email protected]). Present address: Lawler, Matusky, and Skelly Engineers, One Blue Hill Plaza, Pearl River, NY 10965, U.S.A.

Can. J. Fish. Aquat. Sci. 60: 279–285 (2003)

J:\cjfas\cjfas60\cjfas6003\F03-021.vp Thursday, April 10, 2003 8:36:53 AM

doi: 10.1139/F03-021

© 2003 NRC Canada


280

ralized populations of nonnative salmonid species throughout North America, the scale of loss in productivity resulting from negative interactions among salmonids that are sympatric but not coevolved is potentially large (Fausch 1998). The widespread mixing of native and nonnative stocks, coupled with the propensity for naturalized nonnative salmonids to outcompete native species, presents a formidable challenge for fisheries management, particularly for efforts to restore native species. In the northeastern United States, efforts to restore Atlantic salmon (Salmo salar) to historic habitats have been ongoing for over 30 years. The stocking of Atlantic salmon fry into tributaries of large New England rivers has been a focus of the Atlantic Salmon Restoration Program. The streams receiving Atlantic salmon fry typically contain a mix of native brook trout (Salvelinus fontinalis) as well as naturalized populations of nonnative brown (Salmo trutta) or rainbow (Oncorhynchus mykiss) trout, or Atlantic salmon. Previous research has shown that in sympatry with brook (Williams 1981) and brown (Mundie 1974) trout, Atlantic salmon display considerable diet overlap with these species. Further, limited empirical evidence shows that juvenile Atlantic salmon may choose less optimal habitats in the presence of naturalized rainbow trout, resulting in unsuccessful food acquisition for salmon and a presumed reduction in salmon productivity (Hearn and Kynard 1986). Many factors including interspecific interactions and prey availability are likely to affect the productivity of stocked salmon populations and thus the potential success of Atlantic salmon restoration in New England. Because populations of both native and nonnative trout occupy habitats where Atlantic salmon fry are stocked, identifying the potential effects of competition among trout and salmon would contribute to an understanding of factors potentially limiting Atlantic salmon recruitment success. Managers could use this knowledge to focus Atlantic salmon fry stocking efforts on the subset of salmonid streams that yield the highest Atlantic salmon feeding, growth, and survival during the first summer. Ultimately, the goal of the Atlantic salmon restoration effort is for adult salmon to return from the ocean to spawn in Vermont tributaries. However, the annual number of adult Atlantic salmon returning to the Connecticut River to spawn is relatively low and most are taken as broodstock so few are released to spawn naturally. In lieu of large numbers of adults returning to Vermont streams, the success of the program has been measured primarily by the survival of age-0 Atlantic salmon during their first summer. Survival to later life stages may be a better indicator of the number of fish that will migrate as smolts; however, age-0 survival provides an alternative measure of the success of Atlantic salmon stocking in Vermont streams (Jay McMenemy, Vermont Department of Fish and Wildlife (VTFW), 100 Mineral St., Suite 302, Springfield, VT 05156, personal communication). This scenario provided an opportunity to determine if Atlantic salmon growth and survival through the first summer is related to trout or food abundance. Where competitive interactions between Atlantic salmon and other native and nonnative salmonids for available resources (e.g., food and habitat) have been investigated (Fausch 1998), studies have commonly focused on the microhabitat scale, reflecting the tenet that interactions occur at scales im-

Can. J. Fish. Aquat. Sci. Vol. 60, 2003

portant to individual fish. The importance of large-scale studies has been raised because interactions observed at small scales may not directly translate into factors affecting populations at larger scales (i.e., multiple streams or watersheds; Fausch 1998; Folt et al. 1998). In addition, we are aware of no study that has determined how competition through interactions for resources can directly affect the overall fitness (e.g., feeding, growth, and survival) of Atlantic salmon populations at a large scale across two watersheds. Because empirical evidence supporting negative interactions between Atlantic salmon and trout at large scales exists and recent challenges to conclusions drawn from small-scale studies have been made, research across multiple streams or watersheds is warranted. Vermont tributaries of the Connecticut River are monitored yearly to estimate Atlantic salmon and trout abundance, growth, and survival across multiple streams having a range of species assemblages and food and habitat types. This scenario provided the opportunity to determine how species interactions and prey availability are linked to age-0 Atlantic salmon feeding, growth, and survival at a large scale across two Vermont watersheds. In particular, we hypothesized that interactions among populations of native brook trout and nonnative but naturalized populations of rainbow trout could negatively affect the feeding, growth, and survival of age-0 Atlantic salmon stocked in Vermont streams. The direct effect of nonnative brown trout alone was not tested because their densities are much lower than the densities of the other two trout species in most Vermont streams, but their densities were included in the analysis with all trout combined. Our hypothesis was formed under the assumption that any negative interactions of Atlantic salmon and trout are likely the result of competitive interactions for food or territories or both. Our primary objectives were (i) to address how numbers and biomass of trout and total salmonids affect Atlantic salmon feeding, individual growth, and survival and (ii) to determine if interactions between species are related to the abundance of benthic prey. We addressed these objectives by conducting an extensive sampling program of collecting data on Atlantic salmon, trout species, and benthic invertebrates at 24 sites within the West and White rivers, Vermont. Simple linear and multiple-regression analyses were used to determine how trout and benthos were related to survival of Atlantic salmon during the first summer.

Methods Study area Study sites were located on tributaries or mainstem reaches of the West (42°51′59′′N, 72°33′54′′W) and White (43°38′55′′N, 72°18′52′′W) rivers, Vermont. The White River flows through central Vermont (1844 km2 drainage area) and the West River (1096 km2 drainage area) flows through southern Vermont in a northwest to southeast direction. Both rivers are major tributaries of the Connecticut River (29 125 km2 drainage area), which creates the eastern border of Vermont and extends 660 km from Quebec to Long Island Sound. The Connecticut River basin is one of the southernmost river systems in North America that historically supported spawn© 2003 NRC Canada




ing populations of Atlantic salmon (MacCrimmon and Gots 1979). We selected sites based on historic population assessment data documenting salmonid species composition in each tributary (Jay McMenemy, VTFW, and Steven Roy, United States Forest Service (USFS), Green Mountain and Finger Lakes National Forest, 231 North Main Street, Rutland, VT 05701, unpublished data) and because the entire stream was stocked annually (typically by mid-May) with Atlantic salmon fry at equal densities. The index stations selected are sampled annually by state and federal biologists and were initially chosen because each station contains habitats that are representative (proportion of riffle, run, and pool) of those outside of the index station. Thus, we made the basic assumption that fish will immigrate and emigrate at similar rates in these stream reaches. All sites were on first- to fourth-order streams that ranged from 5 to 30 m in width and from 91 to 188 m in length. The amount of riparian vegetation ranged broadly across sites from complete canopy characterized by mixed coniferous (Tsuga spp. and Pinaceae spp.) and deciduous (Acer spp. and Betula spp.) trees to no riparian vegetation in a stream bisecting a town. Fish were collected by VTFW and USFS biologists using standard electrofishing procedures with either backpack or streambank electrofishing units. A net blocking off the upstream and downstream ends of the site allowed for sampling the entire width and length of the area. Sampling was conducted using a two-pass depletion method (Zippin 1958) to assure representative estimates of the fish at each site. After collection, salmonids were anesthetized, identified to species, weighed to the nearest 0.01 g, and measured (total length) to the nearest millimetre. Stomach contents were obtained via a gastric lavage technique from a maximum of 10 age-0 Atlantic salmon (0–100 mm) from each stream for the determination of stomach content dry weights (Raffenberg 1998). We sampled the benthos using a 0.093 m2 frame Surber sampler (Wildlife Supply Company, 95 Botsford Place, Buffalo, NY 14216) with 0.5 mm mesh. A total of three randomly chosen net sets were conducted at each site directly upstream and before fish sampling. The sampler was placed over the substrate, the area within the frame was stirred to a depth of 10 cm, and rocks were scrubbed clean with a stiff brush for 2.5 min (Dedual and Collier 1995). Sample processing Stomach contents were dried at 60°C until a constant weight was obtained, and weights were recorded to the nearest 0.0001 g. Mean dry weight (gram dry food per gram wet fish) of stomach contents were calculated for each stream. We identified Surber samples to their lowest practicable taxonomic classification (primarily family) and enumerated and measured organisms using a dissecting microscope. We subsampled those samples containing a large number of organisms. Homogeneity of the subsampling techniques was verified by replicate subsamples, which indicated close correspondence, confirming that a reduction of up to 25% of the original sample was representative of the prey present (Allan and Russek 1985). Benthic prey density was calculated for each

281

site as the ratio of the number of organisms per area sampled. Calculations and data analysis VTFW and USFS biologists calculated Atlantic salmon and trout population estimates for each species and size class. Total Atlantic salmon abundance was determined by combining Atlantic salmon size classes (i.e., age-0, age-1+, and age-2+). Total trout abundance (brook, rainbow, and brown trout) was calculated in the same manner. Population estimates for all Atlantic salmon, all trout species, and total salmonids (salmon + trout) were standardized to salmon units (number/100 m2) for analysis across sites. In addition to abundance, salmon and trout biomass was calculated for each species and size class and standardized to salmon units (g/100 m2) for analysis across sites. Percent survival of Atlantic salmon was calculated for each site under the assumption that the possibility of immigration and emigration was consistent across all sites. Age-0 survival was determined by calculating the ratio between the population estimate and the initial stocking density. Mean percent survival of small salmon was calculated for the five previous years (1992–1996) with data collected by VTFW and USFS biologists. Growth of age-0 salmon was determined at each site by taking the difference between the mean total length at the time of stocking (25 mm) and mean length at capture. The daily relative rate of increase in size was determined by calculating the ratio between the mean growth and the number of days since stocking. Mean weight-specific dry weights (g) of salmon stomach contents were calculated for each stream. A ratio of food dry weight to fish wet weight was calculated using grams of food per gram of fish body × 100 to allow for the comparison of dry weights across different-sized fish (Parrish and Margraf 1990), providing a standardized measure for regression analysis across streams. To determine if univariate or bivariate relations existed among age-0 salmon feeding (i.e., stomach content dry weights), growth, or survival and the abundance and biomass of trout, salmon, and total salmonids, as well as benthic invertebrate density, we used simple linear and multipleregression analyses (Table 1). Abundance of brook and rainbow trout for each size class and available prey densities were chosen as the primary independent predictor variables of age-0 salmon success. Brown trout abundance was not used as an independent predictor variable because their densities were extremely low in the streams sampled for this research. Brown trout were included in the calculation of total salmonid (salmon + trout) abundance used as a secondary predictor variable to identify density-dependent effects on age-0 Atlantic salmon success. When variables did not meet the assumptions of parametric statistics, they were log10-transformed to make distributions normal or to stabilize variance. To protect against experimentwise error, the significance level of each regression was set at α = 0.003, based on a Bonferroni adjustment in which the alphas for individual test = 1 – (1 – α)1/p and p = 16 is the number of tests (Milliken and Johnson 1984). Standard model-building procedures were used to determine if any bivariate associations existed between salmon response © 2003 NRC Canada



282

Can. J. Fish. Aquat. Sci. Vol. 60, 2003 Table 1. Response and predictor variables used in simple linear and multilinear regression analysis to determine factors affecting Atlantic salmon (Salmo salar) success. Response variables

Predictor variables

Dry weight of stomach contents Age-0 Atlantic salmon survival (%) Five-year mean of age-0 Atlantic salmon survival (%) Growth rate of small Atlantic salmon

Primary Total trout abundance and biomass Brook trout abundance and biomass Rainbow trout abundance and biomass Benthic prey abundance Secondary Total salmonid abundance and biomass Total salmon abundance and biomass

Note: All response variables are associated with Atlantic salmon. Variables related to fish include small, medium, and large sizes unless otherwise specified.

variables and multiple predictor variables. Post hoc power analysis was conducted on all regressions. When power was not sufficient (>80), sample size necessary to achieve sufficient power was calculated (Cohen 1988).

Results Numbers of Atlantic salmon at index sites ranged from ~4/100 m2 at three sites to over 45/100 m2 at two sites (Fig. 1). Atlantic salmon abundance was higher than trout abundance at 15 sites; salmon dominated two sites with >98% of the fish present, and age-0 Atlantic salmon was the most abundant size class collected. Trout rarely exceeded 10/100 m2 at most sites, although one site had a relatively high abundance of 35/100 m2. Atlantic salmon biomass was also variable across all sites ranging from a low of 4 g/100 m2 to well over 370 g/100 m2. At 14 of the 24 sites, Atlantic salmon biomass was 20%. The site with the highest age-0 salmon survival had the highest trout abundance. In contrast, the mean growth (relative rate of increase in total length) of age-0 salmon from the time of stocking to the sample date had the least variation of all variables. The highest rate was 0.56 mm/day at three sites and the lowest rate at one site was 0.39 mm/day. Linear and multiple-regression results Few relationships in our linear regression analysis were significant (most r2 < 0.4; Table 2) and few were negative. Age-0 Atlantic salmon survival had a weak positive relationship with trout abundance (r2 = 0.29, P = 0.02; Fig. 2b).

Fig. 1. Total Atlantic salmon (Salmo salar; solid bar) and trout abundance (Salvelinus fontinalis and Oncorhynchus mykiss; open bar) and percent survival (line) of small Atlantic salmon sampled at 24 sites in the summer of 1996 on tributaries of the West and White rivers, Vermont. Percent survival data are discrete but are linked to illustrate the relation of survival to total abundance estimates.

This relationship was driven by high salmon survival and trout abundance at one site. In addition, small Atlantic salmon survival (r2 = 0.18, P = 0.058; Fig. 3a) and individual growth (r2 = 0.14, P = 0.09; Fig. 3b) had no linear association with benthic prey abundance. The strongest positive linear relationship (r2 = 0.93, P = 0.0001) was between age-0 salmon survival and the total number of salmonids (i.e., salmon and trout combined) (Table 2). The weak relationships observed were typified by the relationship between small Atlantic salmon growth and trout abundance (r2 = 0.075, P = 0.23; Fig. 2a), where there was relatively no linear association. Atlantic salmon feeding (i.e., stomach content dry weights) also had no linear relation to the abundance of trout, Atlantic salmon, or total salmonids (Table 2). A weak positive relationship also existed between age-0 salmon stomach content dry weights and benthic prey abundance (r2 = 0.34). A post-hoc power analysis was conducted to determine if sufficient power existed in the linear regression results (Table 2). We established 80% as the lower power limit for de© 2003 NRC Canada




283

Table 2. Results from simple linear regressions of relationships between Atlantic salmon (Salmo salar), trout (Salvelinus fontinalis and Oncorhynchus mykiss), and total salmonid abundance (number/100 m2) and benthic prey (benthos) abundance (number/0.093 m2) and response variables related to age-0 salmon growth and survival. Predictor variables Response variable Growth rate (mm/day) Dry weight of stomach contents Survival (%) Mean survival (%)

Trout

Atlantic salmon

2

r = 0.075 (+) P = 0.23, power r2 = 0.22 (–) P = 0.03, power r2 = 0.29 (+) P = 0.02, power r2 = 0.008 (+) P = 0.71, power

Total

2

= 30 = 73 = 80 =7

r = 0.007 (+) P = 0.71, power = r2 = 0.079 (+) P = 0.22, power = r2 = 0.66* (+) P = 0.0001, power r2 = 0.45* (+) P = 0.0009, power

2

7 30 = 100 = 98

r = 0.007 (+) P = 0.72, power = 7 r2 = 0.24 (+) P = 0.024, power = 73 r2 = 0.93* (+) P = 0.0001, power = 100 r2 = 0.36* (+) P = 0.0044, power = 90

Benthos r2 = 0.14 (+) P = 0.09, power = 51 r2 = 0.34 (+) P = 0.01, power = 90 r2 = 0.18 (+) P = 0.058, power = 51 r2 = 0.06 (+) P = 0.30, power = 25

Note: All data were collected in the summer 1996 at 24 sites on tributaries of the West and White rivers, Vermont. Plus (+) or minus (–) signs indicate the direction of the linear relationship and an asterisk (*) indicates significant relationships at P < 0.003; power > 80 was considered significant.

Fig. 2. (a) Relationship between small Atlantic salmon (Salmo salar) growth (mm/day) and trout (Salvelinus fontinalis and Oncorhynchus mykiss) abundance (number/100 m2) and (b) survival of small salmon (%/100 m2) and trout abundance (number/ 100 m2) sampled at 24 sites in the summer of 1996 on tributaries of the West and White rivers, Vermont.

termining the strength of the relationships observed. In many cases, especially when no negative relationships were detected, the results were limited by insufficient power (