Provenance & quality
The database was generated through a systematic search of published literature, conducted via Web of Science (1900-4th January 2018). Data was extracted from single and paired catchment studies identified in our search and from papers reviewed by Farley, Jobbágy, & Jackson (2005).
Corrections were applied to control river flow values, based on linear regression, where extracted data had not been corrected for differences in climate and/or catchment between the two raw river flow data sets. Any single catchment studies that used a mechanistic catchment model to predict river flow had to report validation analyses for that model for the study to be incorporated. Studies were not included if treatments were a combination of deforestation and forestation, or if the study was a retrospective investigation of a known decrease in river flow.
Control river flow, treatment river flow, and river flow response to forestation (treatment minus control) are reported in millimetres. River flow response is also reported as a percentage of control river flow in the same year. Additional climate data was extracted from CRU TS4.3 (Harris, Jones, Osborn, & Lister, 2014) using the location and extent of the experimental catchment and are reported in millimetres. The age of the established forest (years) was calculated as an area weighted mean. Where the change in forest area was not reported between two years of known forest area, the rate of expansion was assumed constant. Change in forest area is reported as a cumulative percentage of catchment area for each time series. All other variables are defined in the associated metadata. Some data sources did not report all information required by categorical data fields, and this is indicated by ‘NR’ (not reported). Where river flow and climatic data were reported graphically, these were extracted using PlotDigitizer (Huwaldt, 2015). River flow and precipitation measurements reported at a greater than annual frequency were aggregated to annual time scales prior to entry into the dataset.
For further detail regarding the production of this data set, please refer to Bentley, L. & Coomes, D. (2020) Partial river flow recovery with forest age is rare in the decades following establishment Global Change Biology. DOI 10.1111/gcb.14954.