Hi everyone, thanks so much for being here today! Feel free to say hello in the chat - let us know why you’re interested in this topic and where you’re joining from. We’ll get started in a few minutes

Hi everyone! I’m Ellie from WILDLABS. - excited to see all of you here to chat about movement ecology!

Hi everybody--I'm David Savage, finishing up my PhD at Purdue University and I'm here in part because I think it's interesting and in part because I'm in the middle of applying for a postdoc working on movement ecology and want to get more up to date on it.

Hi everyone! I'm Sareach from Cambodia. It great to join the training

Hi, I'm Tony from Portugal. I am a data scientist and am interested in what new developments there are in ML and conservation

Hi everyone! I am Lars from Denmark, tracking muskoxen and arctic foxes in Greenland!

Hi everyone, my name is Ukarapo from Namibia. Excited to be here and learn more about data analysis.

Hello! I'm Sandrine from France and I'm curretnly working on movement and behaviour ecology of porbeagle sharks

Hello! I am a PhD student at Washington University in St. Louis studying lemur movement ecology

Todd from Washington state in the U.S. I am an environmental researcher studying the application of environmental technology.

Hi everyone-- Im Emma a PhD Studend in Troms

hi, I am Dibyendu from India. working on movement ecology of old world vulture in India, both residents and migratory species.

Hi everyone. I'm Becky Hodgkiss joining from the UK. I work in offshore wind, providing advice to developers on how to assess the impacts of offshore wind on seabirds. I am interested in how advances in tracking technology can help fill evidence gaps and better inform these assessments.

Hello everyone, I'm Tuğçe from Istanbul/Turkey. I'm a master's student and I am interested in migration of bats and bat conservation.

Hi! I'm Sara, just about to finish a PhD at Caltech in computer vision and interested in movement data as a component of multimodal biodiversity monitoring systems!

Hi All, I'm Carrie joining from Cape Town. I work with camera trap, GPS and accelerometer data to quantify animal energetics in the wild (mostly large carnivores!)

Hi there, Robin from the Austrian Woods. Lot´s of wildlife movement data around here ;-)

Hello there! Rick Jordan, system administrator for the St. Louis Zoo, in St. Louis, MO, USA. We have several projects in the works and this discussion is relevant :)

Hi everyone! I am Felipe Bufalo, from Brazil. I am currently studying wild black lion tamarins’ movement and decision making processes during my PhD 🙂

Hi everyone! Shir from Israel, doing my PhD at Tel Aviv university in computer vision and zoology, mainly interested in fish movement and behavior 🙂

This is Ross from Taiwan. I am studying migratory ecology of Chinese Sparrowhawk, and habitat use and movement of Australasian Grass-Owl.

Hello everyone, I am Nick Porter a Fisheries Scientist from Idaho. I work tracking Chinook salmon in the Pacific Northwest to monitor habitat use and survival. I am interested for a chance to cross train and learn from other researchers.

Hello, This is Sankarshan from India. I currently work with WWF-India on Grasslands and Ungulates. My recent inclinations towards Animal behaviour and movement bring me here!

Hey I’'m Oorjit , a high school student based out of India. I am an aspring social entrepreneur looking forward to utilise technology for environmental impact and conservation . Looking forward to having an enriching experience ahead !

Hi! I'm Emmanuel from Israel. Working with ATLAS (reverse-GPS) and fruit bats (Rousettus aegyptiacus) in Israel. Final year PhD. Excited to be here.

Hi, Tobias here from the Firetail team from Unterhaching, Germany (Tracking and Acceleration Analysis software)

Hi everyone, Lisa Davenport, from Cedar Key, Florida and associated with UF Gainesville and sometimes James Cook in Cairns, Australia. I study intra-tropical migrant birds, mostly in South America. About to publish some data, so looking at what newest analyses are.

Hi All! Scott Yanco, postdoc at the Max Planck - Yale Center for Biodiversity Movement and Global Change

Hi! I'm Kenady in the USA. I'm a data scientist and product manager at Wildlife Computers. Lots of telemetry data from all sorts of critters over here :)

Good day from Costa Rica. I am starting a company on data, database applications for nature conservation. See https://rubberbootsdata.com

Hi, Robin from the UK. I'm interested in interpreting accelerometer data in gps trackers to show behaviour on interactive maps.

Hello! I’m Fernanda from Brazil, and I'm starting my Masters with Movement Ecology of the Lear’s Macaw, an endangered species endemic to the Caatinga Dry-forest in Brazil

Hi everyone-- Im Emma a PhD student in Tromsø Norway. I use biotelemetry to study the influence of multi-scale divers on top-predator predator movements and behaviours

Hi everyone, I'm Dimitri a researcher from the University of Pisa. I study migratory ecology of several wader species

Hi everyone, Andrew from Munich Germany. Researching urban biodiversity and green infrastructure.

HiI am Vanessa from Frankfurt Germany.I just finished my master thesis (Data Science) which was on the individual identification of leopards from unlabeled video data, in which I also used the Mega Detector and DeepLabCut and other components.Maybe a future PhD student, if the proposal gets funded for the topic AI in wildlife conservation.Curious to learn more and meet people ;)

Hi, I'm Giacomo: Computer Science Student from Munich, Germany. Very interested in computer based solutions for conservation and movement ecology

Hi all, I'm Samuele, a Master student at the University of Milan in Italy. I study foraging and non-breeding ecology of the Lesser kestrel, a small colonial raptor

I'm António Ferraz, with JPL at Pasadena, CA. I'm a Remote Sensing scientist interested in integrating remote sensing of habitats with animal movement data. Also, currently working on a project that focuses on the gaps in current tag tracking technology and seeks to identify new tech to fill those gaps.

Hello, I'm Callie a postdoc in the US. Studying bird migration

Hi all! I'm Beni Bar-Gera, a master student at the University of Applied Sciences in Zurich, Switzerland, focussing on wildlife management and modeling mammal behavior based on acceleration data.

Good luck Sara!

Our WILDLABS AI queen!

All the best for next week Sara!!

Nice introductions. I am Wouter Vansteelant, previously mostly studying migration behaviour of various birds of prey in relation to atmospheric processes. Recently started a postdoc at University of Groningen (Netherlands) where I’ll use bio logging data to study how Spoonbills and Godwits cope with rapid changes in wetland habitats along the East-Atlantic flyway.

Hi all, I’m Angie, a neuroscientist at The Kavli Foundation. I’m doing my own horizon scanning to understand the state of movement ecology and find potential places we can merge ecology and neuroscience. Always happy to speak about this topic if anyone is willing!

Amazing Angie!

Hi all. Please to be here. I’m Neil Hammerschlag from University of Miami. Focused on movement and behavioral ecology of marine apex predators. https://linktr.ee/DrNeilHammer

I love that the graph starts in the BC!!

Hi all! Don’t forget to drop your questions for speakers right here in the chat - we’ll call on you during the discussion when it’s your turn!

Hi everyone! I’m a prof that works at the intersection of neuroscience, engineering and computer science - joining you from Geneva,🇨🇭🏔

Thank you for DLC, Mackenzie 😄

Hi all, pleased to be here. I am studying dolphin responses to tourism activities filmed with a drone in Sao Miguel, the Azores, and were working on tracking dolphin movements. I am a PhD student in Wageningen University, the Netherlands. https://azoresdelphisproject.wixsite.com/project

🙏 you’re so welcome!

challenge accepted!

Hi all, my name is Silvia, I work on 3D animals from a computer vision perspective, from Italy

The STRANGE framework for mitigating sampling biases in, and improving the generalisability of, animal behaviour research applies to bio-logging studies. We are currently working on an article that explains how/why -- please reach out if you want to find out more. https://doi.org/10.1038/d41586-020-01751-5

Hi Svenja, I remember you! Great to see you here!!!!! Hug Fadia

Thanks for sharing Christian! Christian’s also on our advisory committee for the project :)

Hi all, my name is Dennis Kim, a PhD student at U of Minnesota from Fieberg lab, focusing on how to quantify the effects of memory in animal movement

Sara, can you put that email address in the chat?

We've got a decent framework for incorporating physiological data (eg heart rate) with gps tracking for humans--just ask anybody doing endurance sports; to what extent are folks doing this in the movement ecology space?

Great introduction, thank you

aiforconservation@gmail.com

Yes I also need the email.

aiforconservation@gmail.com

David, how are you attaching the sensors for hearrate monitoring?

Awesome intro thanks Sara!

The CERES wildlife tag https://cereswild.com/ is another new example using realtime analysis. Used by Giraffe Conservation Foundation https://giraffeconservation.org/

Rob, I'm talking about things like smartwatches, fitbits, etc. I'm not a movement ecology person, but I am an endurance athlete and a brief google scholar search didn't reveal a lot in that space for wildlife at least that I found so I was hoping to learn more.

Hey David. There are a number of us working on incorporating heart rate into animal movement -- the biggest limitation for us is usually having to implant heart rate loggers or send them through the digestive system. It's hard to keep HR sensors on animals in a place where they will stay still & keep working -- but it would be amazing if there have been advances on this.

Ok cool, I've had some friends deploy fitbits on animals with mixed results. It's a good idea though. I wonder if others have used something different (or the same) with greater success?

For folks interested in AI we also have an AI for Conservation group on WILDLABS - check it out here: https://www.wildlabs.net/groups/ai-conservation

@David try searching for the StarOddi HR monitor or Vectronic FIWI rumen sensor

We also have an AI Office Hours program set up for one-to-one support for conservation scientists from leaders in the field, coming back later this year: https://www.wildlabs.net/community-announcement/ai-conservation-office-hours-2021-review

Star:Oddi are amazing!

Depends on how big and 'hearty' your study species is. See this study implanting aftermarket HR loggers into black bears: https://link.springer.com/article/10.1186/s40317-018-0157-z

hi everyone, I'm Lilian Musembei. just graduated my masters degree in animals genetics and breeding, university of Nairobi, Kenya . passionate about research and wildlife conservation. interested in pursuing my PhD , though yet to get a scholarship. any help will be highly appreciated . Linnetmusembei@gmail.com here to learn more

Hi all, I'm a PhD student at Stanford studying large predator movement ecology and data science. I'm working to synthesize remote sensing and biologging techniques to understand spatiotemporal habitat partitioning across age classes in Pacific white sharks. Thanks for the great presentation, Sara!

Hi Lars, good to see you again. I see the Star Oddi is an implant...that usually limits what it can be used on, at least in my experience...

Great visualisations!

@Rob oh yes! Lots of limitations! The rumen sensor is only meant for ruminants ;)

@David Savage, it also depends why you want to monitor heart rate - if it is to study "fitness" or physiology then it probably needs to be an implant but if you're looking to measure energy expenditure, activity intensity, or behavioiur then accelerometers can do a better job

It's quite a small looking implant though, relatively speaking...

Ok phew...I was wondering what animals are regularly moving at >150km/hr!

Hmm so could these types of interactive diagrams and analyses be used to determine potential transmission routes and contact rates with animal diseases…?

Rob it’s lovely to have your commentary back we missed it last event

Hi all, I am a machine learning engineer with Global Fishing Watch. I am interested in developing resources to support bycatch research. Thank you for these great talks.

yeah I'm just thinking of how easy it is for me to generate something like a TCX or GPX with a spatial track, plus things like temperature, heart rate, power, cadence, etc. If we can do this for people we should be able to figure out a way to do it for wildlife.

Thanks Talia!

Man, ORTEGA looks awesome!

I agree! A bit more sophisticated than GPS Visualizer (no offence GPS Visualizer)!

@David yes it should be possible but the modern light sensor based HR watch monitors are optimised for rather naked human skin - and then there is the HUGE question of power consumption. How often do people c´charge the GPS watches? Solar power or other forms of energy harvesting could be a neat thing to help.

David, I saw a video on Youtube that showed a perpetual power generator that used magnets and crystals and it seemed pretty convincing to me, so maybe we could use that for power?

sounds like a great plan Rob, I'll put in a grant to the NSF to try it. You want to be on the proposal?

Ok great!

The quality wildlabs collaboration we love to see

ancient aliens were the real tech for wildlife all along

😄

Pyramids...all tracking devices should be shaped like pyramids!

We’ll save screen caps of this conversation for when you guys win an award for revolutionizing science

👍

A miniaturized perpetual power generator? Jump on the DARPA gravy train for that one 🙂

I feel ashamed looking at the quality of these graphs!

I agree Rob these graphics have been fantastic.

Thanks Somi for this great talk and for developing ORTEGA and your other contributions not shown today. BTW, our figure illustrating how resolution affects inference from movement data was published in Science (not Nature), see https://www.science.org/doi/full/10.1126/science.abg1780

Some of our Austrian red deer projects used the FIWI rumen sensors. It´s quite tricky to make them swallow the bolus. They were just to big for my small roe deer ;-)

@Ran! Thanks so much for your comment and for correcting me! Of course.

We have used the FIWI with mixed succes on muskoxen. Some of them spit out the sensor! We are planing to develop a new device to deploy it.

Hi Robin - having spent most of the winter trying to dart red deer from the ground, that sounds very appealing! What sort of information do you get from the FIWI rumen sensors?

@Lars The weight of the Bolus just has to be right to stay at the correct place in the rumen.

our plan is to place it in the rumen with a tube BEFORE they wake up and need to swallow it

Christen, when you say cross validation, is that using actual data from the bear collar (or some other movement/detection data)?

Apologies for talking so long!

Hi @Eilidh lots stuff. The rumen temperatur gives you info on feeding (colder forage) and core body temperatur (energy saving in winter etc). plus the heartbeat

@Christean: Where do we find the tools to run these analyses?

I'm happy for Chris to get into the mathematical mud for me.

No apology necessary. it was interesting.

your stuff was amazing Somayeh!

https://move-ucsb.github.io/covid19-mobility-vis/

Lovely tapirs!!!

No no! @somayeh, it was fantastic!

Dynamovis: https://github.com/move-ucsb/DynamoVis

Thank you all!

Yes all good Somayeh! Thanks so much for the interesting talk, we’ll get to discuss more in the open chat part

phwoar that's pretty powerful stuff!

I was happily watching the clock on my screen saying 8:47, and thinking cool I have more than a minute left. Not knowing that the clock was showing the time 8:47 am!! 🙂

Question for Chris: Do you think we could inform stochastic movement process between GPS locations with other sensors on the animal (like accelerometers and heart rate monitor) to be able to derive the most probable complete track?

We are hearing from Motus next call!

ah MOTUS! noice!

Thanks Robin, I'm looking at disturbance so GPS and accelerometer tech pretty advantageous (but incredibly difficult to deploy in UK), but I wonder if body temperature could indicate flight responses i.e. if they are running, their temperature would go up...

@Neil: GitHub page for the ctmm package: https://github.com/ctmm-initiative/ctmm

Christen, would this important distinction between modeling movement as discrete or continuous in time vanishes once movement is collected at high spatial accuracy and temporal frequency?

@eilidh bear in mind the FIWI measure HR during resting state https://www.vectronic-aerospace.com/heart-rate-logger-fiwi/

@Eilidh That was exactly idea: Lots of disturbance = higher movement rate = higher heartrate and body temperature = more energy needed in the snowy Austrian alps

speaking of accelerometers and complimentary sensors has anyone experimented with dead reckoning? I've been interested in for years, but not aware of much in our fields...

100% re: extracting more scientific insight from the data we have!

Thank you - I LOVE CTMM!

Thank you Christen

Rory wilsons group has been working on dead reckoning in birds, but a magnetometer is needed, too

@Rob nope! It seems like you need quite a high fix rate for correction. The upcoming snapper GPS might be nice for that

you can check out the R package Gundog.Trakcs

@Rob check out the work of Rory Wilson - his lab in Swansea does loads on dead-reckoning on wildlife and is a great human if you want advice/ collaborations

https://animalbiotelemetry.biomedcentral.com/articles/10.1186/s40317-021-00245-z

https://movementecologyjournal.biomedcentral.com/articles/10.1186/s40462-015-0055-4

@Chrusten: what kind of models you suggest to integrate accelerometer data and movement data (e.g., GPS), HMM?

oooh yeah, point and click!

https://biology.umd.edu/movement

@Robin - amazing, will look out for your pubs

Thanks @Carolyn!

Chris -- continuous ACC needs a lot of memory. What sampling rate do you think is minimum?

Ran you’ll be next!

Just a comment - we used ctmm in a 2021 article modeling Blue-throated macaws (Critically threatened, so we haven't released the data on breeding ground locations), and it was easy to use even for our R-challenged team

Has that comparative analysis of home range estimates been published? Link?

@Teague: this really depends on the questions you are asking. Tags typically start from 8Hz and higher, but for wing frequencies and similar questions you will likely need higher rates

@Hannah here's the big HR comparison paper https://esajournals.onlinelibrary.wiley.com/doi/full/10.1002/ecm.1344?casa_token=ZTSMfz8riaEAAAAA%3ALgq-3jRDVSItnjOVT1rum0UMc8bngPjRaHXOgiPPdGuFxr56M_drhg403GyHW-OY2ZIqoyuaL8CqFmM

@Roland, perfect thank you so much!

This is a great recent ‘practical guide' to HR estimates as well https://ecoevorxiv.org/23wq7/

OOHEEE Gunner, R et al. (2021) . . . How often should dead-reckoned animal movement paths be corrected for drift?

bedtime reading!!

@Roland, this review looks super useful, thank you for sending!

if you require acceleration visuals & segmentation with the possiblity to interactively refine predictions you may want to look into Firetail http://www.firetail.de -- in particular for large scale, high res data

Just so everyone is aware - we also include a link on wildlabs where you can access the chat after the call

So if you need to look back later at resources/links etc, you can access it there.

I can highly recommend http://www.firetail.de !

Hi @Christen, thank you for your talk! I was wondering how would you create a HR for a species that has one clear foraging location, and one separate roosting location, with limited locations connecting the two? I have had issues where one large HR is created, despite many data points at the two separate locations. Thank you!

no sound

Here are some websites for ctmm and ctmmweb: https://movementoflife.si.edu/analytical-tools/ https://biology.umd.edu/movement https://cran.r-project.org/web/packages/ctmm/index.html https://github.com/ctmm-initiative/ctmm https://github.com/ctmm-initiative/ctmmweb https://groups.google.com/g/ctmm-user

uh oh...back to Teams...

no joke, that's exactly what my kitten looks like!

man, this is going to put a lot behaviour caoding software in the bin!

All of the models in ctmm are technically HMM models. The models I think would be necessary for integrating accelerometer would be continuous-acceleration models (slightly more complex than the continuous-velocity models in ctmm) and need to be coupled with behavioral switching models, like from Blackwell's group. I have some mathematical results in mind that would make implementing these models computationally efficient (without MCMC).

Rob, we want to see the kitten and analyze its behavior!

Here’s the DeepLabCut community page: https://gitter.im/DeepLabCut/community?at=5dd30a1ac26e8923c44c2acf

Awesome stuff, thank you.

I think the minimum ACC sampling rate depends on the time interval of the animal's stride (or flapping) and also on the scale of the errors.

I would like to know why he loves to attack me all the time so I am keen @Ellie

As a general resource: CTMM, FireTail, DynamoVis, and a bunch of other tools for tracking/bio-logging data are summarized https://www.movebank.org/cms/movebank-content/software

@Christen: wow thank you so much! I will have some literature search to do

Ooh that’s awesome thanks Sarah

Nice talk!!!

Really enjoyed your talk @Christen -- you guys are doing an amazing job at pulling these relatively complex models to an inuitive level!

For the 3D model of the cheetah running did you need to use multiple cameras in order to get the 3D representation of the cheetah moving through space?

Seconding that Tobias! You also explained it all so clearly, I really appreciated it. I was so worried I would have to concentrate really hard to understand 10% of what you would talk about!

The cheetah model partly worked on wild leopard data from camera traps in my project :)

With a bi-modal home range, there are usually two associated movement behaviors, which implies a non-stationary process. Long term, I would like to tackle data like that with behavioral switching models. Short term, I would consider classifying the data (foraging versus roosting), calculating two separate stationary estimates, and then averaging them together with a hierarchical model (all possible now in ctmm).

inspiring talk, Mackenzie! Obviously two of the major limitations to apply these tools for wildlife in the wild are short duration and limited spatial coverage. What would be your most optimistic solutions for these limits (i.e. longest in time and largest in space) in the next 1-2 years?

Mackenzie for your ID success do you need something reasonably well marked, like a zebra, or can any animal be ID'd?

These models are awesome for captive animals like in enclosures in zoos or for farm animal monitoring - but isn’t it of quite limited use for wild animals spending only very little time in front of video cameras?

That was so interesting thank you

@Christen, I'm dealing with coarse resolution data in both space and time and I'm wondering how far I can dig into my datasets with mathematical tools used at higher resolution. (12 hours / 50km to more)

Not that I dont think yours were interesting rob and lars!

I think ran covered your lars to a degree

I get the hint Steph...

You can feed coarse data into ctmm and nothing will go wrong, because the autocorrelation model is selected. You just won't be able to estimate quantities like speed and possibly diffusion rate. That's important, though, because conventionally you would just proceed with dividing straight-line distance by time and calling that a speed estimate.

Rob in my experience and in lots of discussions with the WildMe team, biomarked (striped, spotted, contoured) species are easier to ID and the methods are deployable now, for less clearly marked species there has been some good progress and promising results but the biggest barrier I see is the challenge of building good evaluation sets for species that are hard or impossible for humans to re-ID, because without that we can't tell if our unsupervised or semi-supervised methods are trustworthy

So many interesting questions - if we don’t get to yours or you’d like to continue the conversation hop over to our Biologging group on WILDLABS: https://www.wildlabs.net/groups/biologging

Good points! It could be very valuable in ground truthing acc data for instance

@Vanessa — cool!!

Well, I am sufficiently blown away once again!!

@Rob — no, we show that we can reID even black mice without any markings

whoa! @Mackenzie...that's really exciting!

I can totally see using this on giant otters (that have individual throat marks) to improve our estimates of which subadult is sharing with young-of-the-year. Sometimes you get a ½-second glimpse of a small section of the throat before the share is complete and the otter giving the food away re-submerges. Want to try this with some ancient video 🙂

Thanks all so much for being here today! For those who have to run, make sure to register for our next event on Data Sharing & Archiving in movement ecology on May 25: https://www.eventbrite.co.uk/e/virtual-meetup-data-sharing-archiving-in-movement-ecology-tickets-337552026197

Thank you very much @Christen, I might a bit more questions on the use of ctmm and how to be enough cautious in interpretation!

Really neat and amazing!

Sarah Davidson at Movebank (also on our advisory committee) will be giving the intro for that one, should be a super interesting follow-up to this conversation

Thanks for the wonderful talks!

This is a huge obstacle to collaborative work--just fixing the bad incentive structures of academia. Thanks for emphasizing it Sara.

stupid granting bodies and Unis requiring data silos!

Agreed!

Thanks for coming along, everyone! Make sure you visit our brand-new WILDLABS platform to meet the conservation tech community and continue these conversations! https://www.wildlabs.net/

Woohoo Sara! That's what I'm excited about too. Can't wait to discuss that more at the next meeting ☺️

I didn’t talk about it today, but for joint modeling of data, we have worked on this for continuous time series data —> cebra.ai

We also just launched our new WILDLABS platform so go check that out as well :) Recording will be posted there soon too! www.wildlabs.net

Researchers: 'let's do collaborative work and share our data'Universities: 'But what if we can MONETIZE this???!?!?!?'

@Makeckenzie - Such a valuable tool! Is there a group-size limit to your method? Can we, for example, potentially track many roosting bats/birds?

EXACTLY @DAvid!

It drives me crazy!

I was so lucky to start my roe deer studies

with the Wonderful https://eurodeer.org/

And if you’re interested in sharing your expertise in discussions like these, make sure you check out our Groups and get involved! https://www.wildlabs.net/groups

Thanks everyone for the inspiring talks!

@Emmanuel - no limit; just pixel resolution of the animals 🙂

Just need to be able to see them 🙂

Thank you to all the speakers.

Of course, there are limits to DL approaches (see our primer Mathis 2020 Neuron)

Thanks to all our speakers (including Talia)!

THANKS - all the talks were very inspiring

Nice talks! thank you everyone

Thanks for the great and inspiring talks!

Thank you for organizing this and to all the speakers for such interesting talks!!

Thank you to the speakers for all the amazing talks!

Ha thanks Rob!

Thanks very much.

thanks for a great event all!

thanks

Thanks everyone!!

Thank you

Thanks to al speakers