Wetland responses to non-stationary hydro-climatic dynamics in the context of land cover and land use change

AQUINO, DIEGO SEBASTIÁN; GAVIER?PIZARRO, GREGORIO IGNACIO; RESCIA, ALEJANDRO J.; QUINTANA, RUBÉN D.

Remote Sensing Applications: Society and Environment

Elsevier B.V.

Año: 2024 vol. 34

Most ecological processes typically violate the stationarity assumption as its statistical properties do vary with time. Not only is climate a non-stationary driving phenomenon, but biological systems are intrinsically non-stationary as well. Since wetlands’ structure and functioning are determined by its hydrologic regime, they are not simply driven by temperature and rainfall, but also by seasonal and inter-annual phases of flood and drought. Thus, assessment of non-stationary processes might be particularly accurate in regards to dynamically complex or altered wetland ecosystems. Our aim was to address whether wetland vegetation growth dynamics are intrinsically non-stationary and driven by transient interactions with non-stationary hydro-climatic factors in the context of land use and land cover changes (LULLCs). In order to provide better insight into how and when temporal dynamics guiding ecological transitions occur, we decomposed information from six NDVI time-series depicting differing LULCCs scenarios in a mosaic of wetlands. Thus, to better comprehend the simultaneous underlying processes driving wetland dynamics in the non-insular Lower Delta of the Paraná River, Argentina, we applied the Wavelet Transform. Wavelet analysis is free from the assumption of stationarity and successfully addresses the relationships between two time-series, in the context of gradual changes forced by exogenous variables. In general, our main results show that non-stationary wetland vegetation dynamics can significantly and cyclically alter its periodicity across time in the context of LULCCs and because of its significant interactions with non-stationary hydro-climatic drivers. Our results also show that wavelet analysis can aid in understanding multi-scale non-stationary ecological time-series and reveal features that were either unseen or wrongly assumed otherwise, such as stationarity or constant linear relationships. Implementing the wavelet approach, we have demonstrated that it is possible to study irregular, non-stationary NDVI time-series in wetland ecosystems in order to detect weak and transient interactions between hydro-climatic drivers and wetland vegetation growth dynamics. Particularly through wavelet power spectrums, our results not only identified and quantified the main periodic component of given NDVI time-series, but also assessed its progression through time in the context of complex LULCCs. Principally, we observed the greater the LULCC, the weaker the relationship with most hydro-climatic variables. The aforementioned LULCCs seemingly related to both natural and anthropogenic processes occurring at a given moment in time. Thus, it reflected not only on transient interactions with significant hydro-climatic drivers, but also on changes in the dominant periodicity of wetland vegetation dynamics. Our findings suggests that water management infrastructure exerts significant and irreversible impacts on wetlands. Therefore, a comprehensive approach and strategic planning are necessary to minimize the negative impacts of such infrastructure and to ensure the long-term sustainability of wetland ecosystems.