Metadata

Description
Literature search In March 2018, we performed a review of the academic literature using the Web of Science to identify drivers and ecological responses to abrupt change, as well as the spatial and temporal scales they occurred at using literature published between 2000 - 2018. We targeted both paleoecological and modern ecological peer-reviewed articles using two searches. The first search used the terms “TITLE: (abrupt chang* OR rapid chang* OR sudden chang* OR tipping point*) AND TOPIC: (ecosystem type) from 2000 - 2018” and returned 334 articles. The second search used the terms “TITLE: (regime shift* OR state change* OR alternate stable state* OR phase shift*) AND TOPIC: (foundation species OR keystone species OR top down OR bottom up) AND CATEGORY: (ecology) from 2000-2018” and returned 404 articles. Four articles were found to be duplicates, resulting in a total count of 734 articles between the two searches. Inclusion-exclusion criteria for article screening Due to the relative nature of abrupt changes in ecological research, we developed several exclusionary criteria in order to focus our meta-analysis on papers that captured naturally occurring, environmental change that resulted in identifiable responses that were either ecological or had clear ecological relevance. We excluded all papers that were (1) from a field outside of ecology or an associated field; (2) not an original, peer-reviewed research article (including reviews, meta-analyses, comments, editorials, book chapters, etc.); (3) field- or lab-based experiments; (4) models that either forecasted future changes, or were simulations; and (5) paleoecological studies, and (6) studies where a clear abrupt change could not be identified. Our exclusion criteria were developed with the goal of selecting articles that characterized discrete ecological changes as they naturally occur, resulting in the inclusion of 117 papers. Data extraction For each paper, two rounds of data extraction were done as a means of quality assurance/quality control. A third round was then completed by the lead author to ensure consistency, including the article screening process. We generally followed the meta-analytic framework outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA; Moher et al. 2009). During the third round, screening and data extraction were completed simultaneously. Data extraction began by classifying the general ecosystem as either terrestrial, freshwater, or marine. We then identified the geographic location, including latitude and longitude, and, if present, recorded the definition of abrupt change. Papers were then examined for evidence of drivers and triggers of ecological change and the associated ecological response. A “driver” was defined as an external variable or factor that influenced the dynamics of the system without themselves being affected by the system (Ratajczak et al. 2018), while a “trigger” was defined as a discrete event in time, or a singular event that directly affected the ecological response (Bestelmeyer et al. 2011). A “response” was any ecological or biological parameter where a change was identified (Bestelmeyer et al. 2011). Drivers were binned into one of six categories: climate, disturbance, eutrophication, hydrology, species, or anthropogenic. Anthropogenic refers to changes induced by management, recreation, commercial operations, pollution caused by surface run-off in urban or agricultural areas. For papers where we identified multiple drivers (n = 5), combined driver bins were recorded, rather than identifying them as primary and secondary drivers (e.g. climate-hydrology). Triggers were categorized with the same categories as drivers and only recorded when a trigger could be clearly identified. Responses were binned into one of seven categories: population, community, ecosystem, inorganic matter, organic matter, and primary production. Ecosystem was recorded when changes were measured across multiple components of the system. When authors clearly identified multiple, distinct ecological changes, each change was given a separate entry. When papers only focused on examining drivers of abrupt change, no ecological response was recorded. When authors could not clearly distinguish the driver of abrupt change, no driver was recorded. We then identified the spatial and temporal scales of the driver of the ecological change and associated ecological response. Spatial scales were binned into one of three categories: (1) single site, (2) local, or (3) regional. Single site was recorded when the study clearly took place at a single site. Local spatial scale was recorded when the study was clearly at the scale of an ecosystem or connected landscape, including lakes, and any study that took place at small scales e.g. 10x10m plots. Regional spatial scale was recorded when it was explicitly stated in the paper or when ecological changes could only be identified after aggregating information from multiple sampling locations. Temporal scales were binned into five categories: (1) sub-annual, (2) annual, (3) sub-decadal, or (4) decadal, or (5) centennial. When ecological changes were only documented across space, such as space-for-time substitutions, no temporal scale was recorded. Temporal scales of responses were determined based on the length of time it took for the ecological change to measurably occur, taking into account the temporal resolution of the sampling (e.g. monthly sampling averaged to annual scales). Temporal scales of drivers were determined based on the length of time it took for the driver to have a significant impact on ecosystem state. Lastly, we identified if the ecological change was associated with a “state change”. We defined state change based on how “state” is defined by Ratajczak et al. (2018). A state is, “the characteristics used to describe the state of an ecosystem at a particular domain in space and time”. Therefore our definition of state change was defined as an observed large change in an ecosystem that can be gradual or fast, with the latter being described as an abrupt change, which can or can not lead to a transition to an alternative state, and be used synonymously with phase shift (Ratajczak et al. 2018).