Research
Research themes
The ecological footprint of humans is increasingly apparent: 75% of the land is significantly altered, local crop varieties are disappearing worldwide, species’ fitness and interactions are under pressure to adapt to rapid climate change, among other marked impacts. Through my research, I hope to contribute to understanding and addressing this ecological footprint, by (1) designing practical solutions to reconcile land use incentives and conservation goals, (2) monitoring the sustainability of different biodiversity outcomes and ecosystem functions at large scales, and (3) estimating the causal impact of policy choices on such outcomes. I combine approaches, data, and knowledge from economics and ecology primarily, but also draw from quantitative tools like remote sensing and machine learning, and qualitative tools like interviews. My current research areas include:
Developing solutions to reconcile urgent land conservation goals in agricultural regions with farmer and agricultural stakeholder incentives
Canada is working on scaling up its land conservation efforts. This creates opportunities in agricultural lands, where biodiversity is both at risk and the pillar of critical ecosystem services. I am working with Claire Kremen, Hannah Wittman, and Carolyn Callaghan (CWF), on assessing the biodiversity impacts and economic feasibility of a land restoration strategy on Canadian agricultural lands. This work will be developed in collaboration with local conservation and agriculture organizations, to maximize its relevance and impact.
Measuring and studying species and ecosystems at large scales
In my thesis work, I developed cross-disciplinary approaches to document tree physiology and plant flowering phenology at large scales. The broader goal of this work is to document tree growth resilience to high temperatures, and climate-induced changes in plant-pollinator synchronization.
Establishing causality when ecological outcomes are involved is challenging, in part because data on those outcomes does not necessarily match in space or in time with social outcomes, or with the natural experiments that can be used to obtain causal identification. I am exploring how mixed data sources (herbarium records, citizen science data, disparate monitoring networks, remote sensing) can be used to create panel data on biodiversity. In particular, I am interested in how the data creation process, which likely induces some measurement error, might impact the validity of subsequent causal estimates.
Estimating the causal impact of land use choices and conservation policies on biodiversity
I am working on mapping crop diversity and hedgerows at regional scales. This is part of a broader goal to assess the causal impact of these farming strategies on pollinators, insects, and birds. More broadly, I currently collaborate on projects establishing causal links between social and ecological outcomes: one on the social cost of locust outbreaks, and another on the efficiency of conservation funds for birds.
Publications
Soroco, M., Hempel, J., Xiong, X., Lécuyer, M., & Gantois J. (2023). Flowering Onset Detection: Traditional Learning vs. Deep Learning Performance in a Sparse Label Context. Tackling Climate Change with Machine Learning: workshop at NeurIPS.
Abstract: Detecting temporal shifts in plant flowering times is of increasing importance in a context of climate change, with applications in plant ecology, but also health, agriculture, and ecosystem management. However, scaling up plant-level monitoring is cost prohibitive, and flowering transitions are complex and difficult to model. We develop two sets of approaches to detect the onset of flowering at large-scale and high-resolution. Using fine grain temperature data with domain knowledge based features and traditional machine learning models provides the best performance. Using satellite data, with deep learning to deal with high dimensionality and transfer learning to overcome ground truth label sparsity, is a challenging but promising approach, as it reaches good performance with more systematically available data. Link to paper.
Gantois, J. (2022). New tree‐level temperature response curves document sensitivity of tree growth to high temperatures across a US‐wide climatic gradient. Global Change Biology, 28(20), 6002-6020.
Graphical abstract: In many ecoregions of the United States, radial tree growth responds non-linearly to temperature. In particular, temperatures above a certain level have a strong negative impact on total annual growth. This breakpoint level is identified in the 1°-40°C range based on model predictive performance. The annual and spring-summer (AMJJA) temperature responses are similar. Excluding drought effects shifts the temperature response up, especially in dry ecoregions, but the non-linearity remains. This flexible temperature model can be used to document the impact of sub-lethal high temperatures on tree growth across large spatiotemporal scales.
Working papers
The Value of Monitoring for Disaster Prevention: The Desert Locust. (with Anouch Missirian, Evelina Linnros, Anna Tompsett, Amir Jina, Gordon C. McCord, Eyal G. Frank)
Monitoring shifts in flowering phenology using satellite imagery and deep learning. (with Mathias Lécuyer)
Shifts in the timing of plant flowering are a key signal of an ecosystem’s response to environmental change. Consequences are far-reaching: the timing of plant flowering affects the synchronization between plants and pollinators; it influences the exposure of crops to weather extremes during flowering, the most sensitive stage of development, with implications for later yield. A consistent and large-scale measure tracking the timing of flowering across years would be of great use for assessing risks of plant-pollinator desynchronization or risks of subpar crop yields. Current efforts, ranging from local to regional scales, do not quite achieve this goal: ground observations, while expanding, remain limited in their spatial and temporal coverage; process-based or statistical models are not flexible enough to capture local acclimation or adaptation to environmental factors; and remotely sensed estimates of plant development stages are often disconnected from the ground. We propose a method that combines spectral reflectance data from satellite imagery, with a deep learning model, to create a large-scale, high-resolution proxy for the timing of flowering. We illustrate this approach with two use cases. First, over cropland in Illinois and Iowa, we construct a proxy measure of the evolution of crop flowering—silking for corn, flowering for soybean—across the growing season. Second, over the Eastern US, we construct a proxy measure of the onset of spring flowering, which reflects the phenology of early-flowering, temperature-sensitive species.
