Article: Quantitative effects of grazing on vegetation and soils over a global range of environment
Authors: Milchunas, D.G. and Lauenroth, W.K.
Published: Ecological Monographs, 63(4). 1993. pp. 327-366
We have given significant consideration to Samuel McNaughton’s grazing optimization hypothesis. Essentially, McNaughton argues that grazing can in fact improve ecosystem productivity through compensatory plant responses and other biotic and abiotic factors associated with the effects of grazing animals. He provides compelling evidence to support these claims.
However, most of his research is specific to the African Serengeti. As noted by the authors of the article under consideration, “no quantitative evaluations of the long-term effects of grazing on ANPP (aboveground net primary production) have been made across ecosystems.” Therefore, the objectives of their research “…were to use quantitative techniques to compare the impacts of grazing on plant communities in relation to various grazing, abiotic, and ecosystem variables.” The authors go on to name the variables with which they were most concerned:
Specifically, community variables included plant species composition, abundance of dominant species, aboveground net primary production (ANPP), root biomass, and soil carbon and nitrogen of grazed and ungrazed sites. We asked how these depended upon grazing variables such as level of consumption (intensity) and years of protection from grazing (duration), and upon ecosystem-environmental variables such as mean annual precipitation, high and low temperatures, latitude, ANPP, and evolutionary history of grazing. We then asked whether relationships differed among grasslands, shrublands, forests, deserts, and high-elevation sites.
In order to perform such a comprehensive and global analysis, they searched and compiled information from the scientific literature to create a large data set for statistical analysis. The authors elaborate:
Approximately 500 potential articles were surveyed, with 97 articles representing 276 data sets obtained for analyses. The criteria for selection were that (1) ungrazed controls were compared with grazing treatments, (2) grazing was yearlong or during some part of the growing season, but not only during winter in temperate locations, and (3) species abundances and years of treatment, plus either above- ground net primary production (ANPP) of ungrazed treatment (or peak standing crop) or grazing intensity (either as a percentage of ANPP or as a stocking rate with ANPP) were provided or could be obtained from the authors. Other data compiled included precipitation, mean low and high temperatures for the coldest and warmest months, range of the annual low-high temperatures, ANPP of grazed plant communities, and latitude of the site. Belowground data most commonly reported were root mass, total soil N, and soil organic matter or carbon.
In regards to the nature of the included studies, the authors provide this information:
Some of the studies used in the analyses were from systems with native grazers where populations are not regulated by humans, some were from sites where grazing by domestic livestock was “uncontrolled” or “free- grazing” or represented “overgrazed” situations, but most were studies of controlled levels of grazing by domestic livestock.
To determine a key variable in their statistical models, they turned to their colleagues in the scientific community:
…sixteen scientists with several different areas of expertise and from various parts of the world were asked to categorize the sites according to evolutionary history of grazing, and to rank the certainty of their estimates. The individuals were asked to rank each study site either 1, 2, 3, or 4 for least to greatest evolutionary history of grazing, and to rank their estimate either 1, 2, 3, or 4 for low to high certainty.
Statistical analysis performed on the data relied heavily on the use of multivariate regression models. Various iterations were performed on different statistical models. Models that had poor explanatory power were not included in the results; models using independent and dependent variables with strong relationships were included in the results and discussion of the paper. Since stronger relationships were found in grassland and shrubland communities (which tend to have a longer evolutionary history of grazing), separate “regressions were run for all community classes combined, grasslands-plus-shrublands, grasslands, and shrublands when sufficient data were available.” The statistical models developed are given consideration below.
Aboveground Net Primary Productivity
The response of ecosystem productivity to grazing is a controversial topic that has been under debate for decades. This paper provides strong evidence that grazing can increase or decrease ecosystem productivity. The authors describe the measurement method for ANPP:
In the majority of cases ANPP was estimated as peak standing crop in ungrazed treatment and in temporarily caged grazed treatment; compensatory growth due to current-year defoliation is not accounted for. [emphasis added]
The summarized results of the regression analysis are provided below:
Most of the differences between ANPP of grazed vs. ungrazed plant communities were negative. However, the statistical models for grasslands-plus-shrublands or for grasslands alone predicted positive ANPP responses to grazing in some situations. For the entire data set, 17% of the cases had positive values for the effects of grazing on ANPP and these were generally low levels of consumption and few years of treatment.
The authors elaborate on those conditions where grazing tends to induce a positive productive response:
Conditions under which grazing was more likely to increase or have no or small effect on ANPP were long evolutionary history and low productivity, regardless of the number of years of grazing treatment or the levels of consumption within a range generally not considered abusive “overgrazing.”
In other words, grasslands and shrublands seem to benefit from moderate grazing, while mountains, deserts, and forests are less likely to see an improved productive response due to grazing.
When analyzing below-ground productivity, even across ecotypes, this argument becomes even more compelling:
Whereas the effect of grazing on ANPP averaged -23%, the effect on root mass averaged + 20%. Further, positive effects of grazing on root mass occurred in 61% of the sites where grazing had negative effects on ANPP and in 62% of sites where differences in ANPP were negative, positive, or not known. [And] negative impacts of grazing on aboveground production were accompanied by as many positive as negative responses belowground. The general perception of decreased plant production with grazing may not be as great when viewed at the level of the whole plant
Another critical data point makes this argument more persuasive. As we have seen in previous articles, compensatory growth due to herbivory can be a powerful force driving greater ecosystem productivity. Yet, as mentioned previously, this study did not account for compensatory growth due to defoliation in measures of ANPP.
Of critical importance is the concept that evolutionary history may be a key driver affecting ecosystem response to herbivory and defoliation. The presented data show a very strong relationship between evolutionary history of grazing and ANPP; sites with a long evolutionary history of grazing show greater resilience and in general respond more favorable to herbivory. As the authors point out, “the predominance of ecosystem-environmental variables rather than grazing variables in sensitivity analyses suggests that where we graze may be more important than how we graze.”
The authors address a critical point of weakness in their “meta-analysis” approach:
Our approach entailed combining studies that used different methods, grazing systems, and animals, and varied in topography or weather during the year(s) of study, etc. The studies also spanned a range of strong to weak experimental designs and degrees of accuracy in estimating variables such as aboveground net primary production (ANPP) or consumption.
Also of significance is the fact that most of these studies represent systems under management either by scientists or field managers. In most cases, the natural cyclic patterns of grassland ecology so thoroughly described by McNaughton are not in effect; the critical factor of human management may in fact be the principle phenomenon driving “noise” in the research data. Yet, as the authors note, the “general directions indicated by the relationships and the relative influence of variables in the statistical models” on the whole seem acceptable, especially when tempered by thoughtful analysis.
In my next post, I will present the research results from this paper dealing with changes in species composition, and the relative impact of grazing on shrubland ecosystems.