A Strategy for Increasing Gill-Net Catch Rates and Minimizing Sampling Mortality of Alligator Gars
Abstract
Management of the Alligator Gar Atractosteus spatula has been hampered by the lack of sampling methods that can efficiently reveal population characteristics or minimize incidental sampling mortality. In a 3-year evaluation of the Alligator Gar population in Choke Canyon Reservoir, Texas, we documented changes in sampling methodology that produced an eightfold increase in multifilament gill-net catch rates (from 0.66 to 5.10 fish/h). The increased sampling efficiency was attributed to the development of an aggressive-predator sampling strategy over the study period; our highest catch rates were achieved by (1) observing surfacing fish before net deployment and (2) relocating to a new sampling site when catch rates were less than 1 fish/h. This aggressive strategy resulted in shorter net soak times, which reduced incidental net mortalities; nets soaking less than 30 min produced zero mortalities. Our results suggest that the aggressive-predator strategy is suitable for the majority of Alligator Gar sampling objectives (e.g., collection for age and growth and mark–recapture), which will improve the ability to manage Alligator Gar populations.
Received August 18, 2014; accepted February 4, 2015
Historically, the Alligator Gar Atractosteus spatula was among North America's most disliked fishes, and efforts to eradicate them were undertaken for a century or more (Scarnecchia 1992). The success of these efforts in some systems (Harlan and Speaker 1951; Robinson and Buchanan 1988), coupled with widespread habitat destruction, resulted in decreased abundance or extirpation of the species throughout much of its range (e.g., Gilbert 1992; Etnier and Starnes 1993; Ross et al. 2001). However, attitudes and opinions have changed. The large body size, prominent teeth, and predatory nature of the Alligator Gar are now considered desirable attributes (Scarnecchia 1992). Increased interest in this species among anglers (Ross et al. 2001), as well as support for preserving the biodiversity in aquatic systems, has prompted fisheries managers to actively manage populations.
Unfortunately, efforts to manage Alligator Gar stocks have been hampered by the inability to efficiently sample the species. Inefficient sampling gears, coupled with low population density in many systems (Layher et al. 2008; Inebnit 2009), typically result in catch rates that are too low for precise statistical evaluation of the many population characteristics necessary for management. For example, gill- and trammel-net catch rates are typically less than 0.5 fish/h (Seidensticker and Ott 1989; O'Connell et al. 2007; Brinkman 2008; Layher et al. 2008), and overnight jug lines average less than 1 Alligator Gar for every 10 lines set (Seidensticker and Ott 1989). Other common gears, such as electrofishing, trap nets, and fyke nets, are ineffective for sampling this species due to its large body size. Despite their inefficiencies, gill nets appear to produce higher catch rates than any other gear and thus are used most often (Ferrara 2001; Robertson 2007; Brinkman 2008; Layher et al. 2008). However, before fisheries managers can develop comprehensive population assessment and management programs, gill-net sampling efficiency must be increased.
A predator-based sampling strategy (Vokoun and Rabeni 1999) may be the most viable option for improving gill-net sampling efficiency. This technique uses a nonrandom, cluster-based strategy by targeting areas known to contain large densities of fish (Vokoun and Rabeni 1999) and can be most useful for providing high catch rates (Vokoun and Rabeni 1999; Daugherty and Sutton 2005; Ford et al. 2011) when sampling rare species or populations with low abundance (Philippi 2005; Brinkman 2008). For example, Brinkman (2008) increased gill-net catch rates by using telemetry techniques to locate congregations of Alligator Gars before sampling. Buckmeier et al. (2013) relied on visual observation before using jug lines, rod and reel, and gill nets to increase capture rates of Alligator Gars in the Trinity River, Texas. Developing and implementing this type of sampling strategy may provide a reasonable means of collecting the quantity of data needed to make informed management decisions.
In addition to improving catch rates, minimizing or preventing the incidental sampling mortality of Alligator Gars captured in gill nets is critical. Incidental sampling mortality caused by drowning is known to occur in gill nets (Layher et al. 2008). This is problematic considering that many populations are experiencing declining abundance or are at risk of extirpation (e.g., Gilbert 1992; Etnier and Starnes 1993; Ross et al. 2001; Jelks et al. 2008). As a result, fisheries scientists are often reluctant to sample for fear of contributing to the decline of Alligator Gar populations (Layher et al. 2008). Use of gill nets in a more active-predator sampling strategy may reduce incidental sampling mortality if captured fish are more rapidly removed from the net, as is more likely during active sampling. In this study, we describe the development of an aggressive-predator sampling strategy that resulted in increased catch rates and reduced the incidental gill-net mortality of Alligator Gars in Choke Canyon Reservoir, Texas. Our results will aid fisheries professionals in improving sampling efficiency for Alligator Gar populations throughout their range.
METHODS
Study site.
Choke Canyon Reservoir is a 10,517-ha impoundment of the Frio River in McMullen and Live Oak counties, Texas. The reservoir was created in 1982 for water storage and recreation. Maximum water depth is about 29 m, and water levels vary 3–4.5 m annually (Texas Parks and Wildlife Department, unpublished data). The population of adult Alligator Gars (≥1,100 mm TL) is about 5,400 individuals (95% confidence interval = 3,200–9,200; Binion et al., in press).
Alligator Gar sampling.
Alligator Gars were collected with multifilament gill nets from mid-April through early June 2011 through 2013 (water temperature range, 23–33°C), as part of an Alligator Gar mark–recapture study in the reservoir. Nets were 61 m long and 3 m deep, and were constructed of a 19-mm foam-core float line, #21 black-twine mesh, and a 9-kg lead line. Typically, one to three individual nets with bar-mesh sizes ranging from 89 to 128 mm were deployed within each sampling site (about 1 km2). The sampling sites were primarily located in the upper half of the reservoir in both littoral (e.g., coves and shorelines) and limnetic (e.g., open water flats and main channels) areas in water 2–7 m deep. Occasionally, one or two additional nets with bar-mesh sizes of 140 and 152 mm were set when catch rates were low (e.g., <1 fish/h). Each net was fished so that the float line remained on the water surface, with four to five buoy-style PVC floats attached. Most captured gars disturbed or submersed the float line or floats, facilitating rapid removal and processing of the fish. Upon indication of capture, the net was immediately retrieved and fish were removed. Captured fish were measured for TL, transported approximately 200 m away from the sampling site, and released. The incidental mortality of captured fish was also recorded for each net. Evidence of capture was not always apparent, however; therefore, some nets were allowed to continue fishing despite capturing a fish. The maximum soak time (the interval between deployment and retrieval) for nets without visible evidence of captured fish ranged from 1 to 10 h, depending on sampling strategy.
A primary objective of the concurrent mark–recapture study was to maximize the number of Alligator Gars caught and marked while minimizing mortality. Therefore, our techniques during that study were adapted annually as sampling experience increased. This affected the relative amounts of time spent actively sampling at a site and the time spent selecting sites (i.e., searching for sampling sites and determining where to sample). During 2011, we used a passive-predator sampling strategy similar to the methods used in previous studies (e.g., Seidensticker and Ott 1989; Ferrara 2001; Sakaris et al. 2003; Brinkman 2008; Layher et al. 2008). Sample sites were primarily selected in areas thought to contain high densities of Alligator Gars based on habitat characteristics or previous observations of fish in the area, and the locations were sampled for an extended period of time (e.g., up to 10 h of either continuous soak time/net or continued redeployment after catching a fish), regardless of catch rate. This strategy maximized soak time and minimized the time spent selecting sites. In 2012, we attempted to increase catch rates by employing a more active sampling strategy (hereafter, “intermediate-predator sampling”). Sampling sites were selected based on the same criteria as in 2011; however, after 1–2 h (i.e., nets were allowed to soak up to two continuous hours), we abandoned the original sampling sites and selected new ones when catch rates were less than 1 fish/h. This approach decreased soak time yet increased the amount of time spent selecting sites and the number of sites sampled per day. In 2013, we again modified our sampling procedures and adopted an aggressive sampling strategy, (hereafter, “aggressive-predator sampling”). Sample sites were selected by traversing the reservoir until surfacing Alligator Gars (due to their aerial breathing) were observed; these locations were immediately sampled. When catch rates at these sites decreased to less than 1 fish/h, we abandoned the site after 1 h of continuous fishing and resumed searching for other sites occupied by Alligator Gars. When we were unable to observe surfacing activity throughout the reservoir, we returned to areas where we had previously captured Alligator Gars. If catch rates were less than 1 fish/h, we discontinued sampling for the day or waited until surfacing resumed.
Data analyses.
We quantified the effects of the adaptations to our sampling strategy by examining the relationships between Alligator Gar catch rate, soak time, and incidental gill-net mortality among the three sampling strategies. To account for differing selectivity among mesh sizes, we randomly selected samples within each strategy to balance mesh-specific sample sizes across sampling strategies. Using the mean of the ratios approach (Pollock et al. 1994), mean CPUE (fish/h) was examined across sampling strategies. The catch rate data were not normally distributed; therefore, CPUE was rank-transformed (Daniel 1978) and differences were assessed with a one-way ANOVA. For significant ANOVA effects (P < 0.05), subsequent pairwise comparisons of the means were made with a Tukey's honestly significantly different test. We then used a Spearman's rank correlation to evaluate the relationship of mean CPUE to mean gill-net soak time.
The data used for the mortality analysis were not subsampled because a balanced sampling design was not needed for this objective; therefore, incidental gill-net mortality was expressed as a percentage of the total number of dead fish collected per strategy. We also examined the distribution of mortalities across four soak times (i.e., <30, 30–59, 60–89, and >90 min) to further identify potential methods for minimizing mortality. Differences in gill-net mortality were assessed with logistic regression (Proc GENMOD; SAS Institute 2012). Because no deaths were observed in the <30-min duration, we added one mortality (post hoc) to achieve model convergence. When significant effects were detected (P < 0.05), pairwise comparisons were made with a Tukey–Kramer test.
RESULTS
In the subsampled data that we used for the catch rate analysis, 389 Alligator Gars were sampled in 414 gill-net sets (N = 138 per sampling strategy) comprising 565 cumulative hours of gill-net effort. The mean TL of fish was 1,480 mm (range = 678–2,275 mm). We observed an eightfold increase in mean catch rate by switching from the passive-predator sampling strategy (0.66 ± 0.11 fish/h [mean ± SE]) to the aggressive-predator sampling strategy (5.10 ± 1.06 fish/h; F2, 411 = 16.01, P < 0.001). The improvement in catch rate was negatively correlated to gill-net soak time (Spearman's rank correlation coefficient = −0.47, P < 0.01); soak time decreased from 1.89 h/set to 0.7 h/set (Figure 1). The incidental mortality rate (based on the 665 fish examined) was not statistically different among sampling strategies (1.4 ± 0.8% for aggressive-predator sampling, 1.8 ± 1.0% for passive-predator sampling, and 3.2 ± 1.0% for intermediate-predator sampling); however, mortality was lowest for the shortest gill-net soak time (<30 min; all P < 0.05; Figure 2).
Relationship between gill-net soak time and CPUE for Alligator Gars across three predator-sampling strategies (see text; N = 138 gill-net sets per sampling strategy) for nets set during April–June 2011–2013 in Choke Canyon Reservoir, Texas. Error bars denote the empirically derived ranges from the 5th to the 95th percentiles of the distributions around the means.
Frequency of incidental gill-net mortality of Alligator Gars based on soak time for nets set during April–June 2011–2013 in Choke Canyon Reservoir. The numbers in parentheses are the total numbers of fish collected.
DISCUSSION
The increasing catch rates of Alligator Gars during our study were only partially attributable to reduced soak times. The majority resulted from our selection of sites that we believed to be especially promising and our leaving chosen sites when the catches were low. For passive gears such as gill nets to be effective, they must be set in areas where target fish are located during sampling, and individuals at the site must be active enough to encounter the nets (Hubert et al. 2012). Therefore, selecting sampling sites based on knowledge of the presence of active fish would presumably increase sampling efficiency. Although such information would be unobtainable for most fish species, the surfacing behavior exhibited by Alligator Gars owing to their aerial breathing provided us the opportunity to locate active fish.
The passive-predator sampling strategy employed initially produced low catch rates similar to those in previous studies using such sampling (<1 fish/h; Brinkman 2008; Layher et al. 2008). This strategy relied on the assumption that past visual observations or knowledge of habitat preferences would enable us to predict the current locations of Alligator Gars. Unfortunately, such assumptions often did not lead to accurate predictions of the presence, absence, or activity level of fish during the time of sampling. This “ambush” predator strategy (Wootton 1998) often resulted in sampling suboptimal sites, which led to greater soak times, fewer sites being sampled per day, and lower catch rates. In contrast, the aggressive-predator sampling strategy resulted in greater amounts of time searching for and sampling sites occupied by actively surfacing fish. The catches at such sites were typically immediate and numerous (e.g., up to 14 fish in a net within 8 min of deployment). When sampling was no longer efficient, nets were lifted and the site was abandoned, resulting in shorter soak times. Our results suggest that aggressive-predator sampling greatly improved our sampling efficiency of Alligator Gars. Further knowledge of their spatial distribution (e.g., habitat use) and activity (daily and annual movement patterns, surfacing behavior, etc.) may help refine optimal locations or periods for sampling.
Because Alligator Gars are vulnerable to extirpation across their range (Jelks et al. 2008) and have a low intrinsic rate of population increase (Ferrara 2001), it is important that sampling techniques minimize sampling mortality. We found evidence that soaking nets for less than 30 min reduces or eliminates incidental mortality. Although we observed no statistical differences in incidental mortality rates among sampling strategies (likely due to there being few deaths throughout), our aggressive-predator sampling strategy inherently reduces average soak time; therefore, a lower probability of incidental mortality is likely an added benefit of this strategy. Although our study was not specifically designed to evaluate the effects of gill-net soak time on mortality or the underlying causal mechanisms for it, a plausible explanation for the observed mortality may be the increased likelihood of atmospheric oxygen deprivation for Alligator Gars submerged in gill nets during longer net sets (i.e., >30 min). Gars are typically considered facultative air breathers (Moyle and Cech 2000) and have been shown to receive up to 80% of their oxygen from the atmosphere (Rahn et al. 1971). Metabolic oxygen demands are affected by a number of factors, including activity level, environmental temperature, dissolved oxygen concentration, and physiological stress. Active fish require more oxygen than those at rest, and oxygen demands increase with increasing temperature and physiological stress (Davis and Schreck 1997; Moyle and Cech 2000). Alligator Gar mortality during our sampling (N = 15; TL = 1,100–1,940 mm [mean = 1,509 mm]) occurred at water temperatures ranging from 24°C to 30°C. These relatively high water temperatures, coupled with the physiological stress associated with capture and a lack of atmospheric oxygen, may have resulted in greater metabolic oxygen demands than could be met by gill respiration alone, resulting in drowning. This, combined with our findings, suggests that nets set and retrieved in less than 30 min can minimize or eliminate the incidental net mortality of Alligator Gars.
We believe that the catch rate increases in our study are attributable to improvements in sampling strategies rather than to changes in the Alligator Gar population. Alligator Gars exhibit a periodic life history strategy and have a low intrinsic rate of population increase (Ferrara 2001), suggesting that short-term population changes are unlikely. Further supporting this, a post hoc analysis of mean TL among the three sampling strategies in Choke Canyon Reservoir indicates that the population experienced little change during our study (TL range = 1,542 mm for passive-predator sampling and 1,459 mm for intermediate- and aggressive-predator sampling; ANOVA of mean TL from nets that captured at least one fish: F2, 208 = 2.32, P = 0.10).
We recommend that biologists consider the aggressive-predator sampling strategy where practical. Although this strategy is most commonly used when sampling with active gears (e.g., electrofishing; Vokoun and Rabeni 1999; Daugherty and Sutton 2005; Ford et al. 2011), it can also be applied to passive gears such as gill nets if these gears are fished actively rather than passively. Our aggressive-predator sampling strategy produced catch rates 8 times greater than those of the passive-predator sampling strategy and over 10 times greater than those of any other common sampling strategy (Seidensticker and Ott 1989; Brinkman 2008; Layher et al. 2008). This suggests that aggressive-predator sampling is the most useful method for producing catch rates that are high enough for practical use. However, further improvements in overall sampling efficiency (e.g., efficiency measured in fish/d rather than fish/h) may still be possible. The aggressive-predator sampling strategy primarily relies on the ability to visually identify congregations of fish. While this strategy may be unsuitable for estimating relative abundance, there are some objectives for which it is appropriate (e.g., population estimation, survival analysis, the determination of mean length at age, von Bertalanffy growth parameter estimation, etc.), assuming that the sample of fish collected from a given location is representative of the overall population. The higher catch rates offered by this technique offer more opportunities for collecting information about Alligator Gars and thus may improve our ability to manage them.
ACKNOWLEDGMENTS
We thank the many staff from the Texas Parks and Wildlife Department–Inland Fisheries Division's Heart of the Hills Fisheries Science Center and District 1E for assistance with field data collection. Constructive comments provided by Nate Smith and Michael Baird greatly improved this manuscript. This study was supported by Federal Aid in Sport Fish Restoration grants F-231-R and F-221-M to the Texas Parks and Wildlife Department, Inland Fisheries Division.
