Factors Influencing Bluegill Growth in Small Kansas Impoundments

Bluegill Lepomis macrochirus is a small-bodied panfish that supports fisheries throughout much of the United States (Spotte 2007). As such, information pertaining to physiology, life-history characteristics, biology, and management of the species is abundant in fisheries literature (Werner and Hall 1988; Tomcko and Pierce 2005; Spotte 2007; Bevil and Weber 2018). Effective management of Bluegill can be complex because the species is often managed both as forage for larger piscivorous fishes (e.g., Largemouth Bass Micropterus salmoides) as well as to support quality fisheries (Coble 1988; Trebitz et al. 1997; Reed and Parsons 1999; Rypel 2015). Additionally, the wide geographic distribution of Bluegill creates management complexities attributed to variation in abiotic influences and community structures across its range (Dettmers and Stein 1992; Garvey and Stein 1998; Aday et al. 2003a; Bevil and Weber 2018). Despite difficulties associated with consistent manipulation of Bluegill populations, directed management remains a common practice among fisheries-management agencies (Beard et al. 1997; Reed and Parsons 1999; Schneider and Lockwood 2002; Edison et al. 2006).

Bluegill are relatively short lived, rarely living past 9 y of age, and annual mortality typically exceeds 50% (Spotte 2007). However, spawning behaviors that include breeding synchrony and colonial nesting generally preclude Bluegill populations from being limited by natural reproduction bottlenecks (Gross and MacMillan 1981; Spotte 2007). Bluegill populations that require management are typically characterized by high densities of fast-maturing, slow-growing conspecifics (i.e., stunting; Aday and Graeb 2012). Bluegill also exhibit size-based social interactions that can result in shorter length at maturation in populations that lack large, territorial males (Jennings et al. 1997; Aday et al. 2002, 2006). Given these documented life-history attributes, Bluegill management (e.g., restrictive creel or length limits) is often directed toward increasing size of individuals in the population to promote energy allocation into somatic growth rather than gonadal development.

In addition to intrapopulation dynamics, Bluegill population measures (e.g., size structure, body condition, growth, relative abundance) can also be influenced by abiotic and biotic factors. Examples of abiotic variables that may influence Bluegill populations include watershed land use (Bevil and Weber 2018), water quality (Tomcko and Pierce 2001), and percent littoral area (Claramunt and Wahl 2000; Shoup et al. 2007). Bottom-up biotic control of Bluegill populations has been attributed to zooplankton abundance (Welker et al. 1994; Bevil and Weber 2018) and nutrient availability (Boyd and Sowles 1978; Aday et al. 2006; Bevil and Weber 2018). Top-down biotic structuring of Bluegill populations has been linked to several piscivorous fish species, but most frequently Largemouth Bass (Swingle 1950; Doxtater 1967; Gabelhouse 1987; Guy and Willis 1990). Angler harvest also can alter size structure within a Bluegill population (Coble 1988; Beard and Essington 2000; Rypel 2015). Despite numerous studies that have assessed variation in Bluegill populations, few have addressed how abiotic and biotic factors influence Bluegill growth.

In areas where Bluegill are managed for recreational angling, information about growth patterns is important to determine if management actions (e.g., harvest regulations) are needed to improve angler experience. In most instances, growth is important because it measures the ability of individuals within a population to reach lengths desired by anglers. However, management-based goals for fish lengths can also be established to optimize reproduction or provide forage for other organisms (Noble 1981; Birkeland and Dayton 2005; Isermann and Paukert 2010).

Kansas has historically supported popular Bluegill fisheries in small impoundments with numerous fish > 203 mm (preferred length; Gabelhouse 1984) being present in both fish and creel samples. However, recent fish sampling and creel data suggest that fish > 203 mm have decreased in abundance (S. Steffen, Kansas Department of Wildlife, Parks, and Tourism, personal communication). These declines have prompted questions pertaining to growth of Bluegill at a statewide scale with a focus on identifying factors that contribute to variability in growth rates. To address these questions, we identified two objectives: 1) describe growth of Bluegill in small Kansas impoundments, and 2) identify limnological and fish catch factors that describe interpopulation variability in Bluegill growth.

We sampled Bluegill from 24 small Kansas impoundments that varied from 12 to 148 ha (mean = 51 ha; Figure 1). These impoundments were managed primarily for recreational angling, although some provided municipal water supply or flood control. Fish communities in all impoundments were dominated by Largemouth Bass, White Crappie Pomoxis annularis or Black Crappie Pomoxis nigromaculatus (or both), Channel Catfish Ictalurus punctatus, and Bluegill. In some instances, saugeye Sander vitreus × Sander canadensis, Flathead Catfish Pylodictis olivaris, Gizzard Shad Dorosoma cepedianum, and sunfishes Lepomis spp. other than Bluegill existed in variable numbers.

We collected Bluegill using daytime shoreline electrofishing in autumn 2017. Samples at each impoundment consisted of between two and seven randomly selected 10-min stations depending on impoundment size (i.e., larger impoundments received more effort). We measured all netted Bluegill for total length. We made an attempt to collect sagittal otoliths from five individuals per centimeter group to represent the population; however, instances of less than and more than five individuals per centimeter group were common depending on extraneous factors (e.g., fish availability, different collectors). We air dried, set in epoxy, and transversely sectioned collected otoliths through the nucleus (∼ 0.6-mm section) using a low-speed saw (Long and Grabowski 2017).

We collected limnological and fish catch variables from routine monitoring conducted by state agencies in Kansas (Table 1). The Kansas Department of Health and Environment summarized total nitrogen and total phosphorus (TN and TP; mg/L) as the mean value from all samples collected between 2012 and 2017 as part of the Lake and Wetland Monitoring Program (KDHE 2017). Latitude (lat; decimal degrees) represented a temperature gradient across impoundments and we collected data in concert with maximum depth (max.depth; m) from bathymetric surveys that we conducted. We sampled Bluegill in autumn 2017 with 91 × 183 cm fyke nets fished overnight (Miranda and Boxrucker 2009; Pope et al. 2009). We estimated relative abundance by catch per effort (blg.cpe) and calculated it as the mean number of Bluegill sampled per net-night. We estimated size structure by proportional size distribution of 178-mm fish (blg.psd178) and calculated it by impoundment as the proportion of sampled Bluegill ≥ 80 mm that were also ≥ 178 mm. We sampled Gizzard Shad in autumn 2017 using American Fisheries Society-standard sinking gill nets (24.8 m long, 1.8 m tall, variable mesh size; Miranda and Boxrucker 2009) placed at randomly selected locations with water depths between 1.8 and 5.0 m. We recorded catch rate (gzs.cpe) by impoundment as the mean number of fish captured per net night. We sampled Largemouth Bass in spring 2017 with 10-min shoreline electrofishing stations at randomly selected locations within each impoundment (Miranda and Boxrucker 2009; Pope et al. 2009). We measured stock catch rate as the number of stock-length fish (≥ 200 mm; Gabelhouse 1984) sampled per hour.

We placed Bluegill sagittal otolith sections on a dark background and submersed them in mineral oil for reading. We then magnified them ×20, displayed them on a monitor with μManager software (Edelstein et al. 2014), and assigned them a consensus age by three experienced readers (Koch et al. 2019; Data S1, Supplemental Material). We fit growth models using the nls() function in R with the Gauss–Newton algorithm (R Core Team 2018). We generated starting values for parameter estimates with vbStarts() in the FSA package (Ogle 2016; Ogle et al. 2018). Last, we reported growth as the predicted age at which individuals in the population reached 152 mm (quality length [Gabelhouse 1984]; t₁₅₂) using the Ogle–Isermann parameterization (L_∞, K, and t₁₅₂) of the von Bertalanffy growth model (Ogle and Isermann 2017). We estimated variation in growth models using a nonparametric bootstrap procedure with 1,000 iterations (nlsBoot() function in nlstools package in R [Baty et al. 2015; R Core Team 2018]). We estimated confidence intervals from converged iterations for all model parameters, but we present only those for t₁₅₂ herein.

We took two steps to stabilize variance and standardize limnologic and fish catch values. First, we log₁₀ transformed variables, excluding gzs.cpe and blg.psd178, that were log₁₀ (x + 1) transformed, to stabilize variances. Upon meeting the assumption of normality, we z-score standardized values to facilitate comparisons of individual variables (Grueber et al. 2011). Once transformed and standardized, we included variables in all-subsets normal linear regression models (N = 256) to examine relationships with Bluegill t₁₅₂ using the dredge() function from the MuMIn package in R (Barto 2018). We evaluated and ranked each subset model with Akaike's information criterion corrected for small sample size (AIC_c). We retained models with ΔAIC_c ≤ 2 as the confidence model set, and we rescaled Akaike weights from all models in the confidence set to sum to 1 (Burnham and Anderson 2002). We used these rescaled model weights to calculate weighted means and standard errors for parameter estimates (Symonds and Moussalli 2011). We calculated confidence intervals (95%) by multiplying 1.96 × standard error, then adding (upper 95% confidence interval) and subtracting (lower 95% confidence interval) the product from the weighted mean parameter estimate. When confidence intervals of a given parameter contained zero, the effect could be either positive, negative, or neutral, thus uninterpretable. When confidence intervals did not include zero, we could interpret the parameter as having either a negative or positive effect on the dependent variable (Grueber et al. 2011; Neely et al. 2016).

We assigned ages to 1,323 Bluegill from 24 impoundments in this study. Number of aged fish per population varied from 25 to 72 and the mean among populations was 55 (SD = 14). Ages ranged from 0 to 9 y, although over 95% of individuals were age 4 or younger (Figure 2). We collected age-0 individuals from 15 populations and collected individuals age 5 or older from 16 populations. We sampled age-2 and age-3 individuals from all populations. We identified a minimum of 3 and a maximum of 7 year classes in study populations, with a mean of 5.6 (SD = 1.1).

Bluegill populations in this study exhibited substantial variation in growth rate. The quickest-growing population measured in this study reached 152 mm at 1.6 y and the slowest-growing at 4.9 y. Confidence intervals and proportion of converged bootstrap simulations generally suggested precise estimates of t₁₅₂, although we observed some variation in slower-growing populations and those with fewer year classes present in the sample (Figure 3). When we included all aged fish in a single growth model, Ogle–Isermann parameters were: L_∞ = 228 mm, K = 0.25, and t₁₅₂ = 3.10 y (Figure 4).

When we evaluated all subset models to identify how catch and limnological factors influence Bluegill growth, we retained two with ΔAIC_c scores ≤ 2. Parameters included in those models were lat, blg.psd178, gzs.cpe, and st.lmb.cpe (Table 2). Parameters representing productivity (TN and TP), max.depth, and Bluegill density (blg.cpe) were not present in the confidence model set. The final averaged model explained 40% of variation in Bluegill growth rate observed in this study. Confidence intervals for three parameter estimates did not overlap zero, suggesting an interpretable effect on Bluegill growth: lat, blg.psd178, and st.lmb.cpe. Positive effects (e.g., increased values associated with slower Bluegill growth) were associated with lat and st.lmb.cpe, whereas a negative effect (e.g., increased values associated with more rapid Bluegill growth) was associated with blg.pld178 (Figure 5).

Bluegill growth in small Kansas impoundments was variable, but within bounds of other populations in the United States (Spotte 2007; Jackson et al. 2008; Brouder et al. 2009). Specifically, a randomly selected Bluegill in Kansas would be expected to reach 152 mm at 3.1 y. This growth rate is similar to the 75th percentile for North American Bluegill populations (143 mm at age 3 y; Brouder et al. 2009). When we considered individual populations, only 3 of 24 populations exhibited growth to 152 mm slower than the 50th percentile of North American populations (i.e., reached 152 mm after age 4 y; Brouder et al. 2009). Similarly, 3 of 24 populations exhibited more rapid growth to 152 mm than the 95th percentile of North American populations (i.e., reached 152 mm before age 2 y; Brouder et al. 2009). Generally, Bluegill populations in small Kansas impoundments exhibited more rapid growth at young ages (i.e., up to age 3 y) than other studied populations (Paukert et al. 2002; Spotte 2007; Sammons and Maceina 2009).

Limnological factors (e.g., productivity, temperature regime, water depth) are frequently linked to Bluegill population parameters including size structure, body condition, growth, and relative abundance (Spotte 2007). In the current study, we examined four limnological variables that have been frequently implicated as factors influencing Bluegill populations: TN, TP, latitude, and maximum water depth. Productivity of water is measured by TN, TP, chlorophyll a, or Secchi depth and has also been positively related to Bluegill growth in several studies (Tomcko and Pierce 2001; Hoxmeier et al. 2009). However, neither TN nor TP was related to Bluegill growth rate in the current study. One reason for this lack of association might be the prevalence of high productivity in small Kansas impoundments. Kansas waters are seldom oligotrophic, and most are eutrophic (Dodds et al. 2006). Additionally, median of both TN and TP among study systems exceeded values suggested by Dodds et al. (2006) to classify as eutrophic. These results suggest that although water productivity can influence Bluegill growth, it is likely not limiting in primarily eutrophic small Kansas impoundments. Growing degree days, often approximated by latitude or temperature, has been positively related to Bluegill growth, suggesting adherence to converse Bergmann's rule (Hoxmeier et al. 2009; Rypel 2014). In the current study latitude was the only limnologic variable present in the confidence model set, suggesting that geography of an impoundment significantly affected Bluegill growth rate, specifically, that Bluegill growth was more rapid at lower latitudes as observed in previous studies (Hoxmeier et al. 2009; Rypel 2014). The littoral nature of Bluegill suggests that water depth of a lake or impoundment can influence populations (Spotte 2007). In Minnesota lakes, maximum water depth was negatively related to Bluegill growth (Tomcko and Pierce 2001). Conversely, mean water depth was positively related to Bluegill growth in Iowa impoundments (Schultz et al. 2008). These conflicting results might reflect different processes occurring in natural lakes vs. impoundments, but generally support findings herein that water depth might not be a reliable predictor of Bluegill growth in all instances.

Bluegill populations occupy an intermediary trophic niche in most systems, and as such, can be influenced directly and indirectly by sympatric fish community composition and structure (Spotte 2007). Relative abundance of Bluegill, Gizzard Shad, and Largemouth Bass and size structure of Bluegill populations are among the most cited factors that influence Bluegill growth. Intrapopulation density-dependent effects on Bluegill growth have been identified in some instances (Tomcko and Pierce 2005) but not all (Hoxmeier et al. 2009), including the current study.

Gizzard Shad are frequently noted as being deleterious to Bluegill populations through both direct resource competition and indirect mechanisms (e.g., trophic cascade hypothesis; Dettmers and Stein 1992; Stein et al. 1995; Aday et al. 2003a; Spotte 2007; Oplinger et al. 2013). However, focused Gizzard Shad reductions do not always result in changes to Bluegill growth dynamics (Kim and DeVries 2000; Neely et al. 2018). Relative abundance of Gizzard Shad in the current study was present in the confidence model set to explain variation in Bluegill growth, but the effect was not significant, whereas Bluegill relative abundance was not represented in the confidence model set. The lack of explanatory power of these variables might suggest that they have greater effect on Bluegill growth dynamics when catalyzed by finer-scale interactions (e.g., macrophyte, zooplankton, and benthic invertebrate community composition and abundance) not explored in this study (Welker et al. 1994; Michaletz and Bonneau 2005).

Abundance and size structure of Largemouth Bass can also influence Bluegill growth (Swingle and Smith 1942; Swingle 1950; Aday and Graeb 2012). Absence or poor size structure of Largemouth Bass can result in less Bluegill predation, overabundance of small individuals, and slowed growth (Schramm and Willis 2012). The current study measured relative abundance of stock-length Largemouth Bass (≥ 200 mm) as a proxy for both relative abundance and size structure of the primary predator of Bluegill in these small impoundments. This variable was present in the confidence model set and was identified to have a significant effect on Bluegill growth. Specifically, when densities of stock-length Largemouth Bass increased, Bluegill took a longer time to reach 152 mm. These results contradict previous literature that suggests either a positive relationship between Bluegill growth and relative abundance of Largemouth Bass (Guy and Willis 1990) or no effect (Tomcko and Pierce 2005). The mechanism driving this relationship in the current study was not identified, and additional research is warranted to identify the influence of Largemouth Bass on Bluegill populations in Kansas.

Bluegill males have unique reproductive ecology that directs energy to either somatic growth or gonadal development depending on social structure of the population (Jennings et al. 1997; Aday et al. 2003a, 2003b). In populations with large individuals (e.g., low mortality), male Bluegill will frequently delay maturation in favor of somatic growth. In populations devoid of large individuals (e.g., high mortality), male Bluegill mature earlier at the expense of somatic growth (Jennings et al. 1997; Aday et al. 2003b). In the current study, proportion of Bluegill in the population longer than 178 mm had a significant effect on growth rate. This result supports previously stated theories that intrapopulation size structure, specifically proportion of large individuals, can influence population-level somatic growth.

Although we identified several variables in the current study as influential on Bluegill growth, the final model only explained 40% of the variation. Two notable limitations with the current study included a lack of age-at-maturity and angler-harvest data. Addition of age-at-maturity data would allow testing of the hypothesis that Bluegill populations with more rapid somatic growth rates mature at an older age than individuals from slow-growing populations. Similarly, angler-harvest data would allow examination into the role of angler-induced mortality on population-level growth (Beard and Essington 2000; Audzijonyte et al. 2013). Given the unexplained variability in Bluegill growth in both the current study and previous studies, it is important to acknowledge complexities of aquatic systems when developing management plans for Bluegill populations in Kansas and likely in other small midwestern U.S. impoundments.

Results from this study primarily indicate that growth of Bluegill across Kansas is highly variable. Further, limnological and fish community characteristics that have been implicated in influencing Bluegill growth in other systems were variably related to growth rates in small Kansas impoundments (Shoup et al. 2007; Oplinger et al. 2013). A possible exception was evidence that Bluegill reached 152 mm more rapidly in populations with larger individuals present. However, Bluegill growth was likely influenced by complex interactions of abiotic and biotic factors both included and omitted from the current study (Spotte 2007). Unexplained variability in Bluegill growth in both the current study and previous studies highlight complexities of developing management plans for Bluegill populations in Kansas and likely in other small midwestern U.S. impoundments. Further, additional work is warranted to directly measure how mortality, specifically angler harvest, and age at maturity influence population-level growth of Bluegill.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Data S1. Supplemental data for development of von Bertalanffy growth models is in the tab labeled “agedat.” This contains impoundment (impd), total length at capture (TL), species (spp), and age at capture (age). Supplemental data for construction of the model to examine variation in Bluegill Lepomis macrochirus growth is in the tab labeled “moddat.” This file contains impoundment (impd), age at 152 mm as modeled by the Ogle–Isermann parameterization of the von Bertalanffy growth function (years; t₁₅₂), latitude of the impoundment (decimal degrees; lat), mean total nitrogen (mg/L; TN) of all measurements collected from 2012 to 2017, mean total phosphorus (mg/L; TP) of all measurements collected from 2012 to 2017, maximum depth of the impoundment (m; max.depth), catch per effort of Bluegill from the most recent autumn trap net survey (fish/net-night; blg.cpe), percentage of Bluegill ≥ 80 mm that were also ≥ 178 mm from the most recent autumn trap net survey (blg.psd178), catch per hour of stock-length (≥ 200 mm) Largemouth Bass Micropterus salmoides from the most recent spring electrofishing survey (fish/h; lmb.st.cpe), and catch per effort of Gizzard Shad Dorosoma cepedianum from the most recent autumn gill net survey (fish/net-night; gzs.cpe).

Found at DOI: https://doi.org/10.3996/082019-JFWM-065.S1 (37 KB XLSX).

Funding for this project was provided in part by Sport Fish Restoration Grant F-22-R administered by Kansas Department of Wildlife, Parks, and Tourism. We thank many Kansas Department of Wildlife, Parks, and Tourism and Kansas State University employees for their assistance collecting data for this study. L. Kowalewski, S. Steffen, D. Ogle, M. Colvin, and several anonymous reviewers provided helpful comments on previous versions of this manuscript. There is no conflict of interest declared in this article.

Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.