March 12, 2024

So a Snake and a Centipede Meet on a Trail... - Observation of the Week, 3/12/24

Our Observation of the Day is this meeting of a Chinese Red-headed Centipede (Scolopendra mutilans, 왕지네 in Korean) and an Ussuri Mamushi pit viper (Gloydius ussuriensis, 쇠살모사 in Korean) in South Korea! Seen by @lhurteau.

Originally from Vancouver, British Columbia, Leslie Hurteau has been teaching English in South Korea since 2017 and takes daily morning birding trips on Jeju Island, where he currently lives. 

I always keep my eyes open for any wildlife, especially reptiles and interesting arthropods during the warmer months. I was walking along a trail in a local city park, and some movement caught my eye so I looked down to see a snake (Gloydius ussuriensis) and a Chinese Red-headed Centipede (Scolopendra mutilans). It seemed the snake was interested in the centipede and was eyeing it down. The centipede scuttled under a nearby leaf, at which point the snake prepared itself to strike. 

Unfortunately the excitement ended there, as some other people walked by and caused both the snake and centipede to move off trail and into the side vegetation. So, I didn't see who won the standoff. A friend (@j-j) previously found another Gloydius species consuming the same type of centipede on the mainland, so I think it's possible the snake would have won this, although given it appeared to be a smaller juvenile, it wouldn't surprise me if the centipede would have a strong chance against it.

A fun side note, Gloydius ussuriensis is the only Gloydius species (a group of venomous pit vipers in East Asia) found on Jeju, and its colouration is so different from the mainland populations that it has attracted some potential interest from local researchers.

Not only do Gloydius snakes live in mainland Korea, they range west to the Ural Mountains, into South Asia, and east to Japan. Like other pit vipers, they inject venom through hinged fangs and sport heat-sensing pits on their snouts.

Chinese red-headed centipedes occur in East Asia, including Japan, and average about 20 cm (8 in) in length. Like other centipedes, they’re predatory and they subdue prey by injecting their venom through forcipules, which are modified front legs. 

Leslie (above, scanning for birds on Jeju Island), credits his parents for his interest in nature. They’d take him to parks, especially Stanley Park, in Vancouver, “and my dad would point out the names of different birds and plants that he knew.” He eventually volunteered with the Wildlife Rescue Association of BC and the Stanley Park Ecology Society.

Leslie started using iNat back in Canada, but really got into it after his move to Korea.

After moving to Korea and seeing how scarce information was in English on local fauna and flora, I decided to go all in and document nearly everything I find, with a focus on birds. Uploading every bird photo I could take greatly improved my bird identification skills, helping me learn how to spot various features even in not-so-ideal photos. 

It's been a fun journey, having now seen over 370 different species of birds in Korea, including one species that was a national first for the country. I'm happy to say it's a different situation today in 2024 with many observations in Korea from many different people. After moving to Jeju Island, my favourite part of Korea, I have been working even harder to upload observations to document the wildlife here. With development happening so fast in this country, it can be important to document what is here now in case habitat changes or is lost (which happens often here), or as species ranges change to deal with development and climate change.

iNaturalist has helped improve my identification skills, as well as make connections with other like minded individuals, here in Korea and elsewhere. It's a wonderful globally minded website that I recommend to anyone with an interest in nature.

(Some quotes have been lightly edited for clarity. Photo of Leslie was taken by Jiwone Lee.)


- some giant centipedes eat bats!

- two previous Observations of the Week document centipedes as prey: this one by @msone, and another by @magdastlucia!

Posted on March 12, 2024 09:24 PM by tiwane tiwane | 10 comments | Leave a comment

March 7, 2024

Some thoughts on ML accuracy

The folks on the iNat team who work on the iNaturalist Computer Vision Model and the iNaturalist Geomodel spend a lot of time thinking about how accurate the Machine Learning (ML) models are, and how we can improve them. It's a particular challenge since our computer vision training dataset grows by over a thousand taxa and by over a million photos every month.

We tend to visualize accuracy using charts like you'll see below, and we use these charts to validate our models internally: are they ready for prime time? Is this month's model good enough for release?

Sometimes our accuracy numbers would drift from month to month, and recently we set out to explore why that might be.

Some of this could be due to test set sampling differences - if we test using a batch A of photos for January and batch B for February, then we can expect to get slightly different results. It's also a little tricky to compare models directly to each other since the taxonomies are different. The Feb model knows about taxa that the January model didn't, for example.

Another theoretical reason might be some combination of ML training run differences: random initialization, optimizer, and variance in batch selection. In our experience this doesn't look like a big source of model accuracy difference from month to month but we'll keep an eye on it.

Another reason why our models might show different accuracy numbers from month to month is training set sampling error. We previously described the transfer learning strategy that we are using to train these monthly models. Models share the same foundational knowledge, but we use new photos each month to fine-tune the base model. For commonly observed taxa, we don't train on every image we have. Instead, we select at most 1,000 photos for training. For a taxon like Western Honey Bee, which our users have upload more than 400,000 observations of, every month we select a different fraction of photos to train on. Some months we might get better or worse photos than others, and that might result in better or worse performance.

In order to start exploring this last reason, when we made the export for the 2.11 model, we also made an alternate version of the export with the same taxonomy but selecting a different batch of photos. Then we trained an alternate model on this alternate export, and we did some comparison.

For each of the major groupings shown in the chart below, we selected 1,000 random RG observations not seen by the model during training process and compared the species ID'd by the community to the species predicted by the model suggestions. If the model agreed with the community, we considered the model correct, while if the model disagreed, we considered it incorrect.

Here's a chart showing the two models, accuracy performance compared. The y-axis shows top1 accuracy as a percentage, the x-axis shows major clade or geographical groupings, and the colors represent the different models compared against each other. The *-vision colors are computer vision model accuracy only, and the *-combined colors show accuracy when combining the computer vision model with our geo model.

(Note: I'm red-green colorblind and I have some difficulty distinguishing the 2.11-alternate-vision and 2.11-alternate-combined colors, but they are in the same order in each major grouping, so I can still interpret the chart. If anyone is struggling to interpret the chart, please let me know and I'll see if I can make a more accessible variant.)

For someone who hasn't seen one of these charts before, I'd like to point out a few interesting interpretations or conclusions. First, combining computer vision and geo modeling performance helps a lot, sometimes as much as 20%. Second, we can see our dataset bias in this result: our models know a lot more about taxa in North America and Europe than it does about taxa in South America or Africa. This mirrors our observer community and our dataset of photos to train on. When we have more images from an area, we have better performance. There are some other interesting questions to explore about why our models perform better or worse on some taxonomy groupings like herps or mollusks compared to birds or plants.

However, in the context of understanding this alternate experiment, we can takeaway that two models trained on the same taxonomy but different photos will see a small degree of variance in both computer vision and combined performance on the same test set.

Here's a look at how the 2.11-alternate model improved (or didn't) when compared directly against the 2.11 model:

We can see from this that we can probably expect less than a 2% variance from model run to model run based on sampling.

From this, we can do a better job of characterizing our model accuracy month-to-month: are things improving, getting worse, staying the same? It looks to us like our models are staying about the same in average accuracy while adding 1,000 new taxa a month. We think that's a pretty great result.

Another great way we can use the this experiment is to judge how much a change to model training or technology is really improving model accuracy. If we try out a supposedly better piece of technology to train a future model, but we see less than a 2% accuracy bump, then we should probably be a little suspicious.

We're excited to share the results here with you, and we plan to share accuracy metrics about our models when we release them from now on.

Posted on March 7, 2024 04:26 PM by alexshepard alexshepard | 12 comments | Leave a comment

March 5, 2024

Ant Stacking Assassin - Observation of the Week, 3/5/24

Our Observation of the Day is this ant-encrusted assassin bug nymph (in the genus Inara), seen in Malaysia by @victor0001!

Victor Heng, a teacher at the Natural History Museum in London, got going with iNaturalist (as well as iRecord) in earnest during the lockdowns of 2020. 

I started using it to record everything I was seeing in my local parks. I started using the maps to look for places with few records in my area and set myself a mission of characterizing the plants and invertebrates in those blank areas. Since then I've started to look more for phenology, recording flower plants at early and late points in the season.

For Lunar New Year Victor and his partner visited Victor’s family in Malaysia, where he took the photo you see here.

This was only the 3rd time I've been to Malaysia, and the first time I'd been since picking up the iNaturalist habit, so I was very excited to have an opportunity to explore rainforests with this new perspective. A few of my relatives live in Penang and, knowing that my partner and I like nature, took us up to Penang hill to experience the canopy walkway.

Several of my cousins and younger relatives were along for the walk, and I was inevitably at the back. My photographing of things encouraged them to also take a closer look at what was around, if only to point it out as something I might want to take photos of. The assassin bug was one of these spots. My cousin called me over and pointed it out, wondering if this small black thing was a bit of poop or actually an animal. I initially thought it might be a spider, but once I saw it through the camera lens I realized what it was, and was amazed at all the ants it had stacked on its back. I'd still like to know how it reaches back to get them stacked up, and how the ants stick together up there.

Yunhan (@eggshe11), who works on identifying assassin bugs in Asia, came across Victor’s find and identified it as being in the genus Inara. I contacted Yunhan about the ID and any info they might have about this behavior. 

To my knowledge, there are 2 common genera of carcass-stacking assassin bugs in southeast Asia - Acanthaspis and Inara. Where I'm from (Singapore), both genera have been relatively well documented (see this page on Inara and this one on Acanthaspis for example). As the photographer Nicky Bay (@nickybay) mentions on his page, Inara nymphs are identifiable by being less concealed overall and having a “neater” ant stack. Acanthaspis nymphs tend to be covered in debris including their heads, legs, etc (example, also visible in this popular observation) while for Inara you will always be able to see the bare legs and the head complete with large eyes resembling the adult (for example, this observation). 

For Inara, post-feeding, a nymph will use its front and mid legs to adjust the carcass to a satisfactory shape, subsequently transferring it underneath the existing stack onto the dorsal surface of the abdomen using its hind legs. The abdomen possesses glands which secrete a sticky substance, and these secretions also hold the stack together. After ecdysis, nymphs will transfer the stack off their molt, rather than start a new one. I'm not sure why they keep a stack, although I think the main reasons lean towards (1) attracting/pacifying ants via pheromones from the carcasses (2) predator avoidance/deterrence. 

Victor (above, in Kuala Lumpur), traces his interest in nature back to his childhood, when he’d collect rocks or bring home insects on sticks. Later in life he became interested in social science and teaching, and is happy he can combine those interests in his current position. He’s currently developing the museums new outdoor program, to be held in its redeveloped gardens. One project involves the students analyzing “bycatch” from moth traps, that represent insects from hypothetical farms. 

The early pilots of this session have been really rewarding. Students often have strong reactions when they first put their petrie dish of dead insects under the microscope. Mostly a mix of surprise and fascination, but occasionally disgust. But in the vast majority of cases, all their reactions drift towards fascination at the variety of insect forms. Its been great to be able to see their shift in attitude towards insects, and appreciate their importance to ecology. And understand the importance of ecology to agriculture and humans.

He’s been using iNaturalist in this work with the museum, tracking which invertebrates are found on which plants in the gardens, for example. 

The process of submitting observations, roughly evaluating the accuracy of iNat's suggested identifications, and being able to quickly look up the taxonomy, I've also found to be a great way of getting a feel for the characteristics of different families. I really like going to someplace I've never been, and recognizing plant and animal families. It’s a nice feeling to be walking somewhere and think to myself “I don't exactly know these ones, but I think I know your relatives.”

(Some quotes have been lightly edited for clarity.)


- check out Victor's Knowing Nature podcast, about enviromental education!

- here's some footage of an assassin bug in action.

- take a look at the amazing panoply of assassin bugs on iNat!

Posted on March 5, 2024 11:27 PM by tiwane tiwane | 9 comments | Leave a comment

March 2, 2024

iNaturalist February News Highlights

We can’t believe it’s already March! Here are our February news highlights. If you missed last month's highlights, you can catch up here.

Species Discoveries


Usually we talk about species discoveries by mentioning a few examples from the news. But this month, we wanted to highlight some of the great work partners in the iNaturalist Network are doing to better find and surface iNaturalist observations that represent important discoveries.

  • iNaturalist Canada coordinators at the Canadian Wildlife Federation have set up this project to track and highlight Canada’s most groundbreaking observations.
  • iNaturalist Uruguay site admin ​​@flo_grattarola describes on her blog efforts to surface novel observations posted from Uruguay.
  • This great article by Naturalista Colombia site admin @carolinasoto in the oldest newspaper in Colombia describes all the work she and colleagues at the Humboldt Institute are doing to leverage iNaturalist for conservation impact in Colombia.
  • Similarly, this article highlights @paul_luap’s efforts to lead citizen scientists as site admin for iNaturalist Luxembourg.




    Range Extensions and Distributions


    Wolverines haven’t been spotted on iNaturalist in the Cascade mountains south of Mount Rainier in Washington State. This article describes how the Oregon Department of Fish and Wildlife is investigating rare range extensions south into Oregon and turning to iNaturalist for help.

    There were many articles this month demonstrating how iNaturalist is being used to better understand the distributions and biogeography of species including:


    Invasive Species Science


    Some of the complexities around invasive species are captured in this article about a new study by @christiaan18. Oaks are introduced in South Africa and some are becoming invasive. On the other hand, a spreading invasive beetle is devastating the iconic oaks planted throughout cities like Cape Town.

    Meanwhile, Australia’s iconic native Eucalypts (which just received a huge boost on iNaturalist thanks to contributions from Eucalypt expert @dnicolle) are being devastated by the spread of invasive myrtle rust. This article describes @myrtle_rust_martino’s efforts to prevent the spread in Australia with the help of citizen scientists through the Gum Tree Guardians project.

    Speaking of myrtle rust, an increasing number of natural resource agencies are turning to iNaturalist to help monitor invasive species, as this myrtle rush announcement form the New Zealand Department of Conservation demonstrates. This article describes how iNaturalist is being used to monitor invasive Stinknet in Saguaro National Park as the National Park Service kicks off Invasive Species Awareness Week with both place-based activities and species-focused resources leveraging iNaturalist.


    Conservation


    Monitoring

    This April, Periodical cicadas should emerge in northern Louisiana on their 13-cycle, but scientists fear the brood may be extinct. This article describes how scientists are turning to iNaturalist to help. Other examples of monitoring include:




    Restoration

    The role iNaturalist plays in conservation isn’t restricted to monitoring; it also extends to land preservation and restoration.


    Climate Change Science


    Warmer winters are disrupting the life cycles of insects, and iNaturalist is helping scientists understand this challenge:

    • This article describes @fuzzybumblebee’s efforts to understand the consequences of being active in the winter on honey bee populations in Minnesota.
    • Similarly, this article shows how iNaturalist data is helping scientists understand why sickly, normally dormant Red-based Jezebel butterflies were spotted this winter across Hong Kong.

    If curbing climate change feels overwhelming, this National Geographic article includes iNaturalist as a tool for individual action that can help children feel empowered when confronting climate change.


    Secondary data science


    Last month we highlighted the concept of “secondary data” as all the important data captured in iNaturalist observations beyond the core species occurrence. This month’s examples include this fascinating study by @alexkerr on spider web architecture that resulted from an analysis of over 13,000 iNaturalist observations of Feather-legged Orbweaver.

    • This study used iNaturalist images to document ray stranding events to understand what’s causing them.
    • This study used iNaturalist images to document the mushroom eating habits of native Chilean snails.


    iNaturalist Impact on AI Research

    Alongside satellite imagery and environmental DNA (eDNA) data, citizen science data is one of the only scalable sources of environmental data. Many AI projects are now combining multiple sources of data in exciting new ways. For example, this article describes a study that combines iNaturalist data and satellite imagery to map croplands. Likewise, this Popular Science article on advances in eDNA describes how it will complement iNaturalist biodiversity data. This article describes how AI advances are changing birdwatching and the role iNaturalist plays.

    Bioblitzes and Events



    iNaturalist’s Human Health and Social Science Impact

    Clear air and clean water are important contributors to human health. This article describes how iNaturalist is used to monitor macroinvertebrates as water quality indicators in Colombia. Similarly this article describes how the National Park Service incorporates iNaturalist to help leverage lichens as indicators of air quality.

    iNaturalist’s Education Impact


    We enjoyed this article about Ecoexplore, a Kindergarden through 8th grade environmental science program developed by @jmarchal and colleagues at the North Carolina Arboretum that leverages iNaturalist.


    iNatters in the News


    Thank you to everyone who participated in iNaturalist this February and we look forward to a busy Northern Hemisphere Spring - your support makes it all possible!


    Donate to iNaturalist


    Posted on March 2, 2024 08:51 AM by loarie loarie | 7 comments | Leave a comment

    February 28, 2024

    A School of Rays in Mexico! - Observation of the Week, 2/28/24

    Our Observation of the Week is this school of Golden Cownose Rays (Rhinoptera steindachneri, Raya tecolote in Spanish), seen in Mexico by @flore_boituzat!

    Flore Boituzat grew up in the French countryside and says she’s always been interested in nature. She’s a food engineer, and tells me “I had the opportunity to do one semester in Mexico to do marine biology, that's where I started to dive and go snorkeling.”

    She also met her boyfriend Alberto Alcalá (@alboertoalcala) while there, who’s a diver and an oceanologist, who introduced her to iNaturalist. 

    After her semester in Mexico, Flore had to finish her last year of schooling in France but she recently returned to Mexico. “As a welcome gift, Alberto took me out for a dive where we saw all these beautiful rays. I felt really blessed. :)”

    Golden cownose rays occur in the eastern Pacific Ocean, from Peru up through Mexico, and are often found in groups like this. Like other rays, it doesn’t have teeth but bony plates in its mouth for crushing its prey - mostly bottom dwelling species. It’s listed as Near Threatened by the IUCN, as it’s often caught as bycatch by artisanal gillnet and trawling fisheries. 

    Flore (above) continues to use iNat, and continues to share her nature photos on iNaturalist. 

    I like nature photography and he knows about the biology part, so thanks to iNaturalist I learn about the species I'm seeing, and I can also see where a species is found a well as, it’s habits and characteristics, so it's easier for me to find and to capture photos of it. I think biodiversity is part of the beauty and the magic of this planet, and to take pictures of biodiversity helps me share it; I find it very nice and fun to see what's surrounding us.

    (Some quotes have been lightly edited for clarity. Photo of Flore was taken by Alberto Alcalá.)


    - here’s footage of a golden cownose ray school!

    Posted on February 28, 2024 09:19 PM by tiwane tiwane | 8 comments | Leave a comment

    February 26, 2024

    A second experiment to learn about the accuracy of iNaturalist observations

    Today we’ve launched our second Observation Accuracy Experiment (v0.2). Thanks to everyone helping us conduct these experiments. We're learning a lot about iNaturalist observation accuracy and how to improve it.

    Changes in this Experiment

    We made two changes to the experimental design from v0.1 based on feedback:

    • We changed the validator criteria to be at least 3 improving identifications of the taxon from the same continent since many reported not feeling comfortable identifying taxa outside of their regions of expertise.
    • We messaged candidate validators rather than emailed them since many reported not noticing emails. We also only left a 4 day interval (rather than 2 weeks) between contacting validators and the deadline since last time most validating happened within the first couple days after contacting candidate validators.

    Eventually, we’d like to increase the sample size from 1,000 to 10,000, but we’re sticking with 1,000 until we get a few more kinks out of the methods. The page for this experiment is already live and the stats will update once a day until the validator deadline at the end of the month, but you won’t be able to drill into the bars to see the sample observations or the validators until the deadline has passed.

    New Data Quality Assessment condition for photos unrelated to a single subject

    We also made one change to iNaturalist functionality in response to findings from the study. We added a new “Evidence related to a single subject” condition to the Data Quality Assessment table to make it easier to remove observations with multiple photographs of unrelated subjects from the verifiable pool.

    Two of the incorrect Research Grade observations in Experiment v0.1 were of this type which we estimate to be ~350k observations in the entire iNaturalist dataset. The norm up until now to make these observations casual has been to set an identification to the nearest taxonomic node shared by the multiple subjects and then vote no to “Based on the evidence can the Community Taxon still be confirmed or improved?”, but many found this process to be clunky and confusing. We hope this new Data Quality Assessment condition will make it easier for the community to remove these observations from the verifiable pool where they negatively impact data quality and distort features on iNaturalist (such computer vision model training and the browse photo tool) that assume observation photos all relate a single labeled subject.

    Thank you!

    Thank you to everyone contacted as a candidate validator for participating in this experiment. We expect that considering location may decrease the percentage of samples validated compared to the previous experiment by constraining the candidates available to validate, so we very much appreciate your participation in helping get as much of this sample validated by the end of the month as possible. As always, please share any feedback or thoughts you may have on the topic of data quality. We’re excited to continue learning from these experiments and your feedback about data quality on iNaturalist and what changes we can make to improve it!

    Results (added 2/29/2024)

    Thanks everyone for participating in this 2nd experiment. The validator deadline has now passed, meaning that on the experiment page the stats will no longer update, the validators are now visible, and the “Accuracy results by subset” bar graphs are now clickable allowing you to drill in to see the observations behind the graphs.

    In this second experiment, we estimated the accuracy of the iNaturalist Research Grade observation dataset to be 97% correct and the accuracy of the Needs ID subset to be 79% correct. The graph below shows the first experiment (v0.1) in lighter bars and this experiment (v0.2) in darker bars. The results are very close which is reassuring.

    These estimates of average accuracy of the entire Research Grade observation dataset are in line with our expectations, largely because the iNaturalist dataset is skewed towards a relatively small number of common, easy to identify species (e.g. mallards, monarchs, etc.) that have an outsized impact on the average. Nonetheless, we wanted to touch on three sources of uncertainty in these estimates: validator skill, sample size, and the uncertain category.

    Validator skill

    We are assuming that candidate validators can perfectly validate observations as Correct, Incorrect, or Uncertain. We know this assumption is not exactly correct because there are a small fraction of situations where more than one validator looked at the same observation and they disagreed (e.g. validator 1 says Taricha torosa and validator 2 says Taricha granulosa OR validator 2 says Taricha because you can’t rule out Taricha granulosa). This happened on 1.6% of the time in v0.1 and 1.2% of the time in v0.2. This error might be higher if we are underestimating disagreements because validations aren’t done blind (i.e. validators can see each other’s validations). But the error might also be lower because each observation was validated an average of 4 times, so assuming the validations are mostly independent even if one validator made a mistake it was reviewed an average of 3 more times. In future experiments, we’ll do more work to estimate uncertainty in the labels (Correct, Incorrect, or Uncertain) stemming from imperfect validator skill. But while this uncertainty stemming from validator skill is non-zero, it’s likely close to 0. Furthermore, there’s no reason to assume that this uncertainty would bias towards inflating the accuracy by overestimating the proportion correct, it could just as well bias towards underestimating the proportion correct.

    Sample size

    Because being correct or not is like a coin-flip, we can put confidence intervals on our estimates of average accuracy based on the sample size. As the sample size increases the confidence intervals become narrower. We can compare our estimates and confidence intervals from v0.1 to our estimates and 95% confidence intervals if we pool v0.1 and v0.2 together effectively doubling our sample sizes.

    We already have a large enough sample size (n) to have fairly confident estimates for large subsets of the iNaturalist database such as the Research Grade (RG) Accuracy Estimate. After v0.1 (n=534) our estimate and 95% confidence interval was 0.95 (0.93 - 0.96) and pooling v0.1 and v0.2 (n=1109) it is now 0.96 (0.94 - 0.97).

    However, for smaller subsets such as RG Fungi our confidence intervals are still quite large. After v0.1 (n=6) our estimate was 0.83 (0.36 - 1.00) and pooling v0.1 and v0.2 (n=19) it is still 0.95 (0.74 - 1.00) - so somewhere between 74 and 100% accurate.

    For very small subsets, our sample size is still much too small to provide useful estimates. For example, to estimate the accuracy of RG Rare (taxa with fewer than 1000 observations) African Insects our estimate even after pooling v0.1 and v0.2 (n=5) is 0.6 (0.15 - 0.95). For other subsets (e.g. RG Very Rare (<100 obs) African Insects) we have a sample size of zero and can’t make any estimate.

    The size of our sample is adequate for getting relatively confident estimates of average accuracy for the entire iNaturalist database and for large subsets (e.g. the RG subset, the North American subset, the Insect subset etc.), but these sample sizes are too small to yield confident estimates for more niche subsets (e.g. RG South American Fungi from 2022, etc.). We are very interested in variation in average accuracies across these subsets and look forward to growing the sample size to the point where we can better understand this variation.

    Uncertain Category

    Ideally, we’d be able to label all observations as Correct or Incorrect. But because we don’t have the capacity to get validations on all observations in the sample, some remain Uncertain. This was 3% of the RG subset in each of v0.1 and v0.2. Since we are calculating Accuracy as the percent correct (as opposed to 1 minus the percent incorrect) this Uncertain category is biasing us to underestimate the Accuracy. The true average accuracy of the iNaturalist Research Grade observation dataset is somewhere from 0% to 3% higher than our estimates because of this bias which is on par with the uncertainty interval resulting from sample size.

    Thank you and next steps

    Thanks again for all your help with another successful experiment. We’re amazed by the capacity of this incredible community to validate these samples. We hope you’ll click on the graphs and explore the results here. We plan to run another experiment at the end of March. We may try keeping the methods the same and increasing the sample size from 1000 to 10,000. Or we may make another change to the methods such as more changes validator candidate criteria. Thanks again for making these experiments possible!

    Posted on February 26, 2024 06:58 PM by loarie loarie | 109 comments | Leave a comment

    February 23, 2024

    Identifier Profile: @sofiazed1

    This is the twenty-third entry in an ongoing monthly (or almost monthly!) series profiling the amazing identifiers of iNaturalist.

    I met Sofia Zvolanek and her mother Jenny at an iNaturalist meet-up arranged by @thebeachcomber in Sydney back in 2022 and was really impressed by her passion for nature - especially fungi. She’s currently the top identifier of verifiable fungus observations in Australia with about 80k identifications made, so I thought I’d take the opportunity to profile her here.

    Currently a 23 year old graduate from Sydney, Sofia’s been living in a suburb by the bush for about 18 years, and tells me

    I think I've always been interested in nature - the colours and patterns and shapes and minute details, interacting with things and watching how they behave, seeking out interesting creatures under logs, collecting cool shells and other objects from the beach. I've been picking up bugs since I learned to walk. My interests were nurtured by my Mum - as a kid I spent a lot of time with her gardening, bushwalking, and doing crafts. I also spent a lot of time outdoors up trees or in the bushes.

    Her first memories of mushrooms involve her staining the walls of her cubby by rubbing Coprinellus everywhere (“artistic spirit I guess”), but she really got into fungi over the past few years, her interest sparked by friends who brought them some foraged mushrooms.  “I discovered that mushrooms could actually taste nice,” she says. “From there I gained an interest in identification for foraging purposes, which grew into identification for curiosity's sake after I started using iNaturalist and the information became much more accessible to me.”

    Sofia got hooked on iNat in 2020, when she had to make five observations for a university class. 

    …and from there I did not stop. Then the lockdowns hit, and I was spending a lot of time at home or in the local bush, walking for fun or to travel to uni once classes were back on, while the mushroom season was in full swing. The ephemeral and mysterious nature of mushrooms, along with their interesting and varied appearances, makes them particularly appealing to my novelty-seeking brain, and I decided to answer the endless stream of “Oooh what's that?” by photographing and posting them all to iNaturalist. From there I juggled classes and assignments while spending as much time as possible on identification sprees or cram-studying the variety in fungal morphology by scrolling through the iNaturalist observed species lists order by order or family by family (for example) to get an idea of what exists and how it's arranged in the tree of life. I've learned pretty much everything I know about broader fungal taxonomy through iNaturalist and its users.

    Funny story - I spent my first few weeks on iNaturalist under the mistaken impression that @thebeachcomber was one of our course instructors, because he was so diligent about welcoming my classmates and I to the platform.

    And she gives back by adding a lot of identifications, both on her phone when she has a break or is in transit and via the iNat website. Her current strengths are Agaricomycetes, Xylariales and Lecanoromycetes (mostly Cladoniaceae), as well as a few slime molds (Mycetozoa).

    Most of the time I rely on my memory of patterns observed while cram-studying the observed species lists, but when a species or taxa piques my interest I'll go on research binges to dig up as many relevant taxonomy papers and descriptions as I can find in a few hours to gain a better understanding of morphology and species delineations. Unfortunately most of the Australian compiled resources are years out of date, with old or misapplied names. I'm also not subscribed to any of the journals, so I tend to just put the taxon into Google Scholar and hope for the best.

    These tend to be followed by an hour or more of scanning through the observations of the most relevant overarching taxonomic group to apply the newfound IDs to unidentified observations. I've also recently been making use of the “Similar Taxa” tab to help resolve the frequent mix-ups between similar-looking but often widely taxonomically distributed genera of earthstars and coralloid fungi, along with some other groups.

    And why add all those IDs? “Sorting things and solving problems is satisfying and helps contribute to global knowledge of fungi,” she explains. “Every so often I learn something new or notice something scientifically interesting. Sharing my interests with others is fun and I enjoy people's excitement when they learn something new. Finding friends and a community, and making connections. I feel like I'm doing something worthwhile with my time.”

    Now that she’s graduated from university, Sofia’s current plans are to find a job and get settled into “adult life,” but she may return to school for a graduate degree in the future, hopefully focusing on fungal taxonomy or ecology. She also has a pipe dream of starting a mycological garden “to showcase the beauty of different fungal forms and give the general public a better understanding and appreciation of fungi.”


    Need tips for making identifiable fungus observations? Sofia’s got some advice:

    • Take multiple photos from multiple different angles and make sure they’re in focus. The most important is a wide shot showing the fungus in its habitat. Also, close-ups of the whole fruiting body from different angles (top, side, underside, etc), whether or not there is a stem and what it looks like. Acquiring a small mirror such as a compact or dental mirror will help capture those critical underside and stem angles if you want to leave the fungal fruiting body intact.
    • Other useful things would be closeups of different structures and textures on the fruiting body hymenium structure (gills/pores/teeth/smoothness) and attachment to the stem/substrate, as well as the base of the stem and any other lumps, bumps, tufts or shreds on the fruiting body.
    • Breaking or bruising can cause some fungi to react to air, which can help with identification, and you can also make spore prints (only a thumb-sized part of the hymenium is needed). 

    Some of Sofia’s favorite fungi taxa: Bird's nest fungi (Nidulariaceae), Cannonball fungi (Sphaerobolus), Crown-tipped coral fungi (Artomyces), Pretty-lips (Calostoma), and regular Coral fungi (Ramaria). Taphrina cornu-cervae is my favourite plant pathogen. Psathyrella aquatica is average looking, but pretty cool because it’s (I think) the only known agaric that fruits under water. There are many other aquatic fungi, but most of them are single-celled or have insignificant fruiting bodies. Rosecomb deformities (for example https://www.inaturalist.org/observations/164406178) are pretty wacky looking.


    You can see Sofia (and her mum!) at about nine seconds in on this interview video with Thomas Mesaglio!

    Posted on February 23, 2024 10:38 PM by tiwane tiwane | 45 comments | Leave a comment

    February 21, 2024

    Watering the Plants leads to Rare Beetle Discovery - Observation of the Week, 2/20/24

    Our Observation of the Week is the first Megamerus alvarengai leaf beetle posted to iNaturalist! Seen in Brazil by @alenilson.

    “As a child I had a lot of contact with nature through my grandmother, who was a farmer,” recalls Alenilson Rodrigues. 

    I sometimes accompanied her in her efforts in the field and that's where my passion came from. As time went by I started going to the forests as a way to de-stress and I started taking photographs at the suggestion of my wife. At that moment I was having personal problems that were distracting me and photography came to me as a way to change the direction of my life. Without a shadow of a doubt my wife always deserves to be highlighted for my photography. If it weren't for her suggestion we wouldn't have these images today.

    I developed a special taste for photographing arthropods, especially flies and moths, and I ended up developing a technique that generates images with a white background, giving me the nickname “studio macro”. It's a very laborious method because I only register live animals and after registration I release them in a safe place, most of which I find on my property or nearby.

    He came across this Megamerus alvarengai beetle while at home, and at first didn’t know it was a particularly special find. 

    There is no great story, I simply believe that any living being should be photographed so that there is the possibility of knowing its occurrence as well as a means of protecting it, as I believe that only what is known is respected. While watering some plants I noticed the beetle hidden among the leaves and registered it. It was @borisb who identified both the being and its importance.

    Boris Büche (@borisb) is iNaturalist’s top identifier of beetles, having added IDs to over 650k (!) verifiable observations of them, so I reached out to him about Alenilson’s find to get his thoughts on it. He says that the subfamily to which it belongs (Sagrinae) is an ancient branch of leaf beetle evolution. Most extant members of it are found in the old world.

    When I saw the discovery by @alenilson, my first belief was that it would be something completely unknown. I remember well my identification of Atalasis sagroides four years ago, then found to be the single American representative in the basic checklist by Blackwelder (1945/46).

    M. alvarengai was discovered five years afterwards in the state of Rio Grande do Norte, Brazil (seven examples, found by Moacyr Alvarenga in 1950, scientific description in 1956 by Francisco de Asis Monrós). The beetle was found again only in 2009-2013 during a trap sampling campaign in Caatinga formations in the state of Paraíba (twenty examples, in xerophilous vegetation) - and now, by Alenilson.

    Alenilson found two of these beetles (here’s the other), and both were on coconut palms, but Boris thinks it’s likely not their host plant. Not much is known about the species, but host plants for other members of the subfamily are dicots, not monocots

    Larvae of its Malagasy relatives form galls (unpublished observation), and pupae of two Australian forms (including a Megamerus) have been found within earthen cocoons underground. One iNaturalist observation of a female laying eggs would support the suspicion that larvae of Megamerus are root miners.

    “I use iNaturalist to promote the animals I find in my region,” says Alenilson (above). 

    Here I can meet experts as well as see  the occurrence of different animals in different points of interest. After getting to know the platform I discovered people and places, and of course, several animals that occur here in the state along with other people who share the same interest.

    (Some quotes have been lightly edited for clarity and flow.)


    - you can follow Alenilson on Instagram and YouTube!

    - this observation is the First Known Photographs of Living Specimens project!

    Posted on February 21, 2024 05:13 AM by tiwane tiwane | 17 comments | Leave a comment

    New Computer Vision Model (v2.11) with 1,256 new taxa

    We released a new computer vision model today. It has 84,878 taxa up from 83,622. This new model (v2.11) was trained on data exported on December 31, 2023.

    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.

    Here is a sample of new species added to v2.11:

    Posted on February 21, 2024 12:16 AM by loarie loarie | 39 comments | Leave a comment

    February 6, 2024

    Canary Island Woodlice! - Observation of the Week, 2/6/24

    Our Observation of the Week is the first Porcellio ombrionis posted to iNaturalist, seen in Spain by @amaneko!

    Amanhuy Duque and Andrea Castro (who have the @amaneko account) are both biologists from the Canary Islands, and they specialize in terrestrial biodiversity and conservation. Amanhuy, who is quoted in this piece, tells me that he’s been fascinated by animals for as long as he can remember. 

    My parents always took me for walks in nature and this helped raise awareness about its conservation and the importance of knowing how to look. From that early age it was clear to me that I wanted to be a biologist when I grew up, and that’s what I ended up doing. 

    In Andrea's case, she always had an interest in nature and biology, but her true passion, birds, was discovered during her studies. She learned to mix it with her passion for drawing to make some incredible illustrations.

    Last month, Amanhuy visited the island of El Hierro with his family.

    I had cut myself on a piece of glass on my foot a few days before, so while they were finishing the walk I took the opportunity to search the area some interesting species to photograph. 

    One of the species that caught my attention the most that I found on this walk was this Porcellio woodlouse that showed striking ornamentation. I took several photos trying to show the important characteristics for later determination and the rest was the work of the small group of researchers and isopod specialists who discussed the ID of this observation.

    El Hierro is the second-smallest island in the archipelago, and it is both the one with the fewest number of observations on iNaturalist and one of the least studied at a scientific level. Just in two small trips to the island, including this one, a dozen new records have emerged (which will have to be verified in most cases with the collection of some individuals). 

    I reached out to one of those identifiers, @pepe_rando, about his identification and this observation. Pepe tells me there are twenty-three known species of Porcellio in the Canary Islands, most of which are endemic. 

    Porcellio ombrionis is only present (at least so far) in El Hierro and La Gomera. It was first described by Albert Vandel in 1954. It is characterized (to the naked eye) mainly by its granulations, which are distributed all over the body, and are quite developed. And by its head lobes (the three "horns" the animal has on the head), specially the central one, which in this species is quite developed and curved upwards. This last characteristic is what made me rethink my ID of the observation. I had originally thought that it was Porcellio studienstiftius, which is an extremely similar species, but smaller, and with a central lobe that is not as developed, but when I took a second look at the second picture, it is fairly clear that this specimen's central lobe is way too developed for it to be Porcellio studienstiftius. In general, both species are characteristic enough to automatically rule out any other species from the islands, so it had to be one of those two.

    Oniscideans are an often overlooked group. Maybe this observation is only relevant to a few of us who are addicted to looking at pictures of these creatures for hours trying to figure out exactly what species exactly they are, but I like to think it will be useful for anyone who wants to look for the species in the future, be it to study it, or just for pure pleasure (as is the case with me), and can now do so through these pictures. This observation is (most likely) the first picture ever taken of this species. Maybe seeing pictures of these unique species, rather than some drawings from seventy-five years ago, will spark interest in protecting them, seeing as arthropods are usually underrepresented when it comes to conservation efforts. Many of the Oniscidea species from the Canary Islands still have no pictures that can be found online, but just recently, thanks to iNaturalist, we now have pictures for 8 species (1 Ctenorillo and 7 Porcellio).

    Amanhuy and Andrea joined iNaturalist just over two years ago.

    We discovered iNaturalist a bit by chance, when finishing our master’s degrees. We started by identifying some photos that Andrea had taken over the years and soon I was hooked and ended up buying a camera so I could take photos too. iNaturalist has managed to increase the love we have for nature and, especially in my case, it has made it our obsession to search, investigate and photograph the different species we have on our islands. 

    We owe much of the knowledge about the species we have in our territory, especially arthropods, to iNaturalist and the wonderful people who participate in this platform. I have spent hours reading scientific articles and interpreting keys just to try to identify some of the species we have photographed. We are generalists in terms of the species we photograph, although we have more interest in fauna than flora. Andrea has a special interest in birds, while I have found small life both in the sea and on land fascinating, with a special weakness for nudibranchs and the fly families Asilidae and Tephritidae, as well as lace bugs.

    (Some quotes have been lightly edited for clarity. Photo of Andrea is above, Amanhuy below.)


    - some iNat users are part of the American Isopod and Myriapod Group (AIMG), which has some excellent tips for photographing isopods for identification! Check out their project.

    - take a look at the First Known Photographs of Living Specimens project on iNaturalist!

    Posted on February 6, 2024 08:44 PM by tiwane tiwane | 11 comments | Leave a comment