Spring migration is an important life stage for ducks because their ability to find and acquire nutrients can affect subsequent reproductive success. Therefore, providing sufficient habitat to support the energetic needs of ducks and facilitate efficient feeding is a goal of habitat management and restoration. The rapid, unpredictable flood events that regularly occur in highly modified landscapes can make habitat management challenging and justify diverse management strategies. We examined the effect of habitat management on dabbling duck behavior and distribution during spring migration in southwest Indiana. We investigated three management options for wetlands: active management, passive management, and unmanaged agricultural food plots. We assessed duck behavior and density on 14 wetlands at Patoka River National Wildlife Refuge and Management Area. The agricultural food-plot areas had the lowest estimates of food availability followed by the actively managed areas; the passively managed wetlands had the greatest estimate. Dabbling duck density was greatest on the actively managed wetlands followed by food plots coming in second and passively managed wetlands third. Most dabbling ducks fed more intensively while on the passively managed wetlands followed by the actively managed and food-plot wetlands. Conservation prioritization of passively managed areas would provide larger areas for dabbling ducks to feed, but active management provides habitat regardless of climatic variability. Moving forward, wetland complexes encompassing diverse wetland management approaches would be the best option for spring-migrating waterfowl as these complexes can provide high-quality habitats and buffer against uncontrollable climactic conditions.
The intensity of wetland management for ducks ranges from sites with active management of the hydrology and vegetation to sites with passive or even no management. Active management (sometimes referred to as moist-soil management) generally consists of actively controlling hydrology in the wetland through levees and water control structures, thereby affecting the soils and vegetation (Fredrickson and Taylor 1982; Gray et al. 1992). Typically, habitat managers drain these areas in the spring and summer to facilitate growth of early successional wetland plants that produce abundant seeds, followed by flooding in the fall to provide habitat to migrating and wintering waterfowl (Fredrickson and Taylor 1982). Actively managed wetlands are expensive with initial construction costs ranging from $500 to $37,000/ha plus ongoing mechanical and maintenance costs (Lane and Jensen 1999; Pankau 2008).
Less intensive management strategies are generally much cheaper than active management over the long run as most cost involves the conservation of these lands (Lane and Jensen 1999; Pankau 2008). Passively managed lands are areas that depend on natural water level fluctuations for inundation; however, these areas may have low levees that hold water after initial inundation (Brasher et al. 2007; Pankau 2008; Stafford et al. 2011). Managers can manipulate these areas by planting into row crops, seed-producing perennials, or a combination of the two. Keeping the land in agriculture prevents succession of woody plants and shrubs typical of many passively managed areas. Inundation of passively managed wetlands results in the creation of large foraging areas. Wetlands that go unmanaged are areas where no restoration work has occurred and water level and food abundances are rarely anthropogenically altered (Lane and Jensen 1999; Pankau 2008). Both passive and unmanaged wetlands may depend on unreliable water sources making them less seasonally predictable as a habitat for migrating waterfowl than actively managed wetlands (Brasher et al. 2007; Pankau 2008; Stafford et al. 2011).
Habitat quality assessments regularly inform management strategies for wetlands focused on providing waterfowl habitat. Managers generally assume that food or energy density is the most limiting and thus the most important determinant of habitat quality for nonbreeding waterfowl (Soulliere et al. 2007). Directly measuring food density and other determinants of habitat quality is labor intensive relative to monitoring the distribution of animals; therefore, managers often measure animal use or distribution as an index of habitat quality (Brasher et al. 2007; Pankau 2008). Animal distribution, however, may not reliably indicate habitat quality (Gates and Gysel 1978; Van Horne 1983; Morrison 2002; Pidgeon et al. 2003; Ries and Fagan 2003; Johnson 2007). For example, because managers assume food availability to be the most limiting resource for waterfowl during spring migration, they also assume that waterfowl distribute themselves based primarily on the distribution of food (Soulliere et al. 2007). However, factors other than food density such as predator avoidance, inter- and intraspecific competition, and microclimate also influence bird distributions (Pulliam 2000; Scott et al. 2002). These factors may result in waterfowl using habitats with lower food densities (Morrison 2002). Predation risk evokes a trade-off between vigilance and feeding behavior in foraging animals (Brown 1999) and during an individual foraging bout, the proportion of time an individual spends feeding relative to being vigilant likely depends on both forage density and perceived predation risk (Schoener 1971; Hill and Ellis 1984; McNamara and Houston 1994; Verdolin 2006; Bednekoff 2007; Behney et al. 2018). Because direct assessment of food availability and predation risk is difficult, many birds use habitat characteristics, such as cover, to judge the predation risk and potential food density of a habitat, and therefore the quality of that habitat (van der Wal et al. 1998; Rowcliffe et al. 1999). Previous research has shown mixed results on the effect of cover on predation risk (Davis 1973; Underwood 1982; Behney et al. 2018), and therefore assessing both distribution and behavior can provide insight into perceptions of animals. Behavioral observations may be particularly useful for migratory populations whose numbers are variable and where scheduled bird counts may miss large influxes of birds (Webster et al. 2002).
Waterfowl eat a variety of foods during migration including invertebrates, natural seeds, plant fragments, and waste grain (Pearse et al. 2011). Previous literature suggests that natural, unmanaged wetlands provide less food relative to managed wetlands. In the Upper Mississippi River and Great Lakes Region, Straub et al. (2012) found average food density during the spring was 208 kg/ha for shallow semipermanent and deep marshes. The Upper Mississippi River and Great Lakes Region Joint Venture uses 188 kg/ha (Soulliere et al. 2007) in their biological planning models, which are used to help establish habitat objectives in the region for migrating waterfowl. (Joint Ventures are partnerships established under the North American Waterfowl Management Plan to help conserve the continent's waterfowl populations and habitats.) Most available food estimates are from research conducted in the fall; however, food depletion and decomposition may occur during the fall and winter, leaving lower food densities during spring migration (Stafford et al. 2006; Greer et al. 2009).
Our objective was to increase our understanding of how local management actions influence quality of habitat for migratory waterfowl. We tested the hypothesis that more intensive management leads to higher-quality habitat by comparing the behavior, density, and food availabilities of spring migratory waterfowl across three levels of wetland management intensity. We predicted that actively managed wetlands would have greater food densities than passively managed wetlands. Furthermore, we predicted that waterfowl would be most abundant and feed most intensively on actively managed sites due to the importance of energy acquisition during spring migration.
We conducted this study on the Patoka River National Wildlife Refuge and Management Area (hereafter Refuge) in Pike and Gibson counties in southwestern Indiana. The Refuge encompasses 2,671 ha with an ultimate acquisition boundary of 9,094 ha. The Patoka River has an extensive history of hydrologic alterations (USFWS 2008) including extensive ditching and dredging as well as the creation of a 3,200-ha impoundment known as Lake Patoka upstream of the Refuge. These alterations have contributed to the Patoka River having very rapid, unpredictable flood events that inundate large tracts of the Refuge for varying periods of time (H. Hamilton, U.S. Fish and Wildlife Service, personal communication).
We monitored 14 wetlands encompassing three management classifications: six actively managed, three passively managed, and five unmanaged agricultural food-plot wetlands. Our sample of actively managed and food-plot wetlands included all wetlands on the refuge falling into these categories. For passively managed wetlands, we selected the three that were the least influenced by human hydrologic management. The passively managed wetlands had water levels independent of the river, but the river affected them during out-of-bank flood events. Managers have conserved these wetlands to allow them a chance to revert to a naturally functioning wetland complex without major intervention. The agricultural wetlands were located close to a channelized portion of the river and were prone to winter and spring inundation by out-of-bank flood events. A cooperative farming agreement between the U.S. Fish and Wildlife Service and a local cooperator enabled the cooperator to plant a harvestable cash crop such as corn or soybeans on Refuge property and leave 25% of the field in a wildlife food plot split between leaving the cash crop standing, Japanese millet Echinochloa esculenta, and buckwheat Fagopyrum spp.
At each wetland, we surveyed waterfowl once weekly between late January and early April 2012 and 2013. We surveyed waterfowl during a randomly selected time period (morning, midday, or afternoon) to prevent time-associated behavior bias. We avoided evening surveys to prevent lower detection probabilities caused by low light levels and shadows. Each survey consisted of behavior scan samples every 10 min for 1 h (seven samples per survey) limited to ducks (Altmann 1974), and a total waterfowl count. Surveys commenced after a 30-min rest period to reduce observer bias. We used a variable ×20–60 power spotting scope to scan across the wetland and recorded the species and behavior of all ducks present. We randomly chose scan direction. We surveyed from a vehicle if possible to reduce observer bias on bird behavior because vehicle travel around the Refuge was common. If surveying from a vehicle was not feasible, we chose a vantage point that concealed the observer and allowed for a broad view of the wetland. If large concentrations of ducks were present, we subsampled by dividing the wetland extent from the observer's vantage point into quarters and randomly selected two quarters (Hepworth and Hamilton 2001). If ≥ 500 ducks appeared to be present, we divided the wetland extent into eighths and randomly selected two of those eighths for sampling.
We classified behaviors into 11 categories: 1) feeding on surface, 2) feeding with the head underwater, 3) feeding by upending, 4) feeding by diving, 5) resting, 6) courtship, 7) swimming, 8) self-maintenance, 9) aggression, 10) alert, and 11) flying (Pöysä 1983a, 1983b, 1987; Lovvorn 1989; Guillemain et al. 2002; Arzel and Elmberg 2004), but later combined categories into feeding and not feeding. Scan samples underestimate the amount of time that diving ducks spend feeding (Baldassarre et al. 1988). Therefore, we limited analyses of time budgets to eight dabbling duck species that were most prevalent: American black duck Anas rubripes, American wigeon Mareca americana, blue-winged teal Spatula discors, gadwall Mareca strepera, green-winged teal Anas crecca, mallard Anas platyrhynchos, northern pintail Anas acuta, and northern shoveler Spatula clypeata.
We counted waterfowl immediately following the final scan behavior sample. If extensive emergent vegetation was not present on the wetland, we counted ducks from the survey location. On wetlands with extensive emergent vegetation, we counted waterfowl by flushing to account for birds obscured by the emergent vegetation (Pöysä and Nummi 1992).
To assess food density at each site, we collected five core samples at random locations in each wetland at the beginning (late January–early February) of the 2013 field season. Cores were 10.2 cm in diameter and 10 cm deep (Behney et al. 2014). We washed core samples through a 500-μm mesh sieve bucket in the field, preserved the samples with 10% buffered formalin, and transported them to a lab at Southern Illinois University Carbondale for processing. In the lab, we washed samples through 750- and 500-μm sieves, and picked out all invertebrates and seeds. We dried invertebrates and seeds at 60°C for 48 h, and weighed them to the nearest 0.1 mg.
To compare food biomass among management types, we modeled food biomass (dry mass of seeds and invertebrates combined) from each core sample in a normal linear mixed-effects modeling framework in the lme4 package (Bates et al. 2015) in R (R Core Team 2018). We included management type as a fixed effect and site as a random effect. As suggested by Straub et al. (2012), we present median food biomass as well as means. For all analyses, we compared models using an information-theoretic approach (Burnham and Anderson 1998) with Akaike's Information Criterion adjusted for small sample size (AICc) and model weights, and assessed consistency with assumptions as outlined in Zuur et al. (2009).
To assess predictors of duck density, we used normal linear mixed-effects models (Bates et al. 2015; R Core Team 2018) with dabbling duck density observed at each visit as the dependent variable. We log transformed density because it was a positive-only variable (Gelman and Hill 2007) and it resulted in improved normality and homoscedasticity of residuals. We added 0.1 to all densities to avoid zeros in the dataset. We used a multistage modeling strategy in which we first evaluated temporal covariates to account for changes in duck density based on migration stage. We compared a model including day of year in linear form with models including day of year in quadratic and cubic forms, as well as a null model. In the second stage of modeling, we added habitat variables to the best temporal or null model from stage one. We included three models in this stage: 1) the best model from stage one, 2) the best model from stage one plus management type, and 3) the best model from stage one plus food density. We collected food density data only in 2013 and used these estimates for the same sites in 2012 as well. We included site as a random effect in all models to account for repeated visits to the same sites. We included year as a fixed effect in all models because there were only 2 y (Gelman and Hill 2007). We ran models that grouped all dabbling duck species together in addition to running models for eight duck species individually. We back-transformed all model-predicted density values for presentation.
To examine predictors of the proportion of time ducks spent feeding, we used generalized linear mixed-effects models (Bates et al. 2015; R Core Team 2018). We combined behavior classifications into a feeding vs. not-feeding binary response variable for each observation and used a binomial error distribution with a logit link function. We included species, date, management type, food density, and year as fixed effects and site as a random effect. We included interactions between species and management type and species and food density to test whether management type or food affected species differently. We had to standardize the date and food density variables by subtracting the mean and dividing by the standard deviation to achieve model convergence. Raw data are available as supplemental material (Data S1, Supplemental Material).
We collected 70 core samples. The model of food availability containing management type as a fixed effect (wi = 1) outperformed the null model (ΔAICc = 21.2, wi = 0), indicating that food availability varied among wetland management types. In contrast to our prediction, passively managed wetlands had the greatest estimated food density (mean ± SD: 813.8 ± 159.6 kg/ha, median = 836.8 kg/ha), followed by actively managed wetlands (717.9 ± 112.9 kg/ha, median = 639.2 kg/ha) and agricultural food plot wetlands (575.7 ± 123.6 kg/ha, median = 470.2 kg/ha). Seeds made up most of the overall biomass found in samples (passive = 93%, active = 96%, food plot = 98%) and invertebrates made up a small percentage of overall biomass (passive = 7%, active = 4%, food plot = 2%).
During spring 2012 and 2013, we counted 64,966 ducks on our sites at the Refuge. In 2012, we conducted 125 bird counts, 27 of which were flush counts. In 2013, we conducted 97 bird counts, 23 of which were flush counts. Mallards were the most numerous duck recorded (18,203), followed by northern pintail (15,405), green-winged teal (12,240), northern shoveler (9,484), gadwall (7,577), blue-winged teal (1,022), American wigeon (753), and American black duck (282). For all species combined, green-winged teal, mallard, and northern shoveler, the model including management type was more parsimonious than a model including food density, or the best temporal trend or null model (Table 1; Table S1, Supplemental Material). For American black ducks, American wigeon, blue-winged teal, gadwall, and northern pintail, the null model or temporal trend model outperformed the management type and food density models (Table 1; Table S1). For each species that management type affected, densities were greatest in actively managed wetlands followed by agricultural food-plot wetlands (Figure 1).
We made 48,722 observations of dabbling duck behavior. Feeding and alert behaviors dominated activity budgets (71% of all behavioral observations) and were inversely correlated (r = −0.77). There was very little model selection uncertainty in our behavior model set. The model that achieved all the weight included an interaction between species and management type and additive effect of date (Table 2). We found no evidence that food availability directly influenced behavior (Table 2). The relationship between management type and feeding intensity was different for different species (Figure 2). American black ducks fed most intensively in passively managed wetlands, followed by actively managed wetlands, and lastly, food plots (Figure 2). American wigeon and northern shovelers fed most intensively in actively managed and passively managed wetlands and least in food plots (Figure 2). Gadwall and mallard fed most intensively in passively managed wetlands and green-winged teal fed most in intensively in actively managed wetlands, followed by passively managed wetlands, and lastly, food plots (Figure 2). Feeding intensity decreased with date (β ± SE: −0.04 ± 0.01).
Organisms attempt to increase fitness through habitat selection (DeCesare et al. 2014). If food availability has the greatest impact on fitness, then waterfowl should select sites based on food availability. In general, habitat managers have accepted the paradigm that the distributions of spring migratory waterfowl correspond to food resources, with food availability being the most limiting factor for ducks during the migratory period (Soulliere et al. 2007; Stafford et al. 2014). Our results suggest that spring-migrating dabbling ducks using the Refuge distribute themselves according to management type and not solely based on food availability.
One reason that waterfowl may not be distributing themselves based on food availability is that they were unable to locate food-rich patches. Because waterfowl food resources are typically in the benthic layer of wetlands, obscured by both the water column and the wetland substrate, food is probably difficult to locate by sight. Waterfowl likely use vegetation structure or other feeding individuals (Pöysä 1992) as cues to the availability of food. Although these cues may be of general use, they may only weakly correlate to actual food availability, leading to suboptimal habitat selection (Amano et al. 2006). Additionally, our inability to detect a relationship between duck counts and food was most likely not due to a lack of variation in food availability, as the overall range of food availability among sites was 273–1,135 kg/ha.
Alternatively, the assumption that food availability is limiting to the point that it drives habitat selection may not be appropriate. Predation risk (Pöysä 1987), ability to form and maintain pair bonds (Hepp and Hair 1984), or weather (Jorde et al. 1984) may influence decisions to a greater extent than food availability. Thus, food availability may play only a small role in habitat selection of migratory ducks during spring (Stafford et al. 2014). If food is not a limiting factor, it could lessen importance of areas managed for high food densities occurring in actively managed wetlands and their associated higher costs. Previous literature indicates food resources during spring have been declining for diving ducks (Afton and Anderson 2001), but these same declines may not be occurring for dabbling ducks. A variety of characteristics, such as bill morphology (Nudds et al. 2000) and ideal feeding depth (Pöysä 1983a), have allowed several species of dabbling ducks to coexist, especially in areas with diverse habitat conditions (Elmberg et al. 1993). Dabbling ducks are able to use areas prone to rapid, unpredictable flooding that result in shallow water (Lindstrom 2017). Due to the depths at which diving ducks typically feed, they are limited to larger, more permanent water bodies during migration, which may have less food available than in the past (Anteau and Afton 2009). Therefore, diving ducks may be better candidates to test the relationship between food availability and duck density because they may be more food limited during migration.
Core sampling is commonly used method to estimate food availability; however, it is possible that ducks may perceive food availability differently than what researchers estimate through core sampling. Our core samplers captured the water column but likely missed some invertebrates that were able to evade the sampler as we placed it into the water (Cheal et al. 1993). However, invertebrates represent a small percentage of duck diet during the nonbreeding seasons (Jorde et al. 1983; Combs and Fredrickson 1996). Also, we sampled food at the beginning of spring migration. If ducks depleted sites at different rates during the study period, then our estimates of food availability may be less applicable during mid to late migration. Furthermore, we only sampled food in 2013 and used the site-level estimates for 2012. We acknowledge that food densities may have been different in 2012; however, we think the relative densities among sites would be similar between years and thus, any patterns would still exist.
Management type affected dabbling duck density for three of the eight species and for all dabbling ducks combined, whereas food density did not affect duck density for any species. Duck density was greatest in actively managed wetlands for all dabbling ducks combined and each species affected by habitat management. Actively managed wetlands (i.e., moist-soil impoundments) are typically very good duck foraging habitat because they produce abundant, high-quality food (Bowyer et al. 2005). Based on our results, it appears that duck density was greater in actively managed sites for reasons unrelated to food. Actively managed wetlands are generally shallow throughout, which provides easier access to food and promotes more efficient feeding (Guillemain et al. 2001). However, agricultural food plots were also very shallow and did not receive the same intensity of duck use. Because this study took place during spring migration, ducks had depleted food in our study wetlands during fall migration. It seems likely that if food is an important factor for ducks during migration that they would use cues to select habitats that are indicative of abundant food and good foraging conditions. Abundant seed-producing forbs, grasses, and sedges likely characterize these habitat cues (Naylor et al. 2005), and are typical of actively managed wetlands. If ducks were using the same cues to select a site during spring as they did during fall, they may settle in the same actively managed wetlands that by spring had depleted food sources, which may also explain why we found less food in these types of sites.
Although food availability appeared to have little influence on the distributions of ducks in this study, five of the eight species fed most intensively in plot types with the greatest food density (passively managed wetlands). The lack of a direct linear relationship between our measure of food density and feeding intensity suggests feeding intensity may not be directly related to food density. Theoretically, waterfowl feed at an intensity (proportion of a given amount of time when their head is down and they are actively feeding during a feeding bout) determined by the tradeoff between the level of reward gained by feeding and the risk taken (Lima and Bednekoff 1999). Theory predicts that individuals, while seeking food resources, should increase feeding intensity with an increase in food density or a perceived decrease in predation risk (Lazarus and Symonds 1992). In our study, the majority of duck species in passively managed wetlands fed more intensively and used less vigilance than ducks in wetlands with other management types. While passive wetlands had slightly greater food availability, we found no evidence of a direct relationship between food availability and feeding intensity in our study. Thus, it is possible that waterfowl perceived the passively managed wetlands as being safer from predation. Although we did not formally estimate vegetation structure or density, passively managed wetlands appeared to be more open than actively managed wetlands. Waterfowl may have perceived this more open habitat as safer because they could detect predators at a greater distance (Lazarus and Symonds 1992; Pöysä 1994; Behney et al. 2018).
Currently, the Refuge manages complexes of all wetland types and based on our results, we suggest that this is a valuable strategy moving forward. If the Refuge can maintain high food availabilities on lands that are cheaper to manage (i.e., passively managed wetlands), then the value of the Refuge to migratory ducks should be more robust to fluctuations in funding. However, actively managed wetlands do hold increased value during times of drought or seasonal weather variability (Pankau 2008), so habitat managers should maintain complexes of both actively and passively managed wetlands to provide quality habitat regardless of climatic variability and the exact timing of migration. Areas close to the river that are prone to seasonal flooding could be managed as agricultural food-plot wetlands and passively managed wetlands, utilizing flooding regimes that are unnaturally rapid and unpredictable as a management tool. This can provide large areas of habitat to migrating waterfowl, especially northern pintail, a species of conservation priority, because pintails were most numerous on agricultural food-plot areas.
Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.
Data S1. Data associated with duck counts and duck behavior during 2012 and 2013 at Patoka National Wildlife Refuge and Management Area, Indiana. Separate tabs show duck count and behavior data. Species codes are as follows: ABDU = American black duck Anas rubripes, AMWI = American wigeon Mareca americana, BWTE = blue-winged teal Spatula discors, GADW = gadwall Mareca strepera, GWTE = green-winged teal Anas crecca, MALL = mallard Anas platyrhynchos, NOPI = northern pintail Anas acuta, NSHO = northern shoveler Spatula clypeata.
Found at DOI: https://doi.org/10.3996/062019-JFWM-044.S1 (262 KB XLSX).
Table S1. Full model-selection results including number of parameters (K), ΔAICc, and model weight (wi) for duck density models. First, we found the best temporal model of dabbling duck density by comparing day of year in linear form (Day) with day as quadratic (Day2) and cubic (Day3) forms, as well as with a null model excluding day. We then used the best temporal model (or null model) as the basis of other models including a food-density (kg/ha) variable and a model including management type (actively managed wetland, planted food plot, and passively managed wetlands) on Patoka River National Wildlife Refuge and Management Area in Indiana during spring migration in 2012 and 2013. We modeled all species combined as well as eight species individually. We included site as a random effect and year as a fixed effect in all models. Models including quadratic and cubic terms also include all lower-level forms (e.g., a model with Day2 also includes Day).
Found at DOI: https://doi.org/10.3996/062019-JFWM-044.S2 (21 KB DOCX).
Reference S1. Fredrickson LH, Taylor TS. 1982. Management of seasonally flooded impoundments for wildlife. Washington, D.C.: United States Fish and Wildlife Service Resource Publication 148.
Found at DOI: https://doi.org/10.3996/062019-JFWM-044.S3 (2.56 MB PDF); also available at https://www.fwspubs.org/doi/suppl/10.3996/012014-JFWM-009/suppl_file/012014-jfwm-009.s2.pdf
Reference S2. Gray R, Tuttle R, Wenberg RD. 1992. Wetland restoration, enhancement, or creation. Washington, D.C.: Natural Resource Conservation Service, United States Department of Agriculture Engineering Field Handbook Chapter 13.
Found at DOI: https://doi.org/10.3996/062019-JFWM-044.S4 (2.88 MB PDF); also available at https://pdhonline.com/courses/c707/USDA%20Wetland%20Creation%207313.pdf
Reference S3. Lane JJ, Jensen KC. 1999. Moist-soil impoundments for wetland wildlife. Washington, D.C.: United State Army Corps of Engineers Technical Report EL-99-11.
Found at DOI: https://doi.org/10.3996/062019-JFWM-044.S5 (1.38 MB PDF); also available at https://www.fwspubs.org/doi/suppl/10.3996/072013-JFWM-050/suppl_file/072013-jfwm-050.s5.pdf
Reference S4. Soulliere GJ, Potter BA, Coluccy JM, Gatti RC, Roy CL, Luukkonen DR, Brown PW, Eichholz MW. 2007. Upper Mississippi River and Great Lakes Region Joint Venture waterfowl habitat conservation strategy. Fort Snelling, Minnesota: U.S. Fish and Wildlife Service.
Found at DOI: https://doi.org/10.3996/062019-JFWM-044.S6 (4.24 MB PDF).
Reference S5. [USFWS] U.S. Fish and Wildlife Service. 2008. Patoka River National Wildlife Refuge and Management Area comprehensive conservation plan. Washington, D.C.: U.S. Department of the Interior.
Found at DOI: https://doi.org/10.3996/062019-JFWM-044.S7 (5.93 MB PDF); also available at https://www.fws.gov/midwest/planning/PatokaRiver/index.html
Principle funding for this project was provided by a United States Geological Survey grant. We thank the staff at Patoka River National Wildlife Refuge and Management Area and the Cooperative Wildlife Research Laboratory at Southern Illinois University for resources, guidance, and coordination for field work. We thank the hard working technicians that were able to process our core samples for use in these analyses. Finally, we thank the reviewers and Associate Editor for providing comments that improved this manuscript.
Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Citation: Lindstrom JM, Eichholz MW, Behney AC. 2020. Effect of habitat management on duck behavior and distribution during spring migration in Indiana. Journal of Fish and Wildlife Management 11(1):80–88; e1944–687X. https://doi.org/10.3996/062019-JFWM-044
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.