Divergent density feedback control of migratory predator recovery following sex‐biased perturbations

Abstract Uncertainty in risks posed by emerging stressors such as synthetic hormones impedes conservation efforts for threatened vertebrate populations. Synthetic hormones often induce sex‐biased perturbations in exposed animals by disrupting gonad development and early life‐history stage transitions, potentially diminishing per capita reproductive output of depleted populations and, in turn, being manifest as Allee effects. We use a spatially explicit biophysical model to evaluate how sex‐biased perturbation in life‐history traits of individuals (maternal investment in egg production and male‐skewed sex allocation in offspring) modulates density feedback control of year‐class strength and recovery trajectories of a long‐lived, migratory fish—shovelnose sturgeon (Scaphirhynchus platorynchus)—under spatially and temporally dynamic synthetic androgen exposure and habitat conditions. Simulations show that reduced efficiency of maternal investment in gonad development prolonged maturation time, increased the probability of skipped spawning, and, in turn, shrunk spawner abundance, weakening year‐class strength. However, positive density feedback disappeared (no Allee effect) once the exposure ceased. By contrast, responses to the demographic perturbation manifested as strong positive density feedback; an abrupt shift in year‐class strength and spawner abundance followed after more than two decades owing to persistent negative population growth (a strong Allee effect), reaching an alternative state without any sign of recovery. When combined with the energetic perturbation, positive density feedback of the demographic perturbation was dampened as extended maturation time reduced the frequency of producing male‐biased offspring, allowing the population to maintain positive growth rate (a weak Allee effect) and gradually recover. The emergent patterns in long‐term population projections illustrate that sex‐biased perturbation in life‐history traits can interactively regulate the strength of density feedback in depleted populations such as Scaphirhynchus sturgeon to further diminish reproductive capacity and abundance, posing increasingly greater conservation challenges in chemically altered riverscapes.

One emerging stressor with such effects is synthetic chemicals, which have been increasingly detected in aquatic systems around the globe (Bernhardt, Rosi, & Gessner, 2017). A rising demand for human food production has increased agricultural use of synthetic chemicals (Tilman, 1999;Tilman et al.., 2001), posing a greater risk to freshwater vertebrates (Bernhardt et al, 2017). Synthetic hormones-often used as growth promoters for farmed animals-are especially potent and often induce sex-biased perturbation, impairing development and reproduction by disrupting energy allocation and maturation (del Carmen Alvarez & Fuiman, 2005). Because maternal investment in egg production incurs greater energetic costs than somatic growth, females may more likely experience reproductive impairment than males (Arukwe & Goksøyr, 1998). In prespawning females, even lowdose (often below detection) exposure to synthetic hormones can disrupt maturation processes such as vitellogenin production-a critical step in egg production (Arcand Hoy & Benson, 1998;Nash et al., 2004), which can lead to skipped spawning in some fishes (Rideout, Rose, & Burton, 2005). Chemically induced inefficiency in egg production may, thus, act density-dependently especially in threatened and endangered populations as it reduces the number of females contributing to a reproductive event (known as a "mate-finding" Allee effect, Gascoigne, Berec, Gregory, & Courchamp, 2009).
Synthetic hormones can also irreversibly perturb sex determination in exposed animals, skewing sex allocation in offspring (Mills & Chichester, 2005). Resulting skewed adult sex ratio may manifest as a mate-finding Allee effect (Gascoigne et al., 2009) by impairing reproduction density-dependently (e.g., a decline in mating success), impeding recovery and posing extinction risk for depleted populations as abundance declines further. Such human-mediated modification in demographic traits has been reported in exploited populations with sex-biased mortality (Grüebler et al., 2008;Molnar, Derocher, Lewis, & Taylor, 2007). However, it remains unclear how positive density-dependent processes regulating fitness (also known as "component" Allee effects, Stephens et al., 1999) manifested from modified traits such as skewed sex ratio and delayed maturation act together to influence per capita reproductive output and population recovery in nature.
We here present a simulation study on how sex-biased, density-dependent perturbation in life-history traits can emerge and modify recovery potential of a long-lived, migratory fish exposed to low-level synthetic androgens in a large river. Migratory animals are exposed to spatially and temporally dynamic flow-dependent exposure regimes as the chemicals move throughout networks of streams and rivers (O'Brien et al., 2016); variable exposure and subsequent symptoms among exposed individuals can obscure a causal link, if any, to population-level responses, delaying management actions (Peterson et al., 2003). We test varying exposure and effect scenarios on the Missouri River shovelnose sturgeon (Scaphirhynchus platorynchus, Figure 1a) as an illustrative case study. Population sizes of Scaphirhynchus sturgeon have declined in many of their endemic habitats, and some are facing potential extirpation owing to extensive habitat modification (e.g., impoundments; Bramblett & White, 2001) that disrupt reproductive activities including seasonal spawning migration and aggregation. Some of the Missouri River populations have shown signs of unsuccessful spawning (DeLonay et al., 2009) and year-class failure owing to low larval fish survival (Hrabik, Herzog, Ostendorf, & Petersen, 2007)-a critical process for population persistence (DeLonay et al., 2009).
Because female Scaphirhynchus sturgeon mature later, spawn less frequently, and live longer than males (DeLonay et al., 2009), female-biased disruption in reproductive traits likely regulates the strength of any Allee effects. We use a spatially explicit individual-based biophysical model that accounts for sex-specific life-history traits to evaluate how sex-biased perturbation in energetics (reduced maternal investment in egg production) and demographics (male-skewed sex allocation in offspring)-two commonly reported symptoms of synthetic androgens-disrupts density-dependent reproductive processes to control year-class success and population recovery under spatially and temporally dynamic exposure and habitat conditions.

| Study system
The study system is the lower 162 km of the Platte River (Nebraska, USA), one of the largest tributaries of the Missouri River basin ( Figure 2 and Figure S1), providing spawning and nursing habitats essential for threatened populations such as Scaphirhynchus sturgeon (Peters & Parham, 2008). Owing to its proximity to wastewater treatment plants and agricultural fields, however, the lower Platte River is also one of the impaired water bodies in the region (NDEQ, 2004). Fish that inhabit K E Y W O R D S agent-based modeling, alternative states, early life history, endangered species, endocrine disruptors, spatially explicit models this river and its tributaries, such as Scaphirhynchus sturgeon, are exposed to elevated concentrations of synthetic hormones widely used at concentrated cattle feedlots near the lower Platte River watershed (Huntzinger & Ellis, 1993) and exhibit symptoms of endocrine disruption (e.g., intersexuality and atresia, DeLonay et al., 2009).

| Sturgeon biophysical model
We use a spatially explicit individual-based biophysical model of shovelnose sturgeon previously developed for the lower Platte River population (Figure 1b) and fully described in Goto et al. (2015Goto et al. ( , 2018. The following briefly describes the model according to the Overview, Design concepts, and Details (ODD) protocol (Grimm et al., 2010).

| Purpose
The model is designed to uncover mechanisms underlying early lifehistory processes modified by natural and human-induced stressors (e.g., synthetic chemicals), and their contributions to population persistence in a spatially and temporally dynamic environment.

| Design concepts
1. Basic principles: The model primarily hinges on energetics principles to understand how maternal and environmental influences modulate year-class variability through energy acquisition and allocation (Houde, 1989); In this study, we evaluated sex-biased effects of synthetic androgen perturbation in maternal investment in gonad development and first-year sex allocation. The following briefly describes the essential features of the model relevant to this study.

| Entities, state variables, and scales
The model simulates sturgeon population dynamics with two types of entities; (a) river and (b) sturgeon. The model river consists of 162 equally divided longitudinal rectangular segments (i.e., 1 river km each) with six state variables; daily, depth-averaged discharge (m 3 /s), mean depth (gage height, m), channel width (m), benthic invertebrate prey density (g/m 2 ), water temperature (°C), and photoperiod (hr). Discharge, depth, width, and prey density are spatially explicit, whereas temperature and photoperiod are spatially uniform. The model keeps track of 11 state variables of sturgeon, body length (mm), storage mass (g), structural mass (g), gonadal mass (g), the number of individuals in each superindividual, maturity status, reproductive status, physiological condition (see Growth), sex, age (years), and longitudinal location in the river (rkm), of the entire life cycle (Figure 1b). Early life stages (fertilized eggs, yolk-sac larvae, and feeding larvae, and age-0 juveniles) are represented as superindividuals (Scheffer, Baveco, DeAngelis, Rose, & Nes, 1995), whereas age 1 + juveniles and adults are represented as individuals. Each superindividual is a collection of individuals ("cohorts") from the same mother.

| Process overview and scheduling
The model simulates the population and updates state variables at from late spring to early summer in the study system.

| Submodels
The model consists of one benthic prey submodel and six sturgeon submodels. In this study, simulations were designed to evaluate the effects of synthetic androgen perturbations in vitellogenin production in adult females and sex allocation in offspring (see Synthetic androgen effects below for detailed description) through the following processes in the submodels: a. Benthic prey production is computed using a mass-balance model of standing biomass, daily production, and net migration from neighboring cells as described by Rashleigh and Grossman (2005) with modifications described in Goto et al. (2018). Prey biomass density in each cell is initialized on January 1st of each simulation year with a mean of 18.8 (wet g/m 2 ) + 2.2RN, where RN is a random number drawn from the normal distribution (Whiles & Goldowitz, 2005).
b. Embryonic and yolk-sac larval development is computed using cumulative fractional temperature-dependent nonlinear functions (Goto et al., 2018) fitted to data reported by Wang, Binkowski, and Doroshov (1985). Hatched larvae are assumed to use their yolk sac for ~10 days (at 17-18°C) before exogenous feeding.

F I G U R E 2
Map of the study system, lower Platte River, Nebraska, USA c. Foraging is computed using a modified version of the Beddington-DeAngelis functional response model (Beddington, 1975;DeAngelis, Goldstein, & O'Neill, 1975), a function of encounter rate, handling time, size-dependent prey selectivity, and prey and predator densities. The exact prey length is randomly assigned using a uniform distribution (between minimum and maximum of each size class) each day (Goto et al., 2018).
When structure is less than optimal storage, feeding larvae and juveniles allocate surplus net energy to storage first until optimal storage is reached (Goto et al., 2018). Growth in length is then calculated using modified length-mass functions, where length increases only when structural mass increases. When the assimilated energy is less than the energy expended through metabolism, mass is lost only from storage without any change in structure and length (Goto et al., 2018). Sturgeon mature based on sex-specific probability, which is length-dependent functions derived from field observations of the Platte River population (Goto et al., 2018). Reproductive adults further allocate energy to gonad incrementally between spawning events.
Adults initiate the reproductive cycle only when their physiological condition is above a threshold based on relative storage mass, current storage/optimal storage (Goto et al., 2018).
Nonreproductive adults allocate energy only to structure and storage. The amount of daily energy allocated to gonad (J/day) by healthy adults (based on their physiological condition) is computed by dividing expected gonad energy content on the spawning day by the minimal spawning interval (365 days for males and 1095 days for females) plus the earliest expected spawning day (day of the year). Expected gonad energy is computed by an allometric function derived by calibration for the Platte River population. The amount of daily energy allocated to gonad is further adjusted when the physiological condition is below the threshold (Goto et al., 2018). While the earliest expected spawning day occurs in late March (when daylight hours >12 hr), the realized spawning day is determined by discharge rate, temperature, and relative gonad mass.
e. Spawning is simulated as a stochastic event and occurs only when spawning cues-water temperature, river discharge rate, daylight hours, and gonad development (gonadosomatic index or GSI), concurrently meet the thresholds (Delonay et al., 2016). Female Scaphirhynchus sturgeon spawn at highly variable river discharge rates among Lower Missouri River tributaries (Delonay et al., 2016). Because our sturgeon model showed little sensitivity to this parameter (Goto et al., 2015), a minimum threshold is set at 141.6 m 3 /s, which is expected to reach roughly 50% connectivity (migratory pathway connectedness) in the lower Platte River (Peters & Parham, 2008) and directly influences the sturgeon movement and subsequently habitat conditions for other individual-level actions. A minimum GSI to initiate spawning is set at 0.18 for females and 0.03 for males based on field observations (Wildhaber et al., 2007 (Braaten et al., 2008). Feeding larvae (postsettlement fish, Kynard, Henyey, & Horgan, 2002), juveniles, and adults move in the direction with probability of attain- ing higher habitat quality based on discharge rate (Goto et al., 2018) using the empirical models developed for the lower Platte River population (Peters & Parham, 2008). Habitat suitability is based on relative areas of exposed sandbars, open water, and shallow sandbar complexes. The new realized location is then determined using fish swimming speed and water velocity (Goto et al., 2018), and the maximum distance that fish can travel daily in this study is thus limited within three (above or below the current cell) neighboring cells (or 3.0 rkm).
g. Mortality results from thermal stress (eggs and yolk-sac larvae), predation (eggs, larvae, and juveniles), starvation, and recreational fishing (age 3+). We assume that a constant proportion of eggs and yolk-sac larvae die when water temperature rises above 24°C (Kappenman, Webb, & Greenwood, 2013). Starvation mortality depends on relative storage mass; starvation mortality is evaluated only when a proportion of storage mass relative to total mass drops below the threshold (optimal storage, Goto et al., 2018). Fishing mortality was derived from annual creel surveys in the Platte River (Peters & Parham, 2008). Realized predation, starvation, and fishing mortality are stochastic events. Realized mortality of eggs and larvae is computed by drawing a random number from the binomial distribution with mortality rates as the probability and the number of individuals represented by each superindividual as the number of trials. For age 1+ fish, each fish dies when a random number drawn from a uniform distribution is below the probability set above.

| Input data
Model inputs include water temperature (°C), mean gauge height (depth, m), channel width (m), discharge rate (m 3 /s), and photoperiod (hr). Water temperature, gauge height, channel width, and discharge rate (1995-2011, Figure S1 and Goto et al., 2018) are derived from the USGS National Water Information System (http://water data. usgs.gov/nwis). Further details on derivation of physical and hydrological input data are described in Goto et al. (2018).

| Initialization
We initialized simulations with 60,000 age 1 + individuals on January 1. The initial location of each sturgeon was randomly assigned to a cell with nonextreme water velocity (≤0.6 m/s) and depth (≥0.5 m).
The initial age structure follows a Poisson distribution with mean age of 7, which is truncated at a maximum age of 16 based on field observations (Hamel, Rugg, Pegg, Patiño & Hammen, 2015), and the initial sex ratio is 1:1. Sex is randomly assigned to each fish, assuming a binomial distribution. For age 1 + sturgeon, initial lengthsat-age were computed using the von Bertalanffy growth function (Ricker, 1975) parameterized for the lower Platte River population (Goto et al., 2018). Once spawned, fertilized eggs were initialized as superindividuals; the number of superindividuals is the number of mothers, and the number of individuals in each superindividual is the total number of eggs spawned by each mother. After settlement, feeding larvae were initialized with an initial total length of 15.6 (mm) + 0.84RN (Braaten et al., 2008), where RN is a random number drawn from the normal distribution, and these fish were subsequently simulated with individual-level processes (e.g., growth) and traits (e.g., body size).

| Simulation scenarios for sex-biased perturbations by synthetic androgens
We (c) because the androgen concentration varies with stream flow rate, the realized exposure level for individual fish is spatially and temporally dynamic and is thus adjusted based on flow rate in a grid cell in which the given fish is located.

| Energetic perturbation
We explicitly evaluated the mechanism underlying synthetic androgen-induced reduction in egg production by simulating the dose-dependent reduction of vitellogenin production in individual fish based on an exponential decay function (Ankley et al., 2003). Vitellogenin (R 2 = .99). Reduction in daily energy allocation to gonad (g/day) in exposed females was then computed based on somatic mass on the given day.
Because we lack information on a quantitative relationship between synthetic androgen exposure and Scaphirhynchus sturgeon vitellogenin production, we tested the hypothetical scenarios of 15%, 25%, and 35% reduction in daily vitellogenin production ( Figure S2) in fish that are located at the Elkhorn River-Platte River confluence, and these reduction rates in vitellogenin production approximately correspond to the responses in fish exposed to environmentally realistic, low-level synthetic androgens (e.g., <~15 ng/l 17β-trenbolone, a synthetic androgen commonly used in cattle feedlots, Ankley et al., 2003). For fish that are located in cells below the confluence, the reduction rates were proportionally adjusted by the difference in flow rate (e.g., if flow rates at the confluence and in the cell where a fish is located are 100 and 120 m 3 /s, respectively, the realized reduction rate for this fish is 12.5% under the 15% reduction scenario). Because the model sturgeon cannot detect low-level synthetic androgens in water, we assumed that their movement and energy acquisition (foraging) vary independently of synthetic androgens.

| Demographic perturbation
We implemented synthetic androgen-induced skewness in sex allocation by proportionally reducing the probability of being female.
We tested the hypothetical scenarios of 20%, 60%, and 80% reduction in probability of being females for posthatch fish that remain in the cell at or below the Elkhorn River-Platte River confluence, and these rates approximately correspond to the responses in fish exposed to environmentally realistic concentrations (e.g., <~15 ng/L 17β-trenbolone, Morthorst, Holbech, & Bjerregaard, 2010). We assumed that the probability of being female incrementally declines by remaining in the exposure area each day during the annual six-month exposure period (i.e., 0.11%, 0.33%, and 0.44%, respectively). For fish in cells below the confluence, the probability was proportionally adjusted by the difference in flow rates (as described above).
For all the scenarios, we used hydrological and temperature inputs empirically derived from the lower Platte River field surveys during 1995-2011 (Goto et al., 2018), from which we randomly selected each simulation year; we assumed no systematic change or shift in hydrological and thermal conditions in this study. Further, we assumed no interannual change in turbidity, channel geomorphology, and photoperiod. We ran ten simulations for each exposure scenario and computed summary statistics and % deviations from the baseline simulations in demographic and life-history traits that contribute to per capita reproductive output; adult sex ratio, relative reproductive female number (the proportion of adult females in the reproductive cycle), reproductive female number, spawner biomass, and recruit number.

| Baseline sturgeon population dynamics
Under the baseline conditions, the sturgeon population remained  (Tables S1 and S2) with ~7,100 recruits per year ( Figure S3f). Although the recruit numbers varied highly among years, a nonlinear (hump-shaped) density-dependent relationship emerged between spawners and recruits ( Figure S3g).

| Recovery from reduced vitellogenin production efficiency
Simulated reduction in vitellogenin production rate lowered reproductive capacities of the sturgeon population exposed to synthetic androgens ( Figure 4a). The proportion of energy allocated to gonad mass (10%) or storage mass (24%) in reproductive females under exposure did not differ from those under baseline (Table S1). However, the number of adult females in the reproductive cycle declined by 7.5%-34% during exposure (Figure 4a). Further, spawner biomass and recruit numbers declined, especially when exposed to medium and high levels of the perturbation (6.8%-25% and 23%-38%, respectively, Figure 4a), which were linearly correlated with relative changes in reproductive female abundance (Figure 4b). Lowered reproductive capacities and year class strength resulted in an 8.6%-43% reduction (relative to the baseline) in population biomass (Figure 3a) with positive density dependence in growth rate (Figure 3b) during exposure.
Reproductive females, spawners, and recruits of the population with moderate exposures returned to the baseline levels ( Figure 4a); similar linear relationships among the reproductive traits remained during recovery (Figure 4b). Population size fully recovered (i.e., mean population size returned to the baseline level) within ~5 years ( Figure 3a) and density dependence in per capita growth rate virtually disappeared (Figure 3b). However, the reproductive females, spawners, and recruits of the population with high exposure continued to decline (Figure 4a), ultimately reducing mean population biomass by up to 83% before showing a sign of recovery (~year 42, Figure 3a).

| Recovery from skewed recruit sex ratio
Sturgeon population biomass abruptly declined by up to 67% after ~year 22 (Figure 3a), and per capita growth rate continued to decline with biomass ( Figure

| Recovery from combined effects of reduced vitellogenin production and skewed sex ratio
Concurrently imposing reduction in vitellogenin production and

| D ISCUSS I ON
Simulations revealed plausible mechanisms involving chemically induced, sex-biased perturbations in life-history traits of longlived, migratory vertebrates such as Scaphirhynchus sturgeon; these perturbations can emerge as vastly different response and recovery patterns-being manifest as different strengths of demographic Allee effects (Hutchings, 2015)-in a spatially and temporally dynamic environment. We find that lowered vitellogenin production rate prolongs maturation period and incurs skipped spawning (a relative decline in reproductive female abundance), and, in turn, reduces spawner abundance, ultimately weakening year-class strength. Positive density feedback in per capita growth rate, however, disappeared (no Allee effect) during recovery.
Further, the energetic perturbation operates as exposure-dependent processes-the higher the exposure level, the more severe the damage, thereby resulting in longer recovery time. By contrast, male-skewed sex allocation (an absolute decline in reproductive female abundance) emerges as delayed (by more than two decades) population-level responses, followed by an abrupt shift in year-class strength and population trajectory under higher exposures, ultimately reaching an alternative state without any sign of recovery owing to persistent negative population growth (a strong Allee effect, Gascoigne et al., 2009;Stephens et al., 1999). When combined with the energetic perturbation, however, impacts of the demographic perturbation are dampened as extended maturation time reduces the frequency of producing male-biased offspring, allowing the population to maintain positive growth rate (a weak Allee effect) and gradually recover. The emergent patterns in longterm population projections illustrate that sex-biased perturbation in life-history traits of individuals can amplify or dampen Allee effects in depleted populations such as Scaphirhynchus sturgeon to further diminish reproductive capacity and abundance, posing conservation challenges in chemically altered riverscapes.
Sex-biased differences in energy allocation to gonadal maturation often determine life-history strategies such as foraging and mate selection, shaping sexual dimorphism in animal populations (Huse, 1998;Rennie et al., 2008). Delayed spawning events may, however, increase the probability of year-class failures owing to environmental stochasticity. We find that exposed (energetically challenged) female sturgeon are more likely to abort the reproductive cycle, as they divert energy to meet metabolic needs when poor habitat conditions persist. Although evaluating energetic tradeoffs between migration and fecundity is beyond the scope of our study, animals with extensive spawning migration such as Scaphirhynchus sturgeon likely skip spawning or produce offspring with lower survival rate owing to reduced energy reserves (Glebe & Leggett, 1981). Sufficient energy reserves in eggs are vital to larval development and survival, and maternal investment in egg production determines the nutritional status of posthatching larvae (Kjesbu, Solemdal, Bratland, & Fonn, 1996). Our model as-  Our study shows that recovery from demographic perturbation can pose greater conservation challenges than energetic perturbation. We find that chemically induced, male-skewed adult sex ratio can incur a lasting mate-finding Allee effect (Gascoigne et al., 2009) in long-lived populations such as Scaphirhynchus sturgeon, severely diminishing their reproductive fitness and recovery potential.
Although skewed sex ratio has been widely reported for a variety of vertebrate taxa (Donald, 2007), causes of the skewness vary; it can result from sex-specific phenotypic differences in physiology, dispersal, and natural or human-induced mortality (Goto, 2009;Goto & Wallace, 2010;Pen et al., 2010;Weimerskirch, Lallemand, & Martin, 2005 Long-term, low-level exposure to stressors such as synthetic hormones often obscures causative relationships (Windsor, Ormerod, & Tyler, 2018). Establishing ecologically relevant indicators such as the thresholds of Allee effects (where the shift occurs from positive to negative per capita growth) resulting from modified life-history traits would guide the risk assessment of synthetic chemicals and habitat conservation planning for threatened populations. Further, given the time lags across levels of biological organization, continued biological monitoring of individual-level traits would help detect early signals that may indicate population-level changes such as the strength of Allee effects. Although our study was designed to evaluate consequences of low-level exposure, our findings also indicate larger implications of among-individual variation for population persistence and recovery of threatened and endangered species. Population models widely used in risk assessments often assume no among-individual variation in a population such as sex-, size-, or age-dependent life-history traits (e.g., age at maturity). These models may not only gloss over individual variation in life-history traits that may contribute to population viability but also under-or over-estimate the strength of density feedback, and, in turn, the probability of extinction risk.
Large-river migratory populations such as Scaphirhynchus sturgeon experience multiple synthetic chemicals over agriculturally rich landscapes in regions such as the Midwestern United States (Welcomme, 1995). Further confounding factors in conservation and management efforts can arise in hydrologically altered systems such as the Missouri River. Precipitation, for example, can regulate the amount of agrichemicals released into rivers. Assessing the likelihood of ecological risks of agrichemicals for long-lived species can be facilitated by the development of trait-based predictive models using anticipated scenarios.
With expected trends in rising climate variability, which can have considerable effects on river hydrology and water chemistry, these forecasting tools are critical for fisheries and water resource management in chemically altered ecosystems.

ACK N OWLED G M ENTS
We appreciate comments by Maxime Vaugeois (University of Minnesota) and two anonymous reviewers on earlier versions of this manuscript. This project was partially funded by University of Nebraska-Lincoln and the Nebraska Game and Parks Commission (Project F-180-R).

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R S ' CO NTR I B UTI O N S
DG, VEF, and MAP conceived the ideas; MJH, JJH, and MLR collected and managed the data; DG led the writing of the manuscript; and all authors contributed to revisions and gave final approval for publication.