Structural heterogeneity increases diversity of non‐breeding grassland birds

Abstract

Grassland birds have experienced greater population declines than any other guild of birds in North America, and yet we know little about habitat use and the affects of management during their non‐breeding period on wintering grounds. The paucity of information on wintering grassland birds limits our ability to develop effective conservation strategies. We investigated habitat use by the winter bird community in grasslands with restored heterogeneity resulting from the interactive effects of fire and grazing. We used 500 m line transects distributed across patches (i.e., <13, 13–24, and >24 months post disturbance) resulting from spring burning with growing season grazing (April‐Sept) and quantified avian relative abundance, community structure, and probability of patch occupancy while accounting for imperfect detection. Grassland structure that resulted from the fire‐grazing interaction created heterogeneity among patches that influenced avian habitat use during winter. Generalist birds such as the Savannah Sparrow (Passerculus sandwichensis) and meadowlarks (Sturnella spp.) were relatively common in all patch types while more specialized species such as the Smith's Longspur (Calcarius pictus) and Le Conte's Sparrow (Ammodramus leconteii) reached greatest abundance and probability of occupancy in the patches with the least and greatest time post disturbance, respectively. This research provides novel information on the response of wintering birds to restored ecological processes in grasslands and can improve efforts to create effective conservation strategies. Our findings add to a growing body of literature supporting the use of fire and grazing to create a shifting grassland mosaic that increases vegetation structural and compositional heterogeneity and maximizes native biodiversity within rangeland ecosystems through the conservation of natural patterns and processes.

Introduction

Grassland bird declines are major conservation concern of the 21st century (Brennan and Kuvlesky 2005). Since the inception of the North American breeding bird survey, no other habitat associated group of birds has undergone such drastic population declines (Sauer et al. 2012). While the historic loss of grasslands is undoubtedly a leading cause for grassland bird declines (Samson and Knopf 1996), it is likely that the cumulative effects of afforestation, mismanagement and intensification of production on rangelands, and increased fragmentation have acted in synergy to drive populations to their current low levels (Herkert 1994, Vickery et al. 1999, Brennan and Kuvlesky 2005). The impacts of these factors have typically been investigated using breeding birds, and data on non‐breeding bird communities is generally lacking despite multiple claims that this is the period limiting avian populations (Wiens and Dyer 1975, Rappole and McDonald 1994, Peterjohn 2003). Grassland degradation in the southern United States may limit suitable wintering sites for many temperate grassland birds (Hunter 1990, Lymn and Temple 1991). The lack of data examining habitat use and survival of grassland birds during the non‐breeding season limits our ability to apply effective conservation management (Herkert et al. 1996).

We must improve our knowledge of avian responses to management throughout the year to maximize conservation efforts for grassland birds. Grasslands are disturbance‐dependent habitats that evolved with fire and grazing, which explains the affinity for certain grassland birds to inhabit specific patches that result from interacting fire and grazing disturbances (Knopf 1994, Hoekstra et al. 2005). Grasslands with fire and grazing still occurring in tandem typically apply these processes uniformly across the landscape, which results in homogenous vegetation structure and reduces grassland diversity (Robbins et al. 2002, Reinking 2005). This approach has been termed “management toward the middle” and reflects the intermediate disturbance model that has shaped rangeland management, but has failed to recognize the diverse needs of grassland organisms and the historically diverse structure of grasslands posited to have occurred pre‐European settlement (Fuhlendorf and Engle 2001, Powell 2006, Fuhlendorf et al. 2012). Management aimed at uniform disturbance limits the amount of undisturbed patches across the landscape necessary for certain species life history traits and fulfills the habitat requirements of a limited suite of generalized bird species (Reinking 2005, Fuhlendorf et al. 2006, Powell 2008). The separation of the fire‐grazing interaction can simplify rangeland communities and limit ecosystem structure and function (Fuhlendorf and Engle 2001, Hoekstra et al. 2005). While there is evidence that burning and grazing can act independently to influence grassland plant communities this typically does not result in structural heterogeneity that drives diversity in grassland bird communities (Valone and Kelt 1999, Davis 2004, Coppedge et al. 2008). Diverse communities of species require habitat heterogeneity that includes intensively disturbed habitats (i.e., bare ground and relatively short‐statured vegetation) and habitats with minimal disturbance dispersed as a shifting mosaic across a complex landscape (Knopf 1994, Fuhlendorf et al. 2006, Fuhlendorf et al. 2009). Furthermore, the response of wildlife to disturbance processes, particularly birds, can vary greatly based on the duration and seasonal timing of disturbances (Brawn et al. 2001, Gregory et al. 2010).

Pyric herbivory, the ecological process through which fire drives grazing and grazing determines the probability of future fires, has been an effective strategy for improving habitat for breeding grassland birds (Fuhlendorf et al. 2006, Churchwell et al. 2008, Fuhlendorf et al. 2012, Hovick et al. 2012). Grasslands are inherently heterogeneous systems that vary functionally and structurally across multiple scales (Wiens 1997, Fuhlendorf and Smeins 1999). When fire and grazers are allowed to interact spatially and temporally across the landscape, it results in a shifting mosaic that influences population dynamics and movement patterns of breeding native birds (Brawn et al. 2001, Fuhlendorf et al. 2006). Furthermore, heterogeneous grasslands create habitat that can support a more diverse bird community (Fuhlendorf et al. 2006), and have been shown to improve reproduction for grassland nesting birds (Churchwell et al. 2008, Hovick et al. 2012). Restoring heterogeneity to grasslands could improve over‐wintering conditions for non‐breeding birds by providing structural variation resulting from interacting fire and grazing disturbances that suits habitat needs of specialized species, thereby increasing avian diversity across the landscape.

Most grassland birds in North America are migratory, spending one‐half or more of their annual cycle in migration or wintering areas (Herkert et al. 1996). Although the complexity of avian life cycles complicates conservation and management efforts, the extent and duration of seasonal movements by grassland birds emphasizes the need for information in areas other than the breeding grounds (Igl and Ballard 1999). Monitoring avian communities during winter is challenging. Inclement weather makes conducting surveys impractical, birds almost never vocalize during the non‐breeding season, and flushing birds are hard to identify (Fletcher et al. 2000). These challenges are, in part, why non‐breeding communities of grassland birds have received less research attention than breeding bird communities.

The limited number of studies that have examined winter bird communities typically report habitat use in the absence of disturbance, or they examine the affects of fire and grazing independently (Igl and Ballard 1999, Gordon 2000, Baldwin et al. 2007). Both of these scenarios are not commonly practiced in working grassland landscapes which makes application to conservation difficult. To effectively evaluate non‐breeding bird communities in a way that maximizes the potential effect on management of grassland ecosystems, research should focus on bird use of grasslands managed with fire and grazing. Therefore, we investigated non‐breeding grassland bird habitat use in a landscape that is managed with interacting fire and grazing processes. Prescribed fires for this study took place in during the dormant season in early spring (i.e., March) and grazing took place throughout the growing season (April–September). We hypothesized that the interaction of fire and grazing would result in structural heterogeneity and the resulting differences in patch structure and composition would influence winter grassland bird abundance and patch occupancy. This work provides empirical support for managing grasslands to improve efforts for bird species of conservation concern during the non‐breeding season and increases our general understanding of avian habitat associations during winter.

Methods

Study area

Our study was conducted at The Nature Conservancy's Tallgrass Prairie Preserve (hereafter, the preserve) in north eastern Oklahoma, USA from 2011 to 2013. This area comprises the southern extent of the Flint Hills region of the Great Plains and is part of the largest remaining tallgrass prairie in North America. The preserve is a 16,000 ha area dominated by a tallgrass prairie plant community. Dominant grasses include Andropogon gerardii Vitman, Schizachyrium scoparium Nash, Panicum virgatum L., and Sorghastrum nutans (L.) Nash. Dominant forbs at the preserve include ironweeds (Veronia spp.), milkweeds (Asclepias spp.), and ashy sunflower (Helianthus mollis). The climate of the preserve is temperate with hot summers and cool winters. During the course of this study (2011–2013) average high temperatures for January and February were 9.8°C and 10.2°C, respectively, while average low temperatures were −4.8°C and −3.0°C, respectively. Precipitation totals for each calendar year prior to sampling were 93.6, 80.1, and 85.8 cm for 2010, 2011, and 2012, respectively. Snowfall occurred on three different occasions during the sampling period, but never persisted for >3 days. As a consequence, we avoided sampling during periods of snow cover.

The preserve is managed to restore the fire‐grazing interaction, resulting in grassland structural heterogeneity (Hamilton 2007, Fuhlendorf et al. 2009). The structural heterogeneity becomes present as animals are allowed to select from areas that are recently burned and those that have greater time post fire (Archibald et al. 2005, Allred et al. 2011). Cattle management at the preserve is done across multiple pastures ranging from (430–980 ha), and are stocked at moderate rates (cattle: 2.4 animal unit months^−ha) using stocker steers that graze throughout the growing season (April–September), and all prescribed fires were conducted during spring (i.e., March–April) for this study.

Data collection

We used permanent, 500 m line‐transects to survey grassland birds during January and February from 2011 to 2013. Transects were distributed across three patches in each of four pastures with two transects in each patch (n = 24). Therefore, we examined three different patch types (0–12, 13–24, >24) resulting from time since fire and grazing and each patch‐type was replicated four times (pastures). Patches were large with an average size of 119 ha and ranging from 61 to 227 ha. Transects were randomly placed within patches using ArcGIS 10.0 (ESRI 2011) but constrained so that transects were >150 m apart and >100 m from patch edges to ensure independence between sites and prevent any double counting. Sampling was conducted during daylight hours by one observer on days with winds less than 20 km/h and no precipitation (Igl and Ballard 1999). Surveys were conducted four times in 2011 and 2013 and three times in 2012. Birds can be surveyed throughout the day during the non‐breeding season because daily activities do not peak as they do during the breeding season (i.e., dawn chorus; Fletcher et al. 2000). Each individual or group was identified to species by distinctive flight patterns, call notes, or coloration (Butler et al. 2009). When groups of individuals were encountered, we recorded them as a single detection and recorded the total number of individuals within the group. Additionally, we recorded environmental data at the start of each transect and for each survey day that could affect detection probabilities (wind, cloud cover, temperature, date, and time of day).

We collected vegetation data along each transect at the end of the sampling period. We measured vegetation in three plots situated at the start, middle, and end of transects to characterize the vegetation characteristics of each transect. Each plot was centered on the transect and had 0.5 m² quadrats distributed every 2.5 m for 10 m in each cardinal direction (n = 17). Within each quadrat we measured grass, forb, shrub, bare ground, and litter cover. Additionally, we measured litter depth and vegetation height within each quadrat and took a reading of vertical vegetation structure for the plot using a Nudd's board modified for grassland environments (Nudds 1977, Guthery et al. 1981).

Data analysis

We calculated relative abundances for the eight most frequently detected species. To do so, we divided the total number of detections for each species by the number of transects surveyed in each patch during each year of the study. This method standardized effort across years and allowed for simple patch‐type comparisons across and among all species. We used this method rather than estimating densities using program Distance because we frequently surveyed transects that had no detections and preliminary analysis indicated that several of the gregarious or large species had detection probabilities that did not decrease as distance from the observer increased (Buckland 2001).

We used nonmetric multidimensional scaling (NMDS) to assess similarity in bird species composition across habitat patches resulting from the fire‐grazing interaction. Specifically, we used relative abundances and the “metaMDS” and “envfit” functions in the vegan package of program R to project a two dimensional summary of avian community habitat use (Oksanen et al. 2010). We chose to use the Bray‐Curtis distance metric in NMDS because it is sensitive to differences in the most abundant species and less sensitive to infrequently encountered species (Pillsbury et al. 2011), which works well for the winter bird community because some species are detected very frequently while others are not. NMDS is iterative procedure that maximizes the rank‐order correlation between Euclidean distance in ordination space and the values in the dissimilarity matrix. Therefore, axes are arbitrary and do not convey any real meaning. To interpret NMDS graphics one must know that species with shorter inner‐point distances are more similar (i.e., similar patch use) than those with greater inner‐point distances. Goodness‐of‐fit is measured through stress in NMDS, which is inversely proportional to the rank order of Euclidean distance correlations. We set our analysis at a level of two dimensions and assessed the stress values to make sure our selection of the number of dimensions adequately described these data.

We used an occupancy model framework in program MARK to evaluate site occupancy and detection probabilities for a subset of species of grassland birds during winter (White and Burnham 1999, MacKenzie et al. 2002). MARK uses a likelihood‐based method for estimating site occupancy rates when detection probabilities are <1 and allows for the incorporation of covariates that can influence occupancy and detection rates. We only incorporated species that are considered grassland obligates (per Vickery et al. 1999) and for which we had >25 detections. Models assume there are no false detections and that sites are independent of one another, which we ensured through proper spacing of transects and combining detections for transects within the same patch (MacKenzie et al. 2002).

To assess occupancy dynamics, we created encounter histories for all surveys conducted from 2011 to 2013 indicating a 1 if the species was detected in a particular patch (i.e., time since fire) or a 0 for non‐detections. Because detection is important in determining unbiased estimates of site occupancy, our first modeling procedure was to determine which covariates most influenced detection for each focal species. We examined the effects of temperature, wind, cloud cover, and annual variation on detection. After determining the best detection model, we then assessed the influence of time since fire and grazing on occupancy by entering user‐specified covariates in the MARK design matrix. Because our primary objective was to examine bird habitat use in patches that result from the spatio‐temporal interactions of fire and grazing, the only a priori covariate we examined was time since fire. We used this approach because time post fire and grazing affects plant community structure and composition (Fuhlendorf and Engle 2004, Winter et al. 2012), which is an important predictor of grassland bird site occupancy especially in large, intact grassland landscapes (Davis 2004, Fuhlendorf et al. 2006, Coppedge et al. 2008). Models were ranked using Akaike's information criterion adjusted for small sample sizes (AIC_c) and we followed the general information theoretic approach for assessing model fit (Burnham and Anderson 2002).

Results

We found that the interaction of fire and grazing resulted in a shifting mosaic of grassland patches as areas with greater time since fire and grazing supported taller, denser vegetation with greater litter cover, litter depth, and less bare ground cover than recently burned and grazed areas. In contrast, recently burned and grazed patches had high amounts of bare ground with relatively minimal litter cover and short vegetation structure (Fig. 1).

Observed vegetation characteristics in the non‐growing season in patches that range in time since fire and grazing at the Tallgrass Prairie Preserve, Oklahoma, USA (2011–2013).

We detected 14 bird species across 144 km of transects and detection rates were low with an average of 2.4 detections km⁻¹. Recently burned and grazed patches had the fewest overall detections but the same species richness as patches 13–24 months since fire and grazing; patches with >24 months since fire and grazing had the most detections and the greatest avian richness. There were seven species that were detected in all three patches, one species detected in two patches, and six species that were only detected in one patch type. Many of these species are gregarious and detections averaged >1 individual; group size averages were 1.07 (±0.03 SE) for meadowlark spp. (Sturnella spp.), 1.05 (±0.02) for Savannah Sparrow (Passerculus sandwichensis), 3.40 (±1.01) for American Tree Sparrow (Spizella arborea), and 4.72 (±1.10) for Smith's Longspur (Calcarius pictus). All other species were only detected as single individuals.

Relative abundance data showed that Northern Harrier (Circus cyaneus), meadowlark spp., Savannah Sparrow, and American Tree Sparrow used a range of vegetation patches resulting from time since fire and grazing. In contrast, Smith's Longspur and Le Conte's Sparrow (Ammodramus leconteii), exhibited trends in habitat use with abundance maximized in the most recently disturbed and the most undisturbed patches, respectively (Fig. 2).

Relative abundances (detections per patch ±SE) of the six most commonly encountered bird species in patches that range in time since fire and grazing at Tallgrass Prairie Preserve, Oklahoma, USA (2011–2013).

The NMDS ordination produced a good fit in two dimensions (stress = 0.01; Kruskal 1964) and accounted for 70% of the variance in the grassland bird abundance data. The ordination showed that bird community composition differed among the vegetation patches resulting from time since fire and grazing and patterns observed in relative abundance data were generally supported. Most strikingly, Le Conte's Sparrow, Smith's Longspur, and Sprague's Pipit (Anthus spragueii) utilized specific, opposite patch types. Le Conte's Sparrow was found in patches >24 months since fire and grazing while Smith's Longspur and Sprague's Pipit were in the most recently burned and grazed patches (Fig. 3). However, there was partial overlap in bird communities across patches as seen by clustering of individual species and the overlap of patch‐hulls.

Nonmetric multidimensional scaling plot for all bird species with >5 detections at the Tallgrass Prairie Preserve, Oklahoma, USA (2011–2013). The hulls around the perimeter of each patch are based on the site scores for each of the three patch types and species are projected based on species score. Species projected near each other indicate similar habitat use. Four letter codes represent the following species: NOHA = Northern Harrier, GRPC = Greater Prairie‐Chicken, SPPI = Sprague's Pipit, ATSP = American Tree Sparrow, SAVS = Savannah Sparrow, LESP = Le Conte's Sparrow, SMLO = Smith's Longspur, and MESP = meadowlark species.

Patch occupancy for the four most abundant grassland birds complemented observed relative abundance data and community results, and indicated that some species are generalists whereas other species are specialists, selecting for specific patches that result from time since fire and grazing (Fig. 4). Savannah Sparrows trended towards occupying the most recently burned and grazed patches but this affect was not significant (β = −0.56, CI: −3.15, 2.03), reflecting the generality of habitat selection for that species. Similarly, disturbance did not have a significant effect on meadowlark patch occupancy (β = 0.38, CI: −0.86, 1.62). Smith's Longspur had a higher probability of occurrence in recently burned and grazed patches (β = −0.81, CI: −1.86, 0.23), whereas, Le Conte's sparrows had the highest probability of occupancy in patches with the greatest time since fire and grazing (β = 18.6, CI: 18.2, 19.1). All species had high detection rates and models for detection for two species were improved by inclusion of temporal or weather parameters (Table 1).

Occupancy probabilities for the most abundant grassland bird species in patches that range in time since fire and grazing at the Tallgrass Prairie Preserve, Oklahoma, USA (2011–2012). Four letter codes represent the following species: SAVS = Savannah Sparrow, LESP = Le Conte's Sparrow, SMLO = Smith's Longspur, and MESP = meadowlark species.

The best models examining weather and temporal effects on detection probabilities for the four most commonly encountered grassland bird species at the Tallgrass Prairie Preserve, Oklahoma, USA (2011–2013). In the table, p indicates the detection probability; LCI and UCI indicate the lower confidence interval and upper confidence interval, respectively.

Discussion

Coupling fire and grazing to allow grazers to select among burned and unburned areas of the landscape has significant implications for breeding grassland birds (Fuhlendorf et al. 2006, Powell 2006, Coppedge et al. 2008, With et al. 2008). Our results now demonstrate the effects of restored fire and grazing processes on a winter bird community. Grazing animals preferentially select the most recently burned areas which increase the amount of bare ground and reduce standing biomass, while areas that have gone unburned accumulate litter and standing biomass over time (Fuhlendorf and Engle 2004, Fuhlendorf et al. 2006). The resulting structural heterogeneity provides greater suitability of habitat structure and increases the diversity of winter birds that can occur across the landscape (Fig. 5). Our work has important conservation implications as many have speculated that North American grassland bird populations are limited by habitat during the non‐breeding season (Brooks and Temple 1990, Basili 1997), yet until now there was little empirical support for this. Our results support the hypothesis that the interaction of fire and grazing would result in structural heterogeneity of vegetation and the resulting differences in patch structure and composition would influence winter grassland bird abundances. Our findings uphold management recommendations that a mosaic of patchy disturbance across the landscape would provide suitable winter habitat for a wide range of bird species (Gabrey et al. 1999, Baldwin et al. 2007). Additionally, our conclusions are substantiated by recent work examining winter bird use in tallgrass prairie managed with interacting fire and grazing that found plant structure (i.e., height) mostly determined patch use (Monroe and O'Connell 2014).

Characterization of the response of bird species during the non‐growing season to interacting fire and grazing at the Tallgrass Prairie Preserve, Oklahoma, USA (2011–2013). Art work courtesy of Gary Kerby adapted from Fuhlendorf et al. (2009). Lines for each species are based on observed patch use.

Much of the previous work investigating winter bird community responses to disturbance have viewed fire and grazing as separate disturbances (Gabrey et al. 1999, Gordon 2000, Baldwin et al. 2007, Baldwin et al. 2010), or have examined the bird community in grasslands that are grazed in the absence of fire (Grzybowski 1982). Our study indicates that the spatial and temporal patterns of grazing and fire in combination may be important to conservation of non‐breeding birds. While we did not explore the impacts of fire alone or grazing along, our results indicate that the structure that results from interacting fire and grazing is beneficial to broad suite of winter birds (Fuhlendorf and Engle 2004, Fuhlendorf et al. 2009). As a consequence of the grassland shifting mosaic that occurred in the presence of the fire‐grazing interaction, areas most recently disturbed had the least amount of litter and vertical structure and created habitat for two bird species of conservation concern, Sprague's Pipit and Smith's Longspur. Both of these species have high continental concern rankings by Partners in Flight (PIF). Sprague's Pipit was encountered very infrequently during this study, but is a watchlist species and this information could help improve future management efforts (Partners in Flight Science Committee 2012). Smith's Longspurs were relatively abundant and had the highest probability of patch occupancy in recently disturbed patches.

We observed that patches with the greatest time since fire and grazing had the tallest, densest vegetation with the most litter accumulation, and the greatest avian richness of all patches examined. Additionally, relatively undisturbed patches had the highest probability of occupancy by Le Conte's Sparrows, which occurred almost exclusively in areas of >24 months since fire and grazing. Our findings illustrate how the interaction of fire and grazing influence birds differently than either disturbance in isolation as previous research reported Le Conte's Sparrows reaching their greatest abundance in areas burned within the previous two years (Baldwin et al. 2007). Thus, decoupling these interacting disturbances under which grasslands birds evolved with, can lead to recommendations which may deviate from historic conditions. Our work elucidates the complexity of grassland bird response to disturbance and illustrates that research outcomes can vary dependent on the timing, duration, and type of disturbance being examined. In particular, Le Conte's Sparrow winter ecology is poorly studied, and our results can be used to improve conservation efforts for this species by increasing grassland patches that have gone undisturbed for multiple years.

Patch selection by non‐breeding birds likely reflects a combination of physiological constraints on their behavior and survival strategies (Pulliam and Enders 1971, Pulliam and Mills 1977). For example, Le Conte's Sparrows similar to most Ammodramus, are generally poor flyers that rely more heavily on cryptic behavior and running to try and avoid predators. As a result, they commonly have small and well‐defined home ranges in winter (Baldwin et al. 2010). Additionally, it is possible that this solitary species seeks dense and tall vegetation structure for thermal protection from exposure to windy conditions common in the southern Great Plains during winter. Conversely, more gregarious species such as the Smith's Longspur and American Tree Sparrow reached highest abundances in recently burned and grazed patches. These species likely select patches that are recently disturbed with small amounts of litter and short vertical structure to enable easier access to food resources. They are able to select for open foraging patches as a result of reduced predation risk that occurs as a result of flocking behavior (Grzybowski 1983). Other commonly encountered species during this study showed a range of patch use and probably use some combination of access to resources, thermal needs, and predator avoidance to select patches.

Many of the species that we examined appear to use similar habitat structure during the non‐breeding and the breeding seasons. For example, meadowlark species that we recorded during winter exhibited very generalist behavior, similar to their patch use during the breeding season (Fuhlendorf et al. 2006, Pillsbury et al. 2011). Additionally, Le Conte's Sparrows were most abundant in undisturbed patches similar to the vegetation they are most abundant in during the breeding season (Igl and Johnson 1999), and Savannah Sparrows were widespread across all patch types reflecting their generalist approach to nesting habitat (Bollinger 1995). There may be exceptions to these trends (Igl and Ballard 1999, Baldwin et al. 2010), but based on our data when a range of habitat structure is available, many species seem to have an affinity for the same type of structure during breeding and non‐breeding seasons. If this conclusion is true for a broad suite of species, simply knowing the over‐wintering range for species of conservation concern would allow conservation efforts to focus on the proper structure type. Nonetheless, managing for a mosaic of structure would offer the greatest diversity of habitat types and would likely suit the requirements of the most species, similar to the results we report.

Our findings concur with breeding bird research from other rangeland systems that indicated bird community composition is dependent on variable patterns of fire and grazing (Knopf 1994, Skowno and Bond 2003, Krook et al. 2007, Reinkensmeyer et al. 2007, Gregory et al. 2010). This makes intuitive sense when you consider the historical disturbance regime of grasslands and the affinity of grassland breeding birds to select areas with specific, varying levels of biomass (Knopf 1994). Breeding bird abundance and diversity was increased in grasslands that mimicked historical disturbance regimes compared to management that promoted uniform vegetation structure across the landscape (Fuhlendorf et al. 2006, Coppedge et al. 2008). The consistent response of grassland birds during breeding and non‐breeding seasons is a strong indication of the potential conservation value in managing for a structural mosaic. Moving forward, we argue that because of the long evolutionary history of fire and grazing in grassland ecosystems, conservation research should focus on the interaction of fire and grazing and the resulting shifting mosaic in the landscape (Fuhlendorf et al. 2006, Fuhlendorf et al. 2012).

Acknowledgments

We would like to thank J. Lautenbach for assistance with data collection and entry, and we would like to thank B. Hamilton, T. Brown, and other personnel that manage the Tallgrass Prairie Preserve. This project was supported by funding from USDA‐AFRI Managed Ecosystems grant #2010‐85101‐20457 and by the Oklahoma Agricultural Experiment Station.