Research Article Simulating Northern Bobwhite Population Responses to Nest Predation, Nesting Habitat, and Weather in South Texas MICHAEL J. RADER, 1 Caesar Kleberg Wildlife Research Institute, Texas A&M University-Kingsville, Kingsville, TX 78363, USA LEONARD A. BRENNAN, 2 Caesar Kleberg Wildlife Research Institute, Texas A&M University-Kingsville, Kingsville, TX 78363, USA KYLE A. BRAZIL, 3 Caesar Kleberg Wildlife Research Institute, Texas A&M University-Kingsville, Kingsville, TX 78363, USA FIDEL HERNA ́ NDEZ, Caesar Kleberg Wildlife Research Institute, Texas A&M University-Kingsville, Kingsville, TX 78363, USA NOVA J. SILVY, Department of Wildlife and Fisheries Sciences, Texas A&M University, College Station, TX 77843, USA ABSTRACT Nest predation is thought to be one of the major factors limiting northern bobwhite ( Colinus virginianus ) populations. We examined the relative impact of altering nest-predation rate, nesting habitat, and weather (i.e., temp and precipitation) on northern bobwhite population dynamics in a hypothetical 15,000-ha subtropical-rangeland ecosystem in south Texas using a simulation model. The systems model consisted of a 3-stage (i.e., eggs, juv, and ad) bobwhite population with dynamics influenced by variables affecting production, recruitment, nest predation, and mortality. We based model parameters on data collected from a 3-yr nest-predator study employing infrared-camera technology, from ongoing field research using a radio-marked population of wild bobwhites, and from the literature. The baseline simulated bobwhite population dynamics corresponded closely to empirical data, with no difference between medians of simulated ( n ¼ 30 yr) and observed bobwhite age ratios over a 28-yr period. Similarly, a time-series comparison of simulated and observed age ratios showed most (89%) observed values fell within the 5th and 95th percentiles of the simulated data over the 28-yr period. We created simulated population scenarios representing 1) baseline historical conditions, 2) predator control, 3) low precipitation, 4) low precipitation with predator control, 5) high temperature, 6) high temperature with predator control, 7) reduced nest-clump availability, and 8) reduced nest-clump availability with predator control that resulted in considerably different median bobwhite densities over 10 yr. For example, under simulated predator control, populations increased by about 55% from the baseline scenario, whereas under simulated reduced nest-clump availability, populations decreased by about 75% from the baseline scenario. Comparisons of time-series for each scenario showed that reduced nest-clump availability, low precipitation, and high temperature reduced bobwhite densities to a larger degree compared to a natural nest predation rate. Reduced nest-clump availability resulted in the most substantial decline of simulated bobwhite densities. Simulations suggested that management efforts should focus on maintaining adequate nest-clump availability and then possibly consider nest predator control as a secondary priority. ß 2011 The Wildlife Society. KEY WORDS bobwhite, Colinus virginianus , drought, nesting habitat, nest predation, population, predator control, simulation, south Texas, stochastic modeling. Northern bobwhite ( Colinus virginianus ) populations are potentially limited by many factors, including predation, weather, and habitat (Stoddard 1931, Lehmann 1984, Roseberry and Klimstra 1984, Guthery 2002). In south Texas bobwhites exist in a unique ecological context charac- terized by a diverse predator community, recurrent drought, extensive and contiguous semiarid-subtropical rangeland subject to livestock grazing, and fee-lease hunting (Lehmann 1984; Herna ́ndez et al. 2002; Rader et al. 2007 b ). South Texas is one of the few regions where bob- white populations have not experienced significant long- term declines in North America (Brennan 1991, Church et al. 1993, Link et al. 2008). Recent trends indicate large cattle ranches characteristic of south Texas have increasing economic incentives to emphasize sale of bobwhite and white-tailed deer ( Odocoileus virginianus ) hunting opportu- nities over livestock production. This development has resulted in economic incentives to maintain rangeland in a condition beneficial to bobwhites and a variety of other wildlife (Fulbright and Bryant 2002). Numerous biotic and abiotic factors influence bobwhite production in South Texas. Nest predation is the main cause of bobwhite nest failure and is a recurrent management concern for bobwhites and other ground-nesting galliformes (Stoddard 1931, Klimstra and Roseberry 1975, Lehmann 1984, Hurst et al. 1996, Rollins and Carroll 2001). Apparent bobwhite nest losses to predation generally fall between 37% and 76% (Lehmann 1946, Simpson 1976, Staller et al. 2005, Rader et al. 2007 a , Sandercock et al. 2008). Thus, there is evidence that predation might limit potential bobwhite pro- duction, but the impact of nest predation on fall and spring bobwhite densities is not clear. In fact, predator removal studies have been shown to have minimal positive effects on south Texas bobwhite populations (Beasom 1974, Guthery and Beasom 1977, Lehmann 1984). Sandercock et al. (2008) observed that nest success was fourth in overall importance after chick survival, adult summer, and adult winter survival, with respect to annual population growth. Nesting habitat is another variable potentially amenable to management and a perennial concern for managers in south Texas (Kiel 1976, Guthery 1997, Kopp et al. 1998). Numerous studies have documented bobwhite nest-site selection and susceptibility to predation (Spears et al. 1993, Guthery 1997, Received: 30 April 2010; Accepted: 12 May 2010 1 Present Address: Wisconsin Department of Natural Resources, 5301 Rib Mountain Drive, Wausau, WI 54401, USA. 2 E-mail: leonard.brennan@tamuk.edu 3 Present Address: P.O. Box 1370, Lampasas, TX 76550, USA. Journal of Wildlife Management 75(1):61–70; 2011; DOI: 10.1002/jwmg.4 Rader et al. Bobwhite Population Model 61 Herna ́ndez et al. 2003, Lusk et al. 2006, Rader et al. 2007 a ). Additionally, other studies have highlighted potential delete- rious effects of grazing on bobwhite habitat and production (Cantu and Everett 1982, Wilkins and Swank 1992). Drought is a recurrent feature in south Texas and is thought to be the driving mechanism of the characteristic boom-bust phenomenon exhibited by bobwhite populations in the region (Lehmann 1984, Lusk et al. 2001, Herna ́ndez et al. 2002, Herna ́ndez et al. 2007). Both reduced precipi- tation (Kiel 1976, Bridges et al. 2001, Herna ́ndez et al. 2005) and high operative temperatures (Guthery et al. 1988, 2001, 2005; Forrester et al. 1998) at the nesting substrate or ground level associated with drought negatively impact bobwhite productivity in south Texas. Previous studies in south Texas have not examined the relative impact of nest predation, precipitation, temperature, and nest- ing-habitat availability on bobwhite population dynamics using a systems approach with simulated data. A critical gap in knowledge is the magnitude of the extent that—and under what circumstances—nest predation is potentially limiting to bobwhite populations whose production is simultaneously impacted by weather and nesting-habitat availability in the semiarid, subtropical-rangeland ecosystem of south Texas. Simulation modeling using a systems approach (Grant et al. 1997) is an analytical technique that provides investigators the ability to examine numerous scenarios for a biological system and assess how the system responds to these scenarios. As such, a systems approach to simulation modeling offers a potentially powerful tool for examining the complex inter- relationships of the nest predation process under different management and environmental scenarios. For most game birds it is economically and logistically difficult, if not impossible, to replicate predator management and associated environmental scenarios related to heat and drought using controlled and replicated field experiments. A systems approach to simulation modeling can help improve under- standing of complex systems and allow managers to better understand interactions and relationships among the biotic and abiotic factors that influence a demographic parameter such as nest success and resulting annual changes in popu- lation density. For example, Martinez et al. (2004) used simulation modeling with STELLA software (High Performance Systems, Inc., Lebanon, NH) to clarify the role of nest success in relation to annual, long-term pro- duction of white-winged doves ( Zenaida asiatica ) in the Tamulipan biotic province. Likewise, simulations have been used to analyze bobwhite population response to exploitation (Roseberry 1979), the cause of gray partridge ( Perdix perdix ) population declines (Potts 1986), effect of predators on red grouse ( Lagopus lagopus ) (Macdonald et al. 1999), and prong- horn ( Antilocapra americana ) response to coyote ( Canis latrans ) control (Phillips and White 2003). Such analytical approaches are an excellent way to help understand dynamics of a complex system, as well as develop and refine a series of successive approximations that can provide a basis for estab- lishing management priorities or conducting further field investigations (Sandercock et al. 2008). Our objective for model simulations was to evaluate relative bobwhite population response to nest predation and predator control under historical conditions, low precipitation, high temperature, and poor nesting-cover availability in south Texas. We did not pose a specific research hypothesis at the outset of this study. Rather, we sought to identify how nest predation may affect quail production and abundance in relation to variability in weather and habitat. METHODS Our model represents the dynamics of a hypothetical bob- white population, including abiotic and biotic variables influ- encing nest failure, mortality, production, and recruitment (Fig. 1). We incorporated 3 life stages (i.e., eggs, juv, and ad) in bobwhite life history in the model to allow specific exam- ination of the effect of nest predation on the population, as well as to denote the variable mortality rates experienced at Figure 1. Flow chart and conceptual model representing the relationships and effects of nest predation, precipitation, temperature, and nest-clump availability on adult bobwhite density in a hypothetical, subtropical, rangeland ecosystem in south Texas. 62 The Journal of Wildlife Management 75(1) 19372817, 2011, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.4 by University Of Florida, Wiley Online Library on [24/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License each life stage. Bobwhite egg production was a function of precipitation, temperature, and adult density (bobwhites/ha). Egg (i.e., nest) failure was a function of nest predation rate and abandonment rate. Eggs not lost to failure or predation were recruited into the juvenile category. We alternatively incorporated juvenile mortality as a constant and as a func- tion of nest-clump availability. Juveniles not succumbing to mortality were recruited into the adult age class, where, for modeling purposes, we incorporated mortality as a constant. Although there is emerging evidence that bobwhite mortality may occur in pulses during the year, rather than at a constant annual rate (Burger et al. 1995), we used the medians of constant juvenile and adult mortality rates reported by Guthery (2002) as a first approximation for our model. Model Details We based our simulations on a mathematical, discrete-time, stochastic compartment model that used difference equations with a seasonal (i.e., 3-month) time step. We performed simulations with STELLA Software (Version 8.0). We calculated temporal dynamics of state variables representing eggs (EG), juveniles (JV), and adults (AD) as: EG ð t þ 1 Þ ¼ EG ð t Þ þ ð PRD FLR RC1 Þ JV ð t þ 1 Þ ¼ JV ð t Þ þ ð RC1 RC2 MY1 Þ AD ð t þ 1 Þ ¼ AD ð t Þ þ ð RC2 MY2 Þ where we measured EG, JV, AD, production (PRD), failure (FLR), juvenile recruitment (RC1), adult recruitment (RC2), juvenile mortality (MY1), and adult mortality (MY2) in terms of individuals, and ( t ) represents 3-month time steps. Production.– In our model, we assumed that production (i.e., egg laying) occurred only during summer (May to Aug). Though bobwhites can nest throughout the year in south Texas, most nesting typically occurs from late spring (May) through summer (Aug; Lehmann 1984). The model allowed for multiple clutches per female by incorporating up to 2.3 clutches per female in a summer as a function of annual precipitation (Fig. 2a). We calculated production as: PRD ¼ IHN NH HNP SL CS where we multiplied the initial number of females nesting (IHN), nests per female (NH), final percentage of females nesting due to precipitation (HNP), season length (SL), and clutch size (CS) to give the total number of eggs produced/ female during the nesting season. Initial proportion of females nesting (IHN) was the product of the total number of adults (AD), proportion of breeding females (i.e., 0.45 [Guthery 2002:70]), and a nesting index (NI): IHN ¼ AD 0 : 45 NI ; where NI described the proportional bobwhite population increase as a function of bobwhite density (bobwhite/ha; Figure 2. Parameters we used in the STELLA model: (a) Nests produced per bobwhite female as a function of annual precipitation in south Texas. b: Proportional population increase as a function of bobwhite density in south Texas. c: Proportion of bobwhite females nesting as a function of annual precipitation in south Texas. d: Proportion of realized nesting season as a function of mean maximum July to August temperatures for bobwhites in south Texas. e: Juvenile bobwhite survival as a function of nest-clump availability in south Texas. Rader et al. Bobwhite Population Model 63 19372817, 2011, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.4 by University Of Florida, Wiley Online Library on [24/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Fig. 2b). We incorporated density-dependence into the model because of its existence in wild bobwhite populations and to provide a check on unlimited population growth (Errington 1945, Roseberry and Klimstra 1984, Guthery 2002). We assumed linear density-dependence, in keeping with the logistic model of population growth (Gotelli 1995:33), and a maximum carrying capacity in excellent bobwhite habitat of 7.4 bobwhite/ha (derived from Kellogg et al. 1972). We incorporated annual precipitation into the model as a stochastic variable based on long-term (i.e., 1908–1997) historic data from Brooks County, Texas (National Climatic Data Center 2003), which was the region described by the model. We represented the natural variability inherent to the system by randomly varying annual precipitation based on a normal distribution ( x ¼ 61.59 cm, SD ¼ 19.51 cm). Empirical evidence from south Texas (Herna ́ndez et al. 2005) suggests that NH and HNP are a function of mean annual precipitation, which we incorporated into the model (Fig. 2a,c). We assumed effective breeding season length (SL) was a function of mean maximum July to August temperature (Fig. 2d), derived from Guthery et al. (1988). Mean maxi- mum July to August temperature was a stochastic variable in the model that varied randomly from a normal distribution ( x ¼ 36.9 8 C, SD ¼ 1.3 8 C) based on long-term (1907– 2002) historical data from the region described by the model (NCDC 2005). Although clutch size in birds tends to decrease with sub- sequent nesting attempts during the breeding season, for sake of simplicity we considered clutch size (CS) as a con- stant in the model. We assumed CS to be 12.7 eggs/female (Guthery 2002:70). Nest failure.– Nest failure (i.e., total no. of eggs lost) was the sum of the nest-abandonment rate (ABR) and the nest- predation rate (NPR), multiplied by the total number of eggs produced during the nesting season: FLR ¼ ð ABR þ NPR Þ EG Bobwhites will abandon nests due to a variety of factors, including human interference and sustained, extreme heat (i.e., > 38.7 8 C; Lehmann 1984). Our model assumed a con- stant abandonment rate of 0.14, which is the mean abandon- ment rate observed from a 2002 to 2004 nest predator study in south Texas (Rader 2006). Nest-predation rate was a stochastic variable in the model. We assigned a mean nest-predation rate for each nesting season randomly from a range of 0.35 (Rader 2006; n ¼ 118 nests) to 0.46 (Lehmann 1984; n ¼ 532 nests) in the baseline model. All eggs not relegated to failure were recruited as individuals into the juvenile population for the following model time step. Mortality.– We assumed a baseline juvenile-mortality rate of 0.36 per season, derived from a daily-survival rate of 0.995, which is the median value of the range (i.e., 0.993–0.997) suggested by Guthery (2002:51). Further, we wished to examine effects of variable juvenile mortality on bobwhite population dynamics in relation to alternative nest-clump availability scenarios. We assumed that nesting habitat influ- enced bobwhite productivity in the model (see Taylor et al. 1999, Townsend et al. 2001, Lusk et al. 2006). Specifically, we derived a functional relationship between age ratios and nest-clump availability (Fig. 2e) based on a preliminary model (third-order polynomial; r 2 ¼ 0.654) of bobwhite age ratios versus the number of suitable nest clumps/ha, developed from 2 yr of empirical data in south Texas (Brazil 2006). We derived juvenile survival from age ratios using the formula described in Lusk et al. (2005:391): s j ¼ s a R f R h ; where s j ¼ survival rate of chicks for 3 months, s a ¼ survival rate of adult females for 3 months, R f ¼ age ratio (juv/ad F) in fall, R h ¼ age ratio (juv/ad F) at time of hatch. Data for estimating s a and R f came from Brazil (2006) and for R h came from Guthery (2002). We calculated s a from age ratios using the formula provided by Guthery (2002): S a ¼ 1 ð 1 þ R Þ ; where R is the age ratio. The variable R h was the product of (nesting attempts/ F ¼ 3) (average chicks at hatch/attempt ¼ 12.7) (probability of nest success ¼ 0.30), using the formula from Lusk et al. (2006:391) and values from Guthery (2002:70). Juveniles not succumbing to mortality were recruited into the adult population during the following winter or spring time step. We assumed adult mortality was constant at 0.26 per season, a coefficient derived from a daily-survival rate of 0.9967 (Guthery 2002:49). To convert daily-mortality rates to quarterly mortality rates, we calculated: L ¼ 1 S t where L is the rate at the larger time step, S is rate at the smaller time step, and t is number of smaller time steps in the larger time step. Model Evaluation We evaluated model performance by comparing predictions of the temporal dynamics of bobwhite age ratios from 30 replicate, 28-yr stochastic simulations (initial density of 3.7 bobwhite/ha) with age ratio data from a 28-yr study in south Texas (Lehmann 1984:133). We used the age ratio outputs in conjunction with predictions of abundance to evaluate overall model performance. To further assess whether we could be confident in model predictions, we conducted a sensitivity analysis on a deter- ministic version of the model. We made the model deter- ministic by assigning stochastic variables their mean values (NPR ¼ 0.35, mean annual precipitation ¼ 61.59 cm, mean max. Jul to Aug temp ¼ 36.9 8 C). We analyzed model sensitivity to changes in parameter values individually for nest predation rate, precipitation, temperature, and nest clumps/ha, as we considered these key driving variables that determined production and recruitment from the egg to 64 The Journal of Wildlife Management 75(1) 19372817, 2011, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.4 by University Of Florida, Wiley Online Library on [24/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License adult life-stages and therefore were key determinants of the effect of nest predation on bobwhite population dynamics in south Texas. Our analysis consisted of running individual simulations for 3 variations in each of the indicated variables to examine the effects of alterations on bobwhite density over 5 yr, ostensibly to examine the short-term impacts of model performance. We selected ranges of values based on mini- mum and maximum values derived from empirical data. Simulations of Management and Environmental Scenarios We ran 8 series of 30-replicate, 10-yr, stochastic simulations (initial density of 3.7 bobwhite/ha) representing: 1) baseline historical conditions (A1), 2) predator control (B1), 3) low precipitation (A2), 4) low precipitation with predator control (B2), 5) high temperature (A3), 6) high temperature with predator control (B3), 7) reduced nest-clump availability (A4), and 8) reduced nest-clump availability with predator control (B4). The rationale for simulation of a 10-yr time span was to: 1) identify whether impacts of predator manage- ment could be observed over a short time period, and 2) a 10-yr perspective is a typical planning window used by quail managers in south Texas. For baseline scenario A1, we assumed that mean annual precipitation and mean maximum July to August tempera- ture varied stochastically based on historical data, and that the nest predation rate varied randomly between 0.35 and 0.46. For predator control scenario B1, we assumed 100% elimination of coyote, skunk ( Mephitis mephitis ), and badger ( Taxidea taxus ) nest predators (i.e., NPR ¼ 0.15), the 3 top bobwhite nest predators in south Texas amenable to traditional predator control (Rader 2006). The 100% elim- ination of these 3 predators represents an overall 57% reduction in losses to nest predators in the context of bob- white nesting ecology in south Texas (Rader 2006, Rader et al. 2007 b ); all other model variable values remained unchanged. For low precipitation scenario A2, we simulated drought during the nesting season by reducing annual pre- cipitation to 50 cm (derived from Herna ́ndez et al. 2005), with a standard deviation of 19.51 cm; remaining model variables were unchanged. Scenario B2 (low precipitation with predator control) was identical to A2 with the addition of simulated predator control as described for B1; all other model variable values remained unchanged. For high temperature scenario A3, we simulated extreme heat during the nesting season by increasing mean maximum July to August temperature by 1 standard deviation (i.e., 1.3 8 C) from historical conditions to 38.2 8 C. Temperature remain- ed a stochastic variable with a standard deviation of 1.3; remaining model variables were unchanged. Scenario B3 (high temperature with predator control) was identical to high temperature scenario A3 with the addition of predator control as described for B1; remaining model variables were unchanged. For scenario A4, we simulated reduced nest- clump availability (i.e., 200 nest clumps/ha); other model variables remained as in scenario A1. Scenario B4 (reduced nest-clump density with predator control) was identical to A4, but with the addition of predator control as described for B1. RESULTS Model Evaluation We observed no meaningful differences between simulated and empirical age ratios. A time-series comparison of the median age ratios for each year of the 30 replicate, 28-yr simulations versus Lehmann’s (1984) data showed that most (25 of 28 yr or 89%) of empirical values fell within the 5th and 95th percentiles of simulated data over the 28-yr period (Fig. 3). In fact, when we considered the entire simulated data set, all empirical age ratio values fell within the range of simulated values. That simulated age ratios corresponded with those observed for south Texas provided confidence that the model effectively simulated bobwhite productivity and mortality, based on the ecological factors that drive bobwhite population parameters over the south Texas por- tion of their geographic range. Sensitivity results suggested that mean or baseline values for the selected variables did not produce remarkable densities—either high or low—but indicated the variability (i.e., the high and low extremes) in the parameter values produced the largest magnitude of effect. Our sensitivity results indicated that predictions of bobwhite density over 5 yr using a deterministic form of the model were more sensitive to changes in precipitation and temperature than to nest-clump availability or nest-predation rate (Fig. 4). High precipitation (mean annual precipitation ¼ 90 cm) and mild temperature (mean max. Jul to Aug temp ¼ 35.0 8 C) scenarios produced the greatest increases in bob- white density, resulting in densities > 5–6 bobwhite/ha (Fig. 4a,b). Changes in nest-clump availability and nest- predation rate produced a lower magnitude of density response (Fig. 4c,d). Bobwhite densities under a heavy nest-predation scenario (NPR ¼ 0.50; Fig. 4d) declined to a level similar to the low precipitation (mean annual precipitation ¼ 50 cm; Fig. 4a) and hot temperature (mean max. Jul to Aug temp ¼ 37.2 8 C; Fig. 4b) scenarios. Bobwhite density increased approximately 40%—but only to approximately 4 bobwhite/ha—under simulated predator control (NPR ¼ 0.15; Fig. 4d). 0 1 2 3 4 5 6 7 8 9 10 1 3 5 7 9 11 13 15 17 19 21 23 25 27 Years Age Ratio (juveniles/adult) 5th percentile Median 95th percentile Lehmann Figure 3. Observed (1940–1976; Lehmann 1984:133) and simulated (this study) bobwhite age ratios in south Texas. Graph represents median and 5th to 95th percentiles of 30 stochastic simulations. Rader et al. Bobwhite Population Model 65 19372817, 2011, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.4 by University Of Florida, Wiley Online Library on [24/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Simulation of Management and Environmental Scenarios Outputs from the model revealed a high range of variability and differences in median bobwhite density for 10 yr, among the scenarios we evaluated (Fig. 5). For example, predator control, low precipitation, extreme heat, and reduced nest- clump availability altered bobwhite densities (range ¼ 0.4– 1.8 bobwhite/ha) compared to the baseline scenario. Reduced nest-clump availability had the greatest magnitude of effect, lowering median bobwhite densities by 1.8 bob- white/ha (75%) from the baseline scenario. Predator control increased density by 1.3 bobwhite/ha (54%) from baseline conditions. Predator control also apparently mitigated effects of low precipitation and extreme heat, but these effects were not as pronounced for the reduced nest-clump availability scenario. Over time, bobwhite median densities for all scenarios, except predator control, showed gradual declines from the initial density of 3.7 bobwhite/ha (Fig. 6a). The decline was most evident for the reduced nest-clump availability scenario. It is important to note, however, these results provide no insight into the range or variability in bobwhite densities under each scenario (Fig. 6a). Simulating Potential Efficacy of Predator Control To examine the range of potential population outcomes, both with and without predator control, it is more instruc- tive—and perhaps more biologically meaningful—to examine percentiles of bobwhite density distributions through time after the initial year of simulation. The effects of predator control on bobwhite densities fluctuated through time relative to baseline densities, giving rise to an apparent cyclical pattern of unknown origin (Fig. 6b). It appears that over time—especially from approximately 7 yr on—bobwhite densities under predator control were substantially higher than baseline densities. For example, bobwhite populations subject to predator control relative to baseline populations, increased by 57% at 3 yr, were virtually indistinguishable at 5 yr, and increased by 132% at 8 yr. Similarly, the baseline scenario generally did not estimate densities much > 4 bob- bobwhite/ha, whereas under predator control, densities at times reached 6 bobwhite/ha or greater. Model results suggested that predator control could have a significant positive effect on simulated bobwhite densities, but that this effect was not constant over a 10-yr period. Examination of low precipitation versus low precipitation and predator control scenarios showed a noticeable predator- control effect on bobwhite density at approximately 2.5 yr (Fig. 6c). Simulated drought conditions (scenario A2) caused 0 1 2 3 4 5 6 7 0.00 1.00 2.00 3.00 4.00 5.00 Years Density (quail/ha) low (50 cm) mean (61.59 cm) high (90 cm) 0 1 2 3 4 5 6 0.00 1.00 2.00 3.00 4.00 5.00 Years Density (quail/ha) mild (35.0 °C) mean (36.9 °C) hot (37.2 °C) 0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 Years Density (birds/ha) Baseline 1000 clumps/ha 200 clumps/ha 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0 1 2 3 4 5 Years Density (quail/ha) predator control (0.15) mean (0.35) heavy (0.50) a b d c Figure 4. Results of deterministic model sensitivity analysis. a: Bobwhite density response over 5 yr to 3 levels of annual precipitation. b: 3 levels of mean maximum July to August temperature. c: 3 levels of nest-clump availability. d: 3 levels of nest predation. Scenario B4 A4 B3 A3 B2 A2 B1 A1 Median Density 7 6 5 4 3 2 1 0 -1 Figure 5. Median bobwhite population densities (quail/ha) for each of the 8 30-replicate scenarios we ran in stochastic model simulations. For each box plot, dark lines represent medians, boxes represent interquartile range, and circles represent outliers. Scenarios: A1 ¼ baseline conditions, A2 ¼ low precipitation, A3 ¼ high temperature, A4 ¼ reduced nest-clump availabil- ity, B1 ¼ baseline conditions, predator control, B2 ¼ low precipitation, predator control, B3 ¼ high temperature, predator control, and B4 ¼ reduced nest-clump availability, predator control. 66 The Journal of Wildlife Management 75(1) 19372817, 2011, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.4 by University Of Florida, Wiley Online Library on [24/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License a marked decline in densities, which generally declined for the duration of simulations to < 1 bobwhite/ha. Population fluctuations were more variable for scenario B2 (low precipi- tation with predator control) versus scenario A2 (low pre- cipitation only), with densities reaching as high as 4 bobwhite/ha in the upper 25% of simulations. Again, the effect was more pronounced over time, with the greatest magnitude of effect being observed from approximately the 3-yr point forward. High temperature (scenario A3) seemed to have the same general effect on bobwhite densities as did low precipitation (scenario A2) in the model over the 10-yr simulation period (Fig. 6d). Median densities generally declined from an initial density of 3.7 bobwhite/ha to approximately 1 bobwhite/ha. As in the low precipitation with predator control scenario B2, model results suggested that predator control potentially mitigated against the effect of high temperature and main- tained median densities at approximately 3 bobwhite/ha (the upper 25% of simulations reached 4 bobwhite/ha). For scenario A3 (high temp only), 75% of simulations never exceeded approximately 2 bobwhite/ha after 5 yr, whereas 75% of scenario B3 (high temp with predator control) simulations exceeded approximately 2 bobwhite/ha. As in scenario A2 (low precipitation only) versus scenario B2 (low precipitation with predator control), the predator control effect appeared more pronounced through time. Our results indicated that it took approximately 3 yr for a noticeable difference between scenarios to become apparent. As in low precipitation scenario A2 versus the low precipitation with predator control scenario B2, the predator control scenario (B3) was more variable than the high temperature scenario (A3). Similar to the low precipitation scenarios (A2 vs. B2), predator control seemed to have an earlier, greater, and more consistent positive effect on median bobwhite densities during prolonged environmental stress (i.e., high temp for 10 yr) as compared to baseline conditions. The reduced nest-clump scenario (A4) showed the most conspicuous decline (a 75% reduction) from baseline bob- white densities over 10 yr, as compared to the other scenarios (Fig. 6e). Simulated predator control with reduced nest- clump density (B4) compensated for reduced nest-clump availability, but the trend of population decline remained unchecked. By year 4, median bobwhite densities were < 1 bobwhite/ha without predator control. With control, 0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 10 Years Density (quail/ha) A1 B1 A2 A3 A4 0 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 10 Years Density (quail/ha) 25th percentile median, A1 75th percentile 25th percentile median, B1 75th percentile 0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 10 Years Density (quail/ha) 25th percentile median, A2 75th percentile 25th percentile median, B2 75th percentile 0 1 2 3 4 5 6 1 2 3 4 5 6 7 8 9 10 Years Density (quail/ha) 25th percentile median, A3 75th percentile 25th percentile median, B3 75th percentile a b d c e 0 1 2 3 4 5 1 2 3 4 5 6 7 8 9 10 Years Density (quail/ha) 25th percentile median, A4 75th percentile 25th percentile median, B4 75th percentile Figure 6. a: Median bobwhite densities over 10 yr for 5 stochastic ( n ¼ 30) model scenarios. A1 ¼ baseline conditions, B2 ¼ baseline conditions, predator control, A3 ¼ high temperature, A2 ¼ low precipitation, and A4 ¼ poor nesting cover. b: Percentiles of distributions of bobwhite density over 10 yr for 2 30- replicate, stochastic model scenarios in south Texas. A1 ¼ baseline conditions, B1 ¼ baseline conditions, predator control. c: Percentiles of distributions of bobwhite density over 10 yr for 2 30-replicate, stochastic model scenarios in south Texas. A2 ¼ low precipitation, B2 ¼ low precipitation, predator control. d: Percentiles of distributions of bobwhite density over 10 yr for 2 30-replicate, stochastic model scenarios in south Texas. A3 ¼ high temperature, B3 ¼ high temperature, predator control. e: Percentiles of distributions of bobwhite density over 10 yr for 2 30-replicate, stochastic model scenarios in south Texas. A4 ¼ reduced nest-clump availability, B4 ¼ reduced nest-clump availability, predator control. Rader et al. Bobwhite Population Model 67 19372817, 2011, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.4 by University Of Florida, Wiley Online Library on [24/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License densities did not decline below 1 bobwhite/ha until year 9. By year 10, median bobwhite densities fell to 0.2 bobwhite/ ha without control, whereas the addition of predator control kept densities at a level just < 1 bobwhite/ha. DISCUSSION Factors Limiting Bobwhite Production Although our model representations of the relationships among bobwhite density, nest predation, precipitation, temperature, and nest-clump availability were not necessarily exact, empirical data from south Texas, and our comparisons of simulated versus observed age-ratio data, indicate that the ranges of bobwhite productivity estimated by the model were reasonable for south Texas. Thus, we were confident in the ability of the model to predict potential differences among the alternative scenarios we evaluated. Simulations indicated that reduced nest-clump availability, low precipitation, and high temperature all reduced median bobwhite densities below baseline densities. Such results suggest that these variables may be of more important man- agement concern than nest predation alone. This is sup- ported by several bobwhite field studies. For example, Errington (1945) concluded that predation had minimal impact on bobwhite populations unless they exceeded carry- ing capacity of available habitat. Lehmann (1984) speculated that reduction of bobwhite nest predation could not com- pensate for poor breeding habitat during drought in south Texas. Manipulative predator reduction field experiments by Beasom (1974) and Guthery and Beasom (1977) also support the contention that nest predator reduction has a small impact on bobwhite population density in south Texas. Additionally, Guthery (2002) and Guthery et al. (2001) reasoned that usable habitat space and operative temperature at nest-level—of particular relevance in semiarid, south Texas—may largely explain bobwhite density variation. Efficacy of Simulated Nest Predator Control Our simulation results assumed intensive (i.e., a 57% reduction in nest predation rate) and long-term (i.e., up to 10 yr) predator control. Is this realistic and cost-effective? It is likely that 100% elimination of coyote, skunk, and badger nest depredation—as assumed in the model—is impossible. Additionally, removing one set of risk factors can potentially make other risk factors such as losses to snakes, or perhaps raptors, more important. However, an overall simulated annual nest predation reduction rate of 57% seems reasonable, at least as a first approximation, as we have no information about the actual reduction of predation rates from the experiments conducted in South Texas by Beasom (1974) and Guthery and Beasom (1977), nor do we have data on recolonization rates of mammalian predators after reduction or removal from these or any other studies. From an economic standpoint, the cost of personnel and equipment involved in a full-time, long-term predator-con- trol program on the large ( > 5,000 ha) ranch properties in south Texas would likely be substantial. Nevertheless, the question persists: Is mammalian nest predator control effective in increasing bobwhite densities, and, if so, under what circumstances? Our model results suggested that predator control was capable of increasing bobwhite densities above baseline-historical conditions with no predator control to median densities near 4.5 bobwhite/ ha (i.e., a 55–132% increase), and that control during simu- lated periods of low precipitation, high temperature, and reduced nest-clump availability, could also more modestly increase densities. Trautman et al. (1974) found that intensive, multi-species (i.e., red fox [ Vulpes vulpes ], raccoon [ Procyon lotor ], badger, and skunk [striped ( Mephitis mephitis ) and spotted ( Spilogale putorius )]) predator control resulted in up to a 132% increase in ring-necked pheasant ( Phasianus colchicus ) densities from 1967–1971 in South Dakota. Similarly, predator control was found to increase the size of a simulated (Potts 1986) and actual (Tapper et al. 1996) gray partridge populations in Britain. In particular, Tapper et al. (1996) found predator control increased partridge numbers by 75% and 36% during late summer and spring, respectively. Furthermore, Lawrence and Silvy (1