Final Report for GNC07-081

Assessment of pasture management practices on microbial biomass, composition, and functional diversity

Project Type: Graduate Student
Funds awarded in 2007: $9,924.00
Projected End Date: 12/31/2009
Grant Recipient: University of Wisconsin-Madison
Region: North Central
State: Wisconsin
Graduate Student:
Faculty Advisor:
Randall Jackson
University of Wisconsin-Madison
Expand All

Project Information

Summary:

Soil resilience is a concept deemed important to our understanding of soil quality and sustainability and is thought to be facilitated by functional diversity of soil biota. To better understand the potential of ecosystems to respond to disturbance, more information on microbial community and functional group composition, abundance, and their distribution within soils is necessary. While the diversity of microorganisms in soils is thought to be high, some management practices, such as conventional tillage, can deteriorate soil physical properties, and bring about a decline in total microbial biomass. The objective of this research was to compare soil microbial communities under different pasture management strategies on farms across southern Wisconsin. We used a hybrid phospholipid fatty acid (PLFA) and fatty acid methyl ester (FAME) analysis to measure microbial biomass, fungal to bacterial ratios (f:b), and characterize by lipid biomarkers the following functional guilds; gram-positive and negative bacteria (Gm+ and Gm-), arbuscular mycorrhizal fungi (AMF), and actinomycetes. For multivariate analysis of microbial community structure, principal components analysis (PCA) was performed.

Introduction:

Human activities over the last several centuries have altered earth’s abiotic and biotic processes (Vitousek and Mooney 1997). Alteration of terrestrial surfaces may force changes in ecosystem processes (Randerson et al., 2002). For example, the conversion from prairie to agriculture has reduced soil carbon and nitrogen stocks, which maintain soil fertility (Mann 1986). Grazing herbivores have been shown to influence plant species composition, aboveground plant structure, and quantity and quality of organic inputs (Bardgett et al.,1998), soil N cycling (Gueswell et al., 2005, LeRoux et al., 2003), root biomass (Holland and Detling 1990), carbon (C) enhancement of the rhizosphere by root exudation (Mawdsley and Bardgett 1997), microbial biomass (Bardgett et al., 1997), and f:b ratios (Bardgett et al., 1996).

Soil microorganisms play an important role in soil quality, and understanding the impacts of human activities, including grazing, is important to our understanding of the role microorganisms play in soil processes and maintenance of soil quality. The impact of land use changes on microbial- mediated processes may be influenced by whether the process is performed across a wide range of microbes or whether it is performed by a narrower group (Balser et al., 2006). To better understand the potential of ecosystems to respond to disturbance, more information on microbial community composition and functional groups, and how they are affected is necessary (Balser et al., 2006, Patra et al., 2005). While the impact of livestock production practices on aboveground biomass is well studied, their impacts on belowground heterotrophic communities are not.

Livestock grazing of pastures is a growing phenomenon in the upper Midwest. Jackson-Smith et al. (1996) estimated that about 15% of all dairy operations in Wisconsin maintained some form of grazing as a management strategy. An update to this study (Ostrom and Jackson-Smith 2000) indicated this estimate had grown to 22% by 2000. Grazing dairy livestock in pasture is contrasted with confinement systems where feed is mechanically harvested and transported to a centralized location where livestock spend most if not all of their time. Social and economic benefits of a particular form of grazing management, Management Intensive Rotational Grazing (MIRG), have been demonstrated in many settings (Parker 1992, Fales et al., 1995, Frank et al., 1995, Paine et al., 1999). More specifically, MIRG, which entails livestock grazing in relatively small paddocks at high densities (150 to 250 animal units•ha-1), but for short durations (1 to 3 days), has been touted as beneficial to both graziers and grazers (Undersander et al., 1993, Paine et al., 2000). But while MIRG’s benefits for production economics are increasingly well established, its ecological consequences remain mostly anecdotal and the scientific basis for how these ecosystems function is lacking.

Project Objectives:

The objectives of this research were to:

1) quantify total microbial biomass, f:b ratios, characterize microbial community composition by functional guild, and
2) compare variability of response variables of pasture management treatments across farms.

Cooperators

Click linked name(s) to expand
  • Teresa Balser
  • Randall Jackson

Research

Materials and methods:

We identified farmer participants at the 2004 and 2005 Wisconsin Grazing Conference in post presentations meetings. Our goal was to gain commitment to the project from farms with a range of operation sizes. On-farm experiments are likely to have low within-site replication and substantial variation as dictated by each particular growing condition and management style. However, the advantage of performing research on multiple farms is the ability to validate our hypotheses across a wider geographical range and subject to a broader range of environmental variation. This allows us to make inferences about the relative importance of patterns we observe when scaled to the landscape.

In 2006 and 2007 samples were collected from 8 grass-based farms that were located within 50 km of Madison, Wisconsin (USA). With farmer assistance we identified areas within each farm where management strategies that were similar to our 4 identified treatments – MIRG, continuous grazing (CONT), harvest of biomass for livestock feeding (HARV), and no management (NONE) – were practiced. Lipid biomarkers from microorganism cell membranes were extracted from soil using PLFA/FAME. These “signature” markers were used to determine both microbial biomass, and relative abundance of indicator lipids to characterize microbial community structure (White and Ringleberg, 1998).

Sub-samples within each experimental unit (paddock) were used in ANOVA linear mixed-effects (LME) modeling using a restricted maximum likelihood (REML) algorithm (Pinheiro and Bates, 2000). All modeling was done using S-Plus 7.0 (Insightful Corp., Seattle, WA, 2005). Saturated models were constructed (Burnham and Anderson, 1998) to analyze response variables as a function of management treatments, once we accounted for the random effect of sub-sample nested within farm nested within date (date/farm/sub-sample). Improvements to subsequent models by dropping random effects terms or including a correlation function parameter were assessed with likelihood ratio tests for pairwise model comparison (Crawley, 2002). Once the random effects structure was determined, the significance of fixed effects were determined using Wald-type F-tests and treatment- level differences were inferred from P-values associated with t-tests of significance (P < 0.05) for each parameter estimate (Pinheiro and Bates, 2000). If treatment was significant, treatment levels were sequentially collapsed and subsequent models were compared using the maximum likelihood algorithm using the same model selection approach. Separate model selection procedures were run to test for treatment effects on microbial biomass, F:B, GM+ and Gm- bacteria, AMF, and actinomycetes. To assess community structure, multivariate analysis of arcsin transformed mol fractions of individual lipids was performed with PCA using JMP 5.0 (SAS Institute Inc., Cary, NC). Variability of random effects (date/farm/management/sub-sample) across farms was assessed using a linear mixed-effects model where we fit the variance parameters around the grand mean (intercept). The resulting random effects parameter estimates are the standard deviations of each source of variation. For a given response variable, the variation was standardized by calculating the coefficient of variation (CV%) - dividing the parameter estimate by the grand mean. To relativize variability for a given response variable, CV% was divided by the sum of all CV% for that given component.

Research results and discussion:

Results

Variability across farms

Average variability of random effects (date/farm/management/sub-sample) across farms was quantified by summing relativized CV% for each source of variation, and then dividing by total number of parameter estimates. Average source of variation in parameter estimates was as follows; management (34%), sub-sample (33%), farm (16%), residual (11%), and date (6%). A paired t-test of management and sub-sample revealed that they contributed equally as the greatest source of variance in parameter estimates, t (5) = 0.14, p = 0.45.

Community Analysis

Principal components analysis indicated overall soil biota differed between management treatments. For all PLFA’s combined, MANOVA run on axis PC1 and PC2 indicated that treatment (P = 0.01) had significant effects on community structure with HARV being positive and highly correlated with PC1. All other treatments were negatively correlated with PC1, with both grazing treatments having the highest negative correlation. Along PC2, both HARV and CONT were positive and highly correlated, while the strongest negative correlation was NONE. Principal components 1 and 2 explained 60% of the variation in community structure by treatment. Individual lipid markers that had the highest positive and negative correlation to PC1, were Gm+, and AMF, respectively. The lipid marker most positively correlated with PC2 was Gm-, while the highest negatively correlated markers belonged to actinomycetes and fungi.

Microbial Biomass

Analysis of biomass showed a significant effect of treatment (Wald-type F1,3 = 5.35, P = 0.02). The NONE treatment biomass was 392 ± 24 nmol lipid g soil-1 (mean ± SE), which model selection revealed was significantly greater than MIRG (285 ± 17), CONT (275 ± 24), and HARV (267 ± 18).

Fungal to Bacterial Ratio

Treatment had a significant effect on f:b (Wald-type F1,3 = 2.77, P = 0.04). Fungal to bacterial ratio was higher in the HARV treatment when compared to MIRG and CONT, but was not different than the NONE treatment. The ratio of f:b markers was 0.61 ± 0.09, 0.48 ± 0.06, 0.46 ± 0.04, 0.45 ± 0.03 (mean ± SE), for HARV, NONE, MIRG, and CONT, respectively.

Microbial Guilds

Bacteria

Gram-positive bacteria (Wald-type F1,3 = 4.59, P = 0.004), and actinomycetes (Wald-type F1,3 = 9.57, P = <0.0001) were significantly affected by treatment while Gm- bacteria was not (Wald-type F1,3 = 0.74, P = 0.53). Average mol % of lipid biomarkers indicative of the Gm+ guild were significantly greater in the treatments that included grazing, MIRG and CONT, when compared to management that did not include grazing, HARV and NONE. Model selection showed lipid biomarkers for actinomycetes were significantly less in the NONE treatment compared to MIRG, CONT, and HARV. Fungi Soil extractions on samples from all farms were void of, or had less than 0.5 mol % of lipid biomarkers for saprophytic fungi and were disregarded. Arbuscular mycorrhizal fungi was not affected significantly by treatment (Wald-type F1,3 = 1.37, P = 0.25). Discussion Grazing herbivores can play a considered role in modification of microbial communities in the soil of grasslands. Grazing affects plant species composition, morphological and physiological traits of plants, inputs of organic matter to the soil, and physical, chemical and biological properties of the soil. Both increased and decreased nutrient availability may result from these influences, depending on their relative importance and on the factors limiting primary production in a particular ecosystem. Soil fertility gradients have been shown to impact microbial biomass, both at the landscape scale (Hamilton and Frank, 2001) and at the micro-site scale (Augustine and Frank, 2001). Our results showed that across farm variance could be attributed equally to both farm (site differences), and management (paddock level alterations). Using LME models we have been able to account for variability contributed by site (large scale), and variability contributed by sub-sample (small scale). This has allowed us to ascertain differences in management treatments across a landscape gradient of farms. Our results suggest that pasture management strategies for livestock and livestock removal influence below ground microbial communities. Principal components analysis revealed that type of management had an effect on general soil microbial community structure. In this study the HARV distinguished itself from all other treatments along PC1, while HARV and CONT separated out from MIRG, which also separated from NONE on PC2. These findings are complimentary to work by Clegg et al. (2006), who reported community differences between grazed and ungrazed grasslands of the United Kingdom, and Ingram et al. (2008) who found differences in microbial community structure in the 5 to 15 cm soil depth between grazing exclusion, lightly grazed, and continuously heavily grazed mixed-grass ecosystems in Wyoming, USA. While in California annual grasslands, Steenwerth et al. (2003) found no separation between grazed and ungrazed sites. Effects of herbage removal by large herbivores on soil microbial communities are well recognized, but they are mediated by soil fertility and amount of biomass consumed by herbivores (Bardgett and Wardle, 2003). Herbage removal changes the arrangement of above-ground parts of plants with consequences for above- vs. below-ground carbon allocation and nitrogen-cycling. Bardgett et al. (2001) showed that microbial biomass of soil estimated by PLFA (phospholipids fatty acids) analysis was maximal at low-to-intermediate levels of sheep grazing. In contrast, Tracy and Frank (1998) observed no effect of grazing by large herbivores on microbial biomass in grasslands of Yellowstone National Park, USA. In our study, all treatments that included biomass removal had lower measured umol lipid/g soil, an indicator of total microbial biomass. This runs contrary to studies that found decreased microbial biomass in heavily grazed pastures, but no decrease in lightly grazed (Bardgett et al., 2001; Ingram et al., 2008). On the other hand, our results are supported by studies that found no difference between grazing treatments (Tracy and Frank, 1998; Patra et al., 2005; Clegg, 2006), and increased biomass with management or grazing removal (Grayston et al., 2003; Clegg, 2006; Ingram et al., 2008). The hypothesis that frequent defoliation and manure inputs from herbivores favors fast nutrient cycling, and therefore bacteria, is thought to be the result of more labile C substrate inputs from both above and belowground (Bardgett et al., 1998). This hypothesis was to some degree borne out in our study. Fungal to bacterial ratios were lower in both grazing treatments, MIRG and CONT, when compared to HARV. Contrary to the clear pattern in the biomass removal treatments, we could not discriminate against the grazed or the HARV treatments, when it came to f:b in the NONE treatment. This suggests that there was more labile substrate available in the grazed treatments, while herbage removal and longer periods of growth between defoliation reduced labile substrates in the HARV treatment. The results of others have shown that f:b was lower in a continuously grazed pasture compared to a lightly grazed pasture in mixed-grass (Ingram et al., 2008), and were lower in a grazed (fertilized) when compared to ungrazed (unfertilized) upland grassland site (Bardgett et al., 1996). Guild analysis revealed no differences in Gm- bacteria, but significantly greater Gm+ bacteria under the grazing treatments. Grazing seems to alter the microbial community structure leading to greater overall lipid markers indicative of bacteria (see f:b discussion above), as well as the Gm+ guild. The results from other studies have been equivocal, with Ingram et al. (2008) finding no differences between lightly grazed, heavily grazed, or grazing exclusion in either Gm- or Gm+ bacteria, while Grayston et al. (2004) found increases in some, and decreases in other specific Gm- and G+ bacteria markers when comparing improved (intensively grazed, fertilized) to unimproved (lightly grazed, unfertized) grasslands. Certain Gm- bacteria lipid markers (14:1 w5c, 16:1 w7c) have been suggested as indicative of “faster” nitrogen (N) cycling (Mentzer et al., 2006). The 16:1 w7c was found in high abundance in our grazing treatments, this coupled with greater overall bacteria biomass suggests that N cycling may be faster in the MIRG and CONT treatments. Analysis of the actinomycete lipid biomarker 10Me 18:0 indicated a significant effect of management. This marker was found in less abundance in the management removal treatment, NONE. Studies have found 10Me 18:0 to be an indicator of change in soil conditions which are usually related to management or the implementation of management removal. A study by Patra et al. (2008) found significantly greater abundance of 10Me 18:00 under tall oat grass (Arrhenaterum elatius) when comparing “extensive” (grazed lightly one time per year) to “intensive” (grazed four times and harvested once per year), and in a semi-natural grassland in France using the same treatments (Patra et al., 2005). Clegg et al. (2006) found 10Me 18:0 markers to be greatest in grazed when compared to ungrazed pasture, but Ingram et al. (2008) found no differences in abundance of 10Me fatty acids between grazed and ungrazed treatments in mixed-grass. It has also been shown by clonal indentification that exogenous Actinobacteria in pasture may be partially the result of manure deposition by livestock (Jangrid et al., 2008). In our treatments that included fertilizer, grazing, and biomass harvest, greater abundance of 10Me 18:0 may be an indication of a change in soil conditions, which has been postulated by others (Frostegard et al., 1993; Clegg 2006). Wilkinson hypothesized (1997) that plants which reproduce vegetatively or have short-distance seed dispersal, such as perennial grasses, come into contact with genetically identical AMF that makes mutualism likely to evolve. In the absence of physical disruption of the soil, it is reasonable to expect the symbiosis between plants and AMF to remain constant. Although one might expect negative impacts on AMF abundance as a result of defoliation as more energy is shunted to the aboveground plant parts, in ecosystems of high soil fertility the competition between plants and AMF may be muted (Bardgett and Wardle 2003) such that treatment differences are unrealized. Arbuscular mycorrhizal fungi abundance was not different between treatments in our study. This result is supported by others who found no difference between the following treatments; mowing and grazing in dairy cow pasture in Finland (Mikola et al., 2009), light, moderate, or heavily grazed on short-grass steppe in Mexico (Medina-Roldan et al., 2008), and grazing exclusion, light continuous grazing, and heavy continuous grazing on mixed-grass prairie (Ingram et al., 2008).

Participation Summary

Educational & Outreach Activities

Participation Summary

Education/outreach description:

The results of this study will form part of a PhD dissertation chapter and will be submitted for publication in a peer-reviewed journal.

Project Outcomes

Project outcomes:
Impacts and Contributions/Outcomes

Information generated from this research will directly benefit and expand the knowledge base about management effects on pasture ecosystems giving farmers, agencies, and policy makers a better understanding of the links between pasture management and biotic resources. This information will assist in making management decisions on implementation and type of management. Furthermore, the participatory nature of this research this has fostered greater understanding and respect between researchers and farmer/producers.

Long-term Outcomes:

Improved management of grazed pastures will optimize the trade-off inherent between livestock production and environmental stewardship. Such an optimization will result in more sustainable farming, which should lead to more vital rural communities.

Economic Analysis

Not applicable

Farmer Adoption

This research was conducted to assess microbial community response across a range of southern Wisconsin farms. Farmer cooperation and assistance was essential in treatment identification and facilitation of sampling protocols.

Recommendations:

Areas needing additional study

Useful follow up studies would include treatment within farm monitoring over longer timescales to track potential changes due to management modification or alteration.

Also, to control variability introduced as a result of farm location and management style, a manipulative experiment with blocking for site variation would be important to further explore treatment differences.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.