Machine Learning Undercounts Reproductive Organs on Herbarium Specimens but Accurately Derives Their Quantitative Phenological Status: A Case Study of Streptanthus tortuosus.

Natalie L R Love,Susan J Mazer,Pierre Bonnet,Alexis Joly,Hervé Goëau

doi:10.3390/plants10112471

Abstract

Machine learning (ML) can accelerate the extraction of phenological data from herbarium specimens; however, no studies have assessed whether ML-derived phenological data can be used reliably to evaluate ecological patterns. In this study, 709 herbarium specimens representing a widespread annual herb, Streptanthus tortuosus, were scored both manually by human observers and by a mask R-CNN object detection model to (1) evaluate the concordance between ML and manually-derived phenological data and (2) determine whether ML-derived data can be used to reliably assess phenological patterns. The ML model generally underestimated the number of reproductive structures present on each specimen; however, when these counts were used to provide a quantitative estimate of the phenological stage of plants on a given sheet (i.e., the phenological index or PI), the ML and manually-derived PI’s were highly concordant. Moreover, herbarium specimen age had no effect on the estimated PI of a given sheet. Finally, including ML-derived PIs as predictor variables in phenological models produced estimates of the phenological sensitivity of this species to climate, temporal shifts in flowering time, and the rate of phenological progression that are indistinguishable from those produced by models based on data provided by human observers. This study demonstrates that phenological data extracted using machine learning can be used reliably to estimate the phenological stage of herbarium specimens and to detect phenological patterns.

Highlights

Within and among plant species, the study of phenological traits such as the timing of bud break or flowering can provide key insights into how species respond to climate [1]
The Machine learning (ML) model generally underestimated the number of reproductive structures present on each specimen; when these counts were used to provide a quantitative estimate of the phenological stage of plants on a given sheet, the ML and manually-derived phenological index (PI)’s were highly concordant
The spring mean maximum daily temperature and cumulative winter precipitation experienced at each collection site during the year of specimen collection ranged from 1.4–24.6 ◦C (SD = 4.8 ◦C) and 88–2167 mm (SD = 357.8 mm), respectively

Summary

Introduction

Within and among plant species, the study of phenological traits such as the timing of bud break or flowering can provide key insights into how species respond to climate [1]. While previous studies have documented a wide range of phenological responses to climate and climate change [6,7], our understanding of these responses in many taxa and ecosystems remains incomplete This gap limits our ability to make broad scale predictions of the ecosystem-wide impacts from climate change [8,9]. Natural history collections, such as herbarium specimens, provide a long temporal record of phenology for hundreds of thousands of taxa globally and offer a data-rich resource with which to fill this gap [6]. With large-scale efforts to digitize and image herbarium records, millions of herbarium records are available via large data aggregators, e.g., GBIF (https://www.gbif.org/ (accessed on 1 October 2020))and iDigBio (https://www.idigbio.org/ (accessed on 1 October 2020)), to advance phenological research

Objectives

Methods

Results