Fig 1.
Photographs of the study system in the Cordillera Real of Bolivia (above 4500 m).
A. Wetland in the arid landscape of the puna, B. Proximity of wetlands and glaciers, C. Cushion plants (Distichia muscoides) composing the wetlands, D. Example of a bird species (Chloephaga melanoptera) tightly dependent on wetlands. © Fabien Anthelme (A) and Olivier Dangles (B-D).
Fig 2.
Map of the study site and satellite images.
Map of the study area in the Cordillera Real, Bolivia (A, B) and examples of high elevation wetland delineation from Pléiades image recorded in 2013 (C) and Landsat-TM 2007 (D). The white outlines in (D) show the extension of the wetlands quantified from the 2007 Landsat image. Overlapping these outlines on the 2013 Pléiades image (C) shows the overall good agreement with the extension of the wetland six years later. The blue arrow illustrates the growth of the wetland between the two dates.
Fig 3.
Changes in the total annual wetland surface area and wetland count over the 1984–2011 period in the Cordillera Real, Bolivia.
The uncertainty in wetland cover calculations based on satellite images varies from a few km2 to 10 km2 depending on the year (see Methods).
Fig 4.
A) Relationship between high-elevation wetland surface area and count and B) changes in the distribution of individual high-elevation wetland surface areas from 1984 to 2011 in the Cordillera Real, Bolivia.
Fig 5.
Changes in the total annual high-elevation wetland surface area from 1984 to 2011 in the Cordillera Real, Bolivia.
Changes are presented with regards to A) orientation, B) elevation, C) mean slope, D) initial glacier cover in the catchment (in 1984), E) initial wetland surface area, and F) wetland watershed surface area. For B to F, the two time series and trends in each plot refer to small (lower quartile) vs. large values (higher quartile) of the environmental variables.
Table 1.
Results of the analysis of covariance of time and six environmental features (elevation, orientation, slope, watershed surface area, wetland initial size, and glacier cover in the catchment) on the evolution of wetland cover between 1984 and 2014 in the Cordillera Real (Bolivia).
Fig 6.
Temporal trends in monthly precipitations (A) and glacier and snow-covered surfaces (B) over the period (1984–2011) in the Cordillera Real, Bolivia.
Monthly Precipitation data were obtained from the Zongo valley. Glacier and snow-covered surfaces were measured using LANDSAT images during the dry period (between May and August). Panels C and D give the relationships between wetland cover and precipitation (monthly values over three months prior to the satellite image) and glacier and snow-covered surfaces.
Table 2.
Results of the generalized linear model’s deviance analysis on wetland drying frequency.
Only significant interaction effects are shown. AIC is the Akaike’s Information Criterion for the initial model after removal of the ‘‘effect” term. Likelihood-ratio test (LRT) and associated p-value test the hypothesis that the suppression of the ‘‘effect” term provides no better fit than the initial model.
Fig 7.
Relationships between wetland drying frequency over the period (1984–2011) and 1984 data and several predictors.
% glacier cover in the catchment (A), wetland area (B) and wetland shape index (C), based on 423 wetlands in the Cordillera Real, Bolivia. Data are fitted with linear models. Dotted lines indicate 95% confidence intervals.
Fig 8.
Surface plot of wetland drying index.
The index is given as a function of wetland area and wetland shape index (1984 data), based on 423 wetlands in the Cordillera Real, Bolivia.
Fig 9.
Map of the predictive vulnerability status of 1689 wetlands recorded in 2011 in the Cordillera Real, Bolivia.
Wetlands are represented by small marks that define irregular shape where clustered.