skip to main content


Title: Applying machine learning to investigate long‐term insect–plant interactions preserved on digitized herbarium specimens
Premise

Despite the economic significance of insect damage to plants (i.e., herbivory), long‐term data documenting changes in herbivory are limited. Millions of pressed plant specimens are now available online and can be used to collect big data on plant–insect interactions during the Anthropocene.

Methods

We initiated development of machine learning methods to automate extraction of herbivory data from herbarium specimens by training an insect damage detector and a damage type classifier on two distantly related plant species (Quercus bicolorandOnoclea sensibilis). We experimented with (1) classifying six types of herbivory and two control categories of undamaged leaf, and (2) detecting two of the damage categories for which several hundred annotations were available.

Results

Damage detection results were mixed, with a mean average precision of 45% in the simultaneous detection and classification of two types of damage. However, damage classification on hand‐drawn boxes identified the correct type of herbivory 81.5% of the time in eight categories. The damage classifier was accurate for categories with 100 or more test samples.

Discussion

These tools are a promising first step for the automation of herbivory data collection. We describe ongoing efforts to increase the accuracy of these models, allowing researchers to extract similar data and apply them to biological hypotheses.

 
more » « less
Award ID(s):
1909821
NSF-PAR ID:
10456798
Author(s) / Creator(s):
 ;  ;  ;  
Publisher / Repository:
Wiley Blackwell (John Wiley & Sons)
Date Published:
Journal Name:
Applications in Plant Sciences
Volume:
8
Issue:
6
ISSN:
2168-0450
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Premise

    Herbarium specimens represent an outstanding source of material with which to study plant phenological changes in response to climate change. The fine‐scale phenological annotation of such specimens is nevertheless highly time consuming and requires substantial human investment and expertise, which are difficult to rapidly mobilize.

    Methods

    We trained and evaluated new deep learning models to automate the detection, segmentation, and classification of four reproductive structures ofStreptanthus tortuosus(flower buds, flowers, immature fruits, and mature fruits). We used a training data set of 21 digitized herbarium sheets for which the position and outlines of 1036 reproductive structures were annotated manually. We adjusted the hyperparameters of amask R‐CNN(regional convolutional neural network) to this specific task and evaluated the resulting trained models for their ability to count reproductive structures and estimate their size.

    Results

    The main outcome of our study is that the performance of detection and segmentation can vary significantly with: (i) the type of annotations used for training, (ii) the type of reproductive structures, and (iii) the size of the reproductive structures. In the case ofStreptanthus tortuosus, the method can provide quite accurate estimates (77.9% of cases) of the number of reproductive structures, which is better estimated for flowers than for immature fruits and buds. The size estimation results are also encouraging, showing a difference of only a few millimeters between the predicted and actual sizes of buds and flowers.

    Discussion

    This method has great potential for automating the analysis of reproductive structures in high‐resolution images of herbarium sheets. Deeper investigations regarding the taxonomic scalability of this approach and its potential improvement will be conducted in future work.

     
    more » « less
  2. PREMISE Quantifying how closely related plant species differ in susceptibility to insect herbivory is important for our understanding of variation in plant-insect ecological interactions and evolutionary pressures on plant functional traits. However, empirically measuring in situ variation in herbivory over the entire geographic range where a plant-insect complex occurs is logistically difficult. Recently, new methods have been developed to use herbarium specimens to investigate patterns in plant-insect interactions across geographic areas, and during periods of accelerating anthropogenic change. Such investigations can provide insights into changes in herbivory intensity and phenology in plants that are of ecological and agricultural importance. METHODS Here, we analyze 274 pressed herbarium samples from all 14 species in the economically important plant genus Cucurbita (Cucurbitaceae) to investigate variation in herbivory damage. This collection is comprised of specimens of wild, undomesticated Cucurbita that were collected from across their native range in the Neotropics and subtropics, and Cucurbita cultivars that were collected from both within their native range and from locations where they have been introduced for agriculture in temperate Eastern North America. RESULTS We find that herbivory is common on individuals of all Cucurbita species collected from throughout their geographic ranges; however, estimates of herbivory varied considerably among individuals, with greater damage observed in specimens collected from unmanaged habitat. We also find evidence that mesophytic species accrue more insect damage than xerophytic species. CONCLUSIONS Our study demonstrates that herbarium specimens are a useful resource for understanding ecological interactions between domesticated crop plants and co-evolved insect herbivores. 
    more » « less
  3. PREMISE Quantifying how closely related plant species differ in susceptibility to insect herbivory is important for our understanding of variation in plant-insect ecological interactions and evolutionary pressures on plant functional traits. However, empirically measuring in situ variation in herbivory over the entire geographic range where a plant-insect complex occurs is logistically difficult. Recently, new methods have been developed to use herbarium specimens to investigate patterns in plant-insect interactions across geographic areas, and during periods of accelerating anthropogenic change. Such investigations can provide insights into changes in herbivory intensity and phenology in plants that are of ecological and agricultural importance. METHODS Here, we analyze 274 pressed herbarium samples from all 14 species in the economically important plant genus Cucurbita (Cucurbitaceae) to investigate variation in herbivory damage. This collection is comprised of specimens of wild, undomesticated Cucurbita that were collected from across their native range in the Neotropics and subtropics, and Cucurbita cultivars that were collected from both within their native range and from locations where they have been introduced for agriculture in temperate Eastern North America. RESULTS We find that herbivory is common on individuals of all Cucurbita species collected from throughout their geographic ranges; however, estimates of herbivory varied considerably among individuals, with greater damage observed in specimens collected from unmanaged habitat. We also find evidence that mesophytic species accrue more insect damage than xerophytic species. CONCLUSIONS Our study demonstrates that herbarium specimens are a useful resource for understanding ecological interactions between domesticated crop plants and co-evolved insect herbivores. 
    more » « less
  4. Abstract Question

    It has been established that community biodiversity has consequences for ecosystem function. Yet research assessing these biodiversity–ecosystem function (BEF) relationships usually occurs at only one phylogenetic scale; as such, the dependence ofBEFrelationships on phylogenetic scale has not been characterized. We present a novel framework for considering the consequences of biodiversity across phylogenetic scales, allowing us to ask: Do the consequences of intraspecific and interspecific diversity affect the growth, survival, and leaf herbivory of three temperate tree species?

    Study site

    Salicaceous tree plantation, Minnesota, northern USA.

    Methods

    We established an experimental plantation consisting of trees of three species within the willow (Salicaceae) family. Two aspen (Populus tremuloides,P. alba) and one willow (Salix nigra) species were represented by three unique genotypes such that tree neighborhoods varied both in genotype richness (intraspecific diversity) and species richness (interspecific diversity). We assessed the consequences of tree identity and diversity across these two phylogenetic scales for all trees’ aboveground productivity and survival, and for herbivore damage (onP. tremuloides) at the end of the second full growing season of the experiment.

    Results

    Diversity at any phylogenetic scale had no effect on the growth and survival ofP. albaorS. nigra. However, intraspecific diversity increased the likelihood ofP. tremuloidessurvival while interspecific diversity reducedP. tremuloidessurvival. Intraspecific diversity also reduced leaf removal and galling herbivory onP. tremuloides, while interspecific diversity had no effect on leaf removal and increased galling herbivory. Neither scale of diversity affected leaf mining.

    Conclusions

    Tree diversity within and among populations and species affected plant performance and ecosystem properties differentially, demonstrating thatBEFrelationships shift across phylogenetic scales in a taxon‐specific manner. We call for further experiments that explicitly span these scales by measuring ecosystem and physiological responses to the manipulation of diversity within and among species.

     
    more » « less
  5. 1. Predicting how ecological interactions will respond to global change is a major challenge. Plants and their associated insect herbivores compose much of macroscopic diversity, yet how their interactions have been altered by recent environmental change remains underexplored. 2. To address this gap, we quantified herbivory on herbarium specimens of four plant species with records extending back 112 years. Our study focused on the northeastern US, where temperatures have increased rapidly over the last few decades. This region also represents a range of urban development, a form of global change that has shown variable effects on herbivores in the past studies. 3. Herbarium specimens collected in the early 2000s were 23% more likely to be damaged by herbivores than those collected in the early 1900s. Herbivory was greater following warmer winters and at low latitudes, suggesting that climate warming may drive increasing insect damage over time. In contrast, human population densities were negatively associated with herbivore damage. 4. To explore whether changes in insect occurrence or abundance might explain shifts in herbivory, we used insect observational records to build climate occupancy models for lepidopteran herbivores (butterflies and moths) of our focal plant species. 5. These models show that higher winter temperatures were associated with higher probability of insect herbivore presence, while urbanization was associated with reduced probability of herbivore presence, supporting a link between insect herbivore occurrence and herbivory mediated through environment. 6. Synthesis. Using a temporal record of plant herbivory that spans over a century, we show that both temperature and urbanization influence insect damage to plants, but in very different ways. Our results indicate that damage to plants by insect herbivores will likely continue to increase through time in the northeastern US as global temperatures rise, but that urbanization may disrupt local effects of winter warming on herbivory by excluding certain herbivores. These changes may scale to shape ecosystem processes that are driven by herbivory, including plant productivity. 
    more » « less