Divergent effect of landscape patterns on stream water chemistry and seasonal variations across mountainous watersheds in a Northwest Pacific island

Abstract

The impacts of landscape pattern on water quality of streams are intricate and influenced by spatiotemporal scales. Untangling the interactions between landscapes and water, and identifying the pivotal indicators that drive spatial and temporal changes in nutrient dynamics, are essential steps towards facilitating sustainable water resource management. This study explored the relationship between landscape patterns and annual and seasonal (wet and dry seasons) exports of major cations (Na+, K+, Ca2+, and Mg2+) and anions (Cl-, SO42-, NO3-N, and PO4-P) in 43 watersheds across subtropical mountainous regions of Taiwan during the 2015–2016 period. Multiple stepwise regression models were constructed to elucidate the crucial landscape variables influencing variations in stream ion exports across different time scales, and spatial extent, ranging from watershed to various riparian buffer widths (100, 200, 500, 1000, 1500, and 2000 m). The findings indicated that the buffer width of 200 m emerged as the optimal scale for illustrating the connections between landscape patterns and water chemistry, outperforming larger scales. Nevertheless, there were minor deviations in the optimal scale for Na+ and Cl- (ranging from 100 to 500 m). Landscape-level configuration metrics (PD [patch density], CONTAG [Contagion], and IJI [Interspersion and Juxtaposition]) played a significant role in explaining the majority of data variations in cations. This suggest that overall landscape fragmentation tends to enhance cations exports, regardless of land cover types. On the contrary, class-level configuration metrics (EDBUD [edge density of buildup], PDBAR [patch density of bareland], and IJIAGR [Interspersion and Juxtaposition of agriculture]) played a primary role in interpreting data variations of anions. This indicated that the configurations of specific land cover types, such as buildup, agriculture and bareland, are the primary factors influencing anion exports. Furthermore, the regression models exhibited robust performance during both wet and dry seasons, implying their consistent efficacy in estimating ion exports. This synthesis illustrated the benefits of utilizing this cost-effective approach to assess landscape-water connections in challenging mountainous regions, assisting water resources management.

Keywords

• Landscape configuration
• Landscape-level metrics
• Class-level metrics
• Cations
• Anions
• Stepwise regression model