New publication: How more than a century of climate change affected Joshua tree flowering

More than a century ago, in mid-June 1924, the Los Angeles Times devoted a photo spread and several column inches to the second-hand vacation story of some “Cadillac tourists” who saw Joshua trees bearing fruit. It meant, the Times reported, that rain was coming.

M.C. Ellison of Sacramento, who has been touring Southern California in his Cadillac with his family, told the prophecy to J.E. Clark, sales manager for Don Lee. He did not guarantee its accuracy, but admitted that it sounded interesting at least.

“We met an old prospector who was camping among the Joshua-tree forest just beyond the summit of the Cajon [Pass],” Ellison told Clark. “He called our attention to the extremely heavy crop of seeds which the Joshua trees are bearing this year. Almost every tree is loaded with green seed pods.

“‘That means good rains on the desert next season,’ the prospector told us. ‘I have been traveling the Mojave for almost forty-five years now and I have never seen it fail.'”

Joshua trees’ irregular flowering has been a feature of their public image since long before the advent of automobile road-trips. Joshua tree flowers were not observed until some years after the trees were given the formal scientific name Yucca brevifolia in 1871. The same stand of Joshua trees rarely flowers two years in row, and often goes multiple years between big blooms. When the blooms do come, it’s not clear what triggered them. Sometimes Joshua trees flower after a winter of good soaking rains. Sometimes they flower after a run of really dry years. Some people think they need a hard winter frost. Joshua trees will bloom at one site, and a few kilometers away they won’t produce a single bud. Back in 2018, there was a big bloom in Joshua Tree National Park — but only in Joshua Tree National Park — in November, seven or eight months after flowering usually finishes. The LA Times eventually gave up on the idea that the trees forecast the rainfall for the coming year, but newspapers in and around the Mojave Desert continue to report on big bloom years — and years with no flowers to be seen — as a feature of interest for locals and tourists to this day.

That public interest has given us a way to finally figure out what cues Joshua trees to flower. Making that connection also gives us a new way to study how the trees have experienced climate change trends going back to the early 20th Century, decades before M.C. Ellison and his family left Sacramento to tour the state in their Cadillac. It’s all in a new paper that’s just been published on the website of the journal Ecology Letters, by a team of Joshua Tree Genome Project collaborators.

The new paper takes advantage of the fact that Joshua tree flowers are unmistakable, borne in big clusters at the end of branches. The fruits are just as visible, and fruits or fragments of fruits usually remain on the tree into the next year. That means that you can make a pretty reliable guess as to whether a Joshua tree has flowered in the last year, just from an image of the tree. Given enough images of Joshua trees in different dates and locations, we can start to determine how weather conditions in places and times where the trees flower differ from places and times where they didn’t flower.

We found a terrific collection of useful images on the iNaturalist platform, where volunteer contributors post smartphone images of species for identification and future study. An undergraduate researcher in the lab, Ana Karina Andrade — who’s now a coauthor on the new paper — undertook the task of poring through thousands of iNaturalist records to annotate whether the trees in the attached images showed signs of budding flowers, open flowers, or fruit. Her work generated a data set of more than 10,000 records, distributed over 15 years from 2008 (the year iNaturalist launched) to 2022.

We linked those records to annual weather data from the PRISM database, and then used a machine-learning method called BARTs — Bayesian additive regression trees — to identify relationships between weather conditions and whether or not the iNaturalist records showed Joshua trees had flowered in a given year. Our final “trained” BART model found that the trees were more likely to flower in a year with more precipitation than the previous year (that is, responding to the increased water supply), in years that had lower drought stress, and in years that followed milder winters.

That trained model is the best description of what cues Joshua tree flowering that we’ve been able to put together. The coolest thing about the model, though, is we could use it to predict whether Joshua trees were likely to flower in years and locations for which we have weather data but no direct observations of what the trees were doing. The PRISM database goes back to the late 19th Century — so we reconstructed Joshua tree flowering conditions for every year back to 1900. If that seems to you like a big extrapolation, well, you’re on the same page as the peer reviewers of our paper. We’d already assembled some records of Joshua tree flowering activity to validate the model predictions, mostly from our own field notes. At the reviewers’ prompting we went back and added information from digitized herbarium specimens — and from newspaper accounts of Joshua tree flowering events.

Across all those records, our model predicts a higher probability of flowering in years and places where field notes and herbarium specimens record flowering. The model also successfully predicts higher probability of flowering in years when regional newspapers reported big Joshua tree blooms. That includes the 1924 report of Joshua trees’ bumper crop of fruit in the Cajon Pass reported by the “Cadillac tourists” and their friendly desert prospector.

With that validation, we have some confidence that the model’s reconstruction reflects Joshua trees’ response to the weather they encountered over the course of the 20th Century. The Mojave has experienced measurable change in average climate over this period, to the point that we suspect Joshua tree populations are already suffering from higher temperatures and less reliable rainfall. This is a big part of the reason the State of California passed a law last year specifically to protect the western Joshua tree species. To see how climate change might have impacted Joshua tree flowering, we summarized the predictions of our trained model of Joshua tree flowering as the predicted number of years with conditions favoring flowering in the first three decades of the 20th Century (1900 to 1929) and in the most recent three-decade period (1990-2019) and then compared the two periods. The result was a big unexpected: Conditions favoring flowering were slightly more frequent in the recent period than they were in the early period. Across the whole range of the two Joshua tree species, the model predicts an average of 1.6 more flowering years in 1990-2019 compared to 1900-1929.

There was a lot of variation around that average, though. Parts of the species’ range saw decreased frequency of flowering conditions, and these were not always in places where we’d expected the trees to be experiencing fewer impacts from changing climate. The northwestern Mojave, for instance, shows stable or slightly decreasing frequency of flowering. On the other hand, high-elevation regions in the southwest — Joshua trees get to elevations above 7,000 feet on the eastern slopes of the San Bernardino Mountains — have some of the greatest increases in predicted flowering frequency.

On top of that complexity, it’s important to remember that flowering is only one part of what a Joshua tree population needs to be “healthy.” A stable population requires a steady supply of seeds that grow into seedlings, which then survive to be reproductive-age adults that replace older individuals as they die. Increased frequency of flowering could certainly help provide more seeds, and then seedlings that grow into adult Joshua trees — but it’s only the first step. For one thing, we don’t know how climate change is impacting Joshua trees’ specialized pollinators, yucca moths. If conditions that promote flowering have a negative impact on the moths, all those flowers would go without pollination, and never develop into fruits full of seeds. For another thing, the conditions that promote flowering —  bigger year-to-year differences in precipitation and warmer overall conditions — seem likely to be challenging for seedling Joshua trees, which are much more delicate than full-grown trees. So even if more flowers do produce more seeds, those seeds may not sprout and survive to grow into the next generation of mature trees.

We’re excited that this study provides us with some new insight into how Joshua trees have experienced climate change, even if it’s far from the whole story. We’re also hopeful that our approach might be more broadly useful — iNaturalist has records for a lot more species than Joshua tree, and it should be possible to derive information about flowering activity from those records in many more cases. We’ve got some possibilities already underway, and if you’re interested in applying this approach to your own favorite species, get in touch.

Meanwhile, you can read the full open-access article describing this work on the website of Ecology Letters.