Abstract
We evaluated the dynamics of walleye Sander vitreus population size structure, as indexed by the proportional size distribution (PSD) of quality-length fish, in Escanaba Lake during 1967–2003 and in 204 other lakes in northern Wisconsin during 1990–2011. We estimated PSD from angler-caught walleyes in Escanaba Lake and from spring electrofishing in 204 other lakes, and then related PSD to annual estimates of recruitment to age-3, length at age 3, and annual angling exploitation rate. In Escanaba Lake during 1967–2003, annual estimates of PSD were highly dynamic, growth (positively) explained 35% of PSD variation, recruitment explained only 3% of PSD variation, and exploitation explained only 7% of PSD variation. In 204 other northern Wisconsin lakes during 1990–2011, PSD varied widely among lakes, recruitment (negatively) explained 29% of PSD variation, growth (positively) explained 21% of PSD variation, and exploitation explained only 4% of PSD variation. We conclude that population size structure was most strongly driven by recruitment and growth, rather than exploitation, in northern Wisconsin walleye populations. Studies of other species over wide spatial and temporal ranges of recruitment, growth, and mortality are needed to determine which dynamic rate most strongly influences population size structure of other species. Our findings indicate a need to be cautious about assuming exploitation is a strong driver of walleye population size structure.
Introduction
Fish population size structure, often indexed from length–frequency data, reflects the interaction of recruitment, growth, and mortality, so is often used as a means of detecting effects of varying year–class strength, growth, and exploitation (Willis et al. 1993; Anderson and Neumann 1996; Neumann and Allen 2007; Neumann et al. 2012). In general, population size structure should be inversely related to recruitment and mortality, but positively related to growth (Willis et al. 1993; Neumann et al. 2012). Population size structure shifts toward high numbers of small fish, usually an undesirable state for both human use and ecological balance, when 1) high recruitment increases numbers of small fish, 2) slow growth prevents small fish from growing into larger sizes, or 3) high fishing mortality reduces numbers of large fish faster than they are replaced (Willis et al. 1993; Neumann and Allen 2007). Each such symptom of population imbalance is of importance to fishery managers, who often wish to sustain adequate numbers of fish that are desirable to anglers and a healthy balance between predator and prey species (Isermann and Paukert 2010). Size structure indices, therefore, can be useful and important for interpreting fish population status and for diagnosing problems to be remedied by management actions (Willis et al. 1993).
The obvious connection between a structural attribute such as size structure and underlying dynamic rates of change (recruitment, growth, and mortality) makes size structure indices appealing for fishery managers as indicators of population imbalance (Willis et al. 1993; Anderson and Neumann 1996; Pope et al. 2010; Neumann et al. 2012). For example, when population size structure indicates a problem, a fishery manager could prescribe a management action to remedy the source of the problem, such as prescribing a change in angling regulations (Isermann and Paukert 2010). Such prescriptions of management actions depend on correctly deducing the cause of unbalanced population size structure, often a length-frequency distribution that is skewed toward small fish (Anderson and Neumann 1996; Neumann et al. 2012). Unfortunately, a sample length frequency is shaped by many factors that cloud its meaning, some of which are related to sampling, such as season, gear, site, and sample size, and others of which are related to temporal variation in forces driving recruitment, growth, and mortality, such as predator–prey ratios and environmental fluctuations (Willis et al. 1993; Anderson and Neumann 1996; Neumann et al. 2012).
Though population size structure is clearly driven by recruitment, growth, and mortality, sample length-frequency indices from an individual population cannot be unambiguously interpreted without more information, such as measurements of population density, recruitment, growth, or mortality (Willis et al. 1993; Anderson and Neumann 1996; Neumann and Allen 2007; Pope et al. 2010; Neumann et al. 2012). For example, a sample length frequency that is skewed toward small fish could be a function of sampling with gears in locations or at times where and when only small fish are vulnerable to capture, even though the population size structure is well balanced (Willis et al. 1993). Alternatively, a sample length frequency that is skewed toward small fish could be caused by high recruitment and slow growth, or by high exploitation, though management prescriptions to remedy high recruitment and slow growth (e.g., a high creel limit or low minimum-length limit) would be quite different from those to remedy high exploitation (e.g., a low creel limit or high minimum-length limit; Isermann and Paukert 2010). To determine the force or forces that most strongly drive population size structure, studies across long time spans and broad spatial scales are needed (Willis et al. 1993; Anderson and Neumann 1996; Neumann and Allen 2007; Pope et al. 2010; Neumann et al. 2012).
Our objective was to determine if walleye Sander vitreus population size structure was influenced more strongly by recruitment, growth, or exploitation in Wisconsin. The walleye is the most highly sought fish species in Wisconsin (McClanahan and Hansen 2005), and is also a keystone predator in cool-water ecosystems (Ryder and Kerr 1978). We indexed walleye population size structure as proportional size distribution (PSD; Guy et al. 2007) in a single lake over multiple years and in 204 other lake-years in northern Wisconsin. We then quantified the amount of temporal variation (Escanaba Lake) and spatial variation (204 other lakes) in PSD that was explained by indices of walleye recruitment (number/ha of age-3 walleyes), growth (mean length at age 3), and exploitation (angling exploitation rate). We expected to find that walleye population size structure was influenced by all three dynamics rates—recruitment, growth, and exploitation—rather than by any single dynamic rate.
Methods
To determine the extent to which walleye population size structure was influenced by recruitment, growth, and exploitation within a lake over a long temporal scale and among lakes across a broad spatial scale, we estimated PSD from angler-caught walleyes in Escanaba Lake and spring electrofishing survey data in 204 other lakes, and then related PSD to annual estimates of abundance of age-3 walleyes (an index of recruitment), mean length at age 3 (an index of growth), and annual angling exploitation rate (an index of fishing mortality). We indexed walleye population size structure as PSD (Gabelhouse 1984) in Escanaba Lake, Wisconsin during 1967–2003 and in 204 other lake-years in northern Wisconsin that were surveyed during 1990–2011 (Figure 1):
We then modeled relationships between PSD and indices of walleye recruitment (number/ha), growth (mean length at age 3), and exploitation (angling exploitation rate) to test the relative influence of each dynamic rate on walleye population size structure. Our data were collected within individual survey years, which enabled detection of within-year, short-term effects, but not delayed density-dependent or long-term effects, of recruitment, growth, and mortality on population size structure.
Locations of lakes in which walleye Sander vitreus populations were sampled for population size structure (as indexed by the proportional size distribution), recruitment (number of age-3 walleye/ha), growth (mean length at age 3), and exploitation (angling exploitation fraction) in northern Wisconsin during 1967–2011.
Locations of lakes in which walleye Sander vitreus populations were sampled for population size structure (as indexed by the proportional size distribution), recruitment (number of age-3 walleye/ha), growth (mean length at age 3), and exploitation (angling exploitation fraction) in northern Wisconsin during 1967–2011.
To evaluate the influence of recruitment, growth, and exploitation on walleye population size structure within a population over a long temporal scale, we used data for the intensively studied walleye fishery in Escanaba Lake (Hansen et al. 1991, 1998, 2004, 2011; Newby et al. 2000; Nate et al. 2011). Sampling methods are described elsewhere (Kempinger et al. 1975; Kempinger and Carline 1977), so we provide a brief summary here. Adult walleyes were caught in 19-mm or 25-mm square-mesh fyke nets, tagged with Monel metal jaw tags, fin-clipped, measured in length (to 2.54 mm), and released. A compulsory creel census allowed inspection of all fish caught by anglers for tags and fin clips. Ages were estimated from scales or spines, and all fish were measured in length and weight. Walleyes were fully vulnerable to angling harvest at age 3 during 1967–2003 because harvest was not regulated by daily creel or length limits (Hansen et al. 2011), so we estimated population size structure (indexed as PSD) from lengths of walleyes harvested each angling season (Nate et al. 2011). We indexed recruitment as the abundance of age-3 walleyes, estimated by catch-age analysis (Hansen et al. 2011), and converted this figure into population density by dividing the estimate by 119 ha. We indexed growth as the mean length (mm) of age-3 walleyes caught by anglers each year (Nate et al. 2011). We estimated angling exploitation as the fraction of marked walleyes harvested by anglers during the ensuing year (Nate et al. 2011).
To evaluate the influence of recruitment, growth, and exploitation on walleye population size structure across broad spatial and temporal scales, we used data for the extensively studied walleye fishery in northern Wisconsin. The walleye fishery in northern Wisconsin lakes has been studied as part of a multiagency effort to monitor walleye populations and associated fisheries in the northern third of Wisconsin (Hansen et al. 1991, 2000, 2005, 2010; Beard et al. 1997, 2003a, 2003b; Rogers et al. 2003, 2005; Schoenebeck and Hansen 2005). Sampling methods are described elsewhere, so we provide a brief summary here (Hansen et al. 1991). Lakes were selected at random each year since 1990 from among lakes containing walleyes that were subjected to angling and spearing harvest. We used 204 surveys for which population size structure, recruitment, growth, and angling exploitation could be estimated from survey data. Surveys on each lake included fyke netting in spring to mark and release walleyes, boat electrofishing of the entire lake shoreline to sample the fraction of walleyes marked, and random stratified roving-access creel surveys to estimate angling harvest (Rasmussen et al. 1998). Walleyes were captured for marking in fyke nets that were set shortly after ice-out when mature walleyes were congregated inshore for spawning. Ten percent of the walleyes in each lake were targeted for marking by partial removal of a fin. Electrofishing was used to collect fish for marking if 10% of fish could not be marked by fyke netting alone. Fish were collected 1–2 days after marking by boat electrofishing of the entire shoreline of each lake, to ensure marked and unmarked fish were equally vulnerable to capture. Ages were estimated from scales or spines for a length-stratified subsample of fish (five fish per 12.7-mm length group). We estimated population size structure (indexed as PSD) from lengths of fish captured during spring electrofishing samples, because PSD could not be computed from angler harvest on 81% (165 of 204) of lake-years for which walleye angling was regulated by length limits, and PSD from angler harvest and spring electrofishing were strongly correlated on 36 lake-years for which angling was not regulated by length limits (r = 0.748; P < 0.001). We indexed recruitment as catch/ha of age-3 walleyes caught during spring electrofishing (we estimated the number of age-3 walleyes from the sample length-frequency using an age-length key; Ricker 1975): we converted catch/mile into number/acre by dividing catch/mile by catchability (q = 2.49; 95% CI = 2.46–2.52; Rogers et al. 2003) and then converted into catch/ha (catch/acre divided by 0.404686 ha/acre). We estimated growth as mean length (mm) of age-3 walleyes that were subsampled for age estimation from the total number caught during spring fyke-netting and electrofishing. We estimated angling exploitation of walleyes as the expanded number of marked walleyes harvested by anglers during creel surveys divided by the number of walleyes marked (Beard et al. 2003b). We merged indices for multiple interconnected lakes into a set of indices for the entire lake chain as the area-weighted mean among lakes within each chain of lakes.
To test the relative influence of recruitment, growth, and exploitation on walleye population size structure, we fit power functions with population size structure (PSD) as the dependent variable (y), and indices of recruitment (R = number/ha), growth (G = mean length at age 3), and exploitation (u = exploitation fraction) as independent variables (x) in three separate models:
Because we measured all independent (predictor) variables with error, we used a measurement-error model to estimate bias-corrected parameters of the loge-transformed power function (Fuller 1987; Quinn and Deriso 1999):
The measurement-error model differs from the classical regression model by including a term for measurement error (μ) of each predictor variable, which negatively biases (attenuates) the slope of the classical regression model that assumes the predictor variable is measured without error. The bias-corrected slope can be estimated directly if the ratio of measurement errors (δ = σee/σμμ) between predicted (σee) and predictor variables (σμμ) is known (Fuller 1987:equation 1.3.7):
The bias-corrected slope estimator also includes variance among measurements of the predicted variable (myy), variance among measurements of the predictor variable (mxx), and covariance between measurements of predicted and predictor variables (mxy). We estimated δ as the ratio of the average coefficient of variation (CV = standard error/mean) of loge-transformed PSD (standard error from Equation 24.21, Zar 1999) to the average CV of loge-transformed recruitment (standard error from Zar 1999:equations 25.8 and 25.9), growth (standard deviation from Zar 1999:equation 4.13), and exploitation (standard error from variance estimators in Beard et al. 2003b) indices. We then defined the bias-corrected intercept of the loge-transformed model from the bias-corrected slope and means of each dependent and independent variable. We back-transformed bias-corrected parameters of the loge-transformed power model into scales of original measurement units to depict functional relationships between variables and to compute coefficient of variation (r2) in original measurement unit scales.
To simultaneously compare the influence of recruitment (R), growth (G), and exploitation (u) on walleye population size structure, we used multiple-linear regression to sequentially test predictors and their interactions (loge scale) in a single regression model (Zar 1999). We added variables sequentially to the model, beginning with the variable having the largest partial correlation with loge(PSD) and adding variables until no remaining variables explained significantly more residual variation (P ≤ 0.05). We judged variables to be collinear if tolerance was less than 0.10.
Results
In Escanaba Lake, walleye population size structure was significantly related to growth, but not to recruitment or angling exploitation rate during 1967–2003. Annual estimates of population size structure, indexed as PSD, ranged from 4 in 1989 to 55 in 1974 (CV = 0.528; Figure 2; Table S1, Supplemental Material). Annual estimates of growth (mean length at age-3) ranged from 285 mm in 1988 to 356 mm in 1973 and 2000 (CV = 0.054; Figure 2; Table S1, Supplemental Material). Annual estimates of recruitment (number of age-3 walleyes/ha) ranged from 0.88 age-3 walleyes/ha in 2003 to 60.9 age-3 walleyes/ha in 1997 (CV = 0.701; Figure 2; Table S1, Supplemental Material). Annual estimates of angling exploitation ranged from 10% in 1972 to 51% in 1971 (CV = 0.398; Figure 2; Table S1, Supplemental Material). Growth was significantly and directly related to PSD (P < 0.001), and explained 35% of PSD variation among years (Figure 3; Table 1). Recruitment was inversely, but not significantly related to PSD (P = 0.25), and explained only 3% of PSD variation among years (Figure 3; Table 1). Exploitation was inversely, but not significantly related to PSD (P = 0.11), and explained only 7% of PSD variation among years (Figure 3:Table 1). No other additional variables (R: F1,34 = 0.992, P = 0.323; u: F1,34 = 0.222, P = 0.640) or interactions (G × R: F1,34 = 0.989, P = 0.327; G × u: F1,34 = 0.206, P = 0.652; R × u: F1,34 = 0.002, P = 0.965; G × R × u: F1,34 = 0.001, P = 0.973) explained significantly more residual error than the model that included only growth (F1,35 = 26.33, P < 0.001).
Population size structure (as indexed by the proportional size distribution [PSD]), recruitment (number of age-3 walleye/ha [number/ha]), growth (mean length at age 3 [Length (mm)]), and exploitation (angling exploitation fraction [Exploitation Fraction]) of walleye Sander vitreus in Escanaba Lake, Wisconsin, during 1967–2003. Error bars depict 95% confidence limits of each annual estimate.
Population size structure (as indexed by the proportional size distribution [PSD]), recruitment (number of age-3 walleye/ha [number/ha]), growth (mean length at age 3 [Length (mm)]), and exploitation (angling exploitation fraction [Exploitation Fraction]) of walleye Sander vitreus in Escanaba Lake, Wisconsin, during 1967–2003. Error bars depict 95% confidence limits of each annual estimate.
Relationships between walleye Sander vitreus population size structure (as indexed by the proportional size distribution [PSD]) and recruitment (number of age-3 walleye/ha [Recruitment (N/ha)]; upper pair of panels), growth (mean length at age 3 [Length at Age-3 (mm)]; middle pair of panels), and exploitation (angling exploitation fraction [Exploitation Rate]; bottom pair of panels) in Escanaba Lake during 1967–2003 (left panels) and 204 northern Wisconsin lakes during 1990–2011 (right panels). Error bars depict 95% confidence limits of each estimate. Solid curves depict functional relationships that account for measurement error of the independent variable.
Relationships between walleye Sander vitreus population size structure (as indexed by the proportional size distribution [PSD]) and recruitment (number of age-3 walleye/ha [Recruitment (N/ha)]; upper pair of panels), growth (mean length at age 3 [Length at Age-3 (mm)]; middle pair of panels), and exploitation (angling exploitation fraction [Exploitation Rate]; bottom pair of panels) in Escanaba Lake during 1967–2003 (left panels) and 204 northern Wisconsin lakes during 1990–2011 (right panels). Error bars depict 95% confidence limits of each estimate. Solid curves depict functional relationships that account for measurement error of the independent variable.
Parameter estimates (SE = standard error), t ratios (t), significance (P), and variance explained (R2) of relationships (y = axb) between walleye Sander vitreus population size structure (y = proportional size distribution) and recruitment (x = age-3 number/ha), growth (x = mean length, mm, at age 3), and angling exploitation fraction (x = u) in Escanaba Lake during 1967–2003 and 204 other Wisconsin lakes during 1990–2011.

In 204 northern Wisconsin lake-years, walleye population size structure was significantly related to growth and recruitment, but not to angling exploitation rate. Surveys of 123 lakes included 63 lakes in 1 y, 44 lakes in 2 y, 12 lakes in 3 y, three lakes in 4 y, and one lake in 5 y. Population size structure (indexed as PSD) averaged 42 (CV = 0.68) and ranged from 0 in Snipe Lake in 2000 to 100 in Magnor Lake in 2007 and Big Sand Lake in 2009 (Table S1, Supplemental Material). Growth (mean length at age 3) averaged 305 mm (CV = 0.12) and ranged from 234 mm in Lake Laura in 1998 to 442 mm in Diamond Lake in 2009 (Table S1, Supplemental Material). Recruitment (number of age-3 walleyes/ha), averaged 9.2 age-3 walleyes/ha (CV = 2.46) and ranged from 0.0 age-3 walleyes/ha in Upper St. Croix Lake in 1992, Lower Turtle Lake in 2004, Wheeler Lake in 2008, and Diamond Lake in 2009 to 290.7 age-3 walleyes/ha in the Turtle-Flambeau Flowage in 1997 (Table S1, Supplemental Material). Angling exploitation ranged from 0.45% in Amnicon Lake in 1999 to 65% in Balsam Lake in 2011 (CV = 0.67). Recruitment was significantly and inversely related to PSD (P < 0.001; Table S1, Supplemental Material), and explained 29% of PSD variation among lakes (Figure 3; Table 1). Growth was significantly and directly related to PSD (P < 0.001), and explained 21% of PSD variation among lakes (Figure 3; Table 1). Exploitation was significantly and positively related to PSD (P = 0.013), but explained only 4% of PSD variation among lakes (Figure 3; Table 1). Growth, recruitment, and the interaction between growth and recruitment explained significantly more variation in PSD (R2 = 67%, F3,195 = 13.39, P < 0.001) than either variable alone. However, the interaction between growth and recruitment was collinear (tolerance = 0.0004230) with recruitment (tolerance = 0.0004195), so growth and recruitment without their interaction explained nearly as much variation in PSD as the model that included the interaction (R2 = 63%, F2,196 = 16.45, P < 0.001). No other additional variables (u: F1,195 = 1.621, P = 0.204) or interactions (G × u: F1,195 = 1.617, P = 0.205; R × u: F1,195 = 2.565, P = 0.111; G × R × u: F1,195 = 2.577, P = 0.110) explained significantly more residual variation than the model that included growth and recruitment.
Discussion
We found that PSD was highly dynamic among years in Escanaba Lake, as in other long-term studies of walleye populations in Escanaba Lake, Oneida Lake, and Lake Erie (Serns 1985; Nate et al. 2011). In an earlier analysis of data for Escanaba Lake, wide-ranging interannual variation in PSD during 1956–1982, whether indexed from spring fyke netting (range = 12–83) or angling harvest in May (range = 9–63) or for angling season (range = 13–55), convinced Serns (1985) that PSD was not a useful index for judging balance, if not coupled with age-frequency and age-growth data. Long-term studies of walleye population size structure are rare, but all suggest that PSD is highly dynamic, with interannual variation ranging from 4 to 52 in Escanaba Lake, 22 to 73 in Oneida Lake, and 28 to 98 in Lake Erie (Nate et al. 2011). Large interannual variation in PSD may suggest that any single-year estimate has little value for judging population balance, so PSD should only be used if measured over enough years to represent the equilibrium between population size structure and dynamic rates of recruitment, growth, and mortality (Anderson and Weithman 1978). Alternatively, large interannual variation in PSD around a value that is not within the objective range for the species (e.g., 30–60 for walleye; Anderson and Weithman 1978) may still be useful for deducing imbalance despite wide variation in annual PSD estimates. For example, PSD of walleye in Escanaba Lake varied widely around a long-term mean (18) that was below the objective range for the species, and 95% confidence limits of PSD overlapped the objective range in only 4 of 37 y (1973–1974 and 2001–2002), which supports a conclusion that this population is not in balance, regardless of the year in which PSD was measured (Nate et al. 2011; this study). In contrast, PSD of walleye in Oneida Lake averaged 46 during 1956–2007 and 95% confidence limits of PSD were outside the objective range in only 2 of 50 y (1957 and 2006), which supports a conclusion that this population is in balance, regardless of the year in which PSD was measured (Nate et al. 2011). We therefore conclude that PSD has value for judging walleye population status, despite wide interannual variation, though any single PSD estimate that is too high or too low may not indicate long-term status.
Our finding that recruitment and growth were more influential than exploitation both compares and contrasts with studies of other species (largemouth bass Micropterus salmoides, Carline et al. 1984; northern pike Esox lucius, Willis and Scalet 1989, Pierce et al. 2003; bluegill, Lepomis macrochirus, Novinger and Legler 1978). For example, in Ohio impoundments, largemouth bass recruitment was more influential on PSD than either growth or survival (Carline et al. 1984), unlike our study of walleye in Wisconsin natural lakes, where recruitment and growth were more influential than exploitation. For northern pike in Minnesota and Wisconsin lakes, population density (a correlate of our recruitment index) was inversely related to both PSD and growth (Pierce et al. 2003), which infers that growth was positively related to PSD, as we found for walleye in Wisconsin lakes, and as was also found by Willis and Scalet (1989) for northern pike across eight states. For bluegill in small impoundments, recruitment was inversely related to PSD, and growth was equally, but directly, related to PSD (Novinger and Legler 1978), as we found for walleye in Wisconsin lakes. These examples emphasize a need for studies of other species over a wide range of recruitment, growth, and mortality to determine which dynamic rate most strongly influences population size structure (Willis et al. 1993; Anderson and Neumann 1996; Neumann and Allen 2007; Pope et al. 2010; Neumann et al. 2012).
In an earlier study of the walleye angling fishery in Escanaba Lake during 1956–1982 (Serns 1985), recruitment was negatively correlated to PSD, whereas growth was not correlated to PSD, unlike our study. The difference between our results and those of Serns (1985) may be caused by differences in power to detect significant correlations (years = 1956–1982, Serns [1985]; years = 1967–2003, our study), or because we used different indices of recruitment (age-0 density, Serns [1985]; age-3 density, our study) and growth (annual growth increment for ages 3–4 and 4–5, Serns [1985]; mean length at age 3, our study). The most likely explanation for differences in our findings was power to detect significant correlations, because our analysis included 37 y and Serns' analysis included only 27 y. Similarly, our use of age-3 density, rather than age-0 density 3 y earlier, should have increased power to detect a significant relationship with PSD by increasing contrast (variation) in the predictor variable: age-0 density CV = 0.82 (Serns 1985); age-3 density CV = 2.46 (this study). In contrast, our use of mean length at age 3, rather than mean annual growth increment for ages 3–4 (or 4–5), should have reduced power to detect a significant relationship with PSD by reducing contrast (variation) in the predictor variable: age 3–4 growth increment CV = 0.39 (Serns 1985); mean length at age 3 CV = 0.12 (this study). In addition, our analysis of many lakes across a large landscape over three decades was more powerful than Serns' (1985) analysis of one lake over fewer years for quantifying the relative importance of growth, recruitment, and exploitation on size structure of walleye populations in Wisconsin.
We conclude that population size structure was most strongly driven by growth and recruitment, rather than fishing mortality, for walleye populations in northern Wisconsin lakes. Our findings confirmed Serns' (1985) conclusion that PSD for a walleye population cannot be correctly judged to have been caused by any particular underlying population process (recruitment, growth, or mortality), but our findings also suggest that PSD of walleye is more likely driven by growth and recruitment than by angling exploitation across a broad geographic range (Wisconsin) and over several decades (1990–2011). Elsewhere, managers of recreational walleye fisheries would be wise to avoid concluding that low PSD is primarily caused by high angling exploitation, as has often been cautioned for users of size structure indices (Willis et al. 1993; Anderson and Neumann 1996; Neumann and Allen 2007; Pope et al. 2010; Neumann et al. 2012). Correct diagnosis of underlying mechanisms for any particular population size structure value requires additional information about recruitment, growth, and mortality, to avoid implementing a management prescription aimed to remedy one putative cause that may be the opposite of what is needed to remedy the true cause (Isermann and Paukert 2010). For walleye fisheries in Wisconsin, our findings suggest that a management prescription to reduce angling exploitation, such as a reduced daily-creel limit or high minimum-length limit, to remedy low PSD, would be the wrong management prescription because high recruitment and slow growth is a more likely cause of low PSD (Isermann and Paukert 2010). Nonetheless, our study across a broad geographic range (Wisconsin) and several recent decades (1990–2011), should enable walleye fishery managers to favor management actions that would build balanced population density and growth over management actions aimed at managing angling exploitation. Similar analyses from elsewhere in the walleye's range would help to confirm or deny the generality of our findings for this species, which is usually the most highly sought species in its native range (Quinn 1992; Schmalz et al. 2011), and also a keystone predator in cool-water ecosystems (Ryder and Kerr 1978). Last, we hope that our findings indicate a need to be cautious about assuming exploitation is a strong driver of walleye population size structure.
Supplemental Material
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.
Table S1. Data table of walleye Sander vitreus population size structure (proportional size distribution [PSD], coefficient of variation [CV]), number of age-3 walleye/ha (N/ha, CV), mean length at age 3 (mm, CV), and angling exploitation fraction (u, CV) in northern Wisconsin lakes during 1990–2011.
Found at DOI: http://dx.doi.org/10.3996/092013-JFWM-065.S1 (63 KB DOCX)
Acknowledgments
Steve Serns (deceased), former manager of the Escanaba Lake research station, inspired our analysis, and Steve Newman, his successor (now retired), encouraged us to expand his work into the present study. Jonathan F. Hansen, Daniel A. Isermann, Mark W. Rogers, three anonymous journal reviewers, and a Subject Editor reviewed the manuscript. We are grateful for the assistance provided by past and present staff from the Escanaba Lake research station and the treaty fishery assessment and management program.
Funding was provided by the Wisconsin Department of Natural Resources partly through the Federal Aid in Sportfish Restoration Program. This article is contribution 1836 of the U.S. Geological Survey, Great Lakes Science Center.
Any use of trade, product or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
References
Author notes
Hansen MJ, Nate NA. 2014. Effects of recruitment, growth, and exploitation on walleye population size structure in northern Wisconsin lakes. Journal of Fish and Wildlife Management 5(1):99-108; e1944-687X. doi: 10.3996/092013-JFWM-065
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.