Agonistic Behavior among Three Stocked Trout Species
Abstract
The popularity of reservoirs to support sport fisheries has led to the stocking of species that did not co-evolve, creating novel reservoir fish communities. In Utah, the Bear Lake strain of Bonneville Cutthroat Trout Oncorhynchus clarkii utah and tiger trout (female Brown Trout Salmo trutta × male Brook Trout Salvelinus fontinalis) are being more frequently added to a traditional stocking regimen consisting primarily of Rainbow Trout O. mykiss. Interactions between these three predatory species are not well understood, and studies evaluating community interactions have raised concern for an overall decrease of trout condition. To evaluate the potential for negative interactions among these species, we tested aggression in laboratory aquaria using three-species and pairwise combinations at three densities. Treatments were replicated before and after feeding. During the three-species trials Rainbow Trout initiated 24.8 times more aggressive interactions than Cutthroat Trout and 10.2 times more aggressive interactions than tiger trout, and tiger trout exhibited slightly (1.9 times) more aggressive initiations than Cutthroat Trout. There was no significant difference in behavior before versus after feeding for any species, and no indication of increased aggression at higher densities. Although Rainbow Trout in aquaria may benefit from their bold, aggressive behavior, given observations of decreased relative survival in the field, these benefits may be outweighed in reservoirs, possibly through unnecessary energy expenditure and exposure to predators.
Received October 1, 2014; accepted January 27, 2015
Agonistic behavior in fishes occurs as a means to obtain and defend resources, and establish dominance over other individuals in a community, which can increase the fitness of aggressive individuals or species (Grant 1990; Vøllestad and Quinn 2003; Stamps 2007). Many laboratory experiments have associated increased growth rates and feeding opportunities with aggressive behavior among salmonid species (Li and Brocksen 1977; Abbott and Dill 1989; Höjsejö et al. 2002). Grant (1990) documented similar results in the field and provided evidence that aggressive individuals are able to obtain better territories and have more feeding opportunities than subordinate and nonaggressive individuals. Furthermore, when species occur sympatrically, differences in exhibition of agonistic behavior may allow one species to gain dominance and growth advantage over another (Krueger and May 1991; Gunckel et al. 2002) or limit access of less aggressive species to the best feeding habitat (Grant 1990). An individual's size is also important in determining its level of aggression, larger individuals generally being more aggressive (Newman 1956; Abbott et al. 1985; Berejikian 1995). In addition, the presence of food has been shown to increase agonistic behavior between individuals (Newman 1956; Symons 1968), as well as the density of prey (Slaney and Northcote 1974).
Differences in agonistic behavior between species have important implications for fisheries management (Newman 1956). While aggressiveness may benefit an individual's performance in the hatchery, agonistic behavior may be detrimental in natural environments (Stamps 2007; Biro and Post 2008). Species selected for aggressive behavior and high growth rates in hatcheries (whether purposefully or incidentally) may experience lower survival in nature because of greater energy expenditure (Nilsson and Northcote 1981; Mesa 1991) and higher predation rates (Berejikian 1995; Biro et al. 2004). However, in communities composed of multiple salmonid species, one species may gain growth or competitive advantage over another by establishing dominance through increased aggression (Newman 1956; Gunckel et al. 2002), increasing its fitness and survival (Li and Brocksen 1977; Grant 1990; Deverill et al. 1999).
In Utah, novel communities of predatory salmonids have been created by managers, often in an attempt to hedge the potentially uncontrollable expansion of nongame fish in certain reservoirs and simultaneously provide angling opportunities. For example, exponential growth of the Utah Chub Gila atraria population in Scofield Reservoir prompted managers to shift the stocking program from exclusively Rainbow Trout Oncorhynchus mykiss to include tiger trout (female Brown Trout Salmo trutta × male Brook Trout Salvelinus fontinalis) and the Bear Lake strain of Bonneville Cutthroat Trout O. clarkii utah as pontential biological controls. Utah Chub are native to the Snake River and Bonneville basins of Utah, but not the Colorado River basin, where Scofield Reservoir is located. Even in their native basins, the high growth rate and reproductive potential of Utah Chub poses a problem to valuable sport fisheries (Winters 2014). Substantial niche overlap has been observed between Rainbow Trout, tiger trout, and Cutthroat Trout at top trophic positions, indicating the potential for direct competition for food and space (Winters 2014). Evidence for interspecific competition in Scofield Reservoir is further supported by recent documentation indicating low survival of Rainbow Trout since the initiation of Cutthroat Trout and tiger trout stocking; the reservoir is simultaneously producing many large Cutthroat and tiger trout of state-record quality (Winters 2014). In contrast, in Lost Creek Reservoir (also in Utah), Rainbow Trout do not appear to be experiencing negative effects from competition, but tiger trout are struggling to establish in the reservoir (C. Penne, Utah Division of Wildlife Resources [UDWR], unpublished data). However, data from 16 Utah reservoirs indicates a negative interaction between Rainbow Trout biomass and tiger trout growth and survival but a positive interaction between Cutthroat and tiger trout (R. Oplinger, UDWR, unpublished data).
Collectively, these studies suggest interactions between these three species are affecting species performance. While a competitive advantage of Rainbow Trout over Cutthroat Trout has been documented (Nilsson and Northcote 1981; Kreuger and May 1991), and tiger trout are expected to exhibit aggressive behavior similar to that of their parent species (McClane 1974; Scheerer et al. 1987), we are unaware of any published studies that attempt to explicitly examine the mechanisms behind these interactions. Overall, there appears to be little known about interspecific interactions among this unusual complex of predatory species, especially with regard to tiger trout.
In this study, our goal was to determine the extent of agonistic behavior between tiger, Cutthroat, and Rainbow trout, and explore the possible drivers of such behavior. Tiger trout are currently stocked in approximately 30 Utah reservoirs, generally in combination with Rainbow Trout and (or) Cutthroat Trout (Winters 2014), and other western states have other similar stocking programs (Miller 2010). Because tiger trout are increasingly added to preexisting species assemblages understanding agonistic behavior between top predators will inform management decisions of which fish to stock in order to best meet management objectives.
METHODS
Trials.
We obtained 16 triploid tiger trout, 18 diploid Cutthroat Trout, and 19 diploid Rainbow Trout from Utah's Fountain Green State Fish Hatchery. Upon arrival at our research center, each fish was measured (total length; mm) and weighed (0.1 g). Tiger trout averaged 247 mm TL (SE, 4.0) and 135.7 g (SE, 5.5), Cutthroat Trout were 248 mm TL (SE, 3.1) and 145.3 g (SE, 3.1), and Rainbow Trout were 243 mm (SE, 6.8) and 169.7 g (SE, 14.4). We marked each fish with a plastic anchor tag. We used tags uniquely colored for each species and uniquely numbered for each individual. We separated fish by species into three circular (diameter = 168 cm, depth = 76 cm) holding tanks filled to a depth of 65 cm with well water. Fresh well water was constantly trickling into the tanks to maintain a temperature between 12°C and 14°C. Each tank was aerated with three air stone disks, and minimal necessary flow was maintained with a bilge pump. Once daily, automatic feeders dispensed commercial trout feed to the fish. Use of automatic feeders eliminated human influence on fish feeding behavior. Fish were fed approximately 3% of their body weight daily (Wagner et al. 2006). Prior to behavior experiments we did not feed fish for at least 24 h. Fish arrived at our facility May 10, 2012, and behavioral trials did not commence until June 29, 2012, giving fish time to acclimate to new conditions. Behavioral trials were conducted between June 29 and August 16, 2012.
From preliminary ANOVA power analyses we determined approximately 20–25 trials would be necessary to detect a medium to large effect size in agonistic behavior between the three species at a power of 0.80 and α = 0.05. Hence, we set a target to complete 30 trials, to account for potential within and between group differences we may not have anticipated. After reaching our goal of 30 trials we used our remaining resources to conduct pairwise trials at various densities. Data collected after the original 30 trials were completed were opportunistic and considered supplementary, rather than necessary, to meet our overall objective.
To determine the extent of aggressive behavior between Rainbow, Cutthroat, and tiger trout, we indiscriminately netted (without replacement) one fish of each species to participate in a trial (three-species trials). Fish participating in a trial were placed in an observation tank (diameter = 122, depth = 35 cm) filled with fresh, 12°C well water. Tank size was carefully determined to place fish in close enough proximity that individuals would need to interact with one another, but large enough to allow enough space to distinguish aggressive behaviors and defensive reactions from normal movement. The tank was not aerated to provide maximum visibility during behavioral observation. Dissolved oxygen levels and temperature remained constant throughout the duration of each trial. Before each trial the tank was refilled with fresh well water.
We video-recorded all behavior trials in high definition (1,080 pixel) to minimize effects of human presence on fish behavior (Wagner et al. 2006). We placed the video camera above the tank, began recording after all fish were in the tank, and recorded each trial for 1 h. Our goal was to observe the subject fish before one achieved dominance over the others; once dominance is established agonistic behavior declines (Wagner et al. 2006); however, we also did not want to observe fish while they were still adjusting to conditions in the observation tank. From video analysis of test trials, we observed fish swimming erratically for a short time after introduction into the observation tank, but normal behavior resumed after about 30 min. Thus, we established the first 40 min of each trial as an acclimation period, and behavior was not analyzed during this period (Newman 1956).
During the remaining 20 min we recorded chases and bites– contacts initiated by each species and the recipient species of the aggressive behavior. We define a chase as one fish advancing within one body length of another and provoking a response, and define a bite–contact as any contact between two fish. Halfway (10 min) into the 20-min observation period an automatic feeder dispensed approximately 4 g of commercial trout feed into the observation tank to observe effects of feeding interest on agonistic behavior. We separated aggressive initiations by each species into prefeeding and postfeeding categories. The same observer recorded all behavioral interactions to eliminate observer bias.
To reach our desired sample size we used fish in multiple trials, after participating in a trial a fish was given at least 24 h rest before participating in another trial. On average each tiger trout participated in 4.5 trials (range, 1–9), Cutthroat Trout in 3.7 trials (1–7), and Rainbow Trout in 3.5 trials (1–6). Some fish participated in only one trial because, despite our use of tank covers, they managed to escape their holding tanks and died. In sum, we conducted 30 three-species trials and 36 pairwise (two-species) trials (Table 1). To assess the effects of density and intraspecific interactions, we conducted trials for all seven species combinations at three densities: low (2 fish/tank or 4.9 fish/m3), medium (3 fish/tank or 7.3 fish/m3, and high (4 fish/tank or 9.8 fish/m3 of 4.9 fish/m3). The pairwise trials were at low and high densities, and the three-species trials were at medium density. Though density treatments increased by only one fish, addition of more than four fish to the observation tank would have made it difficult to differentiate between individuals and species during video analysis.
| Participating species | Number of each species | Total fish in tank | Densitya | n |
|---|---|---|---|---|
| TRT, CTT, RWT | 1 | 3 | Medium | 30 |
| TRT, CTT | 2 | 4 | High | 6 |
| TRT, CTT | 1 | 2 | Low | 6 |
| TRT, RWT | 2 | 4 | High | 6 |
| TRT, RWT | 1 | 2 | Low | 6 |
| RWT, CTT | 2 | 4 | High | 6 |
| RWT, CTT | 1 | 2 | Low | 6 |
- a Densities, in fish/m3, were low = 4.9, medium = 7.3, and high = 9.8.
Data analysis.
We tested for statistical differences among species-specific aggression with an ANOVA model using program R (R Development Core Team 2008). We used the untransformed counts of the total number of aggressive initiations by each species in each trial, and conducted post-ANOVA comparisons between species with the Tukey honestly significant difference (HSD) test. To test the effect of individual fish we added a factorial variable for tag ID to our ANOVA model. We used paired t-tests to analyze the data for differences in encounter type (chase or bite–contact) and encounters prefeeding and postfeeding. A paired t-test was also used for each species to determine if it attacked one species more than another (e.g., if Rainbow Trout attacked Cutthroat Trout more than tiger trout).
In order to increase the number of observations, we also incorporated pairwise trials into our analysis by normalizing aggressive initiations. We did so by dividing the total number of aggressive bouts initiated by a fish in a trial by the number of targets to which the fish could initiate aggressive behavior. For example, in the three-species trials any fish had two other fish available to which it could initiate aggression. Therefore, we divided the aggressive initiations of each fish by two. We once again used ANOVA to test for differences in aggression between species and to test for differences in aggression between densities. We used a two-sample t-test for each species to determine if absence of a specific species affected the aggression exhibited by other species (e.g., if absence of Rainbow Trout increased or decreased the aggression of Cutthroat Trout). Lastly, we tested the results from our supplementary, high-density, pairwise trials for intraspecific aggression with a t-test, testing whether a species attacked conspecific fish more than interspecific fish.
RESULTS
We observed significant differences in aggressive behavior between the species during the 30 three-species trials (F87 = 15.45, P < 0.001; Table 1). Rainbow Trout initiated 24.8times the bouts per trial of Cutthroat Trout (P < 0.001) and 16.5 times the bouts per trial of tiger trout (P < 0.001), while there was no significant difference in aggression between Cutthroat Trout and tiger trout (P = 0.952). No species was observed to be more aggressive before or after feeding (Rainbow Trout: t29 = 1.062, P = 0.279; Cutthroat Trout: (t29 = 0.133, P = 0.895; tiger trout: t29 = 0.514, P = 0.611; Figure 1). Rainbow Trout and tiger trout did not show more aggression to one species over another (Rainbow Trout: t29 = 0.65, P = 0.521; tiger trout: t29 = 1.437, P = 0.162), while Cutthroat Trout exhibited a marginally significant (t29 = 1.964, P = 0.059) preference to attack tiger trout, initiating an average of 0.57 aggressive bouts toward tiger trout per trial as opposed to 0.27 aggressive bouts per trial toward Rainbow Trout. The addition of a dummy variable to our model to represent individual fish confirmed that some individuals did indeed exhibit more aggressive behavior than others (P < 0.001). Of these individuals, seven were Rainbow Trout, and one was a Cutthroat Trout. However, the model accounting for individual fish ranked much lower than the model considering only species (ΔAICc = 127.7; Burnham and Anderson 2002).
Mean aggressive initiations by type (chase or bite–contact), prefeeding and postfeeding, and at tested densities for tiger trout, Cutthroat Trout, and Rainbow Trout in experimental trials. Aggressive initiations are normalized by dividing by the number of targets available for each fish to attack at each density. Error bars are ±1SE from the mean. Densities, in fish/m3, were low = 4.9, medium = 7.3, and high = 9.8.
Results were also significant (F163 = 21.84, P < 0.001) for three-species and pairwise trials (n = 66; Table 1) Rainbow Trout initiated 10.2 times the aggressive interactions per trial of tiger trout (P < 0.001) and 24.8 times those of Cutthroat Trout (P < 0.001), while no significant difference was observed between tiger trout and Cutthroat Trout (P = 0.753). At the densities we explored, our analysis did not support a hypothesis of differences in agonistic behavior at different densities (F163 = 1.185, P < 0.308). No species exhibited a significantly higher rate of aggressive behavior toward conspecific fish than interspecific fish (Rainbow Trout: t11 = 1.572, P = 0.144; Cutthroat Trout: t11 = 1.395, P = 0.191; tiger trout: t11 = 0.8, P = 0.442). We also did not detect significant differences in the aggression of any species in the absence of a specific species (Rainbow Trout: t11 = 0.695, P = 0.519; Cutthroat Trout: t11 = 0.136, P = 0.893; tiger trout: t11 = 0.281, P = 0.782). We observed no physical damage to any fish as a result of the behavioral trials.
DISCUSSION
Our study clearly demonstrates that, in an experimental, laboratory setting, Rainbow Trout are more aggressive than Cutthroat Trout and tiger trout. Our results are further supported by and Nilsson and Northcote (1981), who documented a competitive advantage of Rainbow Trout over Cutthroat Trout in aquaria, Rainbow Trout exhibiting more aggression and capturing more prey items than Cutthroat Trout. In contrast, however, the low degree of aggression exhibited by tiger trout was unexpected.
A high degree of aggression in Brook Trout and Brown Trout is widely recognized as contributing to their competitive superiority over many other species (Newman 1956; Townsend 1996), particularly Cutthroat Trout (McHugh and Budy 2005). However, in our study, agonistic behavior by tiger trout, which are thought to be even more aggressive than their parent species (McClane 1974; Scheerer et al. 1987), did not differ from Cutthroat Trout, which displayed the least aggression. We did not know the sex of the fish, so it is possible that differences in sex ratios between the species could account for differences in aggression. Male Brown Trout and Rainbow Trout have been documented as being more aggressive than their female counterparts (Johnsson and Akerman 1998: Johnsson et al. 2001). Another factor to consider is the potential for reduced aggression of triploid fish (Benfey 1999). In our study tiger trout were triploid, while Rainbow Trout and Cutthroat Trout were diploid. Benfey's (1999) review suggests decreased aggression in triploids from sensory and nervous system changes or reduced levels of androgens, but both Wagner et al. (2006) and Budy et al. (2012) observed no difference in aggression between diploid Rainbow Trout and triploid Brook Trout. It is also important to note that aggression rates observed in the laboratory are generally higher than those found in nature because subordinate fish may not be able to escape aggressive displays while enclosed in a tank (Chiszar and Drake 1975). We may not have observed differences in intraspecific aggression or density treatments due to the small sample size of pairwise trials (Table 1), the range of densities tested (e.g., the densities may not have differed enough), or a combination of both.
In contrast to some studies, the introduction of food did not prompt changes in agonistic behavior (Newman 1956; Symons 1968). Although Rainbow Trout fed more vigorously than Cutthroat Trout and tiger trout (both species rarely fed during video observations), when we introduced food into the observation tank, the degree of aggression expressed during and after feeding did not increase (Budy et al. 2012). Similar behavior was noted outside of behavioral observations; Rainbow Trout began feeding voraciously the instant commercial trout feed was added to their tank, but Cutthroat Trout and tiger trout waited for food to sink and consumed it at a more leisurely rate than Rainbow Trout, often leaving a few uneaten pellets at the bottom of the tank. Hatchery feeding conditions may be more advantageous to Rainbow Trout than to Cutthroat Trout or tiger trout, which did not compete for commercial feed with the same vigor as Rainbow Trout. Introduction of a more natural food source may have influenced the feeding behavior of Cutthroat Trout and tiger trout and, thus, their interactions with other species. Nonetheless, if the feeding habits observed in the laboratory correspond to similar habits in a natural setting, increased risk-taking during feeding may reduce the overall survival of Rainbow Trout (Biro et al. 2004, 2006; Biro and Post 2008).
Though we suspect a study examining these three species in natural environments could yield different results, controlled laboratory experiments are useful for identifying species interactions because species combinations can be tightly manipulated and behavior can be closely observed. From our results we highlight some important management implications for novel reservoir fish communities. Although we determined Rainbow Trout to be the most aggressive species in this study, their survival and growth are lower than tiger trout and Cutthroat Trout when stocked sympatrically in many reservoirs in Utah (Winters 2014; Oplinger, unpublished data). In natural environments excessive energy expenditure during aggressive behavior often inhibits trout growth and survival (Nilsson and Northcote 1981), especially for introduced hatchery fish (Deverill et al. 1999). Therefore, the observed low growth and survival of Rainbow Trout observed in Scofield Reservoir may result, at least partially, from their high degree of aggressive behavior. In hatcheries, bold, aggressive behavior allows for higher growth rates in Rainbow Trout, but in natural settings aggressive behavior may be less advantageous. For example, aggression is an important behavior for obtaining and defending a territory, but it simultaneously greatly increases the risk of an individual to predation and decreases the likelihood of survival for a species (Berejikian 1995; Deverill et al. 1999; Biro et al. 2004, 2006). Thus, with larger predaceous fish already established in a reservoir and considering excess energy expenditures associated with this trait, agonistic behavior by Rainbow Trout could be reducing their survival relative to sympatric species. These are important considerations in designing and maintaining a stocking program.
ACKNOWLEDGMENTS
Funding and support of this project was provided by Utah Department of Natural Resources, Division of Wildlife Resources (UDWR; Federal Sport Fish Restoration, Grant F-134-R, Project 1) and U.S. Geological Survey, Utah Cooperative Fish and Wildlife Research Unit, in-kind. Paul Birdsey, Justin Hart, and Craig Walker at UDWR provided project support and financial administration. We would like to thank Gary P. Thiede for logistical help with experiments, graphical support, and for reviewing previous drafts of this manuscript. We would like to thank Lisa Winters for her efforts to the greater project overall and with earlier development of this study. We would like to thank Carl Saunders, Brett Roper, and Susan Durham for their help with analyses. Eric Wagner reviewed and provided constructive criticism on a previous draft of this manuscript. This research was conducted under State of Utah COR number 1COLL8712 and IACUC protocol 1544. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
