Journal archives for May 2024

May 2, 2024

iNaturalist April News Highlights

iNaturalist broke records this April thanks to an amazing City Nature Challenge. For the first time we logged more than 6 million observations and more than 400 thousand participants in a month! If you missed last month’s highlights, you can catch up here.

Species Discoveries

Here are three exciting species discovery highlights from April:

A. In Argentina, @lrubio7 and @typophyllum published a new species of katydid with an ultrasonic call. This article details their challenges capturing a specimen and the role iNaturalist played in describing the Tuyú meadow katydid.

B. In French Guiana, @elendil_c captured the first living photograph of a spectacular mantis species. It was identified by @piskomantis. The female of the species remains unknown.

C. In Panama, while testing a new Automated Monitoring of Insects system, @mlarrivee posted a moth that @neoarctia recognized as a species that hasn’t been recorded since it was described over 100 years ago.


Distributions and Range Extensions

iNaturalist is helping scientists understand species distributions and how they are changing at in real time. Here are three examples from this month:

D. In Trinidad and Tobago, this beautiful tuft moth posted by @rainernd is just one of many country records @matthewcock has found on iNaturalist and compiled in a new paper.

E. In Mexico, @el_maldo, @jorgehvaldez, and @annyperalta published a paper on the expansion of a Blue-eyed Ensign Wasp into Baja Mexico likely following the expansion of its cockroach host. They acknowledge @bdagley and @adeans for identifying the relevant observations such as this one posted by @andrea_navarro20.

F. In Hawai‘i, @dryrunner found the first record of Hairy Tare described in a new paper authored by @kphilley.

Invasive Species Science

Since 1970, invasive species have cost the world economy over 1 trillion dollars. These April examples show how iNaturalist is being used in the early detection and control of invasive species.

G. In California, @gloval75 found an Ocellated Bronze Skink that likely hitchhiked from the Mediterranean via the nursery trade. In a new paper described here, @gregpauly explains how observations like these help trace introduction pathways in order to preempt the future introduction of damaging pests.

H. In Canada, @etienne_normandin explains here how iNaturalist is being used to track and stop the invasion of Wandering Broadhead Planarians through observations like this one from @bmtf.

I. In South Africa, @daverichardson describes in this new paper how iNaturalist observations like this one from @graceemmy are helping understand the invasion of Myoporum shrubs.

Conservation Science

These stories show how iNaturalist is being used to inform conservation decision making and to monitor conservation effectiveness.

J. From Brazil, in this video clip, @felipewalter summarizes a recent paper with @zamoner_maristela @rodrigogoncalves on how iNaturalist is valuable for bee monitoring and conservation.

K. In Texas, this observation by @scincus of a Broad-banded Copperhead and others in the Roadkills of Texas project were used by @jamestracy and colleagues in models that will help mitigate roadkills.

L. In California, the Xerces blue famously went extinct as San Francisco sand dunes were developed. This story describes how @farinosa and colleagues have reintroduced the closely related silvery blue to restored sand dunes in the city and will use iNaturalist to monitor the population.

Climate Change Science

Here are two stories that illustrate how iNaturalist can be used to understand impacts from climate change:

M. In Belgium, @antmacdf and colleagues explain how iNaturalist observations like this one from @tigojac can be used to monitor and mitigate the damaging invasive termites that they predict will expand into many more cities with climate change.

N. In California, @singleuseplanet explains how Nudibranchs response to ocean warming makes them useful indicators of climate change and how iNaturalist can be used to track this response.

Phenology

iNaturalist is a powerful tool for understanding the timing of natural history events. Here are three examples from this month.

O. Across the eastern US, two different Periodical Cicada broods are emerging at the same time. This New York Times article describes the event and how iNaturalist is being used to track observations. This article describes @willkuhn’s view of the emergence from Great Smoky Mountains National Park.

Don’t miss @imperfectfunguy’s “hyper-sexual zombie cicadas that are infected with sexually transmitted fungus” subplot to this story or @cicadamania’s amazing process for using iNaturalist to predict when Cicadas will emerge.

P. In Europe, @tr0scot and colleagues used iNaturalist observations like this one from @konstakal in a new paper revealing the nocturnal behavior of Grass Snake.

Q. In California, a recent study used iNaturalist observations such as this Spotted Sand Bass by @bregnier to understand the massive die off associated with a 2020 Red Tide.



iEcology

The new field of iEcology or imageomics involves extracting information about ecology from images. iNaturalist is providing the scalable supply of annotated biodiversity images that is driving this field. This recent article provides an imageomics workflow for incorporating iNaturalist big data into machine learning models.

R. In Uraguay, this study used iNaturalist observations like this one from @tianal to understand the diet of two South American Caracara species.


AI Naturalist


While much of the AI race is being driven by monetization and an extractive relationship with creators, iNaturalist is committed to an alternative approach to AI research grounded in commons-based peer production and open data principles. Here are some recent AI research papers that make use of the iNaturalist open dataset:

The long-tail nature of iNaturalist observation (few species with many observations and many species with few observations) makes it a useful dataset for investigating methods for dealing with unbalanced data as in these papers by Rangwani et al., Yang et al., and Hong et al.

S. iNaturalist data is also useful for developing Vision Language Models that are increasingly able to explain AI identifications. For example, the model by Chiquier et al. at Columbia University is able to generate explanations like “twisted curved branches” to describe this Greenleaf Manzanita by @sandor_in. Other Vision Language Model studies from April leveraging iNaturalist include Bendou et al. and Tao et al.


Bioblitzes and Events


There were many City Nature Challenge news stories due to the record-breaking 683 cities participating — so many that there are separate projects for North & South America and Eurasia, Africa, & Oceania. This article described how @kestrel and @lhiggins have grown the annual event. This article by @lmata and colleagues in Australia describes their recent Bioscience paper on how CNC is improving local government biodiversity management decision making.

There were many stories about Earth Day Bioblitzes such as this one built on the Martha’s Vineyard Atlas of Life project that quotes organizer @richardcouse.

T. Last but not least, there was ongoing coverage about iNaturalist and the April eclipse including this nice article in the Smithsonian Magazine that included this striking closing okra flower by @owensdc from the 2017 eclipse. We couldn’t resist sharing this map of observations on April 8, 2024 where the path of the eclipse can be clearly seen in the observation activity.


iNaturalist and Human Health

This month, stories describing how iNaturalist is benefiting human well-being ranged from articles on iNaturalist and Biomimetics (drawing inspiration from nature to solve human problems), to foraging culture in Iceland, to tracking the spread of malaria.



iNaturalist’s Education Impact

U. Students participating in Bioblitzes is an important way that iNaturalist impacts education. This photo by @maninder6398 shows students participating in City Nature Challenge 2024: Nanakmatta.

Classroom activities involving iNaturalist owe their success to great teachers. This article describes the mothing expeditions that @cmeckerman leads his Austin Community College students on. This study from Portugal describes use of iNaturalist on university campuses.


iNatters in the News

V. From California, we loved reading @tkestrel’s article on her passion for wildflowers.

W. From Texas, this article on the scrappy badger includes some great quotes from @jonahevans.

X. @wanderingbotanistph’s efforts to promote conservation of endangered Rafflesia species through social media and iNaturalist projects continue to inspire us.

Lastly, we’re so grateful for all the fantastic outreach being done by the iNaturalist community. Three examples include this great podcast featuring @joemdo, @gyrrlfalcon, @catchang, @griffith, and @naturesarchive, this outstanding Guide to iNaturalist by @thebeachcomber, and this pdf guide to herping with iNaturalist by @brittanymmason @tysmith and @coreytcallaghan.



Thank you to everyone who participated this April - our biggest month ever on iNaturalist! You can become an iNaturalist supporter by clicking the link below:


Donate to iNaturalist


Posted on May 2, 2024 10:20 AM by loarie loarie | 17 comments | Leave a comment

May 7, 2024

Observation Accuracy Experiment v0.4 is Underway

We delayed Observation Accuracy Experiment v0.4 a week to not distract from City Nature Challenge 2024. We're asking contacted validators to assess the sample by May 13.

We made 4 changes to this experiment from v0.3:

  1. Validators are now matched to sample observations based on past ID behavior in the same country rather than continent
    Previously, we matched a validator to a sample observation if they had at least 3 improving IDs within the same continent. We're now requiring at least 3 improving IDs within the same country to try to better match observations and validator experience. If we couldn't find any validators for an observation sample within the country, we expanded our search to continent and then globally, but we did this rarely. The samples are getting a bit large for some validators so we hope its not becoming a burden. Please let us know.

  2. We added a disclaimer to the message to comment here rather than replying to the message
    We've been receiving a large number of responses to the messages we send to contact validators that we aren't able to properly read and respond to. We added a disclaimer to the message with a link to this blog post asking people to post their questions and feedback here rather than replying to the message.

  3. We added more details about how to view your sample when viewing the message within the Android app
    Some people trying to access the message from the Message section of the iNaturalist Android app have been having trouble navigating to the sample. We included more details explaining how to do this.

  4. We added a search parameter to view observations included in an experiment sample
    Now that we're sampling 10,000 observations, clicking bars on a completed Experiment page is limited to exploring the first 500 observations. Longer term, we plan to make improvements to that page. But for now, we added a search parameter to construct explore URLs that, similar to how projects work you can construct Explore URLs using observation_accuracy_experiment_id, so you can see the sample of observations included in an experiment. For example, here are URLs for all the experiments so far (remember the version e.g. v0.4 isn't the same as the id e.g. 5):

Other than these three changes, this design and logistics of this experiment are the same as v0.3. As before, the Experiment page is live but the results will be updating daily and won't be finalized until May 13. At that time, we'll update this post with more discussion from the results.

Thanks again to all validators we contacted participating in this experiment. We wouldn't be able to conduct these monthly audits of iNaturalist observation accuracy with out your help!

Results (added 05/13/2024)

The results of this experiment were very similar to the other experiments. The average Research Grade accuracy (fraction correct) was 95%. You can explore the results including clicking through bar charts to observations here.

From all 4 experiments we've conducted, we've now assessed 22,000 observations including 12,464 Research Grade observations. The graph below uses Research Grade observations from this combined sample to estimate accuracy subset by continent, taxon group, and rarity (<100 observations is rare) and sorted by the uncertainty (95% confidence intervals). We now have much better estimates than we had before this experiment, and for many of the common subsets (black) we now have large enough samples to get confident estimates. But for all of the rare subsets (orange) our sample sizes are still too small to be confident in our estimates. As discussed in this thread below, we will probably have to design an experiment with a non-random sample targeting rare taxa to include enough of them to reduce the uncertainty accuracy estimates for these rare subsets.

Thanks again to everyone who participated in this experiment! We know this was a busy time on the heels of City Nature Challenge and very much appreciate your helping improve these accuracy estimates.

Posted on May 7, 2024 06:48 PM by loarie loarie | 119 comments | Leave a comment

May 9, 2024

Welcome, Michelle, Thea, and Jane!

The iNaturalist team is growing! We are pleased to introduce three new members of the iNaturalist team:



Michelle F. Vryn


Head of Development

Michelle, with over 15 years of nonprofit experience, has led both fundraising and communications teams. Her expertise spans major gifts, digital fundraising, and institutional giving at organizations focused on endangered species, disaster relief, and nature education. Before joining iNaturalist, she worked at OneStar and Bat Conservation International. Michelle is a mentor to young nonprofit professionals and serves on the Association of Fundraising Professionals (AFP) Global board of directors.




Thea Skaff


Director of Digital Fundraising and Engagement

Thea is a nonprofit leader with experience in fundraising, marketing, analytics, and tech, driving engagement and growth for mission-driven organizations. She was recently a senior member of the Online Fundraising team at the Wikimedia Foundation (Wikipedia) where she planned and led fundraising campaigns in an international environment including A/B testing on large global data sets. When not working, Thea is often outside biking or hiking with her husband and two boys.




Jane Weeden


Accounting Associate

Jane joins iNaturalist's accounting department, bringing with her experience gained from the dot com era, the startup world, and as a small business owner. She holds an MBA from University of Texas at San Antonio and is enthusiastic about promoting the iNaturalist message. Outside of work, you'll find Jane leading walks in the parks around San Antonio as a Master Naturalist. She’s an avid photographer and enjoys honing her skills documenting the biodiversity around us.

Please join us in welcoming them!

Posted on May 9, 2024 11:49 PM by carrieseltzer carrieseltzer | 34 comments | Leave a comment

May 16, 2024

New Computer Vision Model (v2.13) with 1,656 new taxa

We released a new computer vision model today. It has 88,517 taxa up from 86,861. This new model (v2.13) was trained on data exported on March 31, 2024.

Here's a graph of the models release schedule since early 2022 (segments extend from data export date to model release date) and how the number of species included in each model has increased over time.

Our goal is to try to attain the same accuracy or improve it while adding more taxa to the model. The graph below shows model accuracy estimates using 1,000 random Research Grade observations in each group not seen during training time. The paired bars below compare average accuracy of model 2.12 with the new model 2.13. Each bar shows the accuracy from Computer Vision alone (dark green) and Computer Vision + Geo (green). Overall the average accuracy of 2.13 is 88.2% (statistically the same as 2.12 at 88.1% - as described here we probably expect ~2% variance all other things being equal among experiments).

Here is a sample of new species added to v2.13:

Posted on May 16, 2024 09:47 PM by loarie loarie | 5 comments | Leave a comment

Archives