Tall fescue (Schedonorus phoenix) is an invasive grass with known negative effects on cattle health and hypothesized negative effects on ecological communities. Our project addresses these effects by examining the interrelated ecological-agricultural dynamics of tall fescue in working Midwestern grasslands. Our goals are to evaluate (1) use of tall fescue by foraging cattle, (2) avian and arthropod community response to experimental changes in fescue cover, and (3) landowner capacity to remove tall fescue on private lands.
Our study takes place in the Grand River Grasslands of Ringgold County, Iowa and Harrison County, Missouri. This research has been ongoing since 2006, and the current SARE project utilizes both prior data and new data. Our first field season for this project took place May-Aug. 2016. During the field season, we collected data on cattle grazing preferences in response to tall fescue abundance (Goal 1). Information on foraging preferences will demonstrate the potential economic value of fescue control for farmers and facilitate sward optimization. We also counted grassland birds and collected arthropods on all study sites in order to elucidate the costs and benefits for wildlife of removing fescue (Goal 2). We also designed and mailed surveys that aim to determine landowner willingness to remove tall fescue through (Goal 3). This survey will reveal barriers and opportunities for controlling fescue.
- Improved understanding of the impact of tall fescue abundance on ecological communities
- Improved understanding of the impact of tall fescue removal on ecological communities
- Improved understanding of relationships between tall fescue and cattle foraging
- Understanding of potential barriers to invasive species removal
- Enhanced knowledge in the community about invasive species
- Longitudinal analysis showing change in attitudes (2007-2017)
- Increased interest in sustainable agriculture in a grassland context
- Increased literacy in areas of grassland ecology
The 2017 survey conducted as a part of this SARE project was a partial replication of a survey conducted in the Grand River Grasslands region in 2007, with the survey instrument including replicated questions about livelihoods, biodiversity, demographics, and use of grassland management techniques. New items measured attitudes toward invasive and non-native plants, native wildlife, and factors related to grassland management using prescribed fire, grazing, and herbicide. The 2017 survey instrument was pilot tested by cattle producers in Nebraska (N=8). Cattle producers were asked to evaluate the survey instrument on formatting/readability, content relevance, and clarity of the vocabulary. This pilot test resulted in small changes to language and formatting. The Institutional Review Boards from collaborating institutions, the University of Illinois and Iowa State University, reviewed the instrument and survey protocol and approved the use of human subjects [IRB #16389].
To collect survey responses, we used a multiple contact system with reminders for non-respondents to obtain the highest response rate possible. In addition to the mail-back survey, an online survey was offered as a convenient option to complete the survey. In February 2017, 528 landowners were first contacted with a mail postcard that alerted them, “Survey Coming!” The postcard provided brief information about the purpose of the study and a link where they could take an online version of the survey. At two-week intervals, landowners were contacted again and encouraged to complete the survey, for a total of seven possible contacts (waves) per household. Respondents were also offered the option to complete an electronic survey using the online survey platform, Qualtrics. For the seventh wave, any non-respondents with publically available phone numbers were called as a final reminder.
Of the 528 households sent the first wave postcard, 72 were found to be vacant, not deliverable as addressed, or the addressee was no longer at that address, yielding 456 valid addresses for the following four survey waves. To assess nonresponse bias, during reminder phone calls (N=154), we asked individuals who declined to return a survey basic questions (age, gender, number of acres owned, years of land ownership in Ringgold/Harrison counties). These answers, in combination with census data (demographics) and information from county plat maps (number of acres owned), were used to test whether this sample accurately represented the human population living in and around the Grand River Grasslands.
Over the course of five mailings and one phone call, 32.7% returned the survey (N=149). An additional 3% (N=13) of the total sample answered basic questions over the phone. These served to help us assess non-response bias, defined as a difference between those that turned in the survey and those that did not that affects interpretation of results. Our evaluation of the non-response bias found there was no detectable differences in age between the two groups. Furthermore, an increase in age between the 2007 sample (average 62) and the 2017 sample (average 67) is likely reflecting a broader pattern of aging farm operators in Iowa. The gender composition in the 2017 sample is similar to that of the 2007 sample. Based on this, we believe that non-response bias is not a concerning issue and that our sample accurately reflects GRG landowners.
To determine the impact of tall fescue and tall fescue control on grassland avian communities, we measured avian diversity and relative abundance along survey transects at 18-20 pastures each summer (2015-2018). These measurements are comprised of 3-8 line transects per pasture (depending on pasture size) surveyed 6-10 times from May through August. Between 05:00 and 10:00 on clear days with winds <10 mph, a trained observer walked along the transects, recording all birds within 50 m. Birds in groups >2 will not be counted to exclude the effects of flocking.
To determine the impact of tall fescue and tall fescue control on arthropod communities, we used a modified leaf blower and a sweep net to sample arthropods along 50-m line transects 20 m away from and parallel to the avian transects (2015-2017). We moved along the transects sweeping the vacuum in a 1-m swath (50 swaths/transect). Arthropod sampling occurred in two rounds in each patch, and specimens are currently in the process of being identified to taxonomic order. We will calculate abundance and biomass and then use regression to test whether abundance and biomass per order are impacted (positively or negatively) by tall fescue abundance or removal.
Within four pastures in the Grand River Grasslands, we examined grazing selectivity by cattle among tall fescue, other cool-season grasses, native warm-season grasses, legumes, and non-leguminous forbs. Since 2007, the four pastures have been managed using patch-burn grazing, a system in which each pasture is delineated into three patches of approximately equal size. One patch per pasture is burned in late March or early April on a rotating basis, such that the entire pasture is burned over three-year cycles. A herd of either black Angus or mixed black Angus and Charolais beef cattle had free access to all patches.
All data were collected over two sampling rounds between June and August 2016. In each sampling round, we collected data on forage abundance and use by cattle within 0.1-m2 plots (interior dimensions: 25 x 40 cm). We placed five plots in each patch of each pasture per sampling round (three patches per pasture), thereby sampling a total of 45 plots in the first round and 30 plots in the second. We used a modified version of the point-quadrat method to measure the relative abundance of plants in the five forage categories: tall fescue, other cool-season grasses, native warm-season grasses, legumes, and non-leguminous forbs.
Plots were marked with gridlines creating 25 evenly-spaced grid points. We laid the plot flat on the soil surface, placed a 2-mm-diameter pin in the ground at each of these 25 grid points, and classified the plant rooted nearest to each pin-drop into one of these categories.Within each plot, we calculated the relative abundance of plants in each of the five forage categories as the proportion of all classified plants belonging to each category (i.e., number of plants of each forage type / 25). We then calculated pasture-scale relative abundance of each category as the average relative abundance per category across all plots in a given pasture.
In addition to measuring relative forage abundances, we estimated overall forage use and use of each forage category in each plot. To measure overall forage use, we randomly selected 10 of the 25 grid points per plot (two in each of the five rows of five grid-points) and noted whether the plant rooted nearest to each chosen point had been grazed. To determine this, we examined all shoots on those plants (or all leaves, on plants forming a leaf rosette). If at least two shoots (or leaves) on a given plant were sheared in a straight line, we counted the plant as grazed. We set the threshold at two shoots to reduce the likelihood of false positives (i.e., a shoot appeared grazed but was instead damaged by another cause). If a plant only formed one shoot, as is true of many forbs, we considered it grazed if that shoot was sheared. We calculated overall forage use in each plot as the number of grazed plants divided by 10.
We then measured use of each forage category in each plot. At the same 10 grid points as above, we examined the nearest-rooted plant belonging to each of the five forage categories and documented whether those plants had been grazed. Thus, in every plot we determined the grazing status of up to 10 tall fescue plants, 10 non-fescue cool-season grasses, 10 warm-season grasses, 10 legumes, and 10 non-leguminous forbs. We calculated use of the five forages in each plot as the number of plants in a given category that were grazed divided by the number of plants in that category assessed for grazing (10, if there were at least 10 plants of that category in the plot).
Finally, we quantified use of each forage category by cattle at pasture scale. First, for each category in each pasture we summed the number of plants grazed across all plots to determine total use of each category. We then summed these forage-specific use levels to calculate the total number of plants grazed per pasture (across all categories). Lastly, we divided each forage-specific use level by the total number of plants grazed per pasture to quantify the proportion of all grazed vegetation constituted by each forage category. These values do not refer to biomass consumed, but rather to the proportion of individual plants grazed. We also point out that our measurements of forage use at both spatial scales are not estimates of use by individual animals, but rather of use by the entire cattle herd stocked on each pasture.
Social Data Analysis and Results:
This report summarizes key findings from the 2017 survey and compares selected survey items to a similar survey conducted in 2007. Although landowner demographics were similar between 2007 and 2017, several changes occurred over the ten-year period. Respondents owning land in the GRG for greater than 25 years increased from 59% in 2007 to 67% in 2017. In addition, respondents owning land for 10 years or less dropped from 20% in 2007 to 7% in 2017. This suggests that there has been very little ownership turnover in ten years. GRG landowners in 2017 had on average 315 acres of open pasture and grassland, 147 acres of corn/soybean, 47 acres of woodland including 3 acres of redcedar, and about 5 acres of small grains. Eighty percent reported having one to four ponds on their property. About half of the respondents in both 2007 and 2017 were row-crop farmers. Respondents using their land for a weekend retreat or vacation home more than doubled from 2007 (5%) to 2017 (11%). Although most GRG landowners value income from agriculture, there was a drop in the percentage of landowners rating it as very or extremely important, from 81% in 2007 to 70% in 2017.
Cattle ranching is a major land use in the GRG. Ninety-five percent of cattle ranchers were moderately to extremely satisfied with their growing season forage in 2017. Less than 25% of respondents would adopt reduce beef production per acre regardless if it would result in an increase in gamebirds, reduce tall fescue, protect wildlife, restore grassland, or increase native plants. A few more respondents said they would be very/extremely likely to try this trade-off if it reduced soil erosion (35%) or controlled invasive plants (32%). Twenty-eight percent thought non-native tall fescue must be removed from their land. Cattle producers specifically are very concerned about the effects of having too much fescue in their pastures. If this leads to a willingness to manage and reduce tall fescue in grazing pastures, there is strong potential for both wildlife habitat restoration and benefits to cattle production systems.
Herbicides are a management strategy for removing non-native grasses and managing woody encroachment in fields. While the majority of respondents were concerned about a number of risks to using herbicides, they were most concerned about soil erosion after use (72% moderately/extremely concerned) and herbicide resistance (69%). A high proportion of respondents to the 2017 survey believed that native warm season grasses are good for wildlife and grazing compared to fields of the invasive grass tall fescue.
Game species, especially bobwhite quail, remain highly important to residents, but most non-game wildlife has less importance to landowners in 2017 compared to 2007. The exception to this are bees, which respondents see as very important to have on their lands. It is likely that landowner awareness and concern have been raised by the well-publicized threats to bees in the popular press over the last several years.
There has been a net gain in grassland acres in the region, but large acreages continued to shift back and forth between crop and grassland in the last five years. This may reflect the downturn in the crop and cattle prices that has occurred in recent years.
While positive attitudes toward conservation seem to have eroded in some areas (with exceptions) over the ten-year period, more landowners are taking part in activities that are good for conservation and restoration. For example, fewer landowners believe that grassland restoration on their land is important in 2017 vs. 2007. However, more landowners are using prescribed fire and physically removing woody plants, activities that promote grassland restoration. There are clear advantages to focusing on how management activities related to restoration have the potential to benefit production goals as well as wildlife habitat.
Ecological Data Analysis and Results:
The goal of the avian analyses is to determine the effects herbicide application and grazing method on (1) individual bird species of concern and (2) avian richness. To calculate species densities, we found the maximum number of individuals ever observed on a single visit, for each of the three patches per pasture. We then divided these numbers by the area sampled (transect length x transect width, converted to birds/ha). Group sizes greater than 3 were excluded to control for effects of flocking. In these analyses, we distinguished between grassland obligate and grassland facultative species. Obligate grassland species are species exclusively adapted to grassland habitats and rarely if ever use other habitat types, whereas facultative species use grasslands regularly but do not depend on them entirely. Richness is the total number of species spotted on a pasture or patch over the course of the season, calculated for both obligate species alone and obligate and facultative species combined.
Next, we used general linear mixed models to compare patches with herbicide application (spray), herbicide application and native seeding (spray & seed), and patches without herbicide application (control). In these analyses, we compared single-species density of obligate and facultative species and overall grassland bird richness to these patch treatments for the eight pastures with controlled herbicide application. Pasture was included as a random effect to control for the non-independence of observations taken from the same pasture. Second, we constructed general linear models to examine the effect of to harvest method (i.e., early intensive stocking, continuous stocking, or no grazing) on single-species densities and richness of grassland birds. In 2017, we began looking at multiple years of these data at once to see if there were any lag effects associated with treatment application.
Several obligate grassland birds responded to herbicide application, although these changes differed by species (Fig. 1). Dickcissels were most abundant on spray & seed patches, following by sprayed and then control patches (p=0.002). Conversely, Henslow’s sparrows were most abundant on control patches, with no distinction between spray & spray and spray-only patches (p=0.029). No statistically significant effects were detected for bobolinks, grasshopper sparrows, Henslow’s sparrows, eastern meadowlarks, or sedge wrens (p>0.1). We did not detect any response to these treatments by facultative species (Fig. 2, p>0.1) or total grassland bird richness (p>0.1).
We were able to detect several species-specific effects of harvest method on avian density for obligate species (Fig. 3) and facultative species (Fig. 4). For obligate species, dickcissels were most abundant on pastures with no grazing and least abundant on pastures with intensive early stocking, with continuous stocking fall between these treatments (p=0.029). This pattern is also present for bobolinks (p=0.13) and Henslow’s sparrows (0.12), although these results are not statistically significant at α=0.1. Sedge wrens were also most abundant on ungrazed pastures, and were mostly absent from grazed pastures of either treatment (p<0.001). In contrast, grasshopper sparrows were least abundant on pastures without grazing, and had essentially equivalent densities on the two grazing treatments (p=0.078). Eastern meadowlarks were slightly more abundant on both grazing treatments when compared to ungrazed pastures, but effect sizes are small and not statistically significant (p=0.12).
For facultative species, red-winged blackbirds were most abundant on pastures without grazing, with equivalent densities on the two grazing treatments (p=0.0025), with similar trends also seen for common yellowthroats (p=0.005). No significant trends were observed for eastern kingbirds, brown-headed cowbirds, or field sparrows. There were no detectable effects of harvest method on species richness (p>0.1).
Beginning in 2017, we also examined time-since-treatment over the course of the herbicide experiment (2014=baseline, experiment=2015, 2016, and 2017). We found that grasshopper sparrows decreased in abundance one-year post-treatment compared to both baseline abundances and control patches (p=0.095). However, grasshopper sparrows have since increased in the three years since herbicide and seeding treatment was applied. Conversely, dickcissels increased significantly one-year after treatment and have stabilized at relatively higher abundances three-years after treatment (p<0.0001). It is notable that 2015 was the year we saw these increases since that year was generally low in dickcissel abundances in the region.
These results will be combined with the long-term Grand River Grasslands avian data set for further analysis.
Agricultural Data Analysis and Results:
The goal of these preliminary analyses is to determine how the relative abundance of several vegetation types influences grazing pressure and selectivity. We built generalized linear mixed models (lognormal distribution and identity link, study site as random variable) using the relative abundances of tall fescue, warm-season grasses, cool-season grasses, forbs, and legumes as predictor variables, alongside additional predictor variables such as time-since-fire, sampling round, and sampling date. We were particularly interested in how these predictor variables influenced (1) total grazing pressure across all vegetation types within sampling plots, and (2) grazing pressure specifically on tall fescue, so we conducted separate analyses for each of these two response variables. For each analysis, we constructed a candidate model set of predictor variables potentially explaining grazing pressure and then used an information theoretic approach (Akaike’s Information Criterion corrected for small sample sizes, AICc) to compare the relative fit of candidate models within each set.
First, based on AICc rankings, total grazing pressure within sampling plots was influenced by the relative abundance of warm-season grasses and tall fescue. As warm-season grasses increase from 0 to 0.4 in relative frequency, there is an increase in total percent grazed, from ~20% to ~45%, albeit with wide confidence intervals (Fig. 5). Conversely, as tall fescue increased in relative abundance from 0 to 0.9, the percent of vegetation grazed decreases from ~40% to ~17% (Fig. 6). Time-since-fire and the relative abundance of tall fescue were the top-ranked candidate models explaining grazing pressure on tall fescue. As time since fire increased from 0 years to 2 years, the percent of tall fescue grazed decreased from ~40% to 15%, with clear separation between year 0 and 2, supported by non-overlapping 85% confidence intervals (Fig.7). Additionally, as the relative frequency of tall fescue increased from 0 to 0.9, the percent of tall fescue grazed decreased from ~45% to ~18% (Fig. 8).
Additionally, cattle used some forages disproportionately to their relative abundance (F(9, 27)=9.71, p<0.001). Although tall fescue was the most abundant forage in the pastures, comprising 46.4% of categorized plants (85% confidence interval: 41.5-51.4%), it constituted only 26% (21.7-31.6%) of all grazed plants. In contrast, even though legumes accounted for only 12.3% (7.3-17.3%) of plants, they constituted 24.6% (19.6-29.5%) of all plants grazed—similar to use of both tall fescue and non-fescue cool-season grasses. Non-fescue cool-season grasses, warm-season grasses, and forbs were grazed in proportion with their abundances (i.e., confidence intervals around relative abundance and use estimates overlapped).
Figures 1: GNC 15-201_Figures_AnnualReport
Educational & Outreach Activities
To enhance knowledge in the community about invasive species, grassland ecology, and sustainable agriculture (Social Objective 2, Outreach Objective 1 & 2), we began an internship program that focused on teaching wildlife sampling techniques. In 2016 and 2017, we worked with two interns. They worked 10-30 hrs/wk and had the option of receiving course credit through their universities. All of our interns from this Grand River Grasslands Internship program have received employment or further internships in the wildlife field, including at the Des Moines Zoo, NOAA, and the Pyramid Lake Paiute Tribe reservation.
We also began making connections with high school teachers in the area, and have developed lesson plans centering on sustainable agriculture and biodiversity in collaboration with educators that we are planning to use in summer 2018. We initiated an “Intern for a Day” program, where high school or undergraduate students could come and participate in our research for a day or more. In 2016, we hosted one high school student over four days. In 2017, we engaged two undergraduate students affiliated with The Nature Conservancy in this program and worked with them over two days.
In addition to outreach, our research team has presented or will present results from this project 12 times between Jan 2017 and April 2018. Venues have included: US-Section of the International Association for Landscape Ecology (Baltimore, MD and forthcoming in Chicago, IL), Midwest Ecology and Evolution Conference (Grand Rapids, MI), the Graduate in Ecology and Evolutionary Biology Symposium (Urbana, IL), and several undergraduate student-led presentations at the Undergraduate Researchers’ Initiative Symposium and the Undergraduate Research Symposium.
Most importantly, in summer 2017 we held a meeting in our study region that included past collaborators, land managers (~15 people) from multiple organizations, and local landowners (~10 people). Two presentations at the “Grand River Grasslands Symposium: Ten Years of Research – Key Findings and Future Directions” were based on this current project. After this symposium, two private landowners invited us to count birds on their land, which was completed during summer 2017.
We are currently working on several manuscripts that detail results from this project. One of these is at the stage of submitting to a journal. We are currently revising and submitting a manuscript entitled, “Native and non-native grass composition influences cattle grazing in the context of management with prescribed fire and alternative grazing pressures,” to the academic journal Grass and Forage Science, which we hope will reach managers and producers interested in the topic. After finding in our social research that many cattle farmers and ranchers are concerned that tall fescue could adversely affect their livestock and are open to reducing its abundance on their lands, we wanted to look into this further. Our agricultural results suggest that although cattle certainly do graze tall fescue—even when alternative forages are available—fescue quantity could be substantially reduced in pastures where it is highly abundant without compromising forage quality. This is because fescue constituted a significantly smaller proportion of all vegetation grazed by cattle than expected given its prevalence, and that cattle used areas with low fescue abundance more heavily than areas with high fescue abundance. Given that fescue has been shown to harm grassland wildlife, reducing tall fescue abundance on grazing pastures may provide both agricultural and ecological benefits. However, if land managers do not wish to remove it, our finding that cattle increase their use of tall fescue (and other cool-season grasses) following management with prescribed fire.