Studying long-term, large-scale grassland restoration outcomes to improve seeding methods and reveal knowledge gaps

Studies are increasingly investigating effects of large-scale management activities on grassland restoration outcomes. These studies are providing useful comparisons among currently used management strategies, but not the novel strategies needed to rapidly improve restoration efforts. Here we illustrate how managing restoration projects adaptively can allow promising management innovations to be identified and tested.
We studied 327 Great Plains fields seeded after coal mining. We modelled plant responses to management strategies to identify the most effective previously used strategies for constraining weeds and establishing desired plants. Then, we used the model to predict responses to new strategies our analysis identified as potentially more effective.
Where established, the weed crested wheatgrass (Agropyron cristatum L.) increased through time, indicating a need to manage establishment of this grass. Seeding particular grasses reduced annual weed cover, and because these grasses appeared to become similarly abundant whether sown at low or high rates, low rates could likely be safely used to reduce seeding costs. More importantly, lower than average grass seed rates increased cover of shrubs, the plants most difficult to restore to many grassland ecosystems. After identifying grass seed rates as a driver, we formulated model predictions for rates below the range managers typically use. These predictions require testing but indicated atypically low grass seed rates would further increase shrubs without hindering long-term grass stand development.
Synthesis and applications. Designing management around empirically based predictions is a logical next step towards improving ecological restoration efforts. Our predictions are that reducing grass seed rates to atypically low levels will boost shrubs without compromising grasses. Because these predictions derive from the fitted model, they represent quantitative hypotheses based on current understanding of the system. Generating data needed to test and update these hypotheses will require monitoring responses to shifts in management, specifically shifts to lower grass seed rates. A paucity of data for confronting hypotheses has been a major sticking point hindering adaptive management of most natural resources, but this need not be the case with degraded grasslands, because ongoing restoration efforts around the globe are providing continuous opportunities to monitor and manage processes regulating grassland restoration outcomes.

越来越多的研究正聚焦于大规模管理活动对草地恢复成效的影响。此类研究虽可为当前主流的各类管理策略提供有效的对比参照，却未能涵盖可快速优化恢复工作的创新型管理方案。在此，我们阐释了如何通过适应性管理（adaptive management）来识别并验证具有应用前景的管理创新举措。

本研究针对北美大平原（Great Plains）327处经煤矿开采后进行播种复垦的地块展开。我们构建了植物对管理策略响应的模型，以筛选出在抑制杂草定植与目标植物建植方面效果最优的现有管理策略。随后，我们利用该模型对本研究分析认定的潜在更高效的新型管理策略的响应效果进行预测。

定植后的杂草冰草（Agropyron cristatum L.）覆盖度随时间推移逐渐升高，这表明需对该禾草的定植过程进行管控。播种特定禾草可降低一年生杂草的盖度；且无论播种量高低，这类禾草的最终丰度均无显著差异，因此可采用低播种量以缩减播种成本。更为关键的是，低于平均水平的禾草播种量可提升灌木的盖度——而灌木是多数草地生态系统恢复中最难建植的植物类群。在明确禾草播种量为关键驱动因子后，我们针对管理者常规使用范围以下的播种量构建了模型预测结果。此类预测虽有待验证，但结果显示：异常低的禾草播种量可进一步提升灌木盖度，且不会阻碍长期禾草群落的构建与发展。

综合与应用。基于实证预测设计管理方案，是优化生态恢复工作的合理进阶路径。我们的预测结果为：将禾草播种量降至异常低的水平，可在不损害禾草群落的前提下提升灌木盖度。由于这些预测源自拟合后的模型，它们代表了基于当前系统认知的量化假说。为验证并更新这些假说所需的数据，需通过监测管理调整后的响应来获取，具体而言，即调整至更低的禾草播种量。缺乏用于验证假说的数据，长期以来是阻碍多数自然资源适应性管理的核心瓶颈，但退化草地恢复并非如此——全球范围内持续开展的恢复工作，为监测和调控草地恢复进程提供了源源不断的契机。