Alternative stable states and spatial indicators of critical slowing down along a spatial gradient in a savanna ecosystem. Issue 6 (9th February 2017)

• Record Type: Journal Article
• Authors: Eby, Stephanie; Agrawal, Amit; Majumder, Sabiha; Dobson, Andrew P.; Guttal, Vishwesha
• Other Names: Mayfield Margaret checker.
• Is Part Of: Global ecology & biogeography. Volume 26:Issue 6(2017)
• Page Start: 638
• Page End: 649
• Publication Date: 2017-02-09
• Subjects: alternative stable states -- critical slowing down -- critical transitions -- early warning signals -- grassland–woodland transitions -- savanna ecosystems -- Serengeti–Mara ecosystem -- spatial ecology -- spatial variance; Ecology -- Periodicals; Biogeography -- Periodicals; Biodiversity -- Periodicals; Macroevolution -- Periodicals; 577
• DOI: 10.1111/geb.12570
• Languages: English
• ISSNs: 1466-822X

Abstract

Abstract: Aim: Theory suggests that as ecological systems approach regime shifts, they become increasingly slow in recovering from perturbations. This phenomenon, known as critical slowing down [CSD], leads to spatial and temporal signatures in ecological state variables, thus potentially offering early indicators of regime shifts. Indicators using temporal dynamics have been empirically validated in laboratory microcosms and other well‐mixed systems, but tests of spatial indicators of regime shifts at large spatial scales in the field are rare due to the relative absence of high‐resolution data and difficulties in experimental manipulations. Here, we test theoretical predictions of CSD‐based spatial indicators using large‐scale field data from the Serengeti–Mara grassland–woodland system. Location: Serengeti–Mara ecosystem, Tanzania and Kenya. Time period: Year 2000 Major taxa studied: Vegetation Method: We used a space‐for‐time substitution method to empirically test the validity of CSD‐based spatial indicators, i.e., we computed indicators along a spatial [in lieu of temporal] gradient of ecological states. First we used a model of vegetation dynamics to determine if our space‐for‐time substitution method was appropriate. Then we tested for CSD‐based spatial indicators using high‐resolution spatial vegetation [30 m] and rainfall [2.5 km] data from the Serengeti–Mara ecosystem. Results: Our model predicts that CSD‐based indicators increase along a spatial gradient of alternative vegetation states. Empirical analyses suggest that grasslands and woodlands occur as alternative stable states in the Serengeti–Mara ecosystem with rainfall as one of the potential drivers of transitions between these states. We found that four indices of CSD showed the theoretically expected increasing trends along spatial gradients of grasslands to woodlands: spatial variance, spatial skewness, spatial correlation at lag‐1 and spatial spectra at low frequencies. Main conclusions: Our results suggest that CSD‐based spatial indicators can offer early warning signals of critical transitions in large‐scale ecosystems.