Detection and Attribution of Regional Land-Use and Climate Change Signatures in River Nutrient Time-Series Using Dynamic Factor Analysis

Abstract:

The scientific community concurs that the rate and the intensity of the current environmental changes have been accelerated by anthropogenic activities. Mediterranean freshwater systems are particularly vulnerable to such changes due to their inherent climate-dependent hydrological variability. We aimed to detect common patterns in monthly nutrient concentration time-series (1980-2011) from 50 sampling stations across a Mediterranean river basin, and to attribute their spatiotemporal variability to environmental factors at the basin and regional scales. Dynamic Factor Analysis (DFA) provided the methodological framework to extract underlying common patterns in nutrient time-series with missing observations, a commonly encountered problem in environmental databases. Most importantly, DFA guaranteed the explicit consideration of the inextricable link between temporal and spatial patterns of change necessary to investigate the drivers and processes that shape them. Using complementary methods such as frequency and trend analyses, we sought to further characterize the extracted patterns and identify the drivers behind their variability across time and space. The extracted nitrate concentration patterns described a large proportion of the observed variability at the basin scale. Cycles of 2.5 and 3.5 years identified in nitrate concentration patterns were linked to climatic oscillations. The seasonality of nitrate patterns was either driven by hydrological or phenological processes, depending on the geographical location of the monitoring point. Land uses linked to fertilizer application further modulated the increasing and decreasing nitrate trends scattered across the basin. Conversely, phosphate concentration patterns did not fully describe the behavior of all monitoring points included in the analysis. Nevertheless, decreasing phosphate trends observed across space and time coincided with changes in land-use management practices in the study basin.