Grassland bird response to harvesting switchgrass as a biomass energy crop

• Amber M. Roth^{a, ,} ,
• David W. Sample^b,
• Christine A. Ribic^c,
• Laura Paine^d,
• Daniel J. Undersander^e,
• Gerald A. Bartelt^b

• ^a Wisconsin Department of Natural Resources, 107 Sutliff Avenue, Rhinelander, WI 54501, USA
• ^b Wisconsin Department of Natural Resources, 1350 Femrite Drive, Monona, WI 53716, USA
• ^c USGS Wisconsin Cooperative Wildlife Research Unit, Department of Wildlife Ecology, University of Wisconsin, Madison, WI 53706, USA
• ^d University of Wisconsin Extension-Columbia County, 120 West Conant Street, P.O. Box 567, Portage, WI 53901, USA
• ^e Department of Agronomy, University of Wisconsin, Madison, WI 53706, USA

• Received 11 November 2003. Revised 15 October 2004. Accepted 11 November 2004. Available online 22 January 2005.

• http://dx.doi.org/10.1016/j.biombioe.2004.11.001, How to Cite or Link Using DOI

• Permissions & Reprints

---

Keywords

• Biomass energy;
• Grassland birds;
• Panicum virgatum;
• Switchgrass

---

1. Introduction

The combustion of perennial grass biomass to generate electricity may be a promising renewable energy option in the Midwest [1]. In addition to the air quality benefits of replacing fossil fuels in the combustion process, production of biofuels can have ecological benefits if they replace less sustainable, more intensively managed crops on the landscape. If perennial grasses are established as biomass energy crops and sited in ecologically appropriate locations on the landscape, their incorporation into agricultural systems could provide extensive grassland bird habitat and address soil and water quality concerns, in addition to generating renewable power [2] and [3].

Switchgrass (Panicum virgatum) is well suited economically and ecologically for energy crop production. It is an excellent biofuel because of its high fiber content, high biomass yield, drought resistance, easy establishment, and perennial growth habit [4] and [5]. Forage crops, such as grass and legume hay, are common nesting habitats for grassland birds, but are harvested early and frequently with documented negative effects on grassland bird nest productivity [6], [7] and [8]. In contrast, switchgrass fields are generally harvested for biomass in late summer or fall after completion of all or most of the grassland bird breeding season; disturbance to nesting birds is thus avoided or minimized. As a result, the primary impact that late summer harvest for biomass has for grassland birds is modification of vegetation structure the following year. Switchgrass thus has the potential to provide a cash crop for farmers and undisturbed nest cover for grassland birds [9]. Compared to the limited longevity of idle grassland habitat established through the USDA's Conservation Reserve Program (CRP), in which fields are typically enrolled for a 10-year period, switchgrass grown as a biofuel has the added advantage of having the potential to remain on the landscape for the life of the power plant for which it provides feedstock.

The goal of this study was to determine the impact of harvesting switchgrass on community composition and abundance of Wisconsin grassland bird species of management concern [10]. We predicted that we would see a shift in the grassland bird community from tall-grass species to short- and mid-grass species after the harvest.

This study was part of a larger experiment involving the University of Wisconsin Department of Mechanical Engineering and a local utility, Madison Gas and Electric (MG&E), in Madison, Wisconsin, to investigate the use of switchgrass as a biomass fuel for energy production [11].

2. Methods

The study was conducted on fields enrolled in the CRP in Iowa and LaFayette Counties in southwestern Wisconsin (near lat. 42°52′, long. 90°08′). For combustion purposes we selected fields planted to as pure a switchgrass monoculture as possible. Because relatively few CRP fields in this region were planted to pure switchgrass, our field choices were limited. Efforts were made to choose five fields with similar vegetation structure though the size of field could not be controlled. Five CRP fields were found meeting the selection criteria; field sizes were 8, 14, 36, 85, and 123 ha. All were established as a pure stand of switchgrass 5–10 years before the study began and had been burned within the previous 5 years. The vegetation in the fields varied from relatively pure stands of switchgrass (<20% forbs) to fairly weedy stands (>45% forbs).

All sites were in the Driftless Area of Wisconsin, a region characterized by rolling topography and hillsides with slopes of 2–20%. Soils were well-drained Hapludalfs (Dubuque silt loams) and Arguidolls (Dodgeville silt loams). Land use in the Driftless Area is primarily agricultural, characterized by crop fields (primarily strip cropping of alfalfa (Medicago sativa), soybeans (Glycine max), corn (Zea mays), and oats (Avena sativa)), pastures, and small woodlots. Weather conditions in the study area were very similar between 1996 and 1997 with near normal climatic moisture conditions at the onset of the growing season [12]. During the growing season (May–September), area temperatures averaged nearly 2 °C below the 30-year average (high of 24.1 °C, low of 11.4 °C) in both years and annual precipitation was 90 and 70 mm above normal (465 mm) in 1996 and 1997, respectively.

Before the study, a portion of each CRP field was designated for harvest based on machinery access restrictions. Two transects measuring 200×100 m were laid out in each field, one within the harvest area and one within the unharvested, control area. All transects were located at least 20 m from an edge of the field; within the field, transects were set at least 50 m apart. In all cases the transects were surrounded with similarly managed habitat.

Pre-harvest bird and vegetation surveys were conducted in all transects in May, June, and July of 1996 to establish a baseline data set for both vegetation structure and bird community composition. Between August 15 and 30, 1996 the switchgrass was harvested in pre-designated areas at a height of 15 cm. A minimum 20 m buffer area was cut around the harvested transects to reduce the influence of the unharvested portion of the field on the bird community. The amount of area harvested was limited by the quantity of switchgrass required for the co-firing component of this joint study. The harvested areas of the fields were 4, 6, 8, 20, and 22 ha.

Several harvest schedules have been tested for yield potential and stand management, but little research has been done to investigate the effects of warm season grass harvesting on grassland bird communities. The commonly used harvest schedules include harvesting twice in July and September or harvesting once per season after the grass has gone dormant in October or November [13] and [14]. We tested a third alternative, an August harvest, with the goal of providing some time for regrowth in late summer so that more nesting cover would be available the following spring. Harvesting switchgrass during the first 3 weeks of August was recently recommended for maximizing biomass production in the Midwest [15].

Post-harvest bird and vegetation surveys were conducted in May, June, and July of 1997 in harvested and unharvested transects. In each year, all transects were surveyed four times at 2-week intervals. Bird surveys were conducted between sunrise and 10:00 AM. We avoided surveying on days when wind or rain could interfere with bird activity or observability. For each survey, the observer walked a line down the center of each transect stopping at 50 m intervals and recording all birds seen or heard within the transect during a 5-min period at each stop. The observer was the same in both years, minimizing observation bias. We estimated the number of pairs of each species as the number of males or females, whichever was greater. For species without visible gender differences, any singing individual was considered male. Two non-singing individuals of a sexually monomorphic species found in the same location on two or more visits were considered a pair. We used the highest estimate of the four surveys as the best estimate of the number of pairs for each species in a transect. Relative abundance refers to the number of pairs per transect.

Vegetation surveys were conducted in harvested and unharvested transects at each site within a week of each bird survey in that field. Two observers collected the data together in 1996 to be certain both recorded observations the same way. One of these observers collected all data in 1997. We sampled vegetation along a line transect spanning each bird survey area on each sampling date. Ten sampling plots were located randomly along the transect; data from those plots were averaged across visits for analysis. At each sampling site, we measured litter depth, forb cover, and vegetation height–density within a  m² quadrat frame [16]. For litter depth, a measurement was taken in each quadrant of the quadrat; the four measurements were averaged for the sampling plot. The percent of ground area covered by forbs was estimated for the entire quadrat. For the height–density measurement, a Robel pole marked at 5-cm increments was placed in the center of the quadrat and was viewed from a distance of 4 m and a height of 1.5 m. The level at which the markings on the pole were obscured by vegetation was recorded from the four cardinal directions and these measurements were averaged for the sampling plot [17]. These vegetation variables were a priori chosen to represent breeding habitat variables important to breeding grassland birds [10].

For analysis, we focused on the grassland bird species of management concern [10]. We categorized the bird species of management concern into three groups, based on their general preference for short-, mid- or tall-grass vegetation structure [10]. Scientific names for bird species mentioned in the text are listed in Table 1. We analyzed richness and relative abundance of the species groups. Species group richness was the total number of grassland bird species of that group detected over the four visits. Relative abundance for the species group was the maximum number of pairs for each species in the group in a transect over the four visits.

We analyzed the effect of field size on species richness by regressing species richness against ln(field size). We analyzed the impact of harvest by calculating the differences between years for richness and abundance for each species group and using a split-plot ANOVA to test for a difference between harvested and unharvested transects. We analyzed the impact of harvest on vegetation height–density, forb cover, and litter depth in an analogous manner. The null hypothesis we tested was whether the change in variable between years on the uncut transect equaled the change in variable on the cut transect. All statistical analyses were conducted using S+2000 [18]. We set statistical significance at α=0.05.

3. Results

In 1996 (before harvest), the mean vegetation height-density in the fields was 57 cm, mean litter depth was 3 cm, and mean forb cover was 39% (Table 2). As expected, vegetation height density decreased (Table 2) more in the harvested transects between years compared to the change in unharvested transects (F=30.6, df=1,4, P=0.005). Litter depth decreased (Table 2) between years in the harvested transects but increased in the unharvested ones (F=31.0, df=1,4, P=0.005). Forb cover did not change (Table 2) between years in both harvested and unharvested transects (F=3.8, df=1,4, P=0.12).

We observed 17 bird species in the harvested and unharvested transects during the 2 years of the study, 10 of which were grassland species of management concern (Table 1). Of the 10 species of management concern, six were found in the unharvested fields during 1996, including two of the three tall-grass species; a single tall-grass species, sedge wren, was found in most of the fields (Table 1). There was no effect of field size on species richness or abundance in 1996 (P>0.10 both tests) or 1997 (P>0.10 both tests).

Species richness was similar between years in harvested transects compared to unharvested transects (F=6, df=1,4, P=0.07) (Table 3). Short- and mid-grass species richness increased in harvested transects between years compared to unharvested transects (short-grass: F=36, df=1,4, P=0.004; mid-grass: F=14.2, df=1,4, P=0.02) (Table 3). For short-grass species, this difference was driven mainly by the increase of grasshopper sparrow occurrence in transects after they were harvested (post-harvest). In addition, upland sandpipers and western meadowlarks only occurred in post-harvest transects (Table 1). Three of the four mid-grass species occurred more frequently in the post-harvest transects than in unharvested ones (Table 1). Savannah sparrows only occurred in post-harvest transects. Tall-grass species richness did not change in either treatment between years (F=7.1, df=1,4, P=0.06) (Table 3). However, no sedge wrens or Henslow's sparrows were observed in any post-harvest transect (Table 1).

Between years, the relative abundance of all species of management concern stayed the same in both harvested and unharvested transects (F=1, df=1,4, P=0.36) (Table 3). The relative abundance of short-grass species did not change in harvested transects between years compared to unharvested transects (F=5.1, df=1,4, P=0.09) (Table 3). Though the relative abundance of grasshopper sparrows increased substantially in the harvested transects between years (Table 1), this change was not enough to significantly affect short-grass species relative abundance (Table 3). Between years mid-grass species increased in abundance in harvested transects but not in unharvested transects (F=50, df=1,4, P=0.002) (Table 3). This was mainly due to the higher relative abundance of Savannah sparrows in post-harvest transects (Table 3). The relative abundance of tall-grass species did not change in either treatment between years (F=6.2, df=1,4, P=0.07) (Table 3); this was despite a large decrease in sedge wrens on the harvested transects in 1997 (Table 1).

4. Discussion

Harvesting switchgrass for biofuel in mid- to late-August resulted in changes in vegetation structure and grassland bird species composition the following nesting season. Vegetation height-density was shorter and the litter layer was shallower in the year following harvest. Similar litter layer changes were observed in harvested switchgrass fields in Iowa [14]. No vegetation regrowth was observed immediately following harvest contrary to expectations. In the unharvested sections of the fields, the height–density was slightly lower and the litter depth higher in 1997 relative to 1996. These differences were potentially due to variation in long-term climatic variables not measured in the study and higher rates of grass blow-down and compaction in 1997.

Following harvest, we observed a shift in bird species composition from a community dominated by birds preferring tall-grass structure to birds preferring short- and mid-grass structure. Other studies reported similar results of species’ response to mowing in the previous year. In Missouri, grasshopper sparrows occurred at highest densities the year following mowing and declined with increasing time since disturbance [19]. Delisle and Savidge [20] reported the most consistent numbers of grasshopper sparrows from a CRP field mowed 3 of 4 years in Nebraska. Grasshopper sparrows, upland sandpipers, bobolinks, meadowlark spp., and dickcissels preferred harvested switchgrass fields to unharvested fields in Iowa; only the results for grasshopper sparrows were statistically significant [21]. Savannah sparrows preferred the portions of CRP fields mowed the previous year in North Dakota [22]. Western meadowlarks occurred more frequently in periodically hayed fields that had been mowed in the previous year in Saskatchewan; however, Savannah sparrows were more abundant in unmowed fields [23]. The discrepancy in results for Savannah sparrows may be explained by differences in idle grassland vegetation structure in Saskatchewan and Wisconsin due to climatic factors, see [22].

Of the four short- and mid-grass species that occurred in unharvested transects in our study, three (grasshopper sparrow, eastern meadowlark, and dickcissel) only occurred in the field with the lowest vegetation height–density. An alternative explanation for the presence of dickcissels on unharvested transects only is that, of all the species in the mid-grass group, dickcissels prefer the tallest vegetation and are considered by some researchers to prefer tall-grass vegetation [24], [25] and [26].

In other studies, mowing adversely affected tall-grass species in the following year, especially sedge wrens [21], [22] and [27] and Henslow's sparrows [19] and [28]. Northern harriers typically prefer tall idle grassland over mowed fields [21] and [29]. However, the only harrier in our study occurred in a harvested transect, in 1997. It was only present during one of four visits and was probably not nesting in that habitat. Northern harriers have large home ranges [30] and may use different parts of that range for nesting, foraging, and loafing.

Species richness for all species of management concern in our study was similar between harvested and unharvested transects. However, Sample [31] found that species richness for rare and declining grassland birds in a variety of grassland habitats in southern Wisconsin was negatively correlated with vegetation height-density.

Our results suggest that both harvested and unharvested switchgrass fields have habitat value for grassland birds of management concern. Harvesting switchgrass for biomass in August created habitat the following year which favored grassland bird species that preferred vegetation of short to moderate height and low to moderate density. Unharvested areas provided tall, dense vegetation structure that was especially attractive to tall-grass bird species, such as sedge wren and Henslow's sparrow. When considering wildlife habitat value in harvest management of switchgrass for biofuel, leaving some fields unharvested each year would be a good compromise, providing some habitat for a larger number of grassland bird species of management concern than if all fields were harvested annually. However, in many areas of Wisconsin, most of the idle grassland habitat present on the landscape is in the form of unharvested CRP fields with tall and dense structure. Habitats with short to moderate height vegetation, such as pasture, are comparatively rare or declining in acreage [32]. In such areas in the upper Midwest and elsewhere, harvest of switchgrass for biofuel has the potential to increase the local diversity of grassland birds.

When considering the timing of switchgrass harvest, the nesting phenology of the tall-grass bird community should be evaluated. In southwest Wisconsin, we believe that Henslow's sparrows, dickcissels, and sedge wrens are the species of management concern most likely to be impacted by harvesting switchgrass in August. Nest data from other studies suggest that a mid-August harvest would allow the young from nearly all Henslow's sparrow and sedge wren nests to fledge [33] (Guzy and Ribic, unpublished data). Though waiting until the end of August for the last dickcissels to fledge would be ideal; a mid-August harvest would still allow the potential for at least 90% of the nests to fledge [34].

Future research should include data on nesting productivity to better assess the conservation value to grassland birds of harvesting switchgrass for biofuel, see [21]. Additional studies should evaluate the impacts of diverse seedings of native grasses and forbs harvested for biofuel on both combustion parameters and grassland birds. Increasing the number of fields sampled would improve statistical power and may more clearly show harvest response of bird species occurring in low numbers in this study.