Coastal wetlands of the Laurentian Great Lakes are diverse and productive ecosystems that provide many ecosystem services, but are threatened by anthropogenic factors, including nutrient input, land-use change, invasive species, and climate change. In this study, we examined one component of wetland ecosystem structure – phytoplankton biomass – using the proxy metric of water column chlorophyll-a measured in 514 coastal wetlands across all five Great Lakes as part of the Great Lakes Coastal Wetland Monitoring Program. Mean chlorophyll-a concentrations increased from north-to-south from Lake Superior to Lake Erie, but concentrations varied among sites within lakes. To predict chlorophyll-a concentrations, we developed two random forest models for each lake – one using variables that may directly relate to phytoplankton biomass (“proximate” variables; e.g., dissolved nutrients, temperature, pH) and another using variables with potentially indirect effects on phytoplankton growth (“distal” variables; e.g., land use, fetch). Proximate and distal variable models explained 16–43% and 19–48% of variation in chlorophyll-a, respectively, with models developed for lakes Erie and Michigan having the highest amount of explanatory power and models developed for lakes Ontario, Superior, and Huron having the lowest. Land-use variables were important distal predictors of chlorophyll-a concentrations across all lakes. We found multiple proximate predictors of chlorophyll-a, but there was little consistency among lakes, suggesting that, while chlorophyll-a may be broadly influenced by distal factors such as land use, individual lakes and wetlands have unique characteristics that affect chlorophyll-a concentrations. Our results highlight the importance of responsible land-use planning and watershed-level management for protecting coastal wetlands.

Laurentian Great Lakes 的沿海湿地是多样化和多产的生态系统，提供许多生态系统服务，但受到人为因素的威胁，包括养分输入、土地利用变化、入侵物种和气候变化。在这项研究中，我们使用作为五大湖沿海湿地监测计划的一部分在所有五个五大湖的 514 个沿海湿地中测量的水柱叶绿素a从苏必利尔湖到伊利湖，平均叶绿素a预测叶绿素-a浓度，我们为每个湖泊开发了两个随机森林模型——一个使用可能与浮游植物生物量直接相关的变量（“近似”变量；例如，溶解的养分、温度、pH），另一个使用对浮游植物生长有潜在间接影响的变量（“远端”变量；例如，土地使用、获取）。近端和远端变量模型分别解释了 16-43% 和 19-48% 的叶绿素a变异，为伊利湖和密歇根湖开发的模型具有最高的解释力，为安大略湖、苏必利尔湖和休伦湖开发的模型有最低的。土地利用变量是所有湖泊叶绿素a浓度的重要远端预测因子。我们发现了多个叶绿素的近似预测因子-a，但湖泊之间几乎没有一致性，这表明虽然叶绿素-a可能受到土地利用等远端因素的广泛影响，但个别湖泊和湿地具有影响叶绿素-a浓度的独特特征。我们的研究结果强调了负责任的土地利用规划和流域级管理对保护沿海湿地的重要性。