Regular issues
Monographs
Special Issues



Southeastern Naturalist
    SENA Home
    Range and Scope
    Board of Editors
    Staff
    Editorial Workflow
    Publication Charges
    Subscriptions

Other EH Journals
    Northeastern Naturalist
    Caribbean Naturalist
    Urban Naturalist
    Eastern Paleontologist
    Eastern Biologist
    Journal of the North Atlantic

EH Natural History Home

Influence of Lake Surface Area and Total Phosphorus on Annual Bluegill Growth in Small Impoundments of Central Georgia
Aaron P. Sundmark and Cecil A. Jennings

Southeastern Naturalist, Volume 16, Issue 4 (2017): 546–566

Full-text pdf (Accessible only to subscribers.To subscribe click here.)

 

Site by Bennett Web & Design Co.
Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 546 2017 SOUTHEASTERN NATURALIST 16(4):546–566 Influence of Lake Surface Area and Total Phosphorus on Annual Bluegill Growth in Small Impoundments of Central Georgia Aaron P. Sundmark1,* and Cecil A. Jennings2 Abstract - The relationships between environmental variables and the growth rates of fishes are important and rapidly expanding topics in fisheries ecology. We used an informationtheoretic approach to evaluate the influence of lake surface area and total phosphorus on the age-specific growth rates of Lepomis macrochirus (Bluegill) in 6 small impoundments in central Georgia. We used model averaging to create composite models and determine the relative importance of the variables within each model. Results indicated that surface area was the most important factor in the models predicting growth of Bluegills aged 1–4 years; total phosphorus was also an important predictor for the same age-classes. These results suggest that managers can use water quality and lake morphometry variables to create predictive models specific to their waterbody or region to help develop lake-specific management plans that select for and optimize local-level habitat factors for enhancing Bluegill growth. Introduction Lepomis macrochirus Rafinesque (Bluegill) are aggressive, inquisitive, active, and brightly colored, all of which have made them more recognizable and appreciated by the angling and non-angling public than almost any other species of freshwater fish (Scott and Crossman 1973). Bluegills most likely account for more individual catches than any other sportfish species in North America (Etnier and Starnes 1993). As a result, much research has been directed at understanding Bluegill population structure and the factors that lead to variation in adult body size (Aday et al. 2008). However, differences in ecological productivity among similar lakes complicate efforts to manage for common fisheries goals (e.g., quality–size structure versus abundance). This variation in the productivity of adjacent lakes and populations has led researchers to suggest that local characteristics (e.g., water quality, habitat availability, lake morphometry) can be strong drivers of Bluegill population dynamics (Kratz et al. 1997). Management strategies for fisheries traditionally focus on 3 variables: fish, habitat, and people (Hubert and Quist 2010). Within the broad topic of habitat, lake morphometry is very important in making management decisions, but this aspect of a fishery typically cannot be actively managed. Managers can, however, select 1Warnell School of Forestry and Natural Resources, University of Georgia, Athens, GA 30602. 2US Geological Survey, Georgia Cooperative Fish and Wildlife Research Unit, Warnell School of Forestry and Natural Resources, University of Georgia, Athens, GA 30602. *Corresponding author - sundmark@uga.edu. Manuscript Editor: Andrew Rypel Southeastern Naturalist 547 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 water bodies for specific management strategies such as special regulations or stocking based on water-body morphometry. For example, managers can determine if an impoundment has suitable habitat requirements, such as expansive littoral areas, to create a trophy Bluegill fishery. Large littoral areas tend to support a higher biomass of Bluegill than habitats with limited littoral zones (Barwick 2004). Common goals for recreational fisheries managers often include providing a quality fishing experience and sustainable harvest for the public. In a phone survey of 200 anglers in Missouri, 81% of anglers mentioned the quality of fishing, 76% of anglers mentioned environment, and 41% of anglers mentioned people in their descriptions of memorable fishing trips (Weithman and Anderson 1978). In a 2013 angler-creel survey conducted in central Georgia, 155 Bluegill anglers were questioned as to whether they would prefer that Marben Public Fishing Area (PFA) lakes were managed for more fish, larger fish, or both. Only 12% of the anglers wanted Marben PFA lakes managed primarily for larger fish, 25% mostly wanted to catch more fish, and 50% wanted the lakes to be managed for both (Roop 2015). Weithman and Anderson (1978) and Roop (2015) described the desire for fisheri es managers to focus their attention on managing for as many fast-growing fish as possible to improve angler satisfaction. Fisheries managers potentially can improve the quantity and quality of fish in a population; however, managing for 1 characteristic typically comes at the expense of the other. In our view, a better understanding of Bluegill growth may be needed to evaluate management trade-offs between quality and quantity in populations and the fishery they support. Georgia Department of Natural Resources (GADNR) fisheries managers at Marben PFA are interested in the production and management of a trophy Bluegill fishery. Based on the overall goal of the fisheries managers, the specific objectives of this project were to: (1) identify and quantify the measurements of surface area and levels of total phosphorus that have the greatest influence on Bluegill growth at 6 Marben PFA impoundments, (2) determine the best model for predicting Bluegill growth based on identified characters, (3) provide the GADNR with an indication of the best impoundments for growing large Bluegills, and (4) provide fisheries managers with a framework for developing predictive models based on environmental variables for selecting their most suitable impoundments for trophy Bluegill fisheries. Methods Data collected for each impoundment included: Bluegill biological data, surface area, mean depth, shoreline-development factor (SDF), impoundment age, total phosphorus, carbon and nitrogen profiles, water temperature, pH, dissolved oxygen (DO), conductivity, total dissolved solids (TDS), alkalinity, Secchi-disc depth, chlorophyll a, and mean total monthly fishing ef fort (angler hours). Field-site description Marben PFA is a public outdoor recreation facility located near Mansfield, GA, that is currently managed by the Georgia Department of Natural Resources’ Wildlife Resources Division (GADNRWRD). The 2590-ha facility includes campgrounds, Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 548 archery ranges, firearm ranges, a wildlife management area (WMA) for hunting, as well as ~120 ha of impoundments that constitute Marben PF A (Fig. 1). We conducted the study in 6 Marben PFA impoundments that varied in size from 1.2 ha to 32 ha and had total phosphorus measurements varied as low as 0.34 mg/L and as high as 1.76 mg/L. We also concurrently sampled 2 additional impoundments within Marben PFA for the same variables to be used as post hoc model-assessment impoundments. These impoundments were 1.9 ha and 5.0 ha in size. The small reservoirs at Marben PFA are typically shallow basins (1.5–3.6 m deep) with flooded standing timber and large amounts of woody debris along the shoreline. Aquatic macrophytes were mostly absent during the sampling period and throughout most of the year because of the high clay-content of the soil in the region. As a result excess nutrients, which would typically be utilized for aquatic macrophyte growth, caused large algal blooms that were present during warm portions of the summer. All impoundments in this study were near the top of the watershed, and were surrounded by deciduous forests. Fish collection We sampled Bluegills during July and August 2014, presumably after they had spawned at least once (i.e., in an attempt to obtain an even male–female ratio from each impoundment). We employed a boat-mounted, pulsed-DC Smith-Root® 6A type electrofisher with a Wisconsin Ring electrode array and pulsed direct current (30 min pedal time per sample) to sample both adult and juvenile fish. We adjusted pulse width and voltage to maintain an electrical field (~4–6 amps) that stunned the fish sufficiently for capture with minimal mortality. Fish sampling occurred from sunset until up to 4 h after sunset. This period has been previously identified as the optimal time for Bluegill collection because it is when the species moves to shallow water for foraging or reproduction (Baumann and Kitchell 1974, Dumont and Dennis 1997, Malvestuto and Sonski 1990, Pierce et al. 2001, Sanders 1992). During electrofishing, we measured the total length (TL) of all captured Bluegills to the nearest mm and sorted the fish into 10-mm–length increments to create length–frequency distributions of the total catches from each impoundment. We subsampled 5 Bluegills from each 10-mm–length increment (>50 mm TL) from each impoundment to determine growth rates (Quist et al. 2012, Tomcko and Pierce 2005). We assigned unique identification codes to, measured for TL (mm) and total weight (g) of, euthanized (UGA Animal Use Permit No. A2014-06-023-Y1-A0), placed on ice, and transported the subsampled fish to the University of Georgia for otolith removal. Otolith preparation and age determination We extracted sagittal otoliths by cutting through the ventroanterior surface of the isthmus and opening the cranial cavity with surgical scissors (Bagenal and Tesch 1978, Schneidervin and Hubert 1986). Forceps were used to remove the otoliths from the cranium, and the otoliths were wiped and dried with a paper towel and stored in 4-mL glass vials labeled with the fish’s individual identification code. Bluegills have relatively thin otoliths; therefore, otolith preparation for reading was minimal. We submersed whole otoliths in a water-filled, black-bottomed Southeastern Naturalist 549 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 Figure 1. Map of the state of Georgia and the Charlie Elliot Wildlife Center showing the Marben Public Fishing Area lakes that were sampled during July and August 2014. Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 550 dish placed under a Leica®TM MZ-7 (Leica Microsystems-Wetzlar, Germany) dissecting microscope at 16X magnification and counted annuli; a fiber-optic light source was also used for side illumination to facilitate counting annuli. The dissecting scope was equipped with a Lieca® DFC295 camera that transmitted the image onto a computer monitor, and we took a photograph of individual otoliths. We began our counts at the central focus and moved outward towards the edge of the otolith. We defined the annuli as the outermost margin of the dark, opaque band as in Schramm (1989). In centrarchids, the annuli of the otoliths tend to be relatively narrow, opaque bands that accumulate during the winter (Maceina and Betsill 1987, Schramm and Doerzbacher 1982, Taubert and Coble 1977, Taubert and Tranquilli 1982). Two readers independently counted the annuli from each whole otolith, recorded the age of each fish as the total number of annuli counted, and assigned year class by subtracting the age from the year 2014. Disagreements in age assignments by the 2 readers were resolved by a consensus recount by both readers. If the 2 readers could not come to consensus on an age estimate, we eliminated the otolith in question from the data set. We recorded the percentage of reader agreement after all otoliths had been analyzed. We plotted length–frequency distributions for all Bluegills captured and used age–length keys for all retained Bluegills to estimate size and age structures of the populations in each lake (Quist et al. 2012). We employed the Fraser–Lee method to back-calculate length-at-age for specific fish based on the consistent ratio of annular distances of the hard structure to the total length of the fish, and we derived and plotted by lake mean length-at-age data and associated standard deviations for each year class (Quist et al. 2012). A von Bertalanffy growth curve was fitted to length-at-age data for all 6 study lakes and the 2 assessment lakes (Quist et al. 2012). We calculated annual growth rates of Bluegills in the 6 individual Marben PFA study impoundments and 2 assessment impoundments by averaging the mean growth (mm) among years (Quist et al. 2012). Measuring predictor variables Initially, we measured many environmental variables as potential predictor variables in our modeling of Bluegill growth. We eventually abandoned this approach in favor of a simpler one that included only the predictor variables of lake surface area and total phosphorus. Specifically, many of the traditional water-quality variables such as morphoedaphic index, total alkalinity, secchi depth, and total phosphorus are indicators of lake productivity, and others such as lake surface area and mean depth are measures of lake morphometry. A small number of impoundments have been used to create the models predicting lengths-at-ages of Bluegills; thus, over-parameterizing the models with too many predictor variables became a concern. We addressed this issue by selecting 1 measure of primary productivity and 1 measure of lake morphometry for all study impoundments. We ultimately selected total phosphorus because it is widely considered the largest driver of primary-production dynamics in lakes (Fee 1979, Smith 1979) and is also a strong correlate of fish production (Downing et al. 1990, Hansen and Leggett 1982). We selected the surface area of the lakes Southeastern Naturalist 551 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 as our metric to describe lake morphometry because it is known to strongly influence fish production (Bennett 1971, Hoffmann and Dodson 2005, Hubert and Chamberlain 1996, Jenkins and Oglesby 1982, Tomcko and Pierce 1997, Youngs and Heimbuch 1982, Wagner et al. 2007) and is correlated to fish and zooplankton species richness (Eadie et al. 1986, Kratz et al. 1997). We used a horizontal Van Dorn water sampler biweekly during July, August, and September of 2014 from each lake to collect a columnar, euphotic-zone water sample at a depth of 2 m. From each sample, 250 mL of water was extracted, kept on ice at 4 ºC, and taken to a University of Georgia Lab for environmental analysis. In the lab, we added 1 mL of sulfuric acid to each for preservation. The water was stored in a dark refrigerator at 1–4 ºC. We analyzed water samples for total phosphorus concentration within 60 d of sample collection. We calculated lake surface area from digital orthophotographs available from Google® Earth (Google, Inc. 2013). Data analysis We entered all predictor-variable data from each Marben PFA impoundment into a Microsoft Excel 2013® spreadsheet for export to the statistical software program R: Version 3.1.2® for analysis. We examined relationships between Bluegill growth and lake surface area and total phosphorus by means of multiple-regression techniques; the analysis was conducted for each age class. To counter skewed predictor- variable data, we added 0.1 to data on total phosphorus and log10-transformed the results. We square-root–transformed surface-area data (Sokal and Rohlf 1995). The response variable was the mean back-calculated length-at-age of Bluegills collected from each lake. We developed models using variables to represent possible biological hypotheses predicting Bluegill growth. More specifically, we derived regression models for predicting mean back-calculated length-at-age from the transformed lake-surface area and total phosphorus data. We calculated Akaike’s information criteria (AIC; Akaike 1974), with small sample adjustment (AICc; Hurvich and Tsai 1989) for each model. We considered the model with the lowest ΔAICc valueas the most plausible for the age class. Akaike weights (wi, Burnham and Anderson 2002) were calculated to determine the relative fit of the models, with the best approximating model for the age class possessing the highest wi value. In addition, we used a percent maximum wi to identify the confidence model set; the confidence set of candidate models included models with Akaike weights that were within 10% of the highest, which is comparable to the minimum cutoff point (i.e., 8 or 1/8) suggested by Royall (1997) as a general guideline for evaluating strength of evidence. Once we established a confidence set of models for each age class of Bluegill, we employed model averaging to create a composite model that best described the factors affecting Bluegill growth within each age class for each study impoundment. We assigned a relative importance value to each parameter within each of the age class’ composite models (Burnham and Anderson 2002). Assessment of model accuracy We conducted a post hoc analysis of data collected from 2 Marben PFA impoundments (Greenhouse and Lower Raleigh; separate from the 6 included in the Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 552 original design) to assess the models’ predictive performance. We chose these 2 impoundments for their relatively moderate size and ease of access. To test the accuracy of the models, we collected 5 Bluegill for each 10-mm–length group (>50 mm) from the 2 impoundments selected for model-validation analysis. The Bluegills were collected at about the same time as those collected from the original 6 impoundments and were included in the random selection of lake order when sampling occurred. We collected sets of predictor-variable data at each of these impoundments, as described in the previous sections. We compared the model-validation experimental data to the values predicted from the models. We entered predictor variables from the 2 model-validation lakes into the models created from the original 6 lakes and compared the observed responses from the model-validation lakes with the expected responses to determine mean-squared prediction error (MSPE) from the 2 test lakes. Prediction error for the test lakes was then compared to the mean-squared error (MSE) of the 6 model lakes to assess whether error in predicting new values was significantly larger than error in fitting the models. MSPE is calculated with the formula : n* MSPE = Σ ([Yi - Ŷi ]2) / n*, i = 1 where Yi is the observed mean back-calculated lentgth-at-age for the impoundment, Ŷi is the predicted mean back-calculated length-at-age for the impoundment, and n* is the number of samples. We employed an F-test to compare the MSE and MSPE. If a substantial difference existed between the values of the MSPE from the 2 test lakes and the MSE from the 6 model lakes, we would assume that the 2 test lakes had differences in unrelated characteristics that were causing growth to vary from the original 6 lakes (e.g., competition for prey resources). Such a result would indicate that the model was not a good predictor of factors that influence Bluegill growth in the study ponds. Results Bluegill catch, size structure, and growth metrics We captured and enumerated Bluegills of all sizes during the study. Bluegill total length varied from less than 50 mm in all lakes to 225 mm in Shepherd Lake. We included a total of 2797 Bluegills, captured in July and August of 2014, to create length–frequency distributions for the 6 study impoundments (Fig. 2). Total catch by individual impoundment was lowest in Whitetail Lake (233) and highest in Margery Lake (733) (Table 1), and the mean total catch = 466.2, SD = 170.6. Bluegill catch per unit effort (CPUE; 30-min transects) varied from 111.5 in Whitetail Lake to 480 in Bennett Lake (Table 1), and the mean CPUE = 303.7, SD = 128.7. We harvested a total of 420 Bluegills from the 6 study impoundments, and the age and size structures of Bluegills were variable (Fig. 2) among the 6 study impoundments. We harvested the greatest number of Bluegills in individual impoundments from in Whitetail Lake (49) and the lowest quantity from Bennett Lake (81) (Table 1) and the mean number harvested = 70, SD = 12.1. Southeastern Naturalist 553 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 Initial agreement between both readers for age assignments was 78.2%; however, agreement after a consensus read increased agreement to >99%. Only 3 Bluegills were excluded from further analyses because of reader disagreement or unreadable otoliths. The sample of aged fish was comprised of individuals spanning 5 year-classes (2009–2014), with lake-specific maximum-age classes ranging from age-3 in Margery Lake to age-5 in Fox, Bennett, Shepherd, and Dairy lakes. Mean back-calculated lengths-at-age in individual impoundments varied from 132.4 mm Figure 2. Von Bertalanffy growth models fitted to Bluegill length-at-age data from 8 central Georgia impoundments with model parameters listed (t0 = length at age-0, K = curvature parameter, L∞ = length at infinite years). Lower Raleigh and Greenhouse impoundments were used only for model validation. Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 554 in Dairy Lake to 181.1 mm in Margery Lake for the 3-year-old age class (mean = 151.7 mm, SD = 20.3, n = 6), from 152.2 mm in Dairy Lake to 190.3 mm in Bennett Lake for the 4-year-old age class (mean = 167.8 mm, SD = 15.4, n = 5), and from 169.7 mm in Dairy Lake to 190.1 mm in Fox Lake for the 5-year-old age class (mean = 184.6 mm, SD = 9.9, n = 4) (Table 2). Annual growth rates were variable across the 6 study impoundments. Minimum and mazimum values for annual growth for the Bluegill age classes were as follows: 67.7 mm in Lower Raleigh Lake to 84.6 mm in Margery Lake (mean = 75.3 mm, SD = 5.7) for age-1, 28.7 mm in Shepherd Lake to 48.9 mm in Bennett Lake (mean = 38.5 mm, SD = 7.1) for age-2, 18.3 mm in Shepherd Lake to 31.7 mm in Margery Lake (mean = 25.9 mm, SD = 4.9) for age-3, 14.0 mm in Lower Raleigh Lake to 28.9 mm in Fox Lake (mean = 19.0 mm, SD = 4.4) for age-4, and 6.4 mm in Lower Raleigh Lake to 18.1 mm in Dairy Lake (mean = 15.0 mm, SD = 4.4) for age-5. Table 1. Fish data: # of transects = number of electrofishing transects, Total catch = total number of Bluegills caught, CPUE = catch per unit effort (30-min transects; one unit of effort was 30 min of electrofishing pedal time), and # harvested = the total number of Bluegills harvested. Lake data: Total phos = total phosphorus (mean ± SD) and lake surface area. Data collected from 8 Marben Farms Public Fishing Area impoundments in central Georgia during July and August 2014. † signifies impoundments used for model assessment that were not included during the modeling process. Fish data Lake data Total Surface Impoundment # of transects catch CPUE # harvested Total phos (ppm) area (ha) Fox 1 367 367.0 78 0.34 ± 0.24 32.13 Bennett 1 480 480.0 81 0.63 ± 0.70 26.52 Margery 2 733 366.5 77 0.67 ± 0.80 15.11 Shepherd 2 537 268.5 72 1.76 ± 2.78 4.54 Dairy 2 457 228.5 63 0.37 ± 0.53 2.84 Whitetail 2 223 111.5 49 0.37 ± 0.51 1.27 Greenhouse† 2 287 143.5 63 0.79 ± 0.52 1.93 Lower Raleigh† 2 375 187.5 63 1.25 ± 0.99 4.99 Table 2. Mean back-calculated lengths-at-age (mm) for Bluegills from 8 Marben Farms Public Fishing Area impoundments in central Georgia during July and August 2014. Φ = Bluegills of the specified age-class were not captured in this impoundment. † signifies impoundments used for model assessment that were not included during the modeling process. Mean back-calculated length-at-age (mm) Impoundment Age-1 Age-2 Age-3 Age-4 Age-5 Age-6 Fox 76.6 117.6 150.1 176.2 190.1 Φ Bennett 79.9 134.4 171.5 190.3 188.5 Φ Margery 84.6 136.8 181.1 Φ Φ Φ Shepherd 77.8 111.7 134.8 157.6 190.0 Φ Dairy 69.3 105.7 132.4 152.2 169.7 Φ Whitetail 70.8 108.3 140.0 162.5 Φ Φ Greenhouse† 75.7 113.6 143.6 158.3 176.8 Φ Lower Raleigh† 67.9 113.1 161.6 170.9 167.2 189.7 Southeastern Naturalist 555 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 Lake morphometry and primary productivity data The square-root transformed surface-area data varied from 1.13 at Whitetail Lake to 5.67 at Fox Lake (mean = 3.27, SD = 1.90), and the log10(TP + 0.1)-transformed total-phosphorus data varied from -0.50 in Dairy and Whitetail Lakes to -0.04 in Shepherd Lake (mean = -0.35 ppm, SD = 0.11) (Table 3). Multiple regression and model averaging Multiple regression results indicated that the best predictor variables in a model differed depending on the specific response variable being predicted (i.e., the age class of the fish and the number of predictor variables in the model). We analyzed with multiple regression analysis all possible models containing surface area and total phosphorus (Table 4). The best models for predicting Bluegill lengths at ages 1–4 were single-variable models based on the square root of surface area as the predictor variable. The single variable models of log10(total phosphorus + 0.1) were the second-best models in predicting Bluegill lengths at ages for the age 1–4 models. The log10(total phosphorus + 0.1) and square root of surface-area additive models had the smallest Akaike weights in the predictions of lengths at ages for Bluegills ages 1–5 out of the candidate models. The confidence sets of models to be included in model averaging included both of the 1-variable models for each age class with the exception of the confidence set from age class 4, which included only the 1-variable model of the square root of surface area (Table 4). The confidence set created from the age-1 models included the 1-variable models of square root of surface area and log10(total phosphorus + 0.1) (wi: 0.64, 0.36; Table 4). Model averaging revealed that the square root of impoundment surface area was the best predictor of Bluegill sizes at various age classes (Table 5). Surface area was positively related and had the greatest relative importance in predicting growth of Bluegills at ages 1–4. Total phosphorus was positively related and was also important in predicting growth of Bluegills at ages 1–3. A composite model for predicting Bluegill length at age-5 could not be calculated because of a low sample size of impoundments with age-5 Bluegills. Table 3. Mean log10 (total phosphorus mg/L + 0.1) ± SD and the square root of surface area (Sqrt of SA; ha). Data collected in 2014 from 8 impoundments at Marben Farms Public Fishing Area in central Georgia. † signifies impoundments used for model assessment that were not included during the modeling process. Impoundment log10 (total phosporus mg/L + 0.1) Sqrt of SA Fox -0.40 ± 0.22 5.67 Bennett -0.31 ± 0.45 5.15 Margery -0.35 ± 0.51 3.89 Shepherd -0.04 ± 0.54 2.13 Dairy -0.50 ± 0.39 1.69 Whitetail -0.50 ± 0.40 1.13 Greenhouse† -0.13 ± 0.30 1.39 Lower Raleigh† 0.03 ± 0.33 2.23 Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 556 Table 4. Candidate models for each Bluegill age-class from selected Marben Farms Public Fishing Area impoundments in central Georgia during July and August 2014. Data presented for each model include number of terms within the model (K), Akaike’s information criterion with small sample adjustment (AICc), delta AICc (Δᵢ) and Akaike weight (wi). * indicates models included in the confidence set (i.e., had a wᵢ within at least 10% of the wᵢ of the best model). Model K AICc Δᵢ wᵢ Age-1 models sqrt(surface area)* 3 34.89 0.00 0.64 log10(total phosphorus + 0.1)* 3 36.04 1.15 0.36 sqrt(surface area) + log10(total phosphorus + 0.1) 4 62.46 27.58 0.00 Age-2 models sqrt(surface area)* 3 44.08 0.00 0.87 log10(total phosphorus + 0.1)* 3 47.88 3.10 0.13 sqrt(surface area) + log10(total phosphorus + 0.1) 4 73.93 30.00 0.00 Age-3 models sqrt(surface area)* 3 49.63 0.00 0.84 log10(total phosphorus + 0.1)* 3 53.01 3.38 0.16 sqrt(surface area) + log10(total phosphorus + 0.1) 4 79.61 29.98 0.00 Age-4 models sqrt(surface area)* 3 40.71 0.00 0.98 log10(total phosphorus + 0.1) 3 48.40 7.69 0.02 sqrt(surface area) + log10(total phosphorus + 0.1) 4 70.66 29.95 0.00 Age-5 models sqrt(surface area)* 3 38.45 0.15 0.48 log10(total phosphorus + 0.1)* 3 38.30 0.00 0.51 sqrt(surface area) + log10(total phosphorus + 0.1) 4 47.03 8.72 0.01 Assessment of model accuracy An F-test comparison between the variances of the 2 impoundment-assessment models and the 6 impoundment models revealed that the variances were not significant for ages 1–4, and that they failed to reject the null hypothesis of the F-test (F-ratios for ages 1–4, 3.29, 0.06, 0.67, and 0.93, respectively; F-crit values for ages 1–4, 6.61, 6.61, 6.61, 7.71, and 10.13, respectively). This indicates that there is no evidence of lack of fit of the models that predict the outcome of the assessment lakes. An F-test could not be performed for age-5 Bluegills because a composite model could not be created. Discussion Models that correctly predict variation in fish growth are important tools in fisheries management. The results of this study identified specific environmental variables that were strong predictors of Bluegill growth in small impoundments in central Georgia. Although the impoundments in this study were relatively close in proximity, they displayed considerable variation in lake surface area and total phosphorous concentration. Previous studies elsewhere have linked Bluegill growth to Daphnia spp. biomass (Shoup et al. 2007) and aquatic vegetation diversity (Tomcko and Pereira 2006). Tomcko and Pierce (2001, 2005) found that Southeastern Naturalist 557 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 Bluegill size structure and growth were positively related to Secchi depth, maximum depth, total alkalinity, and water temperature. Our results indicate that Bluegill populations in 6 small impoundments in central Georgia were influenced by lake surface area and total phosphorus. Consistent with the results of previous studies that have investigated Bluegills, the surface area of the impoundments was included in several of our age-specific composite models and had a pronounced positive effect on growth (Bennett 1971, Hoffmann and Dodson 2005, Hubert and Chamberlain 1996, Jenkins and Oglesby 1982, Tomcko and Pierce 1997, Wagner et al. 2007, Youngs and Heimbuch 1982). Surface area explained a portion of the variation in Bluegill mean back-calculated length and was incorporated into composite models of ages 1–4. It had the highest value of relative importance in the predictive models for ages 1–4. Our findings that surface area was positively related to mean back-calculated lengths-at-age for all 4 age-classes of Bluegills with sufficient samples sizes for analysis complement those of Tomcko and Pierce (2001), who found that mean back-calculated lengths-at-ages 5 and 6 were significantly related to surface area. A possible explanation for the profound effect that surface area had on our models could have been its positive relationship to littoral habitat availability. For example, annual growth of Bluegills at ages 2 and 3 was significantly greater in Nebraska sandhill lakes with high percentages of littoral zone area (Porath and Hurley 2005). Model results from our study suggest that surface area may be the most important environmental variable we measured with regard to explaining variation in Bluegill growth in Marben PF A impoundments. Table 5. Composite models and associated parameters created by model averaging for age classes of Bluegill collected at Marben Farms PFA impoundments, 2014. * indicates that composite model could not be created because of small sample size of response variable. Model/model parameter Relative importance β estimate Adjusted SE Age-1 Bluegill growth composite model Intercept 74.75 9.02 sqrt(surface area) 0.64 1.20 1.61 log10(total phosporus + 0.1) 0.36 6.19 14.94 Age-2 Bluegill growth composite model Intercept 105.68 16.21 sqrt(surface area) 0.87 4.28 3.73 log10(total phosporus + 0.1) 0.13 1.80 20.60 Age-3 Bluegill growth composite model Intercept 132.37 24.82 sqrt(surface area) 0.79 5.91 5.82 log10(total phosphorus + 0.1) 0.21 0.26 33.70 Age-4 Bluegill growth composite model Intercept 148.04 8.21 sqrt(surface area) 1.00 6.25 2.24 Age-5 Bluegill growth composite model Intercept * * sqrt(surface area) * * * log10(total phosphorus + 0.1) * * * Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 558 Total phosphorus is often considered to be the main indicator of primary productivity in lentic systems (Downing et al. 1990, Fee 1979, Smith 1979). Elevated phosphorus levels increase available energy that primary producers can consume (Brylinsky and Mann 1973, Dejenie et al. 2008); thus, abundance of many aquatic primary producers, such as phytoplankton, is positively related to total phosphorus concentrations. Primary producers directly consume nutrients (driving primary productivity) and are then consumed by secondary producers in both terrestrial and aquatic systems (Allaby 2004). Our results indicated that the effect of total phosphorus was incorporated into the predictive models in this study for ages 1–4 and was the second-most important factor for predicting Bluegill growth. Overall, we observed the greatest growth in larger lakes with relatively higher total phosphorus concentrations. We did not consider biotic factors such as food-web dynamics or macrophyte coverage in our study because of logistical constraints (e.g., effort and time); however, Theiling (1990) suggested that biotic factors have a relatively large influence on Bluegill growth. Theiling (1990) studied variables such as benthic invertebrate biomass from discrete lake zones, zooplankton density and size distribution, macrophyte density in the littoral zone and throughout the lake, water chemistry and nutrient content, chlorophyll-a concentration, secchi-disk transparency, and lake morphology. He determined that macrophyte density, zooplankton size, and profundal benthos biomass explained 60% of the variation in a Bluegill growth index. In contrast, he found that abiotic variables explained none of the variation in the Bluegill growth he observed. Several other studies considered biotic factors in seeking to explain variation in growth (Aday et al. 2003, Hoxmeier et al. 2009, Schultz et al. 2008, Shoup et al. 2007, Snow and Staggs 1994, Tomcko and Pierce 1997, Tomcko and Pierce 2005, Tomcko and Pereira 2006). Aday et al. (2003) considered the direct and indirect effects of Dorosoma cepedianum (Lesueur) (Gizzard Shad) on Bluegill growth and population size-structure in 20 Illinois reservoirs and demonstrated that the presence of Gizzard Shad was associated with reduced Bluegill growth rates and adult-size structures, and that mechanisms other than direct competition for food resources might have caused this situation. Dorosoma spp. (shad) and other species occupy the same ecological niche within several of the Marben Farms PFA impoundments; thus, the abundance of these species should be included in future studies that model Bluegill growth in those systems. Intraspecific competition can also cause variation in growth because fish of the same species and age generally occupy similar feeding niches. Therefore, high-density Bluegill populations tend to exhibit relatively slow growth and smaller adult-body sizes than low-density populations (Beckman 1950, Osenberg et al. 1988, Weiner and Hanneman 1982). The effect of density dependence on Bluegill growth at all age-classes is widely reported in the published literature, and would be useful to include in future growth models. Swingle and Smith (1942) discussed the influence of macrophyte control, stocking with Micropterus salmoides (Lacepède) (Largemouth Bass), fertilization, and heavy fishing on a stunted Bluegill population in a 1.2-ha Alabama pond. The pond was producing good fishing within 6 months after Southeastern Naturalist 559 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 initiating these management tactics, and after 18 months, seining samples indicated that 1-year-old Bluegills weighed as much as 5-year-old Bluegills had at the beginning of the experiment. The presence of Largemouth Bass and other species should also be documented in future studies that model Bluegill growth because these species occur as predators of Bluegills within several of the Marben Farms PFA impoundments. A final consideration for future research efforts is the inclusion of angler-harvest dynamics. Angling pressure can reduce the number of larger adult Bluegills in a population and can even induce stunting (Beard et al. 1997). However, in some instances, the effects of angler harvest were considered to be negligible (Hoxmeier et al. 2009). This variation in reported effects of angling on Bluegill size is a fruitful avenue for further investigation. Difficulties in consistently estimating Bluegill ages and recruitment patterns were notable challenges to our study. Differences in appearance of otolith centers (opaque or translucent) are commonly caused by differences in the timing of hatching relative to annulus formation (Hales and Belk 1992). The variation in appearance of the otolith centers in this study may have caused the readers to assign incorrect ages to some samples; however, this error was likely minimized in our study by the consensus otolith analyses we used to reconcile disagreements in age estimates when they occurred. During the otolith aging process, we determined that we had not captured any Bluegills over the age of 3 y in Margery Lake. We do not know the reason for the absence of older fish, but it reduced the sample size during the modeling exercise from 6 lakes to 5 lakes for age-4 Bluegills. Also, we did not assign any ages older than 4 years to Bluegills in Whitetail Lake. The lack of age-4 and age-5 Bluegills in these lakes reduced the sample size during the modeling exercise from 6 lakes to 4 lakes. The reduction of the sample size of age-4 and age-5 Bluegills may have reduced the precision of the models. An F-test comparison between the variances of the 2-impoundment assessment models and the 6-impoundment models for age classes 1–4 revealed that the variances were not significant. This result suggests the model’s predictions were appropriate for the data observed from the test lakes. Marben PFA lakes with a large surface area and a relatively high total-phosphorus value likely provide the best conditions for Bluegill growth. Surface area had the highest relative importance among parameters within most of the models; thus, relatively large impoundments such as Bennett or Margery are excellent candidate lakes for establishing a trophy Bluegill fishery. Although, Fox Lake has the largest surface area of all the impoundments within the study, it also had the lowest total phosphorus concentration. Therefore, Fox Lake is not a good candidate impoundment for fostering a trophy Bluegill fishery without intensive fertilization. Fish growth rates are highest in the youngest age classes (including Bluegills); thus, our results on the factors affecting this early growth in Bluegills have implications in areas well beyond our study site at Marben PFA. These results may be useful for informing managers of Bluegill fisheries in small impoundments throughout the species’ range. Similar approaches may assist managers in creating predictive models that include lake- or region-specific water quality and lake morphometry data to develop lake-specific management plans. Such plans would better select for, and optimize, local-level habiSoutheastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 560 tat factors to promote fast-growing Bluegills compared to other generic models. At a minimum, our modeling technique could be a simple and useful tool for identifying lakes with potential for trophy Bluegill management in regions that have a surplus of lentic waterbodies. Acknowledgments We thank the Wildlife Resources Division of the Georgia Department of Natural Resources for their collaboration and funding, as well as allowing the use of the Marben Public Fishing Area during our research. We are grateful to Steve Zimpfer, Peter Dimmick, and Hunter Roop at the University of Georgia for providing technical assistance, and Clint Moore and Douglas Peterson for providing constructive comments that improved the quality of the manuscript. The Georgia Cooperative Fish and Wildlife Research Unit is sponsored jointly by the Georgia Department of Natural Resources, the University of Georgia, the US Fish and Wildlife Service, and the US Geological Survey. This work was conducted under the auspices of the University of Georgia Animal Use Protocol No. A2014—06-023-Y1-A0. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government. Literature Cited Aday, D.D. 2008. Exploring stunted body size: Where have we been, what do we know, and where do we go? Pp. 349–367, In M. Allen, S. Sammons and M. Maceina (Eds.). Balancing Fisheries Management and Water Uses for Impounded River Systems. American Fisheries Society, Bethesda, MD. 697 pp. Aday, D.D., R.J.H. Hoxmeier, and D.H. Wahl. 2003. Direct and indirect effects of Gizzard Shad on Bluegill growth and population size structure. Transactions of the American Fisheries Society 132:47–56. Akaike, H. 1974. A new look at the statistical model identification. Automatic Control, IEEE Transactions 19:716–723. Allaby, M. 2004. The Concise Oxford Dictionary of Ecology, 3rd Edition. Oxford University Press, New York, NY. 415 pp. Bagenal, T.B., and F. W. Tesch. 1978. Age and growth. Pp. 101–136, In T.B. Bagenal (Ed.). Methods for Assessment of Fish Production in Freshwater, 3rd Edition. Blackwell Scientific Publication, Oxford, UK. 384 pp. Barwick, D.H. 2004. Species richness and centrarchid abundance in littoral habitats of three southern US reservoirs. North American Journal of Fisheries Management, 24(1):76–81. Baumann, P.C., and J.F. Kitchell. 1974. Diel patterns of distribution and feeding of Bluegill (Lepomis macrochirus) in Lake Wingra, Wisconsin. Transactions of the American Fisheries Society 103:255–260. Beard J.R., T.D., M.T Drake, J.E. Breck, and N.A. Nate. 1997. Effects of simulated angling regulations on stunting in Bluegill populations. North American Journal of Fisheries Management 17(2):525–532. Beckman, W.C. 1950. Changes in growth rates of fishes following reduction in population densities by winterkill. Transactions of the American Fisheries Society 78:1. Bennett, G.H. 1971. Management of Lakes and Ponds. Van Nostrand Reinhold, Inc., New York, NY. 375 pp. Brylinsky, M., and K. Mann. 1973. An analysis of factors governing productivity in lakes and reservoirs. American Society of Limnology and Oceanography 18(1):1–14. Southeastern Naturalist 561 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Springer Science and Business Media, New York, NY. 488 pp. Dejenie, T., T. Asmelash, L. De Meester, A. Mulugeta, A. Gebrekidan, S. Risch, A. Pals, K. Van derGucht, W. Vyverman, J. Nyssen, and J. Deckers. 2008. Limnological and ecological characteristics of tropical highland reservoirs in Tigray, Northern Ethiopia. Hydrobiologia 610(1):193–209. Downing, J.A., C. Plante, and S. Lalonde. 1990. Fish production correlated with primary productivity, not the morphoedaphic index. Canadian Journal of Fisheries and Aquatic Sciences 47:1929–1936. Dumont, S.C., and J.A. Dennis. 1997. Comparison of day and night electrofishing in Texas reservoirs. North American Journal of Fisheries Management 17:939–946. Eadie, J.M., T.A. Hurly, R.D. Montgomerie, and K.L. Teather. 1986. Lakes and rivers as islands: Species-area relationships in the fish faunas of Ontario. Environmental Biology of Fishes 15:81–89. Etnier, D.A., and W.C. Starnes. 1993. The Fishes of Tennessee. The University of Tennessee Press, Knoxville, TN. 681 pp. Fee, E.J. 1979. A relation between lake morphometry and primary productivity and its use in interpreting whole-lake eutrophication experiments. Limnology and Oceanography 24(3):401–416. Google. 2013. Earth (Version 7.1.2.2041) software., Mountain View, CA. Available online at https://www.google.com/earth/. Accessed 5 November 2013. Hales L.S., Jr., and M.C. Belk. 1992. Validation of otolith annuli of Bluegills in a southeastern thermal reservoir. Transactions of the American Fisheries Society 121(6):823–830. Hanson, J.M., and W.C. Leggett. 1982. Empirical prediction of fish biomass and yield. Canadian Journal of Fisheries and Aquatic Sciences: 39(2):257–263. Hoffmann, M.D., and S.I. Dodson. 2005. Land use, primary productivity, and lake area as descriptors of zooplankton diversity. Ecology 86:255–261. Hoxmeier, R.J.H., D.D. Aday, and D.H. Wahl. 2009. Examining interpopulation variation in Bluegill growth rates and size structure: Effects of harvest, maturation, and environmental variables. Transactions of the American Fisheries Society 138:423–432. Hubert, W.A., and C.B. Chamberlain. 1996. Environmental gradients affect Rainbow Trout populations among lakes and reservoirs in Wyoming. Transactions of the American Fisheries Society 125:925–932. Hubert, W.A., and M. Quist. 2010. Inland fisheries management in North America. American Fisheries Society, Bethesda, MD. 736 pp. Hurvich, C.M., and C.-L. Tsai. 1989. Regression and time-series model selection in small samples. Biometrika 76:297–307. Jenkins, R.M., and R.T. Oglesby. 1982. The morphoedaphic index–concepts and practices: The Morphoedaphic index and reservoir fish production. Transactions of the American Fisheries Society 111:133–140. Kratz, T., K. Webster, C. Bowser, J. Magnuson, and B. Benson. 1997. The influence of landscape position on lakes in northern Wisconsin. Freshwater Biology 37:209–217. Maceina, M., and R. Betsill. 1987. Verification and use of whole otoliths to age White Crappie. Pp. 267–278, In R.C. Summerfelt and G.E. Hall (Eds.). Age and Growth of Fish. Iowa State University Press, Ames, IA. 544 pp. Malvestuto, S.P., and A.J. Sonski. 1990. Catch rate and stock structure: A comparison of daytime versus nighttime electric fishing on West Point Reservoir, Georgia, Alabama. Pp. 210–218, In I.G. Cowx (Ed.). Developments in Electric Fishing. Blackwell Scientific Publications, Oxford, UK. 368 pp. Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 562 Osenberg, C.W., E.E. Werner, G.G. Mittelbach, and B.J. Hall. 1988. Growth patterns in Bluegill (Lepornis rnacrochirus) and Pumpkinseed (L. gibbosus) sunfish: Environmental variation and the importance of ontogenetic niche shifts. Canadian Journal of Fisheries and Aquatic Sciences 45:17–26. Pierce, C.L., A.M. Corcoran, A.N. Gronbach, S. Hsia, B.J. Mullarkey, and A. Schwartzhoff. 2001. Influence of diel period on electrofishing and beach-seining assessments of littoral fish assemblages. North American Journal of Fisheries Management 21:918–926. Porath, M.T., and K.L. Hurley. 2005. Effects of waterbody type and management actions on Bluegill growth rates. North American Journal of Fisheries Management 25:1041–1050. Quist, M.C., M.A. Pegg, and D.R. Devries. 2012. Age and growth. Pp. 677–731, In A. Zale, D. Parrish, and T. Sutton (Eds.). Fisheries Techniques, 3rd Edition. American Fisheries Society, Bethesda, MD. 1009 pp. Roop, H.J. 2015. Angling activity (effort, catch, and harvest), use patterns, and anglers’ evaluations of fishery quality and management at the Marben Public Fishing Area. M.Sc. Thesis. University of Georgia, Athens, GA. Royall, R. 1997. Statistical Evidence: A Likelihood Paradigm. Volume 71. Chapman and Hall, London, UK. 191 pp. Sanders, R.E. 1992. Day versus night electrofishing catches from near-shore waters of the Ohio and Muskingum Rivers. The Ohio Journal of Science 92:51–59. Schneider, J.C. 2000. Manual of Fisheries Survey Methods II: With Periodic Updates. Michigan Department of Natural Resources, Fisheries Division, L ansing, MI. 242 pp. Schneidervin, R.W., and W.A. Hubert. 1986. A rapid technique for otolith removal from salmonids and catostomids. North American Journal of Fisheries Management 6:287–287. Schramm, H.L., Jr. 1989. Formation of annuli in otoliths of Bluegills. Transactions of the American Fisheries Society 118:546–555. Schramm, H.L., Jr., and J.F. Doerzbacher. 1982. Use of otoliths to age Black Crappie from Florida. Proceedings of the Annual Conference of the Southeastern Association of Fish and Wildlife Agencies 36:95–105. Schultz, R.D., Z.J. Jackson, and M.C. Quist. 2008. Relating impoundment morphometry and water quality to Black Crappie, Bluegill, and Largemouth Bass populations in Iowa. American Fisheries Society Symposium 62:479–491. Scott, W.B., and E.J. Crossman. 1973. Freshwater fishes of Canada. Fisheries Research Board of Canada Bulletin 184. 966 pp Shoup, D., S. Callahan, D. Wahl, and C. Pierce. 2007. Size-specific growth of Bluegill, Largemouth Bass, and Channel Catfish in relation to prey availability and limnological variables. Journal of Fish Biology 70:21–34. Smith, V.H. 1979. Nutrient dependence of primary productivity in lakes. Limnology and Oceanography 24(6):1051–1064. Snow, H.E., and M. Staggs. 1994. Factors related to fish growth in northwestern Wisconsin lakes. Wisconsin Department of Natural Resources Research Report 162. Madison, WI. Sokal, R.R., and F.J. Rohlf. 1995. Biometry, 3rd Edition. WH Freeman and Company New York, NY. 880 pp. Swingle, H.S., and EV. Smith. 1942. The management of ponds with stunted fish populations. Transactions of the American Fisheries Society 71(1):102–105. Taubert, B.D., and D.W. Coble. 1977. Daily rings in otoliths of three species of Lepomis and Tilapia mossambica. Journal of the Fisheries Board of Canada 34:332–340. Taubert, B.D., and J.A. Tranquilli. 1982. Verification of the formation of annuli in otoliths of Largemouth Bass. Transactions of the American Fisheries Society 111:531–534. Southeastern Naturalist 563 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 Theiling, C.H. 1990. The relationships between several limnological factors and Bluegill growth in Michigan lakes. Michigan Department of Natural Resources Fisheries Research Report 1970. Ann Arbor, MI. Tomcko, C.M., and D.L. Pereira. 2006. The relationship of Bluegill population dynamics and submerged aquatic vegetation in Minnesota lakes. Minnesota Department of Natural Resources Investigational Report 538. St. Paul, MN. Tomcko, C.M., and R.B. Pierce. 1997. Bluegill growth rates in Minnesota. Minnesota Department of Natural Resources, Section of Fisheries Investigational Report 458, St. Paul, MN. Tomcko, C.M., and R.B. Pierce. 2001. The relationship of Bluegill growth, lake morphometry, and water quality in Minnesota. Transactions of the American Fisheries Society 130:317–321. Tomcko, C.M., and R.B. Pierce. 2005. Bluegill recruitment, growth, population size structure, and associated factors in Minnesota lakes. North American Journal of Fisheries Management 25:171–179. Wagner, T., M.T. Bremigan, K.S. Cheruvelil, P.A. Soranno, N.A. Nate, and J.E. Breck. 2007. A multilevel modeling approach to assessing regional and local landscape features for lake classification and assessment of fish growth rates. Environmental Monitoring and Assessment 130:437–454. Weithman, A.S., and R.O. Anderson. 1978. An analysis of memorable fishing trips by Missouri anglers. Fisheries 3:19–20. Wetzel, R., and G. Likens. 1979. Limnological Analyses. W.B. Saunders Company, Philadelphia, PA. 429 pp. Wiener, J.G., and W.R. Hanneman. 1982. Growth and condition of Bluegills in Wisconsin Lakes: Effects of population density and lake pH. Transactions of the American Fisheries Society 111:761–767. Youngs, W.D., and D.G. Heimbuch. 1982. Another consideration of the morphoedaphic index. Transactions of the American Fisheries Society 111:151–153. Zydlewski, J., and S.D. McCormick. 1997. The ontogeny of salinity tolerance in the American Shad, Alosa sapidissima. Canadian Journal of Fisheries and Aquatic Sciences 54:182–189. Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 564 Appendix A. Morphological data collected in 2014 from impoundments at Marben Farms Public Fishing Area in central Georgia. Age = impoundment age in 2014 (y), SA = surface area (ha), SDF = shoreline development factor, and Mean depth = mean impoundment depth in (m) ± SD. † signifies impoundments used for model assessment that were not included during the modeling process. Impoundment Age SA SDF Mean depth Fox 28 32.13 3.11 3.62 ± 1.99 Bennett 44 26.52 2.09 2.89 ± 1.49 Margery 40 15.11 1.80 2.31 ± 1.21 Shepherd 62 4.54 1.69 1.69 ± 1.07 Dairy 44 2.84 2.16 2.67 ± 1.23 Whitetail 33 1.27 1.56 1.49 ± 0.60 Greenhouse † 56 1.93 1.47 2.46 ± 1.13 Lower Raleigh † 46 4.99 1.69 2.89 ± 1.71 Southeastern Naturalist 565 A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 Appendix B. Mean (± SD) of selected water quality variables collected in 2014 from impoundments at Marben Farms Public Fishing Area in central Georgia. Measurements for total alkalinity were taken at the surface of the water. Measurements for temperature, dissolved oxygen, pH, and conductivity were all taken at a 1-m depth. Measurements for total phosphorus and chlorophyll-a were taken at a 2-m depth. † signifies impoundments used for model assessment that were not included during the modeling process. Total Total Conductivity Temperature phosphorus Secchi alkalinity Impoundment DO (mg/L) pH (μS) (°C) (ppm) Chl-a (ppb) depth (m) (ppm) Fox 6.58 ± 1.75 6.58 ± 1.75 68.35 ± 1.94 29.18 ± 2.03 0.34 ± 0.24 18.56 ± 16.29 1.23 ± 010 30.00 ± 9.88 Bennett 7.84 ± 1.86 9.12 ± 0.58 90.58 ± 4.72 29.68 ± 2.27 0.63 ± 0.70 34.83 ± 15.41 0.87 ± 0.29 34.50 ± 5.58 Margery 7.40 ± 1.30 9.31 ± 0.60 74.03 ± 4.51 29.02 ± 2.36 0.67 ± 0.80 52.45 ± 37.67 0.47 ± 0.08 32.83 ± 11.03 Shepherd 7.05 ± 1.56 9.04 ± 0.69 82.70 ± 7.53 28.72 ± 1.73 1.76 ± 2.78 59.95 ± 19.56 0.76 ± 0.05 30.50 ± 2.95 Dairy 8.57 ± 1.81 9.18 ± 0.66 67.22 ± 5.33 28.58 ± 1.66 0.37 ± 0.53 14.70 ± 17.19 0.69 ± 0.11 27.00 ± 3.52 Whitetail 7.19 ± 0.85 7.42 ± 0.68 111.45 ± 4.67 28.68 ± 2.51 0.37 ± 0.51 22.45 ± 17.04 0.79 ± 0.17 50.67 ± 9.85 Greenhouse† 8.70 ± 0.89 9.37 ± 0.44 96.60 ± 5.42 28.08 ± 1.68 0.79 ± 0.52 31.15 ± 48.71 0.62 ± 0.17 45.00 ± 8.65 L. Raleigh† 7.65 ± 0.83 8.99 ± 0.53 81.77 ± 4.14 29.10 ± 2.02 1.25 ± 0.99 23.87 ± 8.86 1.30 ± 0.20 34.67 ± 2.73 Southeastern Naturalist A.P. Sundmark and C.A. Jennings 2017 Vol. 16, No. 4 566 Appendix C. Summary statistics for 12 morphological and water-quality predictor variables measured at 6 Marben Farms Public Fishing Area impoundments in central Georgia during 2014. Predictor ariables n Mean SD Minimum Maximum Age (y) 6 41.8 11.7 28.0 62.0 Mean depth (m) 6 2.4 0.8 1.5 3.6 Surface area (ha) 6 13.7 13.1 1.3 32.1 Shoreline development factor 6 2.1 0.6 1.6 3.1 Dissolved oxygen (mg/L) 6 7.4 0.7 6.6 8.6 pH 6 8.8 0.7 7.4 9.3 Conductivity (μS) 6 82.4 16.8 67.2 111.5 Temperature (°C) 6 29.0 0.4 28.6 29.7 Total Phophorus (ppm) 6 0.7 0.5 0.3 1.8 Chlorophyll-a (ppb) 6 33.8 18.8 14.7 60.0 Secchi depth (m) 6 0.8 0.3 0.5 1.2 Total alkalinity (ppm) 6 34.3 8.4 27.0 50.7 Appendix D. List of Marben Farms Public Fishing Area impoundments in central Georgia included in the study. X = associated species present and size (ha) are given for each impoundment sampled during July and August 2014. M.s. = Micropteris salmoides (Lacepède) (Largemouth Bass), L. mi. = Lepomis microlophus (Günther) (Redear Sunfish), L. ma.= Lepomis macrochirus (Bluegill), I. p. = Ictalurus punctatus (Rafinesque) (Channel Catfish), and P. n. = Pomoxis nigromaculatus (Lesueur in Cuvier and Valenciennes) (Black Crappie). † signifies impoundments used for model assessment that were not included during the modeling process. Impoundment name Area (ha) M. s. L. mi. L. ma. I. p. P. n. Fox 32.13 x x x x x Bennett 26.52 x x x x x Margery 15.11 x x x x x Shepherd 4.54 x x x x x Dairy 2.84 x x x - - Whitetail 1.27 x x x - - Lower Raleigh† 1.93 x x x - x Greenhouse† 4.99 x x x x -