Rainfall Pulse Patterns Shape Ecosystem Productivity across Contrasting Dryland Hydrological Systems: An Israeli LTER Network Study

Authors:

Abstract
Background: Understanding how ecosystems respond to changing rainfall regimes is a central challenge for predicting ecosystem dynamics under climate change. Rainfall pulse patterns, defined by the timing, intensity, frequency, and duration of precipitation events, regulate water availability and primary productivity in drylands. Dryland ecosystems can be broadly classified as rainfall-dependent, runoff-dependent, or flood-dependent systems, each sustained by distinct hydrological processes. We tested whether ecosystems with contrasting hydrological dependencies require different rainfall pulse regimes to maintain productivity. Guided by the Pulse-Reserve Paradigm, we hypothesized that each hydrological system responds to a unique combination of rainfall pulse characteristics. To test this hypothesis, we developed an integrated framework that combines LTER monitoring, Earth observation, and machine-learning analysis. Methods: The study was conducted across seven sites within the Israeli Long-Term Ecological Research (LTER) network, spanning a climatic gradient from hyper-arid to dry sub-humid conditions (40–700 mm mean annual rainfall). Collectively, these sites represent rainfall-dependent, runoff-dependent, and flood-dependent ecosystems. Rainfall events were monitored between 2018 and 2024 using local meteorological stations, while ecosystem productivity was assessed using Sentinel-2-derived vegetation indices. Rainfall pulse characteristics included event intensity, duration, and timing, as well as cumulative and antecedent rainfall conditions. Productivity dynamics were characterized by the onset, peak, and end of the growing season. Machine-learning models were applied to quantify the relative importance of rainfall pulse characteristics in explaining ecosystem productivity across the different hydrological systems. Results: Distinct rainfall pulse regimes controlled ecosystem productivity in the three hydrological systems. In rainfall-dependent systems, productivity increased with greater cumulative rainfall during the previous two years, a later end to the previous rainy season, an earlier onset of the current rainy season, and higher rainfall intensity per event. Runoff-dependent systems responded positively to longer rainfall events and a later end to the rainy season, but negatively to cumulative rainfall during the previous two years. Flood-dependent systems responded positively to longer intervals between rainfall events, higher rainfall intensity, and greater annual rainfall in both the current and preceding years. These findings demonstrate that ecosystem productivity is governed not only by the amount of rainfall received, but also by system-specific rainfall pulse characteristics linked to dominant hydrological processes. Discussion: The contrasting precipitation-productivity relationships reflect differences in hydrological processes and vegetation characteristics among the three ecosystem types. Rainfall-dependent and flood-dependent systems benefit from intense rainfall events and the carry-over of soil water reserves, whereas runoff-dependent systems are favored by prolonged, moderate rainfall events that extend later into the rainy season. The capacity to absorb, retain and utilize soil moisture reserves emerged as a fundamental characteristic of dryland ecosystems. Conclusion: Our results show that rainfall pulse characteristics, rather than rainfall amount alone, are key determinants of ecosystem productivity in drylands, and that the relevant pulse regime depends on the dominant hydrological pathway sustaining each ecosystem. These findings support the Pulse-Reserve Paradigm while providing new insights into the mechanisms linking rainfall variability to ecosystem dynamics. More broadly, this study demonstrates how integrating LTER monitoring, Earth observation, and machine learning provides a transferable framework for analyzing ecosystem responses to changing rainfall regimes, with the potential to strengthen future long-term ecological assessments and adaptive ecosystem management under climate change.