Back

Temporal analysis of reproduction distributed in space illuminates the climate-change resiliency of toyon (Heteromeles arbutifolia)

Dakduk, D.; Yoder, J. B.

2025-09-15 ecology
10.1101/2025.09.09.675207 bioRxiv
Show abstract

PremiseToyon, Heteromeles arbutifolia (Lindl.) M. Roem. (Rosaceae), is an iconic and ecologically important member of California chaparral and oak woodland communities. Toyons habitat faces changing wildfire regimes, widening variation in annual rainfall, and competition by introduced species. We used a new modeling method, temporal analysis of reproduction distributed in space (TARDIS) to examine how recent climate change alters habitat suitability for toyon. MethodsAs data for TARDIS, we annotated flowering and fruiting in images from 4,105 observations of toyon contributed to the iNaturalist crowdsourcing platform. From these records we trained Bayesian additive regression tree models relating weather to toyon flowering. We used a trained model to hindcast flowering each year back to 1900, and examined trends in hindcast flowering. For comparison, we also modeled changing habitat suitability using a conventional species distribution model (SDM) relating toyon presence to 30-year climate averages. Key resultsToyon flowering is associated with greater winter precipitation and warmer fall and winter temperatures. Our hindcast finds mean flowering intensity has been stable to slightly increasing since 1900, with greater increases at higher elevations, but also at lower latitudes. Variation in flowering intensity has also increased, especially at lower latitudes. Trends in flowering are positively correlated with changes in SDM-predicted suit-ability. ConclusionsTARDIS recovers biologically realistic predictors of toyon flowering, and hindcast changes in flowering intensity indicate the species range remains suitable after 125 years of changing climate. Overall our results indicate toyon populations remain healthy, but may have limited opportunity to migrate northwards as climate change continues.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.