Abstract
With most native prairie now lost, focus has turned to reconstructing grasslands with the aim of replicating ecosystem function for grassland fauna and flora. We report the results of a seven-year study (2008–2014) on the use of four floristically different prairie reconstructions by grassland birds. The study was conducted near Austin, Minnesota on a 64.75-ha (160-acre) Wildlife Management Area. Prairie reconstruction was implemented in 2005 on land previously used for row crops, and the study site consisted of eight identically sized (4.5 hectare), side-by-side of replicates of four distinct planting mixes. We used fixed-point transect surveys to document differences in the number of species and individuals of grassland birds between plot types. We found differences in the abundance of individuals and diversity of bird species between plot types, with fewer bird species and individuals present in lower plant-diversity plot types compared to higher plant-diversity plots. Eighteen species of grassland birds were recorded on the site, with none exclusively restricted to any plot type. Grassland obligate bird species, including Sturnella magna (Eastern Meadowlark), Dolichonyx oryzivorus (Bobolink), Ammodramus savannarum (Grasshopper Sparrow), Ammodramus henslowii (Henslow’s Sparrow), and Passerculus sandwichensis (Savannah Sparrow), were documented using low floristic diversity, cool-season grass plots. Other species, such as Spiza americana (Dickcissel), Melospiza melodia (Song Sparrow) and Spinus tristis (American Goldfinch), showed preferences for high diversity, forb dominated plots. Our results indicate that planting a mosaic of restoration mixes may be the most successful at attracting and retaining the highest diversity of bird species.
- cold-season grasses
- Dolichonyx oryzivorus
- Bobolink
- ecosystem function
- grassland birds
- forbs
- warm-season grasses
Restoration Recap
Bird use of a reconstructed grassland, consisting of plots with four distinct seed mixes planted side-by-side, was compared using fixed-transect surveys across seven breeding seasons.
Grassland songbirds showed variability in their preferences for reconstruction types with no one type hosting all species of birds, nor were any species restricted to one particular plot type.
Obligate grassland species including Sturnella magna (Eastern Meadowlark), Dolichonyx oryzivorus (Bobolink), and Ammodramus henslowii (Henslow's Sparrow) were frequently observed using the lower diversity, non-native, cool-season grass plots.
In order to maximize avian diversity and overall ecosystem function, reconstructions should consider structural diversity of vegetation as much as floristic diversity/rich- ness and the potential value of small, discrete patches of habitat should be considered.
With the loss of nearly two-thirds of native mixed grass and over 99 percent of tall grass prairie in North America (Samson and Knopf 1996, the steep decline in many native prairie bird species is a cause for serious concern (Herkert 1995, Igl and Johnson 1997, Sauer et al. 2014). In Minnesota and Wisconsin, less than one percent of native grasslands remain, largely from habitat conversion to agriculture (Sample et al. 2003). Compounding matters, more recent conversion from grass-based crops (e.g., hay and pasture) to row crops and “fence-to-fence” farming has further limited habitat available to prairie birds (Sample and Mossman 1997, Ribic et al. 2009).
Research on remnant prairies in the Midwest has suggested that remnants provide important habitat for many grassland bird species (Hoffman and Sample 1988, Sample and Hoffman 1989, Herkert and Glass 1999, Fletcher and Koford 2002, Herkert et al. 2003) and may be superior to reconstructions, supporting higher avian density and species richness (Bakker and Higgins 2009, Ramirez-Yáñez et al. 2011). Other work has found that reconstructed grasslands can support higher abundances of specific grassland bird species such as Dolichonyx oryzivorus (Bobolink), Sturnella magna (Eastern meadowlark), and Ammodramus henslowii (Henslow's Sparrow [Ribic et al. 2009]). With declines in remnant prairies, it is necessary to consider whether the floristic composition of reconstructions has a significant impact on their suitability and selection by prairie bird species.
Numerous studies have suggested an association between the increasing area of reconstructed grasslands and an increase in songbird numbers (Johnson and Igl 1995, Herkert 1998, Haroldson et al. 2006, Herkert 2007). Planted grasslands such as those enrolled in the Conservation Reserve Program (CRP) can have multiple positive benefits on birds (Johnson and Igl 1995, Best et al. 1997, Delisle and Savidge 1997, Koford 1999, Johnson 2000, Drum et al. 2015). However, many CRP type installations are planted with a mixture of native warm-season grasses and minor amounts of a few native forbs. In contrast, prairie remnants in the upper Midwest are commonly characterized by more than 100 plant species (Wovcha et al. 1995, Henderson 1998). It has been our observation that even abandoned agricultural lands show re-growth as a mixture of both non-native coolseason grasses and “weedy” native species, containing ten or more species, an observation supported by other authors (Fischer et al. 2012, Navarro and Pereira 2012, Queiroz et al. 2014).
While low diversity installations may be cost effective, these planting recipes tend to create communities that climax as solid stands of grass and structural components that favor a sub-set of grassland nesting birds. There has long been recognition of the value of planting a diversity of grassland species (e.g., Weber 1999), and there is increasing recognition of the need to create plantings with forbs that provide nectar sources for native bees, honeybees and butterflies (e.g., Isaacs et al. 2008, Bennett and Gratton 2013). These so called “pollinator mixes” represent a positive step toward creating more functional habitat, but again the structural components of these plantings will favor some prairie bird species more than others. If the goal of grassland reconstructions is to reestablish habitat, it is critical to consider the importance of both floristic and structural diversity (Schottler et al. 2008, Bakker and Higgins 2009, Murray 2014, DeGolier et al. 2015).
The study presented here made use of a unique grass- land/prairie reconstruction in southern Minnesota. Eight identically sized (4.5 hectare) replicates, representing four floristically distinct reconstructions, planted side-by-side and of the same age, were used to compare bird species preferences for each reconstruction type. Plots differed in both vegetation structure and floristic diversity. Low floristic diversity plots included reconstructions dominated by cool-season and warm-season grasses and contrasted with two higher diversity reconstruction types that were forb-rich but with different proportions of taller forb species. The objectives of this study were to (1) determine the bird diversity and abundance in each of four types of reconstructions and using that information, and (2) compare the grassland bird communities across the mosaic of reconstruction types when individuals are given the opportunity to choose from neighboring habitats. The study site was surrounded by corn and soy row-crops, and thus there was no influence from adjacent grassland habitats.
Methods
This study was conducted on the 64.75-ha Schottler Wildlife Management Area, located 10.5 km (6.5 miles) south of Austin, Minnesota (43°59′97.71″ N and 92°92′45.58″ W). Previously used for row-crop farming, it was replanted as a grassland/prairie using broadcast seeding into soybean stubble during fall 2005 in cooperation with the Minnesota Department of Natural Resources, Pheasants Forever, and private conservation partners. The site had been in a cornsoy rotation for at least 25 years prior to the native prairie reconstruction. All adjacent lands continue as row-crop, corn-soy rotation.
A suite of 12, 4.5-hectare plots were planted with one of four reconstruction mixes (Figure 1). We excluded four plots from the study due to their proximity to a neighboring farmstead and encroachment by a seasonal wet meadow. The reconstruction represents the only grassland habitat in the Wildlife Management Area, other than road ditches, larger than a few acres within a one-mile radius. Eight plots were selected for this study and represented a block of side-by-side reconstructions with duplicate plots of each reconstruction type (Figure 1). Two plot types were dominated by grasses (Table 1). One was predominantly tall (> 1 m), native warm-season grass (WSG) such as Andropogon gerardii (big bluestem) and Sorghastrum nutans (Indian grass). The other grass-dominated plot type consisted of short, non-native, cool-season grasses (CSG) such as Dactylisglomerata (orchard grass), Festuca pratensis (meadow fescue), and Phleumpratense (Timothy grass). The other two plot types were dominated by forbs, one with a greater proportion of tall forb species such as Silphium perfoliatum (cup plant), Silphium laciniatum (compass plant) and Symphyotrichum novae-angliae (New England aster). Other than the differences in height, both forb (F) and tall forb (TF; Figure 1) plots were characterized by similar species such as Monarda fistulosa (bergamot), Zizia aurea (golden Alexander), Solidago rigida (stiff goldenrod), Penstemon digitalis (white beardtongue), Rud- beckia hirta (black-eyed Susan), Ratibidapinnata (yellow coneflower), Liatris pycnostachya (prairie blazing star), and Gentiana alba (cream gentian). The forb-dominated areas had over 45 species, with forbs contributing greater than 70 percent of all rooted plants (Table 1). All the plots were side-by-side and the entire site was bordered by cropland.
Planting design at the Schottler Wildlife Management Area, south of Austin, MN, USA. CSG = coolseason grass-dominated fields, WSG = warm-season grass-dominated fields, F = forb fields lacking Silphium perfoliatum (cup plant) and Silphium Iaciniatum (compass plant), and TF = forb fields characterized by the presence of S. perfoliatum and S. Iaciniatum. Both F and TF have greater than 65% forbs by rooted abundance. Each plot was approximately 4.5 hectares in size. The study utilized approximately half of total Wildlife Management Area. Plots inside the light boundary were used for avian surveys.
Floristic summary of study plots at the Schottler Wildlife Management Area, south of Austin, MN, USA. All plots were broadcast-seeded in the fall of 2005. Floristic inventories were conducted during the sixth growing season, 2011. Metrics presented are based on rooted abundance. The specific details of floristic differences between the plots are less important than the fact that habitat type created in each plot is readily apparent. The species composition and structural differences among the plots is great enough that the boundaries of the plots are visible in aerial photos (see Figure 1). CSG = cool-season grass-dominated fields, WSG = warm-season grass- dominated fields
Floristic Surveys
We conducted post-planting floristic surveys in the summer of 2011, the sixth growing season. Four 100-m transects were randomly selected in each of the four plot types. Because duplicate plot types were sampled for avian use, these same plots were evaluated floristically, thus floristic descriptions are represented by eight transects. We counted rooted abundance of all species in a 31.6 cm × 31.6 cm quadrat placed every 10 m along the 100-m transects. Rooted abundance data of each species were pooled for the four transects and used to calculate Simpson's and Shannon's Diversity indices for each plot type (Table 1). The average number of rooted individuals of all species per unit area provided a comparison of the plant density between the different plantings (Table 1). Estimates of species richness were calculated from pooled transects and augmented by a 15-minute random walk within 100 m of each transect. As an additional qualitative metric to compare the floristic and related structural difference between the plots, we binned the relative abundance of grasses in each quadrat to calculate an estimate of grass dominance (% grasses) in each plot and by difference, the relative abundance of forbs (Table 1).
Avian Surveys
Between 2008 and 2014 we surveyed the bird species present on each of the plots using fixed point transects between sunrise and 11:00 a.m. Surveys were conducted on each plot approximately every two weeks between May 28 and August 13 and consisted of transects extending the width of each plot (approximately 100 m on either side of the transect). By limiting the width of the transect, we minimized the problems associated with long-range detection of some grassland species (Diefenbach et al. 2003). Three points (two points 10 m inside the edges and a center point) within each of the plots represented fixed stopping points along the 200 m transect. Observers stood at each point for five minutes and counted all birds seen or heard within the plot from that location. We walked transects slowly (approximately 1 km/h) to identify birds seen or heard between points. Due to vegetation differences between each plot, it was clear to observers where one plot type ended and the next began. We avoided double counting by recording bird movements. One transect was conducted per plot and was located in the approximate center of the plot.
Statistical Analysis
We hypothesized that floristic (and associated structural) differences in plot types would influence avian species richness and abundance. For each plot type we calculated the mean number of all birds observed as well as species richness. Values are presented as means ± standard error. Bird richness estimates included only species that were considered facultative or obligate grassland species (Vickery et al. 1999). Since all plots were of identical size, we were able to compare means for each plot type directly. We used ANOVAs to determine whether there were differences between years or replicates. When no effect of replicates was observed, plot types were compared using ANOVA with repeated measures. A Bonferroni post hoc test was used to determine where differences in species richness and abundance occurred among grassland species. All statistical tests were considered significant atp < 0.05. Data were analyzed using SPSS (v. 23.0.0.0, IBM Corp., Armonk NY).
Results
Results observed were consistent between years and across replicates of the same plot types with no effect of year or replicate present. Plots dominated by cool-season grasses averaged fewer bird species (CSG; x̄ = 2.37 ± 0.02, n = 70) per census compared to warm-season (WSG; x̄ = 3.38 ± 0.02, n = 70), tall forb (TF; x̄ = 4.40 ± 0.016, n = 70), and forb (F; x̄ = 4.26 ± 0.02, n = 70) fields. The difference in number of species between plots was significant (ANOVA, F3,18 = 20.7, p < 0.01). Only the difference between mean number of species per census in TF compared to F fields was not significant according to Bonferroni post hoc test.
Cool-season grass plots also averaged fewer individuals, with fewer recorded per census (x̄ = 6.24 ± 0.06, n = 70) compared to WSG (x̄ = 9.36 ± 0.07, n = 70), TF (x̄ = 14.31 ± 0.011, n = 70) and F (x̄ = 13.44 ± 0.09, n = 70). All differences were significant (ANOVA, F3,18 =13.5, p < 0.01), except for differences between CSG and WSG and between TF and F (Bonferroni).
Species Richness
A maximum of 15 grassland bird species were observed across the site during at least one of the seven sampling seasons. Overall yearly richness averaged 13.4 ± 0.22 species (n = 7 years). Cool-season grass plots averaged the lowest species richness (x̄ = 7.14 ± 0.17) compared to WSG (x̄ = 9.7 ± 0.33), F (x̄ = 9.86 ± 0.17), and TF (x̄ = 9.86 ± 0.13). The difference between CSG and other plot types was significant (ANOVA, F3,24 = 5.54, p = 0.004). Overall, 18 facultative or obligate grassland species were recorded using the plots during the course of the study.
Species observed in multiple years on all field types included Bobolink, Spiza americana (Dickcissel), Cistotho- rus platensis (Sedge Wren), Geothlypis trichas (Common Yellowthroat), Ammodramus savannarum (Grasshopper Sparrow), Passerculussandwichensis (Savannah Sparrow), Henslows Sparrow, and Melospiza melodia (Song Sparrow). Spinus tristis (American Goldfinch) and Agelaius phoeniceus (Red-winged Blackbird) were observed primarily on forb-dominated plots and WSG with sightings on CSG occurring only in 2013. In the first three years of the study (2008-2010), Eastern Meadowlarks were only documented using CSG grass plots. However, beginning in 2011, individuals were seen regularly on F and TF fields including males using S. laciniatum and volunteer sapling Populus deltoides (cottonwood) as singing perches. Additional species recorded using multiple field types, although not commonly seen, included Phasianus colchicus (Ringnecked Pheasant), Circus cyaneus (Northern Harrier), Pooecetesgramineus (Vesper Sparrow), and Molothrus ater (Brown-headed Cowbird). A single sighting of Bartra- mia longicauda (Upland Sandpiper) was recorded in a TF reconstruction and a Buteo swainsoni (Swainsons Hawk) was observed hunting over an F plot.
Patterns of Note
For all species, individuals were recorded on all plot types across the span of the study (2008-2014), although sightings were rare in some planting mixes. For example, Common Yellowthroat (n = 575) sightings occurred on CSG plots in less than 1% of observations while 80.6% of sightings occurred in either F or TF plots. Of species that were sighted at least 100 times during the study, only Sedge Wrens (ANOVA, F3,24 = 1.58, p = 0.22) and Bobolinks (ANOVA, F3,24 = 2.50, p = 0.08) were not associated with a particular planting mix. American Goldfinches, Red-winged Blackbirds, Song Sparrows, and Common Yellowthroats all exhibited strong associations with F and TF planting mixes (ANOVA, F3,24 = 3.86, 81.6, 16.2, and 33.2, respectively, all p < 0.02). In contrast, Grasshopper Sparrows (50.7%, n = 138) and Savannah Sparrows (63.4%, n = 220) were observed largely in CSG reconstructions (ANOVA, F3,24 = 3.6 and 24.1 respectively, both p < 0.03, Table 2) while Henslows Sparrow appeared on CSG and WSG reconstructions (85.6%, n = 14, NS).
Comparison of preferences for reconstruction type among 14 grassland bird species present on the Schottler Wildlife Management Area, Austin, MN. Averages were calculated using mean number of sightings by plot type for each year of census (2008-2014). Years in which a species was not sighted on the study site were omitted from the calculations. * indicate a significant difference among plots in the number of sightings for that species. Sightings reported as percent of total sightings across all seven seasons. CSG = cool-season grass-dominated fields, WSG = warm-season grass-dominated fields, F = forb fields lacking Silphium perfoliatum (cup plant) and Silphium laciniatum (compass plant), and TF = forb fields characterized by the presence of S. perfoliatum and S. laciniatum.
Dickcissels exhibited an apparent large variation in population size during the study. Present on all four reconstruction types in 2008, and with 79.8% (n = 99) of the individuals recorded on the F (42.4%) or TF (37.4%) plots, the number of sightings declined through 2010 (2009, n = 71; 2010, n = 47) before individuals disappeared from the site entirely in 2011 (Table 3). Dickcissels were most frequently recorded on F and TF plots although they were observed on WSG on 28 ± 4% of surveys between 2008 and 2010. No individuals were recorded between May and August in 2011, with individuals only recorded 16 times in 2012 and once in 2013. A single singing male was observed nearby, but outside of the study area, and not included in the census counts in 2014. Apparent Dickcissel declines may have been related to an increase in Bobolinks numbers or regional variations in population patterns, as has been documented elsewhere (Heckert 1994). A nearby nature center reported similar fluctuations in Dickcissel populations in the absence of Bobolinks (L. Dolphin, Hormel Nature Center, pers. comm.)
Annual variation in individual sightings by plot type for two obligate grassland species recorded on the Schottler Wildlife Management Area. Sightings reported as percent of total sightings within a season. CSG = cool-season grass-dominated fields, WSG = warm-season grass-dominated fields, F = forb fields lacking Silphium perfoliatum (cup plant) and S. laciniatum (compass plant), and TF = forb fields characterized by the presence of S. perfoliatum and S. laciniatum.
Bobolinks were present consistently throughout the study and sightings ranged from a high of 138 in 2010 to a low of 19 in 2008. Bobolinks were present in CSG plots an average of 29% of the time (Table 2) but varied considerably across years. In some years, Bobolinks were more strongly associated with CSG plots, with occurrences on these plots accounting for 40 to 50% of total sightings. However, in some years this preference was less distinct and the percentage of observations on the CSG plot types dropped to as low as 14.5% of total sightings (Table 3). For example, in 2008 Bobolinks were most likely to be found on CSG plots (52% of the sightings) but declined to 17.9% in 2009 and 14.5% in 2010. In years where Bobolinks observations on CSG decreased, there was a simultaneous increase in their observed frequency in F and TF plots (Table 3).
Discussion
The results of this study suggest that all planting types evaluated, including non-native CSG reconstructions, serve as important habitat for obligate grassland species. Nearly all species were seen on all plot types, although some exhibited an apparent preference for particular planting mixes, based on the frequency of sightings. In some cases, associations were pronounced. Common Yellowthroat sightings occurred primarily within forb and tall forb reconstructions with 80.6% (n = 575) of individuals observed in these field types. American Goldfinches (71.4% n = 63), Song Sparrows (84.1%, n = 357), and Dickcissels (83.5%, n = 234) were three other species with notable links to forb-based field types (Table 2).
Our results indicate that in many cases, avian species are not limited to particular reconstruction types, nor do they necessarily favor planting with high floristic diversity, but that planting a diversity of mixes and creating a variety of vegetation structures may be the most successful at attracting and retaining the highest diversity of bird species. This arrangement allows for species to sort themselves according to specific habitat preferences, even within relatively small reconstructions. Jaster et al. (2013) and Murray (2014) have noted that plantings with high densities of bunchgrasses may preclude the use of these plantings by species of conservation concern such as Bobolinks and Eastern Meadowlarks. While we documented lower use of bunchgrass-dominated plots (WSG) by these species, use of these plot types was not statistically different compared to shorter, more open, cool-season grass plantings or forb-dominated fields. Of note is the trend among Eastern Meadowlarks and Grasshopper Sparrows to use shorter vegetation types (CSG and F), matching reported preferences for these species (Herkert 1994; Table 2). While CSG plots had the lowest floristic diversity, they also represented the lowest overall vegetation height of the four plot types and were also the plot type where Savannah Sparrows were most often recorded. This finding supports those of others that found a preference by Savannah Sparrows for shorter vegetation types (Sample 1989, Bakker and Higgins 2009). Other authors have also reported preferences of some bird species for non-native, sparsely planted cool-season grasses (McCoy 2001a, Davis et al. 2013, Murray 2014).
While we found no effect of year on the number of individuals or species recorded on individual plots, we recognize that age of planted grasslands may be a factor (McCoy et al. 2001b). As plant species become established, some disappear, others become established and spread, and they undergo changes in cover type, density, and thickness (Bakker et al. 2004). When the plots were first surveyed in 2008, three years post-planting, Common Yellowthroat were recorded exclusively on forb-dominated fields. By 2010, 22% of sightings were from WSG plots, and by 2014, 29% of sightings were from WSG plots. We believe this increase in use of WSG plots by this species was due at least in part to the establishment of invasive Melilotus albus (white sweetclover) on the site, providing patches of tall, dense habitat attractive to Common Yellowthroat (Bakker et al. 2002, 2004).
While the goal for grassland habitat may be to preserve large, unfragmented habitat (Herkert 1994, Drum et al. 2015), smaller patches along with human-made grasslands represent a significant proportion of the remaining habitat, and these can be equally viable (Weidman and Litvaitis 2011, Lituma et al. 2012, Jaster et al. 2014). For species that are density-dependent or area-sensitive, small-scale plantings may not attract large numbers of these species. However, as this study demonstrates, even smaller-scale reconstructions can have a role in promoting the recovery of a wide variety of species.
We encourage managers to consider planting a diversity of habitat types, even including some that may not be floristically rich but offer a different habitat structure, and even if the area is small. Our research, and that of others, suggests that while habitat requirements for many avian species are somewhat flexible, providing opportunities for species to sort themselves according to a variety of distinct grassland plantings enhances overall diversity in a way that even large, floristically diverse but structurally homogeneous reconstructions cannot.
Meadowlark. Colton B.P. 1903. Zoology: Descriptive and Practical. Boston, MA: D.C. Heath & Co. The Florida Center for Instructional Technology, fcit.usf.edu.
Acknowledgments
We thank the Minnesota Department of Natural Resources for granting us access to the Schottler Wildlife Management area and granting their support and cooperation in reconstructing the grasslands in this area to allow for comparative study. We also thank Douglas Johnson and two anonymous reviewers for their comments that helped to improve the manuscript. Support for this work was provided by Bethel University and the Saint Croix Watershed Research Station.
