Development of an Image Analysis Pipeline to Estimate Sphagnum Colony Density in the Field.

Abstract

Sphagnum peatmosses play an important part in water table management of many peatland ecosystems. Keeping the ecosystem saturated, they slow the breakdown of organic matter and release of greenhouse gases, facilitating peatland's function as a carbon sink rather than a carbon source. Although peatland monitoring and restoration programs have increased recently, there are few tools to quantify traits that Sphagnum species display in their ecosystems. Colony density is often described as an important determinant in the establishment and performance in Sphagnum but detailed evidence for this is limited. In this study, we describe an image analysis pipeline that accurately annotates Sphagnum capitula and estimates plant density using open access computer vision packages. The pipeline was validated using images of different Sphagnum species growing in different habitats, taken on different days and with different smartphones. The developed pipeline achieves high accuracy scores, and we demonstrate its utility by estimating colony densities in the field and detecting intra and inter-specific colony densities and their relationship with habitat. This tool will enable ecologists and conservationists to rapidly acquire accurate estimates of Sphagnum density in the field without the need of specialised equipment.

Plant phenotyping relevance

Sphagnumの植物密度という明示的な形態・生態形質を、画像解析パイプラインで抽出する手法を開発し、異なる種・生育地・撮影条件で検証しているため、植物フェノタイピング手法が中心です。

Code and data availability

The paper's field Sphagnum images and annotation data are deposited publicly (DOI 10.20391/9ba4df8f-2d28-4bce-b9b8-ec6368b636e0), and the authors' Capitula-Counter image analysis pipeline is freely available on the NPPC GitHub. Both are paper-specific, public, and actionable.