Document Type

Thesis - Open Access

Award Date


Degree Name

Master of Science (MS)

Department / School

Natural Resource Management

First Advisor

Alexander J. Smart


Drought can have major impacts on rangeland productivity and remains highly unpredictable. Like many other rangelands in the US, the Tallgrass Prairie of eastern South Dakota contains native prairie where the plant composition includes mostly native species as well as pasture that has been converted to or invaded by cool-season introduced species. So how are these two plant communities impacted by drought? The specific objective of this study was to compare drought resistance and drought resilience of native prairie to introduced cool-season pasture (Smooth brome - Bromus inermis and Kentucky bluegrass - Poa pratensis). Our Hypothesis was that native prairie would be more resistant to drought than introduced cool-season pasture, but not more resilient when average precipitation returns. This is because of the Insurance Hypothesis: native prairie contains more plant species and functional groups with a greater variety of adaptations that allows native prairie to collectively resist drought (Lawton and Brown 1993; Yachi and Loreau 1999). Two sites near Volga, SD were used for the study: a Non-Native Site with introduced cool-season grass site dominated by Bromus inermis and Poa pratensis and a Native Site with primarily native cool- and warm-season grasses, and some forbs. Three automated rainout shelters at each site simulated drought conditions by intercepting rainfall. There were 8, 1-m2 study plots under each shelter and 2 addition study plots outside each shelter. Treatments in 2013 under the rainout shelters included 50%, 75%, 100%, and 125% of the 30-year average growing season precipitation (Brookings CO-OP 2010). There were 2 replicates under each shelter for each treatment and 6 total replicates at each site. The two additional treatments included: an Ambient treatment (precipitation from that year with no moisture added) and a Well-watered treatment (abundant moisture, about 290% of the 30-year average precipitation). There were 3 replicates at each site. In 2014, some of the treatments were altered. The 100%, Ambient, Well-watered, and half of the 50%, 75%, and 125% plots received the same amount of moisture as 2013. However the other half of the 50%, 75%, and 125% plots received 100% of the 30-year average precipitation. The study was composed of three separate experiments. Experiment I looked at drought resistance among the treatments (50%, 75%, 100%, 125%, Ambient, and Wellwatered. Drought resistance was determined by comparing the current year’s biomass production for each treatment during the drought to the biomass production of the 100% treatment (30-year average). Experiment II continued to monitor drought resistance in 2014 with a two-year drought, again comparing the treatments. The focus of Experiment III was to examine drought resilience among the treatments when they received 100% of the 30-year average in 2014 following a drought in 2013. Supplemental water was applied on a weekly schedule based on the 30-year average precipitation and the amount of rainfall not excluded by the shelters. Biomass samples were clipped using a 0.25 m2 quadrat, sorted by species, dried, and weighed at the end of the growing season in 2013 and 2014. Results for all three experiments indicated that a linear trend in the data was observed from the 50% treatment up to the 125% treatment; Experiment I (p = 0.01), Experiment II (p =0.009), and Experiment III (p < 0.0001), and there was a significant difference between the treatments; Experiment I (p = 0.01), Experiment II (p =0.04), and Experiment III (p < 0.0001). There was no significant difference between the sites overall; Experiment (p = 0.3), Experiment II (p =0.8), and Experiment III (p = 0.6). There was no significant difference between each site by treatment for Experiments I and II (p = 0.3 and p =0.6, respectively), however Experiment III did show a significant difference in site by treatment (p = 0.0009). The three plant species at the Non-Native Site were examined individually, Bromus inermis (Smooth brome), Poa pratensis (Kentucky bluegrass), and Cirsium arvense (Canada thistle). There were significant differences by treatment in B. inermis (p = 0.03) and P. pratensis (p = 0.02) in Experiment I in 2013. However the disparity occurred in the 75% treatment and the other 5 treatments (50%, 100%, 125%, Ambient, and Well-watered) had similar values. There was no significant difference by treatment for B. inermis and P. pratensis in 2014 for Experiments II and III; B. inermis (p =0.6, p = 0.3, respectively) and P. pratensis (p = 0.5, p = 0.1, respectively). There was no linear trend from the 50% treatment to the 125% treatment; Experiments I-III: B. inermis (p =0.5, p = 0.2, p = 0.6, respectively) and P. pratensis p = 0.4, p = 0.2, p = 0.9, respectively). Cirsium arvense had very low abundance and was not present in every treatment, so statistical analysis could not produce p-value for treatment effect or linear trend. Functional plant groups were also examined at the Native Site, including warmseason native grasses, cool-season introduced grasses, cool-season introduced grasses, and forbs. However, no significant difference was found in these groups by treatment in for Experiments I-III in 2013 and 2014; warm-season native grass (p =0.2, p = 0.1, p = 0.2, respectively), cool-season introduced grass (p = 0.4, p = 0.1, p = 0.2, respectively), cool-season native grass p = 0.3, p = 0.6, p = 0.2, respectively), and forbs p =0.6, p = 0.9, p =0.6, respectively). There was a linear increase from the 50% treatment to the 125% treatment in Experiment II for warm-season native grasses (p = 0.03), but not for the 3 other functional groups; cool-season introduced grasses (p =0.2), cool-season native grasses (p =0.3), and forbs (p =0.8). A linear trend was not observed in Experiment I and III; warm-season native grass (p =0.2, p = 0.4, respectively), cool-season introduced grass (p = 0.4, p = 0.6, respectively), cool-season native grass (p = 0.3, p = 0.2, respectively), and forbs (p =0.4, p = 0.4, respectively). Based on the results of the study, there was a clear treatment effect in all three experiments, meaning that the treatments were different enough from one another to show have an impact on biomass production. However, the sites were not different, meaning that they were equally impacted by drought and recovery from drought. These data did not support our hypothesis that the diverse plant community of the Native Site would be more drought resistant. Part of this is that unfortunately the treatments for Experiments I and II did not reach the desired goal during the 2 years of research (2013- 2014). However, in Experiment III the drought treatments did not recover to the level of biomass produced by the 100% treatment, so this shows that there was a treatment effect. The inaccuracy of the treatments was mainly due to difficulties and malfunction in the operation of the Rainout Shelters. An increase in the number of study plots and replicates for each treatment may have provided more conclusive results as well. With modifications and additional years of research, more definitive conclusions may be attainable regarding the drought resistance and resilience of these 2 plant communities in South Dakota.


Includes bibliographical references (pages 48-51)



Number of Pages



South Dakota State University



Rights Statement

In Copyright