Fig 1.
Study area showing the Brazilian Pantanal (gray) with the headwaters (white) and the transect areas along the Paraguay River.
Black triangles indicate unburned areas, and white triangles indicated areas burned in 2001, 2005, and 2009.
Fig 2.
Levels of inundation of the Paraguay River between 2000 and 2013 (m) and monthly mean of rain (cm) in the municipalities of Corumbá and Ladário, MS, Brazil.
Data provided by Brazilian Navy and INMET (Brazilian National Institute of Meteorology).
Fig 3.
Maximum and minimum levels attained by Paraguay River (m) between 1974 and 2011 in relation to the zero of Ladário gauge, showing the study period (brace) and the years selected as the basis for the fire investigation (arrows).
The dashed line indicates a general threshold for inundation of the Paraguay River.
Fig 4.
Relationship between the three dependent variables (richness, abundance and basal area) and the interaction of time of inundation and fire occurrence in riparian forests of the Paraguay River.
Graphic performed with Generalized Linear Model only for visualization purposes once the results of GLM and GLMM were very similar. The continuous and dotted lines are burned and unburned areas respectively. The shaded areas in both lines are confidence intervals.
Table 1.
List of shrub and tree species found in the riparian forest of the Paraguay River in burned and unburned areas between 2001 and 2011.
Table 2.
Results for the three GLMM models.
Interactions were found to be significant for richness and abundance but not for basal area. The applied distribution type is listed beneath the name of the respective dependent variable. Numbers in brackets denote standard errors. Pseudo-R² calculation followed Nakagawa & Schielzeth [40]. The marginal R2 shows the variation explained only by the fixed effects while the conditional R2 shows the variation of the fixed and random effects. For the negative-binomial abundance model calculation of R² was not possible hence we used the R2 of a Poisson model as a proxy.
Fig 5.
Direct ordination of burned and unburned plots in the riparian forest of Paraguay River, showing the yearly mean time of inundation.
Grey bars correspond to burned sites and black bars to unburned sites.
Table 3.
Indicator species analysis for all combinations of the fire regime and two inundation categories.
Of the 24 combinations, 5 had species with significant group equalized indicator values (IndValg), and 13 of the 39 tree and shrub species were indicative for a factor combination. Low inundation refers to areas of 4.32–4.85 m, while high inundation refers to areas of 3.79–4.32 m in relation to the zero of the Ladário gauge.
Fig 6.
Species responses to the elevation gradient for (a) burned and (b) unburned areas. Optima of the specific model types are depicted as symbols, while error bars indicate the tolerance range per species. Response model type IV (unimodal); V (unimodal left skewed); VI (unimodal right skewed) and VII (multimodal). Species names were abbreviated to eight letter codes. Elevation is relative to the Ladário gauge.