The Pennsylvania State University. The Graduate School. Department of Ecosystem Science and Management

Transcription

1 The Pennsylvania State University The Graduate School Department of Ecosystem Science and Management EFFECT OF ENRICHMENT IN THE HATCHERY ON THE PERFORMANCE OF BROOK TROUT AND ATLANTIC SALMON A Thesis in Wildlife and Fisheries Science by Bryan D. Ferguson 2015 Bryan D. Ferguson Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science May 2015

2 ii The thesis of Bryan Ferguson was reviewed and approved* by the following: Victoria A. Braithwaite Professor of Fisheries and Biology Thesis Advisor C. Paola Ferreri Associate Professor of Fisheries Management Matt Marshall Adjunct Assistant Professor of Wildlife Conservation National Park Service Eastern Rivers and Mountains Network Program Manager Michael G. Messina Head of the Department of Ecosystem Science and Management *Signatures are on file in the Graduate School

3 iii ABSTRACT Fish stocking has been an important tool utilized by fisheries professionals for many decades. By artificially spawning and rearing fish in a controlled hatchery environment, it is possible to produce large numbers of fish that have a high survival rate while in the hatchery. This method can result in thousands of fish available for supplementing or reestablishing fisheries. However, many of these fish do not survive when stocked into the wild. If you compare the environments in which we raise fish for stocking and the environments we expect them to survive in, there are many apparent differences. Previous research suggests that hatchery fish exposed to altered rearing environments behave more flexibly. Fish raised in plain hatchery environments have little exposure to complex environments that more closely resemble natural streams and lakes. By creating environmental complexity and variability during a fish s residence in a hatchery environment, fish may have the ability to learn and alter behavior in ways that increase poststocking survival. Here I investigate the addition of variability in the environment of hatcheryreared brook trout (Salvelinus fontinalis) and Atlantic salmon (Salmo salar), by assessing behavioral changes and post-stocking survival in several Adirondack lakes. My findings show several changes in behavior prior to stocking, as well as a survival advantage in a controlled laboratory setting. However, it did not result in increased recapture rates when the brook trout and Atlantic salmon were stocked into Adirondack lakes and ponds.

4 iv TABLE OF CONTENTS LIST OF FIGURES... vi LIST OF TABLES... ix ACKNOWLEDGEMENTS... x Chapter 1 Introduction... 1 Background... 1 Approach/Rationale... 6 Chapter 2 The effects of enrichment in the hatchery environment on the pre-release behavior of Brook Trout (Salvelinus fontinalis) and Atlantic Salmon (Salmo salar)... 8 Introduction... 8 Methods Brook Trout Atlantic Salmon Results Brook Trout Atlantic Salmon Discussion Chapter 3 The effects of enrichment in the hatchery environment: Can we increase the survival of stocked Brook Trout (Salvelinus fontinalis) and Atlantic Salmon (Salmo salar) in the wild? Introduction Methods Brook Trout Atlantic Salmon Results Brook Trout Atlantic Salmon Discussion Chapter 4 Discussion Summary Final Conclusions and Future Directions Appendix A Weekly Enrichment Rotation Appendix B Catch Per Unit Effort for Brook Trout and Atlantic Salmon... 62

5 v Appendix C Summary Table of Differences Between Treatments REFERENCES... 64

6 vi LIST OF FIGURES Figure 2.1: Image of an enriched hatchery tank on the left and a standard hatchery tank on the right Figure 2.2: (A) Mean length of brook trout in enriched and standard rearing environments of each year class. (B) Mean weight of brook trout in enriched and standard rearing environments for each year class. (C) Mean condition factor of brook trout for each rearing environment by year class Figure 2.3: Mean amount of time it took brook trout of the 2011 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error Figure 2.4 Mean amount of time it took brook trout of the 2012 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error Figure 2.5: Mean amount of time it took brook trout of the 2013 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error Figure 2.6: Mean amount of time it took brook trout from different rearing environments to go from start area to the cover zone. Bars represent ± standard error Figure 2.7: Mean distance the observer is from the tank when the fish begin to startle for each tank with 2011 year class brook trout. Bars represent ± standard error Figure 2.8: Mean percentage of brook trout that returned to their original positions in home tanks after the aggressive startle occurred for the 2011 year class. Bars represent ± standard error Figure 2.9: Mean distance the observer is from the tank when the fish begin to startle for each tank with 2012 year class brook trout. Bars represent ± standard error Figure 2.10: Mean percentage of brook trout that returned to their original positions in home tanks after the aggressive startle occurred for the 2012 year class. Bars represent ± standard error Figure 2.11: Proportion of fish that survived predation from the enriched and standard rearing environments over 26 trial periods. Bars represent ± standard error Figure 2.12: (A) Mean weight of Atlantic salmon from each rearing environment after 12 months in differing rearing environments. Bars represent ± standard error. (B) Mean length of Atlantic salmon from each rearing environment after 12 months in differing rearing environments. Bars represent ± standard error. (C) Mean condition factor of Atlantic salmon

7 vii from each rearing environment after 12 months of rearing in different environments. Bars represent ± standard error..27 Figure 2.13: Mean amount of time it took Atlantic Salmon of the 2011 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error Figure 2.14: Mean amount of time it took Atlantic Salmon of the 2012 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error...28 Figure 2.15: Mean amount of time it took Atlantic salmon from different rearing environments to go from the start area to the cover zone. Bars represent ± standard error. 29 Figure 2.16: Mean distance of observer from each tank when the fish begin to startle for the 2011 year class Atlantic salmon. Bars represent ± standard error Figure 2.17: Mean percentage of Atlantic salmon that returned to their original positions in home tanks after the aggressive startle occurred for the 2011 year class. Bars represent ± standard error...30 Figure 2.18: Mean distance the observer is from the tank when the fish begin to startle for the 2012 year class Atlantic salmon Figure 2.19: Mean percentage of Atlantic salmon that returned to their original positions in home tanks after the aggressive startle occurred for the 2012 year class Figure 3.1: Mean proportion of brook trout recaptured (bars show ± standard error) from enriched and standard rearing environments in the 8 Adirondack lakes Figure 3.2: Total number of brook trout recaptured in each lake from the enriched and standard rearing environment Figure 3.3: Mean proportion of 2011 year class enriched environment and standard environment fish recaptured in Fourth Bisby Lake (± standard error) Figure 3.4: Mean proportion of 2013 year class brook trout recaptured (bars show ± standard error) from enriched and standard rearing environments in Chambers and Fourth Bisby Lakes Figure 3.5: Total number of 2013 year class brook trout recaptured in each lake from the enriched and standard rearing environment Figure 3.6: Mean proportion of Atlantic salmon recaptured (bars show ± standard error) from enriched and standard rearing environments in 3 Adirondack lakes

8 Figure 3.7: Total number of Atlantic salmon recaptured in each lake from the enriched and standard rearing environment viii

9 ix LIST OF TABLES Table 2.1: Timeline of important dates in the hatchery and of behavior trials for brook trout in 2011, 2012, and Table 2.2: Number of fish excluded from time to cover trials by year and species, and the sample size (n) for each time to cover trial Table 2.3: Timeline of important dates in the hatchery and of behavior trials for Atlantic salmon in 2011 and Table 3.1: Size, mean depth, max depth, water color, and whether the lake thermally stratifies or not in the summer is shown for the 10 Adirondack lakes in this study Table 3.2: Fish species present in each lake as indicated by an X Table 3.3: Lakes that were sampled for brook trout recapture, along with the stocking rate and amount of fish stocked in each lake Table 3.4: Dates sampled for 2011 year class brook trout and gear types used in each lake. Not all lakes had a second or third sample date Table 3.5: Atlantic salmon stocking rates, lake size, and number of fish stocked from each environment in the three lakes where recapture was attempted Table 3.6: Number of brook trout captured in Mountain Pond and Lower Sylvan Pond by treatment and gear type Table 3.7: Mean lengths in millimeters of brook trout from the standard and enriched environments, as well as standard error and p-values for each lake Table 3.8: Mean lengths in millimeters of 2013 year class brook trout from the standard and enriched environments, as well as standard error and p-values for each lake Table 3.9: Capture numbers of Atlantic salmon by gear type and treatment in Panther Lake and Lower Sylvan Pond Table 3.10: Mean lengths in millimeters of Atlantic salmon from the standard and enriched environments, as well as standard error and p-values for each lake

10 x ACKNOWLEDGEMENTS I would like to thank my thesis advisor Victoria Braithwaite. I have been fortunate enough to be a part of many research projects while working for her over the past several years. Through her guidance, encouragement, and support I have gained knowledge and developed professionally in my time at Penn State. I am truly grateful for the opportunity to work on this research and to be able to develop it into my thesis. The research in this thesis would not have been possible without the help of Dan Josephson, Cliff Kraft, and the staff working for the Adirondack Fishery Research Program. Not only were they responsible for the general fish care while I was at Penn State but they also provided equipment, lodging, field assistance, and extensive knowledge of the Adirondack lakes and ponds used in this research. Without their commitment to this project it would not have been possible, and I am thankful for the many forms of support they provided. I would like to thank Paola Ferreri and Matt Marshall for being part of my thesis committee. I also would like to thank the members of the Braithwaite Lab. All of the comments and feedback I received throughout this process were greatly appreciated and helped significantly in the preparation of this manuscript. Finally, I would like to thank my Fiancé, Samantha Kutskel, for putting up with me throughout this process and providing me with encouragement or distraction when necessary. I am grateful she understands how it important it is for me to conduct fieldwork for extended periods of time in beautiful, far away places.

11 1 Chapter 1 Introduction Background Stocking fish has been used as a tool of fisheries managers for decades. Welcomme and Bartley (1998) estimate that over 250 fish species are stocked worldwide and that most countries participate in some form of stocking. Stocking is done for many reasons; the main ones include mitigation, enhancement, restoration, and the creation of new fisheries (Cowx 1994). Stocking for mitigation purposes is often done when the habitat is being disturbed or altered, such as nearby road or dam construction. In these situations, fish are stocked to compensate for losses to fish populations because of the disturbance such activities generate. The goal of enhancement stocking is therefore to maintain or improve fish stocks in a waterway. Sometimes, this is done as an effort to increase the quality of fishing with a put and take type of program, or as an attempt to balance the harvest of an exploited species (Halverson 2008). Stocking for restoration or reintroduction becomes important when the native fish species have been removed or reduced. Activities that accompany the stocking can include the removal of non-native predators, water quality improvement, and habitat improvement. When fish are stocked into waterways where the limiting factor has been addressed, the fish are expected to survive and reproduce, and ultimately build a self-sustaining population. A further reason stocking is conducted can be for the creation of new fisheries. This is done when a fish species is not present in a water body, however it is desirable to create a population of that species. Stocking of this kind is particularly controversial (Cowx 1994; Araki and Schmid 2010).

12 2 Whether or not we should be stocking is not the focus of this thesis. Certainly in some put-and-take situations, survival of stocked fish may not be of much concern to the fisheries manager. However in many scenarios, particularly for restoration or reintroduction purposes as well as when establishing new fisheries, survival of the stocked fish is essential to achieve the goal of the stocking program. Hatchery fish are expensive to raise and to stock (Wiley et al. 1993), therefore low survival rates can be detrimental to a stocking program s budget, and could result in an unsuccessful attempt at reintroduction. For decades now, hatchery fish survival has been investigated and different ways to reform the hatchery have been contemplated (Schuck 1948; Wales 1954). Unfortunately, low survival rates seem to remain common among stocked fishes (Suboski & Templeton 1989; Olla et al. 1998; Einum and Fleming 2001; Kostow 2004). The highest mortality of these fish occurs shortly after their release into the wild (Howell 1994; Olla et al. 1994, 1998). Some reasons for such low survival includes the fish having difficulty at adapting to the natural diet found in the wild (Ersback and Haase 1983; Ellis et al. 2002; Orlav et al. 2006; Larsson et al. 2011), and fish falling victim to predation after stocking in a novel environment (Howell 1994). These reasons for low survival may be indirectly related, with fish weakened by hunger being more likely to experience higher predation mortality (Brown and Laland 2001). If we view wild fish as the model that we want our stocked fish to duplicate, it seems the hatchery fish have many deficiencies. When comparing hatchery salmonids to wild fish, hatchery fish select habitat and forage differently (Bachman 1984; Mesa 1991; Deverill et al. 1999), they are more aggressive (Mesa 1991; Berejikian et al. 1996, 1999), and have poorer predator avoidance (Berejikian 1995; Johnsson et al. 2001). Hatchery fish can differ genetically from wild fish, owing to artificial selection as part of the hatchery process (Einum and Fleming 1997). Such artificial selection can result in the expression of traits that would be selected against in the wild

13 3 (Weber and Fausch 2003), and this may also contribute to a decline in fitness (Araki et al. 2008). Genetic differences can be addressed by using wild fish as brood stock or by continually introducing wild fish into brood stock breeding programs (Kostow 2009). However, how to overcome behavioral deficiencies is less clear. In an attempt to increase survival rates, several hatchery pre-release methods have been investigated. Some of these methods include predator training, pre-releases, foraging training and enrichment in the rearing environment (Hesthagen and Johnsen 1989; Berejikian et al. 1999; Brown et al. 2003; Brockmark et al. 2007). One pre-release strategy used to prepare naive hatchery fish for the wild is predator training. This is done to try and increase a hatchery fish s awareness of the dangers posed by predators, and often involves exposing fish to cues from a natural, common predator prior to stocking. This predator cue is combined with an aversive stimulus such as an alarm signal released by injured conspecifics. Several studies have found an increase in anti-predator response for fish that go through such predator recognition training (Brown and Smith 1997; Berejikian et al. 1999; Mirza and Chivers 2000; Berejikian et al. 2003; Petersson et al. 2014). In addition to these increased anti-predator responses, studies have shown that a learned recognition can increase survival when fish are tested against live predators in a controlled environment (Mirza and Chivers 2000; Darwish et al. 2005; Vilhunen 2006). However, such increased anti-predator responses have not led to a consistent increase in survival when fish are stocked into the natural environment (Kanayama 1968; Berejikian et al. 1999, Hawkins et al. 2007). Brown et al. (2013) provide an overview of why these differences may not carry over into the field, describing the many ways that learning to recognize and avoid predators is an incredibly complex system. The authors ultimately suggest that the question of how long prey retains learned information is just as critical to prey species as the learning itself. Petersson et al. (2014) provide further explanation by showing that although the anti-predator behavior of the fish is positively affected, this adaptation

14 4 still does not make the fish behave in the same way as their wild conspecifics. However, even though there may not be a clear survival advantage in the wild, this research has shown that it is possible to alter the behavior of hatchery fish prior to stocking. Predator training of hatchery fish, therefore, might be one way to attempt increased post-release survival for some systems, and appears worthy of further study. Another approach addressing hatchery fish deficiencies is the use of environmental enrichment. As has been shown in several terrestrial animals, environmental enrichment and variability may be important for the development of hatchery fish behavior (Price 1999). By adding enrichment and some type of variability in the hatchery environment, it could be possible to help fish prepare fish for the frequently changing, novel environments into which they are stocked (Braithwaite and Salvanes 2005). Enrichment in the hatchery has also been shown to improve the dorsal fin quality in juvenile steelhead, suggesting it may produce a higher quality fish (Berejikian amd Tezak 2005). While this approach has shown behavioral differences in many different fish species in the lab, knowledge of what occurs after stocking is limited to few studies (Fast et al. 2008; Heenan et al. 2009; Tatara et al 2009). Survival studies show mixed results when investigating whether differing hatchery environments translate to survival advantages in the wild (Fast et al. 2008; Brouwer et al. 2014). Rearing density has also been shown to be an important factor for fish in the hatchery environment (Brockmark et al. 2007). Although Soderberg and Mead (1987) found no effect of rearing density on growth, survival, or fin condition of Atlantic salmon in the hatchery, there may be other behavioral components that rearing in high densities promote which might influence post release survival. Brockmark and Johnson (2010) found brown trout (Salmo trutta) reared at lower densities grew faster and were more dominant than trout reared at higher densities. Low densityreared trout also survived better when stocked into the stream. In a review by Banks (1990),

15 5 several reports of decreased post release survival of Chinook salmon (Oncorhynchus tshawytscha) were associated with increased rearing density. From this literature, it seems that rearing density may play an important role in the post-release performance of many fish species. Other pre-release methods that have attempted to improve hatchery fish for release into the wild include pre-stocking and foraging training. Pre-stocking allows fish to be exposed to the area they are to be stocked into while protecting them from predation. This gives fish an acclimation period, free of predators, to adjust to their surroundings and some food sources. Variations of this method have been shown to be successful with brown trout. Jonsson et al. (1999), found improved growth rates and recapture rates for fish that were held in enclosures at the release site for 6 days prior to release. Cresswell and Williams (1983) found similar results after holding brown trout for just 24 hours in enclosures at the release site. Hestagen and Johnsen (1989) pre-stocked brown trout in a predator free pond prior to moving them to their final lake destination, and they found that there was a higher recapture rate of pond fish than traditional hatchery fish in each of the two lakes they stocked. Foraging skills are behaviors that can be refined and improved, but the way they are learned can be complex (Warburton 2003). While I am unaware of any study that investigates only prior exposure to live prey on foraging experience, Brown et al. (2003) investigated a combination of enriched environments and prior prey experience on the foraging behavior of Atlantic salmon. They found only the salmon with the prior prey experience and enrichment exposure had improved foraging behavior. Rodewald et al. (2011) compared foraging of both altered hatchery rearing environments of Atlantic salmon as well as the origin of the fish (wild/ hatchery strain). They showed that enrichment led to higher foraging rates in wild strain fish, suggesting both hatchery enrichment and the fish s origin play important roles in foraging behavior. Several other studies use some combination of pre-release training to explore their effects on the behavior of fish. Berijikian et al. (1999) used a

16 6 combination of predator conditioning and enrichment in the hatchery to investigate anti-predator behavior and survival of Chinook salmon. This study found a survival increase with predator conditioning but not with enriched rearing environments. Brockmark et al. (2007) used a combination of density and environmental enrichment to investigate the post-release survival of Atlantic salmon, but they found these treatments had no effect on post-release survival. Approach/Rationale For this study, I chose to focus on providing enrichment and variability in the hatchery environment of brook trout and landlocked Atlantic salmon. Many of the alterations to the hatchery environment described above are labor intensive, however the addition of environmental enrichment might be a simple, yet effective, method. Objects are placed in the tank and moved from time to time to create a variable three-dimensional environment. My goal was to keep my methods realistic for a large-scale production hatchery. While I did not combine the hatchery enrichment with any other pre-release method, it is important to note that our fish were reared in optimal rearing conditions. The fish were raised at relatively low densities for a production style hatchery and the hatchery is fed with nearby lake water which exposes fish in the hatchery to small, live organisms that they will encounter in the wild such as zooplankton, and can even wash in occasional larger animals such as smelt. The fish used in my experiments were taken from wild strain broodstock, which is not typical of many hatchery programs. The brook trout were F1 wild Temsicamie strain from Cat Pond in the northern Adirondack Mountains. The landlocked Atlantic salmon were F1 wild West Grand Lake strain from the Grand Lake Hatchery, Maine. Based on previous findings in other fish species, I expected the enriched environment fish to behave differently than standard rearing environment fish. My investigation included assays of

17 7 behavior prior to release, and field surveys to test for post-release survival. Several studies suggest that enrichment may promote behaviors that lead to an increase in post-stocking survival, however few investigations have been able to test this hypothesis (Berejikian et al. 1999; Berejikian et al. 2000; Braithwaite and Salvanes 2005; Strand et al. 2010; Roberts et al. 2011; Rodewald et al. 2011). To my knowledge, this is the first study of its kind to investigate the behavioral responses to enrichment in combination with post-release survival of two different species stocked into freshwater lake and pond systems. Because this study includes two species and multiple year classes for each species, the Methods and Results of Chapters 2 and 3 are presented by species. This layout results in some repetition, but my hope is it results in improved clarity for the reader.

18 8 Chapter 2 The effects of enrichment in the hatchery environment on the pre-release behavior of Brook Trout (Salvelinus fontinalis) and Atlantic Salmon (Salmo salar) Introduction Many reintroduction projects are unsuccessful because captive-reared animals lack the skills necessary to survive in the wild (McNeil 1991; Snyder et al. 1996; Price 1999). As discussed in chapter one, these behavioral deficiencies likely result from a number of issues with being reared in a captive environment. One aspect in fish restocking programs that may be the least complicated for hatchery managers to address is the early rearing experience of these animals. The early rearing environment of captive animals, especially those raised in production style facilities, is typically very plain and devoid of environmental variability. This environment is a complete contrast to the environment that fish in the wild experience. Studies in several different fish species have shown that enrichment of the rearing environment can promote behavioral flexibility, which is proposed to help fish more readily adapt to their environment upon stocking (Odling-Smee and Braithwaite 2003; Brown et al. 2003; Braithwaite and Salvanes 2005; Lee and Berejekian 2008; Strand et al. 2010). If we are able to alter the behavior of fish in the hatchery in a positive way, it is possible that we could increase survival in the wild after stocking. To address this, a research design was developed to evaluate the effect of hatchery environment enrichment on the behavior of brook trout and landlocked Atlantic salmon. Multiple tests were designed to determine whether the addition of variability into the hatchery environment would promote the development of behavioral flexibility. The first measure tested the time it took

19 9 the fish to seek out shelter in a novel environment. Such a behavioral response could be important particularly during the stocking process. Predators will often hone in and attack newly stocked fish soon after stocking, while the fish are still naïve to the threat posed by predators (Willette et al. 2001). Thus fish that move to cover quickly are likely to have a survival advantage. The second test investigated the fish s response and recovery to a startle. Fish can respond to perceived threats in several different ways; they may freeze, school more tightly, flee, or dive more deeply (Helfman 1989; Litvak 1993). The fish then slowly recover from the startle and eventually resume what they were doing before the startle event. To quantify the startle response, the approach of a person moving toward the tank was assessed. The fish that respond to a startle sooner and stay away from the threat longer may experience higher survival rates in the wild. The final test was a survival experiment. This was conducted to see whether the enriched environment fish truly develop a survival advantage when they are in a situation where they come across a natural predator. Studies that have explored the effects of changing the hatchery rearing experiences typically investigate a single species. Here, we chose to compare the effects of adding tank enrichment in two different species; brook trout and landlocked Atlantic salmon. We chose to use brook trout because it is native to the eastern United States and is an important species for reintroduction and restocking programs owing to population declines throughout its native range (Hudy and Thieling 2008). Additionally, hundreds of thousands of brook trout are stocked annually in New York State (New York State Department of Conservation 2014). As the populations decline, restocking is often turned to as an attempt to reestablish brook trout populations but the success of these stocking programs is highly variable (Fleming and Petersson 2001). Landlocked Atlantic salmon were also native to several areas in New York, and programs attempting to restore extirpated landlocked Atlantic salmon to New York s waters have been in

20 10 place for nearly 50 years (Hulbert et. al 1990). Therefore, we also included landlocked Atlantic salmon in this study. While we were unable to find any previous work trying to alter brook trout behavior in the hatchery environment, some similar work has been done with the Atlantic salmon. Several studies have shown that various aspects of Atlantic salmon behavior can be changed in the hatchery environment, including foraging (Brown et al. 2003; Rodewald et al. 2011), spatial learning (Salvanes et al. 2013), and risk-taking behavior (Roberts et al. 2011). The assays we developed were designed to complement and expand on these earlier observations. We modified the hatchery environment and conducted behavioral assays for three consecutive year classes. Methods Brook Trout Brook Trout Rearing Brook trout (n=6400 per year) were first generation Temiscamie strain fish originating from a wild broodstock in Cat Pond located near Paul Smiths, NY. All fish rearing was done at the Little Moose Field Station, near Old Forge, NY. For each year of the study, after hatch out in January, the fish were first housed in circular 4-foot diameter tanks for a period of approximately 3-4 months until they were large enough to transfer to larger tanks. At this stage (approximately late May), the brook trout were transferred to four circular 10-foot diameter tanks. Enrichment was added to two of the tanks and the other two were left as the control. We refer to these treatments as enriched for the tanks with enrichment and variation, and standard for the tanks left in a conventional hatchery format.

21 11 The enriched tanks were furnished with artificial structures that provided the fish with places to hide, to try to simulate conditions that were closer to a natural environment. We added a section of straight, 5-inch diameter PVC pipe, a 5-inch diameter PVC pipe elbow, a covered area created by corrugated roofing plastic on 12 inch PVC legs, and an aquamat made of multiple fronds that rest on the bottom of the tank and simulate aquatic vegetation (Figure 2.1). The position of the enrichment items in the tanks was changed weekly, during the tank cleaning, to provide a changing environment (Appendix). Water current was also altered in the enriched tanks by adjusting the water supply to produce a directional current around the tank or to have no current by directing the flow into the wall of the tank. The standard tanks were left devoid of any structure inside the tank. Current was not altered in the standard tanks; it was producing a directional current throughout the rearing period. Figure 2.1. Images of an enriched hatchery tank on the left and a standard hatchery tank on the right. Floats around the air stones were used in all tanks to prevent fish from jumping out. The fish were fed commercial fish feed pellets produced by Bio-Oregon. Aquatic Ecosystems, Inc. belt feeders were used to provide consistent daily feeding. The feed schedule and location of feed was left the same daily for fish in the standard treatment. The time of day feed was given varied for the enriched fish. This was accomplished by rotating percentages of

22 12 daily feed amounts in 8-hour feeding blocks. For example, on day one fish may be fed all of their food late in the day and the following day they may get their food fed to them constantly throughout the eight-hour day. The location of the belt feeder was also changed weekly for the enriched fish, at the same time the enrichment was changed. The fish were reared and maintained in this way until stocking 3-4 months later (Table 2.1). Table 2.1 Timeline of important dates in the hatchery and of behavior trials for brook trout in 2011, 2012, and Date Age (Days) Date Age (Days) Date Age (Days) Fish Hatched January 21 0 January 31 0 February 7 0 Reared in 4-foot tanks January February February May 27 3-May May 26 Transferred to 10-foot May May May tanks Objects added to June July June enrichment tanks Time to cover trials September September 225 September 222 conducted Startle trials conducted June July N/A N/A July 27 August 3 Predator trials conducted N/A N/A September 14- October N/A N/A Fish stocked September N/A N/A September Size Comparison Prior to Stocking In order to assess whether the rearing environment resulted in a size or condition difference, a minimum of 50 fish from the enriched and the standard rearing environments were weighed and measured prior to stocking.

23 13 Pre-release Behavior Screening 1. Time to Cover Trials To quantify how the fish respond to cover and open space in a novel environment, a behavioral assay was used. These tests were done prior to stocking, approximately 3 months after the trout were transferred to the 10-foot diameter tanks. To conduct these trials, a fish was removed from one of the standard or enriched tanks by dip net and then transferred to a clearsided, 5-inch diameter, start cylinder in a test tank positioned in the center of a start area. The test tank was 8 feet long and 18 inches wide with a water depth of 5 inches. It was divided into a start area (6 inches long) and at the other end of the tank dense cover was put in place using 4 plastic plants and 2 upturned plastic pots, which provided shelter possibilities. Cobble was used to hold this cover in place and a line of cobble was placed just outside of the cover, 12 inches from the back of the tank, to mark the start of the cover zone. Each fish was allowed to settle for 2 minutes, it was then released into the test tank using a cord attached above the start cylinder so it could be raised remotely, allowing the observer to remain hidden throughout the trial. The time to leave the start area, defined by when the fish s entire head crossed a line of small rocks on the bottom of the tank, was then recorded as a measure of boldness. Time to reach the cover zone was defined as when the fish s head crossed the line of small rocks at the start of the cover area. Fish were observed up to 300 seconds or until they reached the cover zone. A minimum of 32 standard and 32 enriched fish were screened this way each year. 2. Startle Response Trials In the first and second year of the study (the 2011 and 2012 year classes) trials were conducted to assess whether the fish exposed to the contrasting rearing conditions differed in their startle response. These trials were conducted in the 10-foot hatchery rearing tanks and all of the

24 14 fish in the tank were observed. The fish s response to a person approaching the tank, and the recovery time to this response was monitored. An observer began by standing in a standard spot and recording the location of all of the fish throughout the tank (gauging their approximate position in the tank and water column). Next, the observer approached the tank at a slow, steady rate and recorded how far they were from the tank when the fish began to react to the approach by darting away from the observer. After this initial minor startle the new position of the fish was recorded and a more animated startle was conducted by quickly moving towards the tank and waving a meter stick over the tank twice. This startle resulted in the majority of fish fleeing to the opposite end of the tank. After the second startle, the observer returned to the standard starting spot and the location of the fish and their subsequent recovery was recorded at 30, 90, and finally at 150 seconds. This recovery was measured by the percentage of fish that had returned to their original location. These trials were usually performed at 3 points across the day on 10 different days over a period of three and a half weeks. The 3 different time points across the day were conducted at approximately the same time each day and for simplicity are referred to as early, mid-day and late. Due to time constraints, it was not possible to make all three of the observations on some days. 3. Predator Trials Predator trials were conducted in the second year of the study with the 2012-year class fish. These trials were used to evaluate whether enriched fish have an advantage when exposed to a real predator. Smallmouth bass, Micropterus dolomieu, were collected by angling on nearby Woodhull Lake. These are typical predators on brook trout within many Adirondack Mountain region lakes. The bass were transferred to the Little Moose Field Station for trials. The smallmouth bass were placed into four 10-foot diameter concrete flow through tanks and allowed

25 15 a day to acclimate. In each tank, a small rock structure was constructed which the brook trout could freely swim in and out of but the smallmouth bass could not access due to its size. This provided a refuge for the trout to use when the smallmouth bass made predation attempts. Brook trout were fin clipped to designate treatment (upper caudal fin clip for enriched fish, and lower caudal fin clip for standard fish), and then 5 fish of each treatment group were placed into the tank at the same time with the smallmouth bass (i.e. 10 trout and 2 to 4 bass per replicate). Trials lasted for up to 72 hours and 2 to 4 smallmouth bass were used in one trial. The variation in the number of smallmouth bass used was due to some mortality of smallmouth bass once brought into the hatchery. On completion of a trial, the brook trout were removed from the tank and the number and treatment group of any fish consumed by the predators was recorded. Twenty-six separate trials were conducted. Statistical Analysis Time to cover trial data were analyzed in two ways. In the first analysis the mean time to cover of each rearing tanks sample was compared using an analysis of variance (ANOVA). Because fish were reared in a true hatchery setting, this study was limited to two tanks for each replicate, which resulted in very low statistical power. With this issue in mind, the data were also analyzed at the level of individual fish by testing mean time to cover of each treatment group for differences using an ANOVA. Analysis of this type has been used in previous studies with similar limitations (Braithwaite & Salvanes 2005). In both analyses equality of variance was tested, and if necessary the response variables were log transformed. Fish that never left the start area during a 300 second trial were removed from the dataset, resulting in a minimum sample size of n=22 for the individual fish analysis (Table 2.2).

26 16 Table 2.2. Number of fish excluded from time to cover trials by year and species, and the sample size (n) for each time to cover trial. Number of Fish Used for Analysis Number of Fish Excluded (n) Species Year Class Standard Enriched Standard Enriched Brook Trout Atlantic Salmon Startle response trial data were analyzed using an ANOVA. In 2011, the early observations were conducted 10 times, the mid-day 8 times, and the late 5 times. In 2012, the early observations were conducted 6 times, the mid-day 5 times, and the late 4 times. In order to account for these different sample sizes, the overall mean percentage of fish recovered at 30 seconds, 90 seconds, and 150 seconds after the aggressive startle was calculated. A mean percent of fish recovered was then calculated for each tank and compared using ANOVA. A similar method was used to investigate observer distance from the tank when the fish first startled. In this analysis, the mean distance of each day was taken and a mean of these distances was calculated for each tank. An ANOVA was used to compare the tank means and discover if the treatments showed different responses over the 10 days of trials. Predator trials data were analyzed using binomial logistic regression because of the binomial distribution of the data (0 = consumed by predator, 1 = survived trial). This analysis allowed us to incorporate every trial, including those in which fish were not consumed, resulting in a total number of 130 fish from each treatment, and 26 trials. Growth in the hatchery setting was analyzed by comparing mean lengths and weights of each treatment using an ANOVA. The sample sizes compared were n= 50 enriched and n=50 standard for 2011 year class, n=100 enriched and n=100 standard for 2012 year class and n=50 enriched and n=50 standard for 2013 year class. The Fulton condition factor was calculated to

27 17 compare the condition of fish from each treatment, as described by Anderson and Neumann (1996): K = (W/L 3 ) * 100,000 where K is the condition factor, W is the weight in grams and L is the length in millimeters. All statistical analyses were conducted in StatView (version 5.0.1), with the exception of the predator trials which were analyzed in R (version 3.0), and significance was tested at α = Atlantic Salmon Atlantic Salmon Rearing Landlocked Atlantic salmon (n=2600 per year class) were first generation West Grand Lake strain fish originating from a wild broodstock and supplied by Grand Lake Hatchery, Maine. All fish rearing was done at the Little Moose Field Station, near Old Forge, NY and followed the same basic protocol as stated for brook trout. One important difference from the brook trout protocol is the amount of time spent in the rearing environments. For the 2011 year class, once enrichment was added to the enriched rearing tanks the enrichment and feeding schedule continued for a period of approximately 12 months. The 2012 year class was reared in this way for 3 months, until September, at which point behavior was tested and the enrichment was removed ending the study of Atlantic salmon (Table 2.3). Size Comparison Prior to Stocking In order to assess whether the different rearing environments generated a size difference,

28 fish from the enriched environment and 100 fish from the standard environment were weighed and measured for length just prior to stocking in June for the 2011 year class. A condition factor was calculated for each of these fish using the same methods described for the brook trout. Pre-stocking growth data was not obtained for the 2012 year class. Table 2.3 Timeline of important dates in the hatchery and of behavior trials for Atlantic salmon in 2011 and Date Age (Days) Date Age (Days) Fish Hatched March 21 0 March 31 0 Reared in 4-foot tanks March April June 23 June 9 Transferred to 10-foot June June 9 70 tanks Enrichment added June July Time to cover trials May September 166 conducted 13 Startle trials June July conducted July 27 August 3 Fish stocked June N/A N/A Pre-release Behavior Screening 1. Time to Cover Trials Time to cover trials were conducted by using the same protocol and materials as described for brook trout. For the 2011 year class, the tests were done just prior to stocking, in June, after fish were exposed to different rearing environments for a period of about 12 months. The trials were repeated on 41 standard and 41 enriched fish of the 2011 year class. Time to cover trials for the 2012 year class were conducted in September of 2012, after only 3 months of

29 19 exposure to the different rearing environments. For the 2012 year class fish trials were completed on 32 standard and 32 enriched fish. 2. Startle Response Trials Startle response trials for Atlantic salmon followed the same methods as for the brook trout. Atlantic salmon trials were done on the same days, and during the same time periods as the brook trout, so the salmon were younger and had been in the enrichment environment for only 1 to 7 weeks during the trials. Startle response trials were completed for both the 2011 and 2012 year classes. Statistical Analysis Time to cover trials were compared using ANOVA in the same way as the brook trout. Equality of variance was tested, and the data met the assumptions for analysis. After 300 seconds fish that had not left the start area were excluded from the analysis, giving a sample size of n=29 enriched and n=22 standard for the 2011 year class, and n=31 enriched and n=31 standard for the 2012 year class (Table 2.1). Startle response trial data and growth in the hatchery setting data were analyzed in the same manner as stated for the brook trout. All analyses were done using StatView (version 5.0.1).

30 20 Results Brook Trout Size Comparison Prior to Stocking The 2011-year class enriched treatment fish were significantly larger than the control fish in both mean length (F 1,98 =15.66, p=0.01, Figure 2.2A) and mean weight (F 1,98 =8.036, p=0.01, Figure 2.2B). The measured lengths for enriched fish ranged from 71 to 106 mm, and measured weights ranged from 3-12 grams. The standard fish ranged from mm and 2-13 grams. The 2012-year class enriched fish were significantly smaller than the standard fish in both mean length (7.39 mm smaller, F 1,198 = 16.40, p<0.01) and weight (1.27 grams smaller, F 1,198 =10.105, p<0.01). Measured sizes ranged from mm and 3-12 grams for enriched and mm and 3-15 grams for standard. The 2013-year class enriched fish were significantly larger than the standard fish (mean length 6.28 mm larger, F 1,98 =10.570, p=.0016, mean weight 0.86 grams larger, F 1,98 =11.728, p=.0009). Enriched fish sizes measured ranged from mm and 1-7 grams and standard from mm and 1-5 grams. The condition factor value was not significantly different in the 2011 (F 1,98 =2.88, p=.093), 2012 (F 1,198 =1.23, p=.270), or 2013 (F 1,98 =.052, p=.820) year classes (Figure 2.2C).

31 21 A) Enriched Standard Lemgth (millimeters) Year Class B) Enriched Standard Weight (grams) Year Class C) Enriched Standard CondiEon Factor Year Class Figure 2.2. (A) Mean length of brook trout in enriched and standard rearing environments of each year class. Bars represent ± standard error. (B) Mean weight of brook trout in enriched and standard rearing environments for each year class. Bars represent ± standard error. (C) Mean condition factor of brook trout for each rearing environment by year class. Bars represent ± standard error.

32 22 Time to Cover Trials Comparing the mean time to cover for each tank lacked enough statistical power to show a statistical difference for the 2011 (F 1,2 = 2.95, p= 0.23, Figure 2.3), 2012 (F 1,2 = 0.47, p= 0.56, Figure 2.4 ), and 2013 (F 1,2 = 0.68, p= 0.50, Figure 2.5) year classes. Seconds to enter cover zone Tank 2 Tank 6 Tank 4 Tank 8 Standard Rearing Tank Enriched Figure 2.3. Mean amount of time it took brook trout of the 2011 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error. Seconds to enter cover zone Tank 4 Tank 8 Tank 2 Tank 6 Standard Rearing Tank Enriched Figure 2.4. Mean amount of time it took brook trout of the 2012 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error.

33 23 Seconds to enter cover zone Tank 2 Tank 4 Tank 6 Tank 8 Standard Rearing Tank Enriched Figure 2.5. Mean amount of time it took brook trout of the 2013 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error. When analysis is taken to the level of the individual fish, the 2011-year class fish from the standard rearing environment took significantly longer to reach the cover zone than fish raised with enrichment (F 1,46 = 4.45, p=0.04, Figure 2.6 ). The 2012-year class fish showed no difference between rearing environments in their time to reach cover (F 1,72 = 0.28, p=0.60, Figure 2.6). The 2013-year class standard rearing environment fish again took longer to enter the cover zone (F 1,49 = 4.92, p=0.03, Figure 2.6). There was no difference between fish of the two rearing environments for the time to leave start in any of the three year classes. Enriched Standard Seconds to enter cover zone Year Class Figure 2.6. Mean amount of time it took brook trout from different rearing environments to go from start area to the cover zone. Bars represent ± standard error.

34 24 Startle Response Trials For the 2011 year class, the startle response of brook trout in the enriched environment was faster than the brook trout reared in the standard environment. The enriched fish startled sooner; the observer was further away from the tank when the trout startled for the enriched fish compared to the standard fish (F 1,2 = 28.82, p=0.03, Figure 2.7). After the aggressive startle the enriched fish were slower to return to their original positions, resulting in a higher percentage of standard fish recovering within 150 seconds compared to the enriched fish (F 1,2 = 60.63, p=0.02, Figure 2.8). Startle Distance Tank 2 Tank 6 Tank 4 Tank 8 Standard Rearing Tank Enriched Figure 2.7. Mean distance the observer is from the tank when the fish begin to startle for each tank with 2011 year class brook trout. Bars represent ± standard error. Percent Recovery Tank 2 Tank 6 Tank 4 Tank 8 Standard Rearing Tank Enriched Figure 2.8. Mean percentage of brook trout that returned to their original positions in home tanks after the aggressive startle occurred for the 2011 year class. Bars represent ± standard error.

35 25 The 2012 year class showed no differences in observer distance from the tank before the fish startled (F 1,2 =0.09, p=0.79, Figure 2.9) or in the rate of recovery after the aggressive startle (F 1,2 =1.04, p=0.42, Figure 2.10). Startle Distance Tank 4 Tank 8 Tank 2 Tank 6 Standard Rearing Tank Enriched Figure 2.9. Mean distance the observer is from the tank when the fish begin to startle for each tank with 2012 year class brook trout. Bars represent ± standard error. Percent Recovery Tank 4 Tank 8 Tank 2 Tank 6 Standard Rearing Tank Enriched Figure Mean percentage of brook trout that returned to their original positions in home tanks after the aggressive startle occurred for the 2012 year class. Bars represent ± standard error. Predator Trials When exposed to smallmouth bass predation in the hatchery tanks, more fish reared in the enriched environment avoided predation then fish reared in the standard environment (p=0.03,

36 26 Figure 2.11). 85 % of enriched fish survived the trials with smallmouth bass, compared with only 74 % of the standard fish. A total of 56 brook trout were consumed. The rate of predation was calculated as one brook trout per 20.6 hours of trial time. Figure Proportion of fish that survived predation from the enriched and standard rearing environments over 26 trial periods. Bars represent ± standard error. Atlantic Salmon Size Comparison Prior to Stocking Landlocked Atlantic salmon from the standard rearing environment weighed significantly more than fish from the enriched environment (F 1,198 = 6.95, p=0.01, Figure 2.12A). Although, the mean length of the standard fish was also larger than the length of the enriched fish, this difference was not significant (F 1, 198 =2.84, p=0.09, Figure 2.12B). There were also no significant differences in the condition factor between fish from each rearing environment (F 1,198 =3.28, p=.072. Figure 2.12C).

37 27 A) Enriched Standard 20 Weight (grams) B) Enriched Standard Length (millimeters) C) Enriched Standard CondiEon Factor Figure (A) Mean weight of Atlantic salmon from each rearing environment after 12 months in differing rearing environments. Bars represent ± standard error. (B) Mean length of Atlantic salmon from each rearing environment after 12 months in differing rearing environments. Bars represent ± standard error. (C) Mean condition factor of Atlantic salmon from each rearing environment after 12 months of rearing in different environments. Bars represent ± standard error.

38 28 Time to Cover Trials Comparing the mean time to cover for each tank showed enriched fish moving to cover significantly quicker then standard fish for the 2011 year class (F 1,2 = , p= 0.02, Figure 2.13), and no difference in time to reach cover for the 2012 Atlantic Salmon year class (F 1,2 = 3.72, p= 0.19, Figure 2.14 ). Seconds to enter cover zone Tank 1 Tank 5 Tank 3 Tank 7 Standard Rearing Tank Enriched Figure Mean amount of time it took Atlantic Salmon of the 2011 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error. Seconds to enter cover zone Tank 3 Tank 7 Tank 1 Tank 5 Standard Rearing Tank Enriched Figure Mean amount of time it took Atlantic Salmon of the 2012 year class from each tank with different rearing environments to go from start area to the cover zone. Bars represent ± standard error.

39 29 When analysis was conducted using fish as the level of replication, the 2011 year class standard rearing environment Atlantic salmon took significantly longer to reach the cover zone compared to fish reared in the enriched environment (F 1,54 =8.23, p=0.01, Figure 2.15). The 2012 year class salmon showed no differences between rearing environments in the time to reach cover (F 1,60 =1.93, p=0.17, Figure 2.15). There was no difference among rearing environments for the time to leave the start area for either year class. Enriched Standard Seconds to enter cover zone Year Class Figure Mean amount of time it took Atlantic salmon from different rearing environments to go from the start area to the cover zone. Bars represent ± standard error. Startle Response Trials There were no significant differences amongst rearing environments for the distance the observer was from the tank when fish first startled (F 1,2 =0.20, p=0.70, Figure 2.16) or the percent recovery of fish after an aggressive startle (F 1,2 =0.002, p=0.97, Figure 2.17) for salmon of the 2011 year class.

40 30 Startle Distance Tank 1 Tank 5 Tank 3 Tank 7 Standard Rearing Tank Enriched Figure Mean distance of observer from each tank when the fish begin to startle for the 2011 year class Atlantic salmon. Bars represent ± standard error. Percent Recovery Tank 1 Tank 5 Tank 3 Tank 7 Standard Rearing Tank Enriched Figure Mean percentage of Atlantic salmon that returned to their original positions in home tanks after the aggressive startle occurred for the 2011 year class. Bars represent ± standard error. The 2012 year class Atlantic salmon were only reared in 2 tanks total (one for each rearing environment), therefore statistical analysis could not be done. Mean values for the distance observer is from tank when fish startle are shown in Figure 2.18, and Figure 2.19 shows mean percentages of fish recovered after an aggressive startle.

41 31 Mean observer distance from tank when fish startle (cm) Enriched Standard Figure Mean distance the observer is from the tank when the fish begin to startle for the 2012 year class Atlantic salmon. Enriched Standard Mean % of fish returned to original posieons in tank Figure Mean percentage of Atlantic salmon that returned to their original positions in home tanks after the aggressive startle occurred for the 2012 year class. Discussion Size Comparison Prior to Stocking Brook trout of the 2011 and 2013 year classes that were raised in the enriched environment were larger than the brook trout raised in the standard environment. This difference is encouraging because if a fish can get to a larger size in the hatchery environment they may

42 32 have some survival advantages (Sogard 1997). However, the 2012 year class showed the opposite effect. In 2012, brook trout from the enriched environment were smaller than brook trout from the standard environment. Why the opposite effect occurred for the 2012 year class is unknown, however there was some high mortality in the hatchery that may have contributed to this. Unlike in 2011 and 2013, when the majority of fish were healthy and there was little mortality in the hatchery, 2012 fish experienced an outbreak of bacterial gill disease. This resulted in a higher than average amount of mortality in both rearing environments. Interestingly, more dead brook trout were noted in the enriched environment tanks (D. Josephson, personal communication). However, when stocking time came and fish were counted it appeared that more brook trout had died in the standard rearing environment tanks. It is possible the reason for this could be many of the weakened, sick, standard fish were consumed by other fish in the standard rearing treatments whereas in the enriched rearing environment the weakened fish were able to utilize cover while in the weakened state, where they could either avoid cannibalism or end up dying, being noted and removed from the tank. If standard fish did exhibit a higher rate of cannibalism it would explain the larger mean growth rate for standard fish in the 2012 year class. Unfortunately we do not have any solid quantitative data to support these speculations in the hatchery. Another way to investigate size differences amongst the two rearing environments is by utilizing Fulton s condition factor. The condition factor allows you to compare the length-weight relationship of each fish, as an index of wellbeing. The mean condition factor was not different between each rearing environment for any of the three-year classes of brook trout, suggesting that fish are not at a disadvantage as far as their state of wellbeing in either rearing environment. Atlantic salmon of the 2011 year class from the standard rearing environment were larger than salmon from the enriched rearing environment. This contradicts what we would expect based on the three years of brook trout data that were obtained. It is possible that there are species

43 33 differences between the brook trout and Atlantic salmon. Both the brook trout and Atlantic salmon data contradict other studies that have assessed growth in different rearing environments in the hatchery. Brockmark and Johnson (2010) found no difference in growth among brown trout in different rearing environments. Berejekian et al. (2000) found no difference in percent daily growth rate amongst rainbow trout raised in different rearing environments and Smith (2011) found no growth difference amongst westslope cutthroat trout or rainbow trout raised in different rearing environments. Why I found differences and studies of other salmonids did not, is not clear. Time to Cover Trials The 2011 and 2013 year classes again showed differences with the enriched environment brook trout moving to cover faster than the standard environment fish. Being able to move to cover and seek refuge in a novel environment could help fish to escape any initial predation that occurs, which may help with survival just after fish are stocked into the wild. However, the 2012 year class fish behaved differently again. For the 2012 year class fish, there was no difference between the enriched and standard rearing environment fish for the time to reach cover. Why this occurred in this one year, is unknown, but it is interesting that it goes along with the year of the other abnormal hatchery size data, and the bacterial gill disease in the hatchery. In the first year of Atlantic salmon trials (2011 year class) they showed similar results to the brook trout of the 2011 and 2013-year classes, with the salmon from the enriched rearing environment taking less time to reach cover then salmon from the standard rearing environment. The 2012 year class Atlantic salmon did not show significant differences in their time to reach cover. These trials were done when the salmon were at a younger age and were exposed to the different rearing environment for

44 34 a shorter period of time. It would have been ideal to redo the time to cover trials after 12 months of exposure to the rearing environments, and just prior to stocking as was done previously, however logistical constraints made this not possible. It is possible that the time to cover difference may develop later in the Atlantic salmon, or that the difference was not repeatable in the second year of the study. I am unaware of any other studies with salmonids that have investigated this time to cover measure in the same way that we did. Smith (2011) reported results with cutthroat trout (Oncorhynchus clarki) and rainbow trout (Oncorhynchus mykiss) from enriched environments and showed that they were more likely to use and stay in cover compared to fish from standard environments. However, Berejikian et al. (2000) found no difference in juvenile steelhead that had been reared in different environments and were tested in a complex habitat over a 72-hour period. Their study was slightly different because they used semi natural streams and a longer period before measuring cover usage in the streams. Smith s (2011) study, and the results reported in this thesis, collected data in a controlled laboratory setting within a short time period. It is possible that after a longer period, over days, fish from each environment could adjust and use cover similarly. More research should be directed at determining how long it takes the fish to adapt to the novel environment and behave in the same way, regardless of rearing environment. Startle Response Trials The 2011 brook trout startle response trials indicate that enriched environment fish startle sooner than the standard environment fish, as well as the enriched environment fish take longer to recover after an aggressive startle. This difference was developed after just six weeks in the different rearing environments, suggesting that it may be possible to change behavior quickly at

45 35 this age. In 2012, the brook trout showed no differences in the distance the observer was from the tank, and there were no differences in the time taken to recover from the startle. Fewer trials were conducted than in 2011, and they were completed earlier, before the fish had experienced the rearing environments for 6 weeks. It is possible that differences might have developed had the trials continued for longer. Atlantic salmon did not show any difference in startle response or time to return after an aggressive startle in 2011 either. It is however worth noting that during the last days of the trials differences seem to be occurring for the first time. In 2012 Atlantic salmon were only reared in two tanks (one with enrichment and one standard) so statistical analysis was not done for any of the startle trials. However, the means from each of the tanks looked similar to the 2011 salmon with no clear differences. The Atlantic salmon were considerably younger than the brook trout when placed into the different rearing environments and it is possible that it takes longer to develop behavioral differences at a younger age. It is also possible this is a species difference where the salmon require longer in the treatment before they develop differences, or they perhaps do not develop differences at all. It would be interesting to extend the trials several more weeks to observe what changes develop. Based on the brook trout results and other salmonid studies found in the literature, I would expect the Atlantic salmon to develop differences if the trials were continued for a longer period (Roadhouse et al. 1986; Smith 2011). The brook trout results support the findings of Roadhouse et al. (1986), who showed that lake trout (Salvelinus namaycush) reared in environments with cover exhibited an increased set of startle behaviors, and Smith (2011) also found that both rainbow and westslope cutthroat trout from enriched environments exhibited an increased fright response.

46 36 Predator Trials Brook trout from the enriched rearing environment survived smallmouth bass predation better in the controlled hatchery setting than brook trout from the standard rearing environment. This result suggests that enriched environment fish may experience higher survival in the wild than standard environment fish. While Smith (2011) did not find any differences in survival of rainbow trout when exposed to brown trout (Salmo trutta) in a laboratory setting, he reported that the enriched fish had the lowest odds of being consumed by a predator but the analysis was weakened by high variation amongst the trials. From this, he suggested the enriched fish may have a survival advantage. I am unaware of another study that compares predation survival in a controlled setting with fish exclusively from different rearing environments. Summary In controlled laboratory trials, the rearing environment appears to have altered the behavior of both the brook trout and Atlantic salmon. Brook trout differences were seen in latency to swim along an open channel to reach cover for the 2011 and 2013 year class, as well as in startle response sensitivity, recovery time after a startle, and survival when housed in the presence of a predator. Atlantic salmon also showed differences in time to cover prior to stocking for the 2011 year class. Future studies would benefit by continuing for more year classes of both species, to help determine how repeatable these behavioral changes are. These results support those of earlier studies with salmonids that have shown one or more behavioral trait that has been altered by the early rearing environment of fish (Berejikian et al. 1999; Berejekian et al. 2000; Lee and Berejikian 2008).

47 37 Chapter 3 The effects of enrichment in the hatchery environment: Can we increase the survival of stocked Brook Trout (Salvelinus fontinalis) and Atlantic Salmon (Salmo salar) in the wild? Introduction Many studies in recent years investigate the idea of pre-release training for fish as an effort to better prepare fish for stocking into the wild (reviewed in Chapters 1 and 2). However, studying the link between creating behavioral flexibility in the laboratory and the ability to increase the survival rate once stocked into the wild is much less common. The logistics that are involved and expected low recapture rates of any stocked fish make this type of study difficult to accomplish. I have found few studies where survival rate is assessed in salmonids after any form of pre-release training (Berejikian et al. 1999; Brockmark et al. 2007; Fast et al. 2008; Tatara et al. 2009; Roberts et al. 2011). Only two of these studies investigate survival after simple exposure to enriched or standard rearing environments. Fast et al. (2008) conducted a 5 year study with Chinook salmon where they compared the post-release survival of fish from enriched and conventional treatments. They found no differences in survival from rearing treatments over this time period. Similar results were found by Tatara et al. (2009), who investigated Steelhead reared in enriched and conventional treatments. No differences were found in survival or growth between treatments in the wild. I am unaware of any study investigating the survival of brook trout or landlocked Atlantic salmon after exposure to differing rearing environments. As stated in Chapter 2, brook trout and Atlantic salmon are heavily stocked in the eastern United States and

48 38 restocking or reintroduction programs are becoming increasingly common. Therefore, after investigating the behavioral adaptations of brook trout and Atlantic salmon as described in Chapter 2, the second component of this study was to determine whether these behavioral differences would translate to an increase in survival in the wild. This study is unique not only because it investigates the survival of two different salmonid species but also because survival is assessed in a freshwater lake system. The Steelhead study by Tatara et al. stocked fish into experimental stream sections and survival was assessed on juveniles in these stream sections prior to their smolt migration. Fast and colleagues (2008), assessed survival on the Chinook salmon smolt migration out of the river as well as the adults returning to the river for spawning. Therefore, this is the first study assessing the survival of any non-anadromous salmonid species after exposure to differing hatchery environments in an uncontrolled natural setting. Study System In this study, 10 different lakes were used for the survival assessment of stocked brook trout, Atlantic salmon or both. The lakes were located in a small geographic area in the southwestern Adirondack Mountain region of New York State. The lakes had a wide range of characteristics, such as size, depth, fish community, habitat, etc. and therefore represent the wide range of Adirondack lakes. Study lake characteristics are shown in Table 3.1 and Table 3.2 as described by Josephson and Kraft (2011). Larger fish found in these bodies of water are not the only source of predation for the stocked fall fingerlings (age 0). Common loons and common mergansers also frequent many of these lakes as well as mink and river otters that may be found along the shorelines of many waterways. Most of these lakes are very remote, many of them accessible only by hiking. All the lakes are located on private property, where angler harvest is

49 39 monitored and known to be very low from decades of creel card data collected by the Adirondack Fishery Research Program. Because of the variety of waters to be stocked, the known low amount of angling mortality, and the varying lake systems, this was an ideal set of lakes for this study. Table 3.1. Size, mean depth, max depth, water color, and whether the lake thermally stratifies or not in the summer is shown for the 10 Adirondack lakes in this study. Lake Size (Acres) Mean Depth Max Depth Water Color Thermally Stratify? Canachagala Lake Clear Yes Chambers Lake Dark Yes Fourth Bisby Lake Dark Yes Goose Lake Dark Yes Green Lake Clear Yes Little Moose Lake Clear Yes Lower Sylvan Pond Clear No Mountain Pond Clear Yes Panther Lake Clear No Pinchnose Pond Clear Yes Table 3.2. Fish species present in each lake as indicated by an X. Blacknose Dace Common Shiner White Sucker Slimy Sculpin Lake Trout Smallmouth Bass Rainbow Smelt Pumpkinseed Sunfish Mudminnow Brown Bullhead Creek Chub Wild Atlantic Salmon Stocked Atlantic Salmon Stocked Brook Trout Wild Brook Trout Canachagala Lake X X X X Chambers Lake X X Fourth Bisby Lake X X X Goose Lake X X Green Lake X X X X X Little Moose Lake X X X X X X X X X X X X X X Lower Sylvan Pond X X X X Mountain Pond X Panther Lake X X X Pinchnose Pond X

50 40 Methods Brook Trout Marking After being reared in different environments for a 3 month period (Chapter 2), individual brook trout (n=6400 for 2011 year class, n=860 for 2013 year class) were marked to indicate which environment they were from. Visible implant elastomer tags (VIE) from Northwest Marine Technology, were used to provide this designation. A red VIE tag was injected beneath the skin behind the left eye in order to distinguish a fish from the standard rearing environment, fish from the enriched environment were marked with a red VIE tag behind the right eye. VIE tags were put in place 10 days prior to stocking to assure tags had extra time to harden and the brook trout skin could adequately heal over. All 2011 year class brook trout were also given an adipose fin clip at this time, which was standard protocol for the Adirondack Fishery Research Program to designate the fish as a 2011 year class hatchery fish. The 2013 year class fish were given a right pectoral fin clip in addition to the VIE tag to designate them as a 2013 year class hatchery fish. This secondary mark was also critical for this study, allowing a measure of tag retention in the field and giving an easy indicator to look more in depth for VIE tags that had become difficult to see.

51 41 Stocking Marked brook trout (2011 year class) were stocked into 12 different lakes and ponds by float plane on September 22, Marked brook trout of the 2013 year class were stocked into 2 lakes (Chambers and Fourth Bisby) on September 19, As discussed in this chapter s introduction, the lakes and ponds varied in size and attributes but were all located within a small geographic area near Old Forge, NY. All lakes were stocked at a rate of approximately 10 fish per acre with the exception of Pinchnose Pond and Lower Sylvan Pond (Table 3.3). The same number (paired plants) of enriched environment and standard environment fish were stocked into each lake/pond. For example, if a pond required 100 fish to be stocked the pond would receive 50 enriched fish and 50 standard fish. Table 3.3. Lakes that were sampled for brook trout recapture, along with the stocking rate and amount of fish stocked in each lake. Lake Size (acres) Stocking Rate (fish/acre) Enriched Fish Stocked Standard Fish Stocked Canachagala Lake Chambers Lake Fourth Bisby Lake Goose Lake Green Lake Lower Sylvan Pond Mountain Pond Pinchnose Pond Field Recapture Between September 19, 2012 and October 12, 2012, approximately 1 year after stocking, field sampling was conducted to recapture the 2011 year class brook trout. Due to time

52 42 constraints, only 8 of the 12 lakes were sampled. Most lakes were intensively sampled for one day, however some lakes received more effort with multiple days of sampling (Table 3.4). The 2013 year class brook trout were sampled between September 29, 2014 and October 8, Chambers Lake was sampled in 2014 from September 29 to October 3 and Fourth Bisby Lake was sampled from October 4, 2014 to October 8, Table 3.4. Dates sampled for 2011 year class brook trout and gear types used in each lake. Not all lakes had a second or third sample date. Sample Date 1 Gear Type Sample Date 2 Gear Type Sample Date 3 Gear Type Canachagala Lake October 10 th Gillnet Chambers Lake October 7 th October 8 th Fourth Bisby Lake Goose Lake Green Lake Lower Sylvan Pond Mountain Pond Pinchnose Pond Gillnet September 24 th Gillnet October 1 st Gillnet October 9 th Gillnet October 1 st -5 th Trapnet September 19 th Gillnet September 21 st Gillnet Gillnet October 4 th Gillnet October 5 th Gillnet October 8 th -12 th Trapnet October 12 th Gillnet Most lakes were sampled using gill nets, however in Mountain Pond and Lower Sylvan Pond trap nets were used in conjunction with the gill nets for sampling. The gill nets we used had 38 mm mesh, were 50 meters in length, and stretched up 2.5 meters from the lake bottom. Short, one hour, sets were used to assure there was no mortality of fish caught in the nets. Nets were set perpendicular to shore and a systematic sampling approach was used. In many of the smaller lakes, we were able to work our way around the entire shoreline with these one hour net sets. In large lakes we targeted shallow coves, outflow areas, areas with water sources coming in, or

53 43 anywhere else we suspected would result in a high catch. A wide variety of habitat and depths were sampled in each lake using this method. When trap nets were utilized, Oneida style trap nets with a 23 meter lead and 13 mm mesh with a box 1.2 m high were used. Trap nets were set perpendicular to shore in set locations known to catch fish from previous trap netting data. Trap nets were set over a 48 hour period before being checked. Regardless of how the fish were captured, all fish were given an upper caudal fin clip to allow us to check for recaptures and be sure the same fish was not counted twice. VIE tags were read in the field with a VI light from Northwest Marine Technology under a black plastic shroud to create a environment dark enough to fluoresce the tags. Statistical Analysis Recapture rate was analyzed using binomial logistic regression due to the binary structure of the data (1 = recapture, 0 = not recaptured). In order to look for an overall effect of rearing environment these data were analyzed with lake as a random effect in the model, to account for the variability among lakes. The recapture proportion of each lake was then analyzed with a simple binomial logistic regression to compare mean recapture percentages of fish from each rearing environment. While I am looking for an overall effect, the individual lake analysis allows detection of any differences that are occurring in individual lakes. Brook trout sizes at recapture were tested for differences between the rearing environments using a one way ANOVA. Equality of variance was tested, and if necessary they were log transformed. Like with the pre-stocking data, a condition factor was calculated to compare the condition of fish from each rearing environment using the following Fulton condition factor formula, as described by Anderson and Neumann (1996): K = (W/L 3 ) * 100,000

54 44 where K is the condition factor, W is the weight in grams and L is the length in millimeters. The mean condition factors were then compared using a one way ANOVA. Size analyses were conducted in StatView (version 5.0.1). Recapture data was analyzed using R (version 3.0), and significance was tested at α = Atlantic Salmon Marking Atlantic salmon of the 2011 year class were marked in the same way as the brook trout. After being reared in different environments for a 12 month period (see chapter 2), all individual Atlantic salmon (n=2600) were marked to indicate which rearing environment they were raised in. Each fish received an adipose fin clip and a red VIE tag behind the eye (behind left eye designates standard rearing environment, behind right eye designates enriched rearing environment). Atlantic salmon were clipped and marked 23 days prior to stocking, to assure the tag is solidified and the fish have fully recovered. Stocking Atlantic salmon were stocked into 6 lakes by truck on June 18 and June 20, Stocking rates varied by lake as shown in Table 3.5. Each lake was stocked with an equal number of enriched and standard fish. Table 3.5. Atlantic salmon lake size, stocking rates and number of fish stocked from each environment in the three lakes where recapture was attempted. Lake Size (acres) Stocking Rate (fish/acre) Enriched Fish Stocked Standard Fish Stocked Little Moose Lake Lower Sylvan Pond Panther Lake

55 45 Field Recapture Of the 6 lakes stocked with marked Atlantic salmon, only 3 were sampled in fall Little Moose Lake was sampled by boat electrofishing on the nights of September 26 th and 27 th, Lower Sylvan pond was sampled with gillnets on October 5 th, 2012 and with trap nets from October 1 st to 5 th Panther Lake was sampled by trap nets from October 1 st to 4 th 2012, and by gillnets on October 10 th, Gillnet and trapnet sampling methods were the same as described for the brook trout sampling, and the same gear was used. Boat electrofishing was systematically done around the entire perimeter of Little Moose Lake in the dark, over a period of two nights. Statistical Analysis Statistical analysis of the Atlantic salmon recapture data was conducted using binomial logistic regression, following the same methods for overall analysis and individual lake analysis as described for the brook trout. A one way ANOVA was used to compare the mean condition factors, lengths, and weights of Atlantic salmon following the methods previously described for the brook trout. On Little Moose Lake weights were not able to be measured, therefore analysis for weight or condition factor was not possible.

56 46 Results Brook Trout Brook trout were recaptured in all lakes where they were sampled, but overall recapture numbers were low year class recapture rates varied from a low of 2 fish in Canachagala Lake to a high of 30 fish in Chambers Lake. When combining data from all 8 lakes in the random effect model there was no effect of rearing environment on the recapture rate of 2011 year class brook trout (Figure 3.1). A total of 138 brook trout (2011 year class) were captured; 59 from the enriched rearing environment and 79 from the standard rearing environment. There was considerable variation in the recapture results from each lake (Figure 3.2). No differences were detected for recapture rate when analyzing each individual lake separately with the exception of Fourth Bisby Lake. In Fourth Bisby Lake more standard environment fish were recaptured than enriched environment fish for the 2011 year class (p<0.01, Figure 3.3). Table 3.6 shows recapture numbers of each treatment by gear type for Mountain Pond and Lower Sylvan Pond, where both gillnets and trapnets were used. Figure 3.1. Mean proportion of brook trout recaptured (bars show ± standard error) from enriched and standard rearing environments in the 8 Adirondack lakes.

57 47 Enriched Standard Canachagala Lake Lower Sylvan Lake Chambers Lake Goose Lake Fourth Bisby Pinchnose Lake Mountain Pond Number Recaptured Figure 3.2. Total number of brook trout recaptured in each lake from the enriched and standard rearing environment. Figure 3.3. Mean proportion of 2011 year class enriched environment and standard environment fish recaptured in Fourth Bisby Lake (± standard error).

58 48 Table 3.6. Number of brook trout captured in Mountain Pond and Lower Sylvan Pond by treatment and gear type. Lake Treatment Gillnet Caught Trapnet Caught Mountain Pond Enriched 3 5 Standard 7 9 Lower Sylvan Pond Enriched 7 3 Standard 13 3 More sampling effort was given in 2014 for Chambers and Fourth Bisby Lakes, resulting in a slightly higher recapture of the 2013 year class brook trout. A total of 82 brook trout of the 2013 year class were recaptured from these two lakes. Overall numbers included 43 enriched environment brook trout and 39 standard environment brook trout. There was no significant difference in overall recapture (Figure 3.4) or in the recapture rate when analyzed for each individual lake. Recapture numbers of the 2013 year class brook trout in each individual lake is shown in Figure 3.5. Figure 3.4. Mean proportion of 2013 year class brook trout recaptured (bars show ± standard error) from enriched and standard rearing environments in Chambers and Fourth Bisby Lakes.

59 49 Enriched Plain Chambers Lake Fourth Bisby Lake Number Recaptured Figure 3.5. Total number of 2013 year class brook trout recaptured in each lake from the enriched and standard rearing environment. Size at Recapture As expected brook trout size at recapture varied between each lake. Therefore, each lake s brook trout sizes were analyzed individually. Analysis was not possible on Canachagala Lake, with a sample size of only 2 enriched fish and 0 standard fish. There were no differences between the mean condition factors or mean weights of each treatment in any lake. There were differences in mean lengths of 2011 year class brook trout in Goose Lake (F 1,3 = 15.18, p=0.03) and Chambers Lake (F 1,28 = 4.25, p=0.05), but not in any other lakes (Table 3.7). The 2013 year class brook trout showed no differences in length, weight, or condition factor (Table 3.8).

60 50 Table 3.7. Mean lengths in millimeters of 2011 year class brook trout from the standard and enriched environments, as well as standard error and p-values for each lake. Lake Treatment Mean Length (mm) Standard Error P-Value Mountain Pond Enriched Standard Pinchnose Pond Enriched Standard Fourth Bisby Lake Enriched Standard Goose lake Enriched * Standard Lower Sylvan Pond Enriched Standard Chambers Lake Enriched * Standard Green Lake Enriched Standard Table 3.8 Mean lengths in millimeters of 2013 year class brook trout from the standard and enriched environments, as well as standard error and p-values for each lake. Lake Treatment Mean Length (mm) Standard Error P-Value Chambers Enriched Standard Fourth Bisby Enriched Standard Atlantic Salmon Atlantic salmon were recaptured in all 3 of the lakes where recapture was attempted. Across the three lakes sampled 61 Atlantic salmon were captured; 26 enriched fish and 35 standard fish. The overall analysis showed no differences between the rearing environments for the proportion of Atlantic salmon recaptured (Figure 3.6). There was also no difference when looking at recapture rates of individual lakes. The number of fish recaptured from each rearing environment are shown by individual lake in Figure 3.7. Table 3.9 shows the number of fish

61 51 recaptured by each gear type (gillnet and trapnet) in Panther lake and Lower Sylvan Pond. Figure 3.6. Mean proportion of Atlantic salmon recaptured (bars show ± standard error) from enriched and standard rearing environments in 3 Adirondack lakes. Enriched Standard Lower Sylvan Pond Panther Lake LiLle Moose lake Number Recaptured Figure 3.7. Total number of Atlantic salmon recaptured in each lake from the enriched and standard rearing environment.