New publication: An “unprecedented” map of Joshua tree populations

The satellite images provided in Google Maps have sufficient resolution to pick out Joshua trees, especially if the image captures the shadows cast by the trees’ distinctive branching. Esque et al. (2023) used Google Maps imagery to map the distribution of Joshua trees in exceptionally fine detail. Left panel: Google Maps image near Mojave, California, with large Joshua trees indicated by arrows. Right panel: Joshua trees at the same site in spring 2023 (jby).

One of the biggest challenges for studying biodiversity is answering a seemingly simple question, where does this species live? If we know where a species occurs, we can describe the habitat that it needs, assess how large and stable its populations are, and make informed predictions about what will happen to those populations if we develop its habitat, or as climate change alters the conditions in that habitat. But even for a species as distinctive and visible as Joshua trees — Yucca brevifolia and Y. jaegeriana, which are the biggest plants in most of the landscapes where they grow — our records may be surprisingly spotty.

In the case of Joshua trees, we’ve long relied on very general range maps, especially this one, which was published by the US Geological Survey in 2003, and derived from a hand-drawn map dating back to the 1970s. It covers major regions of the trees’ habitat, but the resolution is crude compared to the environmental data we have for this region, and in places it’s straight-up incorrect. We can use partial data like that map and higher-density point observations of the trees on the landscape to extrapolate their complete range using species distribution models — but these are still not complete data. As the Joshua Tree Genome Project collaborators noted in our recent review of conservation issues across the Mojave Desert, poor knowledge of the trees’ distribution limits our ability to plan for their protection as climate change reshapes the Mojave, and as we build out renewable wind and solar energy capacity in the region.

Range map of Joshua trees published by Cole et al. (2003), via Data Basin.

Joshua Tree Genome Project collaborator Todd Esque, an ecologist at the USGS Western Ecological Research Center, realized a few years ago that there might be a way to map Joshua trees’ distribution right under our noses: in the satellite images that Google Maps provides as a base layer for navigation. Because Joshua trees are so much bigger than most other Mojave Desert plants — and because their outreaching, twisty branches cast distinctive shadows, as you can see in the image at the top of this post — it’s possible to spot individual trees in a Google Maps satellite image.

Todd and his research team divided the Mojave into hundreds of thousands of grid cells 500 meters × 500 meters and systematically searched for Joshua trees in the Google Maps imagery in each cell, validating their findings with on-the-ground surveys in more than 29,000 cells. This produced an empirical map of Joshua trees’ range for most of the Mojave, at the resolution of the quarter-square-kilometer (less than 62-acre) grid cells. For a big region in southern Nevada, though, Google Maps satellite imagery is obfuscated — because this region contains Nellis Air Force Base and the Nevada Nuclear Test Site. To cope with that, Todd’s team used the empirical observations from more than 600,000 grid cells to build a species distribution model that let them interpolate Joshua trees’ distribution within the region where they couldn’t obtain good satellite imagery.

An ultra-fine-scale map of Joshua tree habitat, predicted by a species distribution model using records of the trees’ presence and absence in Google Maps imagery across 672,043 grid cells. (Figure 4 from Esque et al. 2023)

The resulting range map (above) is an unprecedented resolution of Joshua trees’ geographic range — there are, I think, few species as widespread as the eastern and western Joshua trees whose ranges are known to this degree of precision. It aligns pretty well with our prior understanding of where Joshua trees grow, but it points up populations in remote regions of the Mojave that haven’t been well documented because they’re so hard to get to. Todd and the USGS team also applied the data to compare the habitats of the two Joshua tree species, and identified quite a bit of overlap, but also substantial differences.

I was very happy to be a small part of putting it all together for a peer-reviewed article, and I’m extremely excited for the many other ways we’ll be able to use this data. For more details, take a look over the full paper, published Open Access in Frontiers in Evolution and Ecology.