Improving Landsat predictions of rangeland fractional cover with multitask learning and uncertainty

Allred, B. W.; Bestelmeyer, B. T.; Boyd, C. S.; Brown, C.; Davies, K. W.; Ellsworth, L. M.; Erickson, T. A.; Fuhlendorf, S. D.; Griffiths, T. V.; Jansen, V.; Jones, M. O.; Karl, J.; Maestas, J. D.; Maynard, J. J.; McCord, S. E.; Naugle, D. E.; Starns, H. D.; Twidwell, D.; Uden, D. R.

doi:10.1101/2020.06.10.142489

bioRxiv · 10.1101/2020.06.10.142489

Improving Landsat predictions of rangeland fractional cover with multitask learning and uncertainty

Allred, B. W.·Bestelmeyer, B. T.·Boyd, C. S.·Brown, C.·Davies, K. W.·Ellsworth, L. M.·Erickson, T. A.·Fuhlendorf, S. D.·Griffiths, T. V.·Jansen, V.·Jones, M. O.·Karl, J.·Maestas, J. D.·Maynard, J. J.·McCord, S. E.·Naugle, D. E.·Starns, H. D.·Twidwell, D.·Uden, D. R.

View original source ↗Export metadata ↓

Abstract

O_LIOperational satellite remote sensing products are transforming rangeland management and science. Advancements in computation, data storage, and processing have removed barriers that previously blocked or hindered the development and use of remote sensing products. When combined with local data and knowledge, remote sensing products can inform decision making at multiple scales. C_LIO_LIWe used temporal convolutional networks to produce a fractional cover product that spans western United States rangelands. We trained the model with 52,012 on-the-ground vegetation plots to simultaneously predict fractional cover for annual forbs and grasses, perennial forbs and grasses, shrubs, trees, litter, and bare ground. To assist interpretation and to provide a measure of prediction confidence, we also produced spatiotemporal-explicit, pixel-level estimates of uncertainty. We evaluated the model with 5,780 on-the-ground vegetation plots removed from the training data. C_LIO_LIModel evaluation averaged 6.3% mean absolute error and 9.6% root mean squared error. Evaluation with additional datasets that were not part of the training dataset, and that varied in geographic range, method of collection, scope, and size, revealed similar metrics. Model performance increased across all functional groups compared to the previously produced fractional product. C_LIO_LIThe advancements achieved with the new rangeland fractional cover product expand the management toolbox with improved predictions of fractional cover and pixel-level uncertainty. The new product is available on the Rangeland Analysis Platform (https://rangelands.app/), an interactive web application that tracks rangeland vegetation through time. This product is intended to be used alongside local on-the-ground data, expert knowledge, land use history, scientific literature, and other sources of information when making interpretations. When being used to inform decision-making, remotely sensed products should be evaluated and utilized according to the context of the decision and not be used in isolation. C_LI

Source connections

Published article ↗

Explore related subjects

ecology Preprint

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Allred, B. W., Bestelmeyer, B. T., Boyd, C. S., Brown, C., Davies, K. W., Ellsworth, L. M., Erickson, T. A., Fuhlendorf, S. D., Griffiths, T. V., Jansen, V., Jones, M. O., Karl, J., Maestas, J. D., Maynard, J. J., McCord, S. E., Naugle, D. E., Starns, H. D., Twidwell, D., Uden, D. R.. 2020-06-12. Improving Landsat predictions of rangeland fractional cover with multitask learning and uncertainty. https://doi.org/10.1101/2020.06.10.142489

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Invasion analysis for structured populations: analytical principles and numerical workflows

All real populations show some form of structure--differences among individuals, such as life stage or physiological condition, that shape a species' ability to persist. Yet it remains challenging to incorporate population structure into models of species interactions. We provide a general framework for studying competition in structured populations by quantifying species' invasion growth rates. The crux of our approach is separation of timescales: the initial dynamics of a structured population depend on how exactly it is introduced, but these transient differences do not affect long-term invasion success. Depending on the research question, the relevant long-term growth rate when rare may equivalently be derived using analytical separation of timescales, matrix population theory, or numerical simulations. Using case studies spanning developmental stage, environmental state, and infection status, we illustrate how our methodology derives simple, ecologically meaningful invasion conditions for a variety of systems. Finally, we extend the approach to fluctuating resident dynamics, showing how to use simulations or a general mathematical framework known as Floquet theory to correctly quantify invasion growth. By unifying the quantitative methods of population biology and community ecology, our framework will allow ecologists to study species persistence and coexistence across a greater variety of demographically complex systems.

ecology↗

bioRxiv2026

Interspecific social contacts spread gut microbiota in wild rodent communities

Background: The mammalian gut microbiota influences a wide array of host phenotypes and varies substantially among individuals. Among conspecifics, microbiota composition is strongly shaped by microbial transmission through social interactions, yet the extent to which such transmission also occurs across species boundaries remains poorly understood. Here, we combine high-resolution tracking data collected over more than 10 months with gut microbiota profiling to examine social associations, space use, and microbiota similarity among individuals of three sympatric rodent species - wood mice (Apodemus sylvaticus), yellow-necked mice (A. flavicollis) and bank voles (Myodes glareolus), in two woodlands. By integrating multi-species social networks and spatial overlap with individual microbiota data, we test how both conspecific and interspecific interactions shape gut microbiota composition. Results: At both study sites, social associations strongly predicted microbiota similarity among conspecific individuals, whereas spatial proximity consistently predicted microbiota similarity among interspecific pairs. In addition, social associations between mice of different species predicted microbiota similarity, although this effect was weaker than that observed within species and was asymmetric, with evidence for stronger microbial transmission from wood mice to yellow-necked mice than vice versa. By focusing on species-indicative bacterial taxa, we further show that wood mice with more social connections to yellow-necked mice harboured a greater richness of gut bacteria typical of yellow-necked mice. Conclusions: Together, these results demonstrate that individual-level social behaviour can influence gut microbial transmission across species boundaries, particularly among closely related species with overlapping ecological niches. Interspecific social interactions may therefore represent an underappreciated pathway for the acquisition of novel gut symbionts in multi-species communities.

ecology↗

bioRxiv2026

Energetic and functional development of soil food webs along secondary succession

1. As ecosystems develop, aboveground biomass and belowground organic matter accumulate, while energy turnover declines. This trend is largely supported by vegetation studies, but remains unexplored for soil food webs that channel the majority of energy in terrestrial ecosystems. 2. Here, we tested whether successional theory applies to soil food webs across size-based compartments, ranging from microfauna (nematodes) to macrofauna (soil arthropods and earthworms), along a secondary succession in southwestern China, using an energy-flux approach. To assess whether forest restoration by planting follows natural succession, we also compared spruce and larch plantations with secondary birch forest of the same successional stage. 3. Supporting expectations of reduced energy turnover, the results showed an increasing animal biomass-to-energy flux ratio and a functional shift from herbivory-dominated (green) to detritivory-dominated (brown) energy channelling along succession. Soil food web compartments showed contrasting trends: macro-food webs shifted towards bacterial energy channels with decreasing predation, while micro-food webs maintained similar bacterial/fungal energy flux ratios and tended towards increasing predation. The larch plantation supported lower animal biomass than the secondary birch forest, whereas the spruce plantation did not differ. We further found that energy flux correlated positively with microbial biomass and soil animal diversity, but negatively with litter biomass during succession. 4. Overall, we expand successional theory by revealing the transformations in energy use and pathways within soil food webs along secondary succession, thereby advancing understanding of ecosystem functioning and guiding ecosystem restoration. Our results further suggest that tree plantations do not necessarily replicate the belowground development of natural succession, which should be considered in ecosystem restoration.

ecology↗