Search for generality in grassland responses to climate change: replicated experiment in three prairie sites

Background/Question/Methods

Few generalizations have emerged from numerous experiments investigating grassland responses to shifts in temperature and precipitation associated with climate change. To understand if there are general patterns of responses, as opposed to a world of special cases, we need replicated experiments utilizing consistent methodology, as has been called for pervasively. Ideally, such studies would allow us to also identify underlying drivers. Further, as grasslands are typically managed systems, grazing intensity should be included as a potentially interacting factor. To address these needs, we conducted a three-year, factorial experiment manipulating precipitation, temperature, and defoliation in one site in each prairie province: Alberta, Saskatchewan, and Manitoba. We used rain-out shelters to decrease precipitation by 60%, open-top chambers to raise air temperature by ~3°C, and hand-clipping and mowing to defoliate. Response measures included plant diversity, and aboveground and belowground biomass. We also measured a host of intermediary factors potentially involved in indirect effects, including soil nitrogen, plant litter, and soil moisture. To evaluate the direct effects of the treatments, as well as their indirect effects, we used structural equation modeling. We developed a structural equation model linking the treatment, intermediary, and response variables, and tested strengths of paths between variables for each site.

Results/Conclusions

Diversity and aboveground biomass were responsive to treatments at all three sites, while belowground biomass was largely unresponsive. A common finding was that temperature consistently had no effect on response variables in any location. In Saskatchewan, the driest and least productive site, responses were driven by clipping, while in Manitoba, the wettest and most productive site, responses were driven by precipitation; both precipitation and clipping contributed to responses in Alberta. The specific indirect pathways and the overall importance of indirect versus direct effects varied highly among sites. However, among all sites, nitrogen did not contribute to any indirect effects on the response variables. Although site-specific characteristics likely caused both the predominant drivers and the indirect pathways to vary among sites, we also identified temperature and nitrogen as factors that had no effect across all sites. Whether the commonalities observed among these three superficially similar sites can be extended as generalities among temperate grassland sites remains a question to be answered. However, comprehensive predictions of ecosystem response to current and future global change may remain challenging, but frameworks such as this one that are capable of  identifying any common responses that do manifest are a step in the right direction.