Establishing nutrient thresholds to protect coral reefs is difficult because water quality is dynamic and shifts with many environmental factors. We examined the contribution of natural and human factors in predicting water quality at the base of 34 streams on a high tropical Pacific island. Mixed regression models revealed that rainfall, sea-surface temperature, and windspeed were fixed factors predicting dissolved inorganic nitrogen (DIN) concentrations at the base of all watersheds. In contrast, human influences were captured as random components of variation associated with site-based differences. The novel modeling approach using temporal and spatial data provided daily-loading simulations that were used to evaluate exceedance criteria, defined as the percent of time each watershed exceeded a suite of DIN thresholds. Exceedance criteria were considered alongside biological data to recommend a 0.1 to 0.15 mg l⁻¹ benchmark to protect coastal water quality and coral reefs surrounding Tutuila, American Samoa.

由于水质是动态的并且会随着许多环境因素而变化，因此很难建立营养阈值来保护珊瑚礁。我们检查了自然因素和人为因素在预测这个热带高太平洋岛屿上34条河流的水质基础上的贡献。混合回归模型显示，降雨，海面温度和风速是预测所有流域底部溶解无机氮（DIN）浓度的固定因素。相反，人类影响被捕获为与基于地点的差异相关的变异的随机成分。使用时空数据的新颖建模方法提供了每日负荷模拟，用于评估超标标准，超标标准定义为每个流域超过一套DIN阈值的时间百分比。^-1基准，用于保护美属萨摩亚图图伊拉附近的沿海水质和珊瑚礁。