Climate mediates land-use and within-lake drivers of seston stoichiometry: a large-scale assessment across tropical shallow lakes

Abstract Seston stoichiometry links nutrient availability to the structure and functioning of aquatic food webs, yet how in-lake conditions interact with broader climatic, geomorphological and land-use drivers to shape seston carbon-to-phosphorus (C:P) ratios remains poorly understood in tropical systems. We used regression tree analysis (RTA) to evaluate the relative importance of in-lake, lake–catchment and climate variables in explaining seston C:P variation across 95 shallow lakes and reservoirs spanning a semi-arid-to-humid climatic gradient in northeastern Brazil. The RTA explained nearly half of the variance in seston C:P (49.6%) and revealed a hierarchical structure among predictors. Cumulative precipitation over the two years preceding sampling was the primary split, separating the lower- and higher-precipitation portions of the sampled gradient at 882 mm. Within the higher-precipitation branch, total dissolved phosphorus (TDP) was the only additional predictor, and the highest C:P ratios occurred where TDP was also high. Within the lower-precipitation branch, the partial pressure of CO 2 ( p CO 2 ) and land use further partitioned C:P variation. A complementary Random Forest analysis also ranked precipitation as the strongest global predictor, although overall out-of-bag predictive performance was low. Because bulk seston integrates algal and non-algal particles, these associations cannot be assigned to a single physiological mechanism. Instead, our results show that antecedent hydrological context changes which local and catchment variables are associated with seston stoichiometry, highlighting the importance of integrating climate, landscape and in-lake conditions when predicting freshwater elemental balance under climate and land-use change.