Computer simulation helps test land management techniques to help bumblebees

https://www.exeter.ac.uk/news/research/title_872443_en.html

by rbanffy • 5 years ago

143 28 5 years ago

exeter.ac.uk

mutagen• 5 years ago

Website and models available at https://beehave-model.net/

Original model implemented in NetLogo (as described on the page).

Some papers I found looking for the model:

https://besjournals.onlinelibrary.wiley.com/doi/epdf/10.1111...

https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111...

Wish I had some time to explore making a game out of something like this while preserving as much of the accurate simulation as possible.

throw149102• 5 years ago

I spent a few minutes looking at this "BEE-Steward" and it's corresponding models, BEEHAVE and BEESCOUT, and the underlying NetLogo language that is used to model these things.

I guess I immediately have a few questions - How do we know that the model corresponds to reality at all? I mean, I know we have experiments to confirm what impacts the bees, but it seems like there could be a massive amount of variety depending on different factors. I'd love to see some research showing that BEE-Steward can work in real scenarios. Right now, they just have a simulation showing "hey, if we change x and y we can get 5x more bees", but does that actually correspond to reality?

Secondly, how do we make farmers actually want to use this? Do they get paid for taking care of bees on their land? I know this is just a policy question, but it's still important to answer. Do farmers need any training to use this? Can we collect data in an automated fashion to reduce the work the farmers have to do? This almost sounds like a deep learning problem - use a camera to map + classify areas of farmland, how many pollinators they have, the presence of bees, etc.

Finally, what is the fundamental problem of bee modeling? Is it fundamentally a computational problem (where we know the rules, but computing the model can be hard) or is it a data problem (where we know how to compute how the bees will behave, we can do it efficiently, but coming up with rules that bees actually follow is hard)? I've noted that the NetLogo environment by default will only take up 1GB of RAM, which is already not very much considering the JVM itself will take a pretty big chunk to run just about anything.

It would be interesting to start building some sort of "National bee model", rather than just working farmland by farmland. Is there any project like this in the works?

Either way, I'm sure all of my questions and more are already being asked/answered by people much more involved with the conservationist field. Just my 2 cents.

rolleiflex2• 5 years ago

Do they get paid for taking care of bees on their land?

In the U.K. at least, they do. The government pays you to leave some part of your field unused and untouched. It’s called ‘wilding’ and it’s well paid enough that the subsidy sometimes is worth more than any reasonable crop that would grow on that patch of land.

Jeremy Clarkson’s Farm (of Top Gear fame) had a segment on this. My take is that farming is a very subsidy-heavy business.

acatnamedjoe• 5 years ago

Adding some slight nuance to this as it sounds like Jeremy Clarkson might be oversimplifying a tad.

Nobody in the UK (or in England at least, i'm not as familiar with devolved policy) gets paid for full rewilding (in the sense of taking land completely out of production and leaving it, a philosophy most famously espoused in England by the Knepp estate). English agri-environment policy expects some kind of active management of the land taken out of production. There are a range of options farmers can choose which they have to discuss and agree based on the land type and context with a specialist adviser. For bees this might be cultivating a plot of plant species chosen for pollen and nectar, for example.

The payment is calculated on a model based on "income foregone plus costs" so in theory should be worth exactly the same as an equivalent crop on that patch of land (although in practice this obviously varies a bit depending on how marginal the land is and fluctuations in the market). Generally the amount of admin and time required, and the limited funds available, means that nobody is making very much money off this.

The much more broken UK subsidy is the Basic Payment System - a leftover from the EU Common Agricultural Policy. Here farmers are paid a flat rate per hectare for any land used for agriculture - just cash handed over with no incentive to either improve productivity or manage for the environment. For obvious reasons this disproportionately benefits massive landowners. This is thankfully currently being phased out and the funds shifted towards the agri-environment model, which has its problems but at least delivers something.

SamBam3• 5 years ago

Agreed. Allow me to be skeptical, but I find it hard to believe that less than 4000 lines of NetLogo are going to accurately model a system as complex as bee ecosystems.

I say this as someone who has spent many years writing ecosystem models for schools (some of them with NetLogo). Toy models often require more than that to be useful.

You'll have a set of assumptions. You'll have a number of formulae that you can individually say there is experimental evidence they roughly models reality, if everything else is held constant. But you almost certainly do not have the experimental evidence to say that when these six formulae are interacting together, they will behave in known, deterministic ways.

The paper does not appear to show any applications where researchers were able to predict something novel using the simulation, test it in the real world, and see how well it matched the results. Their "results" (if I'm interpreting correctly) appear to be mostly summarized in figure 3, which seems to be created by trying different settings until the model very, very roughly matched some historical dataset. Someone correct me if I'm wrong.

troelsSteegin2• 5 years ago

Is there any literature you'd recommend on the disjunction between credible ecosystem models and toy models? I don't disagree, I just want to know more about the middle space between toy on one hand and authoritatively callibrated on the other.

If there is one single person to read on system dynamics modeling, it's John Sterman. Also, the Systems Dynamics Review (SDR), https://systemdynamics.org/system-dynamics-review/ , is what I think of as the journal of record. I am also aware of some work by the Concord Consortium for simulation in education. What seems to get written up is either high effort/high cred (SDR) or edu/toy.

artificialLimbs2• 5 years ago

> Do they get paid for taking care of bees on their land?

A friend of mine is a beekeeper in one of the northern US states. Bee keepers there keep bees on farmland in exchange for most of the honey. The farmers generally get more than they can personally eat. Sometimes keepers will take a farm from another keeper by offering more honey/a sweeter deal (ha!). Farmers also get crops pollinated for free, so it’s a win/win for them.

mcdonje• 5 years ago

Cool project. Confusing headline. They made models to help research land management strategies that would be potentially helpful for bees.

shapefrog2• 5 years ago

Or did they create a matrix for the bees? Somewhere we could upload their consciousness to where they can live happy bee lives, presumably every bee gets to be the queen in this safe space.

ASpaceCowboi• 5 years ago

I'll tell ya whats next, VR Blockchain for Bees!

WaitWaitWha• 5 years ago

All bumblebees (genus Bombus) are queens. You are thinking honeybees (genus Apis), where mostly there is a single queen to a hive.

They use a different matrix.

dang• 5 years ago

We've updated the title to try to make it more accurate. Thanks!

mlindner• 5 years ago

Horrible naming. It is simultaneously, inaccurate, misrepresents things, and then also brings up unrelated political concepts to mind all at the same time.

hinkley• 5 years ago

This brings to mind the old quip about engineering and designing a bridge that stays up - barely.

There's a fine line between improving survival and finding the absolute bare minimum to prevent loss. Situations change, and red-lining nature is always, always going to end in unpleasant surprised.

Rochus3• 5 years ago

Well, it's a simulation. You can also use your time and resources to provide suitable plants for the bees in reality; http://hummelgarten.ch/dokumente/ has been very useful to me; there are tutorials and tables on how to best plant your green spaces that the bees are well taken care of throughout the year; this also works in the city.

artificialLimbs2• 5 years ago

I can’t imagine planting enough flowering plants (that don’t get eaten by… something…) to make a difference to the bees. I also can’t read, what is that, German? I live in the southern US. Bees have a 3-5 mile flight radius for collecting pollen when they’re in a hive, and upward of 60-80k bees in a decent size hive. I imagine you would have to do a LOT of planting to make a difference.

adrianN• 5 years ago

Small islands of appropriate flora where insects can recharge can provide valuable bridges between fractured habitats and can help stabilize populations.

Rochus3• 5 years ago

That might be true if you focus on honeybees; but e.g. bumblebees only have 20 to 200 animals per colony and are also the much better pollinators than honey bees; bumblebees also work in bad weather and collect on more plant types (many agricultural plants such as tomatoes are pollinated only by bumblebees); and honeybees - unlike bumblebees - usually have someone to take care of them and pour sugar water when they don't have enough to eat. So you may very well make a difference if the bumblebees find useful flowers in your garden at a time when everything else is mowed down. The referenced site is in German, but it has also some English documents and tables, such as e.g. http://hummelgarten.ch/wp-content/uploads/2018/11/Anleitung-....

slavik81• 5 years ago

A few years ago, I helped build a (tiny) game based on this subject for a hackathon: http://www.diversibee.org

Computer simulation helps test land management techniques to help bumblebees

Related Stories