Institute of Applied Mathematics, University of British Columbia, Vancouver, V6T 1Z2, Canada.
Department of Biosciences, Wallace Building, Swansea University, Swansea, SA2 8PP, UK.
Sci Rep. 2018 Jan 9;8(1):132. doi: 10.1038/s41598-017-17346-6.
As climate change and other anthropogenic factors increase the uncertainty of vegetation ecosystem persistence, the ability to rapidly assess their dynamics is paramount. Vegetation and sessile communities form a variety of striking regular spatial patterns such as stripes, spots and labyrinths, that have been used as indicators of ecosystem current state, through qualitative analysis of simple models. Here we describe a new method for rigorous quantitative estimation of biological parameters from a single spatial snapshot. We formulate a synthetic likelihood through consideration of the expected change in the correlation structure of the spatial pattern. This then allows Bayesian inference to be performed on the model parameters, which includes providing parameter uncertainty. The method was validated against simulated data and then applied to real data in the form of aerial photographs of seagrass banding. The inferred parameters were found to be able to reproduce similar patterns to those observed and able to detect strength of spatial competition, competition-induced mortality and the local range of reproduction. This technique points to a way of performing rapid inference of spatial competition and ecological stability from a single spatial snapshots of sessile communities.
随着气候变化和其他人为因素增加了植被生态系统持续存在的不确定性，快速评估其动态的能力至关重要。植被和固着生物群落形成了各种引人注目的规则空间模式，如条纹、斑点和迷宫，通过对简单模型的定性分析，这些模式已被用作生态系统当前状态的指标。在这里，我们描述了一种从单个空间快照中进行严格定量估计生物参数的新方法。我们通过考虑空间模式相关结构的预期变化来构建一个综合似然函数。然后，这允许对模型参数进行贝叶斯推断，包括提供参数不确定性。该方法经过模拟数据验证后，以海草草带航空照片的形式应用于实际数据。推断出的参数能够再现与观察到的相似的模式，并能够检测空间竞争的强度、竞争引起的死亡率和局部繁殖范围。这项技术为从固着生物群落的单个空间快照中快速推断空间竞争和生态稳定性指明了一种方法。
