Printer Friendly

Topographic patterns of nest placement and habitat quality for grassland birds in tallgrass prairie.


Grassland bird populations have exhibited widespread declines in North America over the past 50 y, presumably because of the wholesale loss and degradation of their breeding habitat (Askins, 1993; Herkert, 1995; Peterjohn and Sauer, 1999). Less than 5% of tallgrass prairie remains (Samson and Knopf, 1994), and much of that is located in hilly regions such as the Flint Hills of Kansas and Oklahoma, the Sand Hills of Nebraska, and the Loess Hills of Iowa. Despite the presumed conservation importance of these areas for grassland birds, the effect of topography on habitat quality and the settlement patterns of nesting birds has rarely been considered. In the Flint Hills, for example, topographic position has been associated with variation in habitat quality for Dickcissels (Spiza americana). Males settled first in old agricultural fields with taller denser vegetation (i.e., lowland sites) and had more females and nests per territory than males that arrived later and/or defended a territory in upland pastures (Zimmerman, 1971, 1982). In southwestern Wisconsin, greater densities of Bobolink (Dolichonyx orzivorus) and Savannah Sparrow (Passerculus sandwichensis) were found in large upland pastures than in smaller lowland pastures (Renfrew and Ribic, 2002). Topography has also been identified as a factor increasing bird diversity on a tallgrass prairie preserve in west-central Iowa because of the increased vegetation diversity associated with variable topography (Laubach, 1984).

The Flint Hills contain the largest contiguous expanse of unplowed tallgrass prairie remaining in North America (Knapp and Seastedt, 1998); it was this hilly rocky terrain that allowed much of the region to remain unplowed. Topography may contribute to habitat heterogeneity in tallgrass prairie through effects on soils (Ransom et al., 1998; Hook and Burke, 2000), plant productivity (Abrams et al., 1986; Briggs and Knapp, 1995; gaaapp et al., 1998) and the composition and structure of vegetation communities (Abrams and Hulbert, 1987). These effects may be further modified by land-management practices. In the Flint Hills, grasslands are predominantly managed as rangeland for cattle, which involves annual spring burning to increase forage production and quality (Zimmerman, 1997). Topography influences the distribution and grazing patterns of cattle (Gillen et al., 1984; Senft et al., 1985; Pinchak et al., 1991), which in turn interacts with fire to affect habitat structure and heterogeneity (Abrams and Hulbert, 1987). Grazing directly influences habitat structure by reducing litter depth and increasing the patchiness of litter (i.e., dead, non-standing vegetation; Knapp et al., 1999), as well as increasing the local diversity and abundance of preferred forage, such as forbs (Vinton et al., 1993; Hartnett et al., 1996; Towne et al., 2005). Land-management practices geared toward cattle production should thus be influenced by topographic variation, thus, affecting the structure and quality of grassland bird habitat within the Flint Hills.

Our objectives in this study were, thus, to examine topographic effects on the nesting ecology of Dickcissel, Grasshopper Sparrows (Ammodramus savannarum) and Eastern Meadowlarks (Sturnella magna). These species represent the "core" of the avian community in the Flint Hills (Zimmerman, 1993). The specific questions we addressed were: (1) Does the distribution of nesting birds vary among topographic positions, and if so, are these patterns consistent throughout the breeding season?; (2) To what extent is nest-site selection (the non-random placement of nests with respect to available habitat) influenced by topography?; and (3) Does topographic position directly or indirectly (e.g., through local nest-site vegetation) affect nest survival of these grassland birds?



We conducted this study in the Flint Hills of Kansas and northeastern Oklahoma during the 2004 breeding season (25 Apr.-27 Jul.). The Flint Hills comprises approximately 1.6 million ha of rangeland (Knapp and Seastedt, 1998). Dominant vegetation of the region consists of the warm-season grasses big bluestem (Andropogon gerardii), little bluestem (Schizachyrium scoparium), Indian grass (Sorghastrum nutans) and switchgrass (Panicum virgatum) (Pieper, 2005). Some representative forb and woody species of the region include western ragweed (Ambrosia psilostachya), lead plant (Amorpha canescens), smooth sumac (Rhus glabra) and buckbrush (Symphoricarpos orbiculatus) (Abrams et al., 1986; Hartnett et al., 1996). Terrain of the Flint Hills is characterized by high local topographic relief, where distinct hilltops and steep sloping midlands commonly occur in areas outside the major river drainages (Fig. 1). In the Flint Hills, 85% of managed grasslands in Kansas (USDA, 2004a) and 98% of those in Oklahoma (Osage Co.; USDA, 2004b) were used for cattle production. Throughout the region, prescribed spring burns (mid-Apr.) are commonly implemented by ranchers to increase the productivity (Anderson et al., 1970) and nutritional value of grass for cattle (Owensby et al., 1995; Pieper, 2005).


We selected 18 sites that encompassed the major grazing practices within the Flint Hills, with six sites in each of the northern, central and southern regions (Fig. 1). Site treatments consisted of intensive early-stocking (grazed approximately Apr. 15-Jul. 15 at a stocking density of -1.25 cattle/ha) and burned in Apr. just before the 2004 field season (IESB) or season long-stocking (grazed approximately Apr. 15-Oct. 15 at a stocking density of ~0.6 cattle/ha) and burned (SLSB) or not burned (SLSU) just before the 2004 field season (Owensby et al., 1995). Within sites, we placed 5-ha plots (320 m x 160 m) so they incorporated the representative topographic relief of each pasture.


We found nests by rope-dragging to flush incubating females and occasionally through observation of parents or incidental to other field activities. We actively searched for nests every 7-10 d, starting 26 Apr. in the southern region 10 May in the central region and 25 May in the northern region. To ensure a uniform search effort, the entire 5-ha plot was dragged during each nest search. We marked nests at a distance of 5 m (5 paces) with a flag and relocated them on subsequent visits using a hand-held Global Positioning System unit. We visited nests every 3-4 d (rarely 5 d) to determine nest contents and nest fate. Nests were considered successful if they fledged at least a single host young (i.e., we excluded nests that fledged only Brown-headed Cowbird [Molothrus ater] young) in the interval preceding the final visit, and there was no evidence of predation (broken egg shells, disturbed nest lining) or trampling by cattle.


We measured vegetation structure at nest sites at the end of each nesting attempt. At each nest site, we quantified vertical vegetation structure by placing a Robel pole directly next to the nest and obtaining four visual obstruction readings (VOR; Robel et al., 1970) taken from the four cardinal directions. To estimate horizontal cover of vegetation, a 0.25-[m.sup.2] square frame was centered on the nest in which we estimated proportion of cover for live grass, forbs (broad-leaved vegetation) and bare ground (exposed ground without litter). We estimated litter depth (dead, non-standing vegetation) by averaging four measurements of litter depth (cm) taken at the corners of the frame.

We also characterized vegetation within each study plot by surveying vegetation at 40-m intervals along four transects that were spaced 40-m apart (a 280 m x 120 m grid = 8 points/transect x 4 transects = 32 points/plot). We sampled vegetation points early (May 24-Jun. 4) and late (Jun. 21-Jul. 3) in the field season to capture seasonal variation in vegetation. We estimated vegetation structure at points using the same methodology as that for nest sites.


To determine the topographic position of nests and site vegetation points, we downloaded Digital Elevation Models (30 m x 30 m grid cells) from the National Elevation Dataset ( and the USGS "Seamless Data Distribution" system ( We used ArcInfo 9.1 (ESRI, Redding, California) to convert these into 2-m shaded contour maps to highlight terrain and elevation differences on sites. We projected locations of nests and vegetation points onto maps and categorized the topographic position of each using characteristics of the terrain. We used a generalized topographic model of the Flint Hills to conceptualize a lowland, midland and upland sequence (see Fig 4.1 in Ransom et al., 1998). Lowlands were identified as relatively flat areas at the base of the hill and uplands were identified as the relatively flat hilltop areas. Midlands were identified as the steep side slopes ("breaks") between upland and lowland areas, which were illustrated by narrow contour lines on the maps. Vegetation points and nests occurring on two of the sites were categorized as uplands because these sites were positioned entirely in flatter areas upslope of most of the surrounding landscape.


Nest distribution.--We used Cochran-Mantel-Hanzel statistics within a contingency table analysis to explore the distribution of nests relative to topographic position (lowland, midland, upland), season (early, late) and management treatment (SLSU, SLSB, IESB). We treated topographic position as an ordinal variable, and only included data from sites with all three topographic positions present for this analysis (16 of our 18 sites). To define early- and late-season groups of nests, we categorized individual nests as "early" or "late" season based on the date the nest was located and by identifying a midpoint in the nesting season for each species. Most nests were found during incubation, so nest-location dates provide a reasonable approximation of initiation date within season. For these analyses, we used 6 Jun. as the seasonal midpoint for Eastern Meadowlark, 16 Jun. as the midpoint for Grasshopper Sparrow and 3 Jul. as the midpoint for Dickcissel. Our analyses addressed both three-factor (topographic position, season and management treatment) and two-factor (topographic position and season or topographic position and management treatment) contingency tables. Alpha ([alpha]) was set at 0.10 for all statistical analyses. Unless otherwise noted, all analyses were conducted using SAS 9.1 (SAS Institute Inc., 2004).

Vegetation.--We estimated vertical vegetation structure by calculating the median of the four VOR values (Robel et al., 1970). We used median rather than mean VOR because the former is less affected by extreme values (i.e., presence or absence of vegetation at a given height). We derived composite variables describing vegetation structure using Principal Components Analysis (PCA) performed on the mean scores for each of the other four variables we measured at each point: proportion of live grass, live forb, bare ground and litter depth. Components were estimated from the correlation matrix and were retained if eigenvalues were [greater than or equal to] 1. We viewed component loadings greater than 0.30 or less than -0.30 as "significant" and loadings greater than 0.50 or less than -0.50 as "very significant" in terms of the variation in the distribution explained by the eigenvector (McGarigal et al., 2000).

We performed analysis of covariance to examine how site vegetation varied with topographic position, management and time of season. In the site vegetation analysis, vertical vegetation structure (VOR) and principal components from the vegetation analysis were modeled as a function of topographic position (lowland, midland, upland), management treatment (IESB, SLSU, SLSB) and time of season (early vs. late). We obtained a single model using a backward selection procedure (with P > 0.10 as the removal criterion); topographic position was always retained as a predictor to test for topographic effects on vegetation. We computed mean vegetation values from the final model and compared these using two-sample t-tests. All F-tests and t-tests were made using Type III sums of squares, [alpha] of 0.10, and Satterthwaite degrees of freedom.

Daily nest survival.--For each species, we modeled variation in daily nest survival (DNS) using the Nest Survival Model in Program MARK (White and Burnham, 1999). Our nest survival model included the effects of nest topographic position, management treatment, time within season and nest-site variables such as VOR and nest vegetation principal components identified by PCA (see below). These factors were selected because topographic position and management treatment influence vegetation and grazing activity on sites, time is associated with seasonal changes in vegetation (biomass increases during the season) and nest-site vegetation may affect microclimate and predation risk. In MARK, we fit models using a logit link and by incorporating management as a group covariate with topographic position, VOR and principal components as individual covariates. Topographic position was included as an ordinal covariate in this analysis and time of season was included as a linear effect.

The candidate set of models included the global model, which we defined as the model with the management treatment by topographic position interaction, an additive time effect and nest-site vegetation covariates (topo * rang + time + nest vegetation). We considered all combinations of these factors during model comparisons, including the constant model. We compared models using Akaike's Information Criterion adjusted for small sample size ([AIC.sub.c]; Burnham and Anderson, 1998). To account for uncertainty in model selection, we used Akaike Weights ([AIC.sub.c] [w.sub.i]) to obtain model-averaged estimates of DNS. If the addition of a parameter to the model did not result in a reduction in the deviance, the model was removed from the candidate set before model averaging (Guthery et al., 2005). Final estimates of DNS for each species were calculated based on the top models with a cumulative [AIC.sub.c] [w.sub.i] [greater than or equal to] 0.80. Daily nest survival estimates and unconditional standard errors were obtained across a range of covariate values for important nest-site covariates and topographic position and/or time if they were present in the top models.

Nest-site selection.--To explore what vegetation features were important in nest-site selection, we selected response variables that were meaningful indicators of nest-site quality from the DNS analysis (see RESULTS, DAILY NEST SURVIVAL) by identifying which model containing a vegetation covariate had the lowest [AIC.sub.c]. We used DNS as an index of habitat quality, consistent with the expectation that demographic rates such as fecundity (a function of DNS) will be higher in good-quality than poor-quality habitat. If a principal component describing nest-site vegetation was identified as a meaningful indicator of nest-site quality, we performed PCA on site and nest vegetation data simultaneously to ensure component scores were comparable. The effect of topography on nest-site selection was examined by conducting an analysis similar to that of the site vegetation analysis, except that for this analysis we limited our inferences to statistical tests of slope parameters and comparisons between nest and site vegetation at each topographic position in each management treatment. Averages of the early- and late-season vegetation surveys were calculated for each site vegetation point before this analysis. A predictor named "point type" (nest or site) was created in order to compare nest and site-wide vegetation. The full factorial model was fit with the interaction between topographic position, management treatment and point type. Mean models were fit to obtain slope estimates and statistical tests for regression lines and two sample t-tests were used to compare nest and site vegetation. Throughout the results, we report means [+ or -] SE.



Topographic position had a significant effect on nest distribution, although the distribution shifted during the season for some species (topo*season; Table 1). Early in the season a greater proportion of Dickcissel nests were found in lowlands, whereas later in the season proportionately more nests were placed in uplands than lowlands (Fig. 2a). In contrast, most Grasshopper Sparrow nests were found in midlands regardless of time of season (Fig. 2b). However, a greater proportion of nests were placed in uplands early in the season compared to later in the season (Fig. 2b). Eastern Meadowlarks also nested predominantly in midlands, but they shifted in their use of lowlands relative to uplands over the course of the season. Proportionately more nests were placed in lowlands early in the season, whereas later in the season a greater proportion of nests were placed in uplands (Fig. 2c).



Topographic position did not have a significant effect on vertical vegetation structure (VOR; Table 2). Instead, management interacted significantly with season to influence VOR, which was most evident in season-long grazed pastures (P < 0.01, Fig. 3a). The four measures of vegetation cover (grass, forb, litter and bare ground) on sites could be described by the first and second components retained from PC& (Table 3). The first component (PC1) described a gradient from litter and grass cover to bare ground, in which larger scores primarily represented greater litter and (to a lesser extent) grass cover. The second component (PC2) described forb cover exclusively, with larger component scores associated with greater forb cover {Table 3).


Litter and grass cover (PC1) were significantly affected by both management and the interaction between topographic position and time (Table 2). In terms of management effects, litter and grass cover were greatest in season-long stocked pastures that were unburned (SLSU; 0.91 [+ or -] 0.34), intermediate in season-long stocked pastures that were burned (SLSB; -0.37 [+ or -] 0.34) and least in intensive-early stocked pastures that were burned (IESB; -0.54 [+ or -] 0.34). Comparisons among management types indicate that grass and litter cover was significantly lower in IESB and SLSB than in SLSU (IESB vs. SLSB: [ts.sub.15] = -2.98, P = 0.01; IESB vs. SLSU: [ts.sub.15] = -2.65, P = 0.02). The interactive effect of topography and time is explained by the apparent increase in litter and grass cover with increasing elevation during the early part of the season (Fig. 3b), which was most pronounced in midlands and uplands (lowland: [t.sub.559] = 1.97, P = 0.05; midland: [ts.sub.558] = 5.58, P = 0.01; upland: [ts.sub.564] = 5.11, P = 0.01; Fig. 3b).

Forb cover (PC2) was likewise affected by management and the interaction between topographic position and time (Table 2). In terms of management, forb cover was greatest in IESB (0.41 [+ or -] 0.16), intermediate in SLSB (-0.12 [+ or -] 0.16) and least in SLSU (-0.27 [+ or -] 0.16). Comparisons among management types revealed that forb cover was significantly different in IESB compared to SLSB ([ts.sub.15.1] - 2.36, P - 0.03) and SLSU ([ts.sub.15] = 2.99, P = 0.01). Topographic effects on forb cover changed over the season, with higher tbrb cover in lowlands early in the season ([ts.sub.525] = 1.76, P = 0.08), but higher forb cover in uplands later in the season ([ts.sub.530] = -2.72, P = 0.01; Fig. 3c).


Nest vegetation at Dickcissel nests was best described by the first principal component (PC1), which described an inverse relationship between grass/litter cover and forb/bare ground cover (Table 3). The first two principal components were retained to describe nestsite characteristics of Grasshopper Sparrows and Eastern Meadowlarks (Table 3). For these two species, components had a similar interpretation. For PC1, litter cover increased as bare ground decreased, and as PC2 increased, cover of grass and forb increased (Table 3).


Topographic position was of little importance to Dickcissel DNS in the models we examined relative to VOR at nests and time of season (Table 4). Dickcissel DNS was described well by a constant survival rate, but there were several models that were equally parsimonious ([DELTA] [AIC.sub.c] [less than or equal to] 2.0) and the model with VOR had only a slightly smaller Akaike weight ([AIC.sub.c] [w.sub.i]) than the constant model (Table 5). Daily nest survival was predicted to decrease 4-6%/day across all VOR values after model averaging (Fig. 4a). This translates to a 6.5% vs. 22% overall nest success between nests that had the lowest observed VOR (0 cm) vs. those with the highest observed VOR (62.5 era), respectively.

Topographic position also had a minimal effect on Grasshopper Sparrow DNS compared to the effect of litter and bare ground cover (PC1) at nests and time of season (Table 4). Litter and bare ground cover (PC1) occurred in all three of the top models indicating that it was an important variable in this analysis (Table 4). The top two models also included additive effects for time of season (Table 4). Model-averaged estimates indicated an increase of 6-11% in DNS between the observed extremes in litter and grass cover (Fig. 4b). There is a 4% vs. 27% probability of nest success when observed cover of litter (PC1) was low (2.0) vs. high (-3.0), respectively. Daily nest survival for Grasshopper Sparrows also declined throughout the season, and this effect was more pronounced than that observed for Dickcissels (Fig. 4b).

Topographic position was the most important covariate considered for Eastern Meadowlarks; however, the best description of DNS was the constant rate (Table 4). Model-averaged estimates showed little change from the constant rate for any of the covariates considered likely due to the relatively high [AIC.sub.c] weight ([w.sub.i]) for the constant model (Table 4; Fig. 4c).


Vertical vegetation structure (VOR) was the most important nest-site vegetation characteristic influencing DNS for Dickcissels (see RESULTS, DAILY NEST SURVIVAL). We, therefore, included VOR as the response variable in this analysis because it represents an important habitat feature likely to affect nest-site selection in this species. Topographic position, management and type of point (nest or site vegetation) had a significant interactive effect on nest-site selection for Dickcissels (three-way interaction: [F.sub.2,715] = 3.07, P = 0.05). Nest and site VOR were different across all topographic positions for each management treatment (Table 6). Dickcissels selected nest sites with greater VOR than was typically available within sites (Fig. 5a). The mean model revealed that there was a significant decline in VOR with increasing topographic position at nest locations of Dickcissels (slope = - 3.62, [ts.sub.715] = -3.03, P < 0.01) in season-long stocked pastures that were unburned (SLSU; Fig. 5a).

Cover of litter and bareground (PC1) was the most important nest-site vegetation feature associated with nest survival for Grasshopper Sparrows (see RESULTS, DAILY NEST SURVIVAL). These features were also represented by the first principal component (PC1) derived from the combined Grasshopper Sparrow nest and site data (Table 6). In season-long stocked pastures (SLSB and SLSU), nests had greater litter cover than site averages, and these differences increased with increasing elevation (Table 5; Fig. 5b).

Although nest-site vegetation did not have much influence on model-averaged estimates of DNS for Eastern Meadowlarks, model selection indicated VOR was still the vegetation characteristic with the greatest association with DNS (Table 4). For nearly all comparisons within topographic positions, VOR at nests was greater than site averages (Table 5), which indicates that Eastern Meadowlarks were selecting sites with greater VOR than was typically available within sites. Differences between nests and sites decreased with increasing elevation (Fig. 5c), although this trend was significant only in SLSB pastures (slope = -4.62, [ts.sub.644] = -1.73, P = 0.08; Fig. 5c).


In the Flint Hills, topography differentially influenced the nesting ecology of Dickcissels, Grasshopper Sparrows and Eastern Meadowlarks. A greater proportion of Dickcissel nests were placed in lowlands than either midlands or uplands early in the season. This is consistent with previous work that suggested that Dickcissels settled earlier on lowland sites because these areas had taller vegetation and more forb cover, which are preferred nesting habitat for Dickcissels (Zimmerman, 1971). When considering this preference for forbs, the seasonal shift in Dickcissel nest placement corresponded well with the seasonal changes in the distribution of forbs among topographic positions that we observed, where forb cover was higher in lowlands early in the season but was greater in uplands later in the season (compare Fig. 2a and 3c).

For Grasshopper Sparrows and Eastern Meadowlarks, nest placement was greatest in midlands. However, proportionately more Grasshopper Sparrow nests were found in uplands early in the season compared to later, which was the opposite of what was observed for Eastern Meadowlarks. Like Dickcissels, Eastern Meadowlarks nested more frequently in lowlands early in the season (albeit secondary to midlands), but shifted to higher use of uplands as the season progressed (Fig. 2c). In contrast to Dickcissels, Grasshopper Sparrows generally occupy areas with moderate litter cover and depth (Wiens, 1973), which may explain their greater use of uplands early in the season in our study, where litter cover was greatest at that time (Fig. 4b).


Although topographic position in the Flint Hills has been identified as an important factor influencing vegetation density and productivity (Abrams et al., 1986), we did not find a significant effect of topography on vertical vegetation structure (VOR). This could be due to the frequent rainfall that occurred throughout the region in late spring and early summer of the 2004 growing season. Stronger topographic effects might be expected in years of lower rainfall, when runoff from infrequent rains has little chance to infiltrate shallow soils of uplands (Ransom et al., 1988), but can collect in lowlands, increasing the difference in soil moisture levels and productivity between lowlands and uplands. Thus, in years of frequent rains, water stress should be less of an issue in uplands, resulting in similar productivity across topographic positions. Stocking intensity could also have a substantial influence on vegetation patterns across topographic positions because cattle tend to concentrate activity at lower topographic positions and on gradual slopes (Gillen et al., 1984; Senti et al., 1985; Pinchak et al., 1991). Therefore, concentrated grazing in lowlands could also lessen the effect of topographic position on vegetation biomass.

Our results suggest that variation in nest survival cannot be attributed to topographic position for Dickcissels and Grasshopper Sparrows, but nest survival may be marginally influenced by topographic position for Eastern Meadowlarks. For Dickcissels and especially Grasshopper Sparrows, DNS was highly dependent on specific nest-site characteristics. Vegetation structure near nests provides visual concealment (Davis, 2005), and shelter from wind and sun (With and Webb, 1993), and may explain why vegetation at nests is so important for nest smwival. Dickcissels build their nests off the ground and increased vertical vegetation density around the nest probably reduces exposure to visual predators. In contrast, Grasshopper Sparrows often build domed nests on the ground, and thus exposure is probably lowest in areas with high litter cover. Grasshopper Sparrows selected sites with greater litter and grass cover in midlands and uplands of season-long stocked sites (Table 5). When considering how DNS was associated with greater vertical vegetation structure and litter cover in this species, there thus appears to be at least an indirect link between topographic position and nest survival.

In summary, topographic position may be an important factor influencing habitat quality of these grassland birds in the Flint Hills because of how it influences temporal patterns in vegetative cover, particularly of forbs, litter and grass. The distribution of nests within sites corresponded with seasonal shifts in the topographic variation of vegetation. Although topography may not have a direct effect on DNS, it may have indirect effects mediated through nest-site vegetation as a consequence of selective nest placement. Accounting for topographic variation in habitat is particularly relevant in the Flint Hills because this is one of the last great expanses of native tallgrass prairie remaining, and thus it offers perhaps the best opportunity for the management and conservation of grassland birds in this region. Many of the prairie remnants in North America are hilly areas that are poor cropland (e.g., Loess Hills of Iowa, Sand Hills of Nebraska), so understanding how topographic variation affects habitat quality may also be important for grassland management and the conservation of grassland birds beyond the Flint Hills, especially where management actions might be principally directed at a given topographic position (e.g., lowlands or slopes).

Acknowledgments.--This project was funded by a grant awarded to K. A. With (#2003-35101-13714) from the Managed Ecosystems Program of the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service. This work would not have been possible without the support of numerous private landowners throughout the Flint Hills who granted us access to their pastures. We thank Greg Smith for allowing us to operate out of Emporia State University's Ross Natural History Preserve, and Bob Hamilton for providing access and operational support at The Nature Conservancy's Tallgrass Prairie Preserve in Oklahoma. We are indebted to the dozen or so field technicians who assisted us with data collection for this project. Thanks are also due to Tammi Johnson and Nancy Leathers for GIS technical support and to Jeffrey Pontius and Brett Sandercock for assistance with statistical analyses. We dedicate this paper to the memory of Jeff Pontius, who was a dedicated mentor that motivated many students to develop a sound approach to statistical problems and helped many ecologists aspire to be better statisticians.



ABRAMS, n. D., A. K. KNAPP AND L. C. HULBERT. 1986. A ten-year record of aboveground biomass in a Kansas tallgrass prairie. Am. J. Bot., 73:1509-1515.

--AND L. C. HULBERT. 1987. Effect of topographic position and fire on species composition in tallgrass prairie in Northeast Kansas. Am. Midl. Nat., 117:442-445.

ANDERSON, M. D., E. F. SMITH AND C. E. OWENSBY. 1970. Burning bluestem range. J. Range Manage., 23:81-92.

ASKINS, R. A. 1993. Population trends in grassland, shrubland, and forest birds in eastern North America. Curt. Ornithol., 11:1-34.

BRIGGS, J. M. AND A. K. KNAPP. 1995. Interannual variability in primary production in Tallgrass Prairie: climate, soil moisture, topographic position, and fire as determinants of aboveground biomass. Am. J. Bot., 82:1024-1030.

BURNHAM, K. P. AND D. R. ANDERSON. 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York. 488 p.

Davis, S. K. 2005. Nest-site selection patterns and the influence of vegetation on nest survival of mixedgrass prairie passerines. Condor, 107:605-616.

GILLEN, R. L., W. C. KRUEGER AND R. F. MILLER. 1984. Cattle distribution on mountain rangeland in northeastern Oregon. J. Range Manage., 37:549-553.

GUTHERY, F. S., L. A. BRENNAN, M.J. PETERSON AND J. J. LUSK. 2005. Information theory in wildlife science: critique and viewpoint, J. Wildlife Manage., 69:457-465.

HARTNETT, D. C., K. R. HICKMAN AND L. E. FISHER-WALTER. 1996. Effects of bison grazing, fire, and topography on floristic diversity in tallgrass prairie. J. Range Manage., 49:413-420.

HERKERT, J. R. 1995. An analysis of Midwestern breeding bird population trends: 1966-1993. Am. Midl. Nat., 134:41-50.

HOOK, P. B. AND I. C. BURKE. 2000. Biogeochemistry in a shortgrass landscape: control by topography, soil texture, and microclimate. Ecology, 81:2686-2703.

KNAPP, A. K. AND T. R. SEASTEDT. 1998. Grasslands, Konza prairie, and long-term ecological research, p. 3-15.

In: A. K. Knapp, J. M. Briggs, D. C. Hartnett and S. L. Collins (eds.). Grassland dynamics: Longterm ecological research in tallgrass prairie. Oxford University Press, New York. 364 p.

--, J. M. BRIGGS, J. M. BLAIR AND C. L. TURNER. 1998. Patterns and controls of aboveground net primary production in tallgrass prairie, p. 193-221. In: A. K. Knapp, J. M. Briggs, D. C. Hartnett and S. L. Collins (eds.). Grassland dynamics: Long-term ecological research in tallgrass prairie. Oxford University Press, New York. 364 p.

--, J. M. BLAIR, J. M. BRIGGS, S. L. COLLINS, D. C. HARTNETT, L. C. JOHNSON AND E. G. TOWNE. 1999. The keystone role of bison in North American tallgrass prairie. Bioscience, 49:39-50.

LAUBACH, R. 1984. Breeding birds of Sheeder Prairie Preserve, west-central Iowa. Pwc. Iowa Acad. Sci., 91:153-163.

McGARIGAL, K., S. CUSHMAN AND S. STAFFORD. 2000. Multivariate statistics for wildlife and ecology research. Springer, New York. 283 p.

OWENSBY, C., R. COCHRAN, K. ANDERSON, E. SMITH AND E. VANZANT. 1995. 50-years of range research revisited. Cooperative Extension Service, Kansas State University, Manhattan, Kansas. 57 p. Available online: (6 Dec. 2007).

PETERJOHN, B. G. AND J. R. SAUER. 1999. Population status of North American grassland birds from the North America Breeding Bird Survey, 1966-1996. ,Stud. Avian Biol., 19:27-44.

PIEPER, R. D. 2005. Grasslands of central North America, p. 221-263. In: J. M. Suttie, S. G. Reynolds and C. Batello (eds.). Grasslands of the world. Food and Agricultural Organization of the United Nations, Rome, Italy. 514 p.

PINCHAK, W. E., M. A. SMITH, R. H. HART AND J. W. WAGGONER, JR. 1991. Beef cattle distribution patterns on foothill range. J. Range Manage., 44:267-275.

RANSOM, M. D., C. W. RICE, T. C. TODD AND W. A. WEHMUELLER. 1998. Soils and soil biota, p. 48-66. In: A. K. Knapp, J. M. Briggs, D. C. Hartnett and S. L. Collins (eds.). Grassland dynamics: Long-term ecological research in tallgrass prairie. Oxford University Press, New York. 364 p.

RENFREW, R. B. AND C. A. RIBIC. 2002. Influence of topography on density of grassland passerines in pastures. Am. Midl. Nat., 147:315-325.

ROBEL, R.J., J. N. BRIGGS, A. D. DAYTON AND L. C. HULBERT. 1970. Relationships between visual obstruction measurements and weight of grassland vegetation. J. Range Manage., 23:295-297.

SAMSON, F. AND F. KNOPF. 1994. Prairie conservation in North America. BioScience, 44:418-421.

SENFT, R. L., L. R. RITTENHOUSE AND R. G. WOODMANSEE. 1985. Factors influencing patterns of cattle grazing behavior on shortgrass steppe. J. Range Manage., 38:82-87.

TOWNE, E. G., D. C. HARTNETT AND R. C. COCHRAN. 2005. Vegetation trends in tallgrass prairie from bison and cattle grazing. Ecol. Appl., 15:1550-1559.

U.S. DEPARTMENT OF AGRICULTURE. 2004a. 2002 Census of agriculture. Kansas State and County Data. Vol. 1, Geographic Area Series Part 16. USDA, National Agricultural Statistics Service, Washington, D.C. 711 p. Available online: KSVolume104.pdf (6 Dec. 2007).

--. 2004b. 2002 Census of agriculture. Oklahoma State and County Data, Volume 1, Geographic Area Series Part 36. USDA, National Agricultural Statistics Center, Washington, D.C.Available online: (6 Dec. 2007).

VINTON, M. A., D. C. HARTNETT, E.J. FINCK AND J. M. BRIGGS. 1993. Interactive effects of fire, bison (Bison bison) grazing and plant community composition in tallgrass prairie. Am. Midl. Nat., 129:10-18.

WHITE, G. C. AND K. P. BURNHAM. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study, 46(Suppl.):S120-S139.

WIENS, J. A. 1973. Pattern and process in grassland bird communities. Ecol. Monogr., 43:237-270.

WITH, K. A. AND D. R. WEBB. 1993. Microclimate of ground nests: the relative importance of radiative cover and wind breaks for three grassland species. Condor, 91:401-413.

ZIMMERMAN, J. L. 1971. The territory and its density-dependent effect in Spiza americana. Auk, 88:591-612.

--. 1982. Nesting success of Dickcissels in preferred and less preferred habitats. Auk, 99:292-298.

--. 1993. The birds of Konza: The avian ecology of the tallgrass prairie. University of Kansas Press, Kansas. 198 p.

--. 1997. Avian community responses to fire, grazing and drought in the tallgrass prairie, p. 167-180. In: F. L. Knopf and F. B. Samson (eds.). Ecology and conservation of Great Plains vertebrates. Springer-Verlag, New York. 320 p.


Division of Biology, Kansas State University, Manhattan 66506

(1) Present address: Natural Resources and Environmental Sciences/MS 186, 1000 Valley Road, University of Nevada, Reno 89512. Telephone: (775)784-6558; e-mail:

(2) Present address: Department of Biological Sciences, Campus Box 4050, Emporia State University, Emporia, Kansas 66801. Telephone: (620)341-5339; e-mail:

(3) Corresponding author: Telephone: (785)532-5040; FAX (785)532-5652; e-mail:
TABLE 1.--Summary of contingency table analysis (Cochran-Mantel-Hanzel
correlation statistic, [X.sup.2]) of factors affecting the
distribution of nests for three grassland birds in the Flint Hills

Species                n    Factors (a)       [X.sup.2]   df    P

Dickcissel            105   Topo*Season*Mng   3.36        1    0.05
                            Topo*Season       2.74        1    0.10
                            Topo*Mng          0.86 (b)    2    0.65

Grasshopper Sparrow   102   Topo*Season*Mng   2.66        1    0.10
                            Topo*Season       2.58        1    0.11
                            Topo*Mng          4.92 (b)    2    0.09

Eastern Meadowlark     42   Topo*Season*Mng   2.50        1    0.11
                            Topo*Season       3.09        1    0.08
                            Topo*Mng          10.36 (b)   2    0.01

(a) Topo = topographic position (lowland, midland, upland), Season =
early vs. late, Mng = management treatment (IESB, SLSB, SLSU)

(b) Row mean scores statistic

TABLE 2.--Summary of factors affecting site vegetation on grazed
grasslands in the Flint Hills

characteristic   Model (a)   Factors (b)       F      df (c)     P

VOR (cm)         Factorial   Topo*Mng*Season   0.77   2, 581    0.46
                   Final     Topo              2.14   1, 598    0.14
                             Mng*Season        9.85   2, 577   <0.01

Grass-Litter     Factorial   Topo*Mng*Season   0.92   2, 561    0.40
(PC1)              Final     Topo*Season       2.79   1, 563    0.10
                             Mng               5.34    2, 15    0.02

Forb (PC2)       Factorial   Topo*Mng*Season   1.82   2, 529    0.16
                   Final     Topo*Season       7.77   1, 529   <0.01
                             Mng               4.97    2, 15    0.02

(a) All models were fit with all lower-level interactions and main
effects; only tests for higher-order effects are shown

(b) Topo = topographic position (lowland, midland, upland); Mng =
management treatment (IESB, SLSB, SLSU); Season = early vs. late

(c) df = Sattertbwaite degrees of freedom (numerator, denominator)

TABLE 3.--Principal Component Analysis of site vegetation and nest
vegetation data for grassland birds nesting in the Flint Hills.
DICK = Dickcissel nests, GRSP = Grasshopper Sparrow nests,
EAME = Eastern Meadowlark nests. Only retained components
(PC1 and/or PC2) are displayed. Bold values indicate significant

              Site (n = 1139)   DICK (n = 147)
measure         PC1     PC2           PC1

Grass          0.40#   0.13         -0.41#
Forb           0.05    0.99#         0.48#
Litter         0.64#   0.09         -0.55#
Ground        -0.66#  -0.07          0.55#
Eigenvalue     1.71    1.00          2.15
  explained    0.43    0.25          0.54

               GRSP (n = 115)     EAME (n = 75)
measure          PC1      PC2      PC1      PC2

Grass          -0.06     0.77#    0.06     0.82#
Forb            0.23     0.62#    0.37     0.50#
Litter         -0.69#   -0.05    -0.68#    0.08
Ground          0.69#   -0.17     0.64#   -0.29
Eigenvalue      1.60     1.11     1.70     1.08
  explained     0.40     0.28     0.43     0.27

Note: Bold values indicated with #.

TABLE 4.--Top daily nest survival models for Dickcissels, Grasshopper
Sparrows and Eastern Meadowlarks nesting in grazed grasslands of the
Flint Hills

                                          Model statistics (a)

Species             Model         k    Dev      [DELTA]     [w.sub.i]
                siructure (b)                 [AIC.sub.c]

Dickcissel        Constant        1   430.9       0.0         0.30
(n = 147)            VOR          2   429.0       0.1         0.28
                 VOR + Time       3   427.2       0.4         0.25
                    Time          2   430.0       1.2         0.17

Grasshopper      Time + PC1       3   313.1       0.0         0.47
Sparrow       Time + PC1 + Topo   4   312.1       1.0         0.29
(n = 115)            PC1          2   316.5       1.3         0.24

Eastern           Constant        1   254.8       0.0         0.39
Meadowlark          Topo          2   253.5       0.7         0.27
(n = 75)            Time          2   254.4       1.7         0.17
                     VOR          2   254.5       1.7         0.17

(a) Dev = deviance, k = number of parameters, Topo = topographic
position (lowland, midland, upland)

(b) Time = linear time trend; VOR = visual obstruction reading (cm),
a measure of vertical vegetation structure; PCI = Principal Component
1 from analysis of nest data; Constant = point estimate from all data

TABLE 5.--Comparison of nest vegetation and site vegetation at
different topographic positions in the Flint Hills for three
grassland birds. IESB = intensive early-stocked burned, SLSB =
season-long stocked burned, SLSU = season-long stocked unburned
pasture. DICK = Dickcissel, GRSP = Grasshopper Sparrow, EAME =
Eastern Meadowlark. P-values are ns if >0.10, * [less than or equal
to] 0.10, ** [less than or equal to] 0.05, and *** [less than or equal
to] 0.005


                         Veg measure (a)            Statistics

Species   Topo (b)      Nest         Site       t     df (c)    P

DICK        low      25.0 (3.1)   15.2 (2.6)    5.6    716     ***
            mid      26.0 (2.8)   15.1 (2.7)   10.1    718     ***
             up      27.0 (3.1)   14.9 (2.6)    6.7    716     ***
GRSP        low      -0.5 (0.4)   -0.7 (0.4)    1.1    692     ns
            mid      -0.5 (0.4)   -0.6 (0.3)    1.2    692     ns
             up      -0.5 (0.4)   -0.6 (0.4)    0.3    692     ns
EAME        low      23.8 (3.2)   15.4 (2.6)    4.5    643     ***
            mid      22.8 (2.9)   15.1 (2.5)    5.8    644     ***
            tip      21.7 (3.4)   14.7 (2.6)    3.4    643     ***


                         Veg measure (a)            Statistics

Species   Topo (b)      Nest         Site       t     df (c)    P

DICK        low      26.6 (3.7)   15.5 (2.7)    4.6    714     ***
            mid      27.9 (2.9)   15.8 (2.6)    9.2    716     ***
             up      29.2 (2.8)   16.1 (2.6)   10.3    722     ***
GRSP        low      -0.1 (0.6)   -0.5 (0.4)    0.9    693     ns
            mid       0.1 (0.4)   -0.5 (0.4)    2.4    694     **
             up       0.3 (0.4)   -0.4 (0.4)    3.8    694     ***
EAME        low      29.3 (6.5)   16.4 (2.6)    2.6    642     **
            mid      24.7 (3.9)   16.0 (2.5)    3.4    643     ***
            tip      20.1 (3.1)   15.6 (2.5)    2.7    645     **


                         Veg measure (a)            Statistics

Species   Topo (b)      Nest         Site       t     df (c)    P

DICK        low      39.3 (3.1)   20.6 (2.7)   10.9    715     ***
            mid      35.7 (2.8)   20.0 (2.7)   13.6    717     ***
             up      32.1 (3.2)   19.4 (2.6)    7.0    717     ***
GRSP        low       1.0 (0.4)    0.9 (0.4)    0.1    693     ns
            mid       1.2 (0.4)    1.0 (0.4)    1.7    694
             up       1.4 (0.4)    1.0 (0.4)    1.8    694
EAME        low      25.2 (3.7)   20.7 (2.6)    1.9    642
            mid      25.3 (2.8)   20.0 (2.5)    4.4    642     ***
            tip      25.3 (3.2)   19.3 (2.6)    3.3    642     ***

(a) Vegetation measures are means (SE) and vain by species: DICK =
vertical vegetation structure (VOR), GRSP = grass-litter cover (PC1,
Table 6), EAME = VOR

(b) Topo = topographic position (low = lowland, mid = midland, up =

(c) df = Satterthwaite degrees of freedom

TABLE 6.--Principal Component Analysis of pooled site and Grasshopper
Sparrow nest vegetation data (n = 1245). Boldface values indicate
significant loadings

Vegetation measure                  PC1        PC:2

Grass                              0.39#        0.24
Forb                               0.05         0.96#
Litter                             0.64#       -0.11
Ground                            -0.66#        0.10
Eigenvalue                         1.67         1.01
Proportion variation explained     0.42         0.25

Note: Boldface values indicate significant loadings indicated with #.
COPYRIGHT 2008 University of Notre Dame, Department of Biological Sciences
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2008 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Frey, Christopher M.; Jensen, William E.; With, Kimberly A.
Publication:The American Midland Naturalist
Article Type:Report
Geographic Code:1USA
Date:Jul 1, 2008
Previous Article:Predominant use of windthrows by nesting eastern woodrats (Neotoma floridana) in the South Carolina coastal plain.
Next Article:Geographic variation in flowering responses to fire and season of clipping in a fire-adapted plant.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters