Modelling of potential vegetation identifies diverging expectable outcomes of river floodplain widening

Abstract

River re-naturalisations are at the forefront of conservation efforts. The hope is that these interventions will benefit both local ecosystems and facilitate flood mitigation. While hydrological modelling has been a standard procedure in assessing the outcomes of river re-naturalisations, vegetation modelling has not always been performed as part of these assessments. We hypothesised that the use of potential vegetation modelling, i.e. the modelling of self-sustainable vegetation that can survive after the intervention, can provide insight into vegetation outcome of river re-naturalisation and can thus support vegetation restoration and conservation planning. We investigated the utility of potential vegetation modelling under a specific, widespread element of river re-naturalisation: river floodplain widening at three study sites along the Tisza River in Hungary. We applied potential vegetation modelling, in addition to hydrological and groundwater modelling, to assess the expected vegetation outcome of the river floodplain widening. Flood frequency and duration were assessed by the HEC-RAS hydrological model. Based on the output this hydrological model provided, expected values of the water-related explanatory variables (including groundwater level) were calculated. Statistical relationships encompassed by the existing mulitple potential vegetation (MPV) models of Hungary were applied to the environmental variable sets corresponding to conditions before and after floodplain widening (pre- and post-treatment, respectively), including water-related and other explanatory variables. This resulted in predicted potential vegetation distribution for pre-treatment and post-treatment conditions, which were then compared via ordinations and PERMANOVA. At two of the study sites, post-treatment potential vegetation prediction typically showed vegetation types requiring wetter and less saline conditions when compared to the pre-treatment potential vegetation distribution. This pattern corresponds to general expectations given river floodplain widening. However, at one of the sites, the potentiality of saline and non-saline steppe vegetation was actually more pronounced under the expected post-treatment conditions than those before widening. The strengthening of the potentiality of dry vegetation types can be explained by the minor environmental differences related to the microrelief. As the studied sites were all located in the lowlands where geomorphological variation is small, the effect of these minor geomorphological differences on post-treatment potential vegetation would have remained hidden without applying MPV models. In conclusion, scenarios that employ MPV models help predict river restoration outcomes more accurately and can help identify factors that might otherwise be overlooked. Thus, when combined with physical modelling of river flow, their use can aid in the restoration and landscape planning decisions in river re-naturalisation projects.