Login

Join for Free!
112509 members


Conditions for a stable ecosystem? (for a simulation)

Discussion of the distribution and abundance of living organisms and how these properties are affected by interactions between the organisms and their environment

Moderator: BioTeam

Conditions for a stable ecosystem? (for a simulation)

Postby Smig » Wed Dec 10, 2008 1:04 pm

First of all, sorry if this is the wrong category, I have absolutly no academic knowledge of biology, I simply have an interest in the subject.

I've been developing an application in the past week that would simulate a stable ecosystem where many independent elements would need to compete for resources to survive and reproduce. I also wish to simulate evolution by introducing slight random mutations in their qualities and hopefully natural selection would play it's role. This is already happening to some extent, but my problem is still the most basic one, I can't get the ecosystem stable.

I have 3 layers, plants, herbivores and carnivores. The first results I had were predictable, without a constraint on the resources that plants had, all 3 layers would grow exponentially until my cpu was slowed to death. Then I've limited the amount of plants depending on a ratio with the screen space, this would simulate the space and non-organic resources for plants. After that, it seems that the results of tuning their caracteristics and beginning ratios, are always one of two. Either they all get extinct in the first generation or two, or herbivores thrive until plants reach their limit and cannot sustain them, when herbivores get a higher ratio than plants (and plants number start to drop exponentially) carnivores enter that spike too due to abumdance of food, eventually depleting herbivores capacity to reproduce at that speed. At that point herbivores are already dropping massively in numbers due to lack of food and surplus of predators, followed by the predators. Now, some times they all survive the first cycle by just a few units and a new cycle is possible but usually one layer gets completly wiped out destroying the ecosystem.

I thought this was simply a matter of fine tuning but if I start to restrict some conditions the oposite happens and they can't sustain their own growth so it seems I can't find any point in between, either they're successful and will only stop growing when it's too late or they're not successful and die.

I've been trying to avoid the temptation of introducing artificial restraints (like not allowing reproduction of carnivores if their ratio to herbivores is dangerous to the system) so I've been thinking, what am I missing here? What is there in place in nature that I cannot simulate? Is the reproduction rate of the top layers somewhat fine-tuned so that no matter the amount of food they have, they won't overflow the pyramid (since this would ensure their own survival)? Or maybe this destructive cycles I'm experiencing are much more common than I think?

This also introduced another question to me. Are really the top layers supposed to act as mechanisms of control to the bottom layers? How relevant is that factor? Would an ecosystem be destroyed if the top layer was removed or is it supposed to survive?
Last edited by Smig on Thu Mar 26, 2009 4:27 pm, edited 1 time in total.
Smig
Garter
Garter
 
Posts: 19
Joined: Wed Dec 10, 2008 11:52 am
Location: Portugal

Postby biohazard » Wed Dec 10, 2008 1:27 pm

Wow, you've got a pretty ambitious project going... good luck with that!

In the nature, some populations indeed experience such peaks and subsequent collapses, which go like you described: the carnivore population follows the same trend than herbivores, but with some degree of lag. However, complete wipeouts rarely happen, because the nature has had millions of years of time to fine-tune its systems.

One crucial factor there seems to be missing is disease: the more dense a population is, the more susceptible it is to various forms of diseases. An ecosystem does not always "need" an apex predator (or any carnivore), lack of food and/or diseases alone limit the maximum number of herbivores. Heck, an ecosystem doesn't even need herbivores if we go nitpicking. But if it has herbivores, then when their numbers start to decline because of the forementioned reasons, plants have more time to regain their numbers. On some isolated/small areas it has sometimes been described that the herbivores actually manage to chew out all the plants and the whole population dies. Then it is later refilled by same or other species from other reservoirs. On a general rule, though, predators help keep the herbivore populations in reasonable numbers and thus also in reasonable health - the fit survive and the ill get eaten, et cetera.

All in all, any ecosystem probably has a myriad of variables no matter how simple it is, so simulating one successfully is a real challenge!
User avatar
biohazard
King Cobra
King Cobra
 
Posts: 776
Joined: Sun Apr 08, 2007 6:45 pm

Postby biohazard » Wed Dec 10, 2008 2:01 pm

About the scenario where a whole population of animals manages to wipe themselves out:

There is an island off the coast of Canada (fairly large, fairly far away from the mainland - unfortunately I don't remember it's name), where frequently deer and wolves do this feat.

During the winter, some deer cross the ice and thus end up to the island. The ice then melts and the deer get isolated. The remaining population then greatly increases because there are no predators, but eventually the deer eat out all the available plants. This may take a few years, and if the food runs out during the summer when the island is cut off the mainland, the deer die of famine.

So what about the wolves? On occasion, a pack of wolves happens to wander on the island during the winter. There is a lot of deer available, so the wolves stick around. And the same happens again, but this time only the wolves hunt down all their prey and then die themselves of hunger.

Finally, the island's plant ecosystem is also severely damaged by the deer invasion, so it takes again couple of years before there is enough of vegetation for the deer to actually stay on the island for longer times, so this scenario can only happen now and then.

I tried to google the actual article behind this story but didn't find it. If someone knows the url, go ahead an post it here. But anyway, maybe your ecosystem is a bit like this at the moment, Smig? ;)
User avatar
biohazard
King Cobra
King Cobra
 
Posts: 776
Joined: Sun Apr 08, 2007 6:45 pm


Re: Conditions for a stable ecosystem?

Postby Treefrog » Tue Dec 23, 2008 10:15 am

Hi Smig,

Wow that does sound ambitious!! I've just started learning about ecosystem modelling so you'll have to bear with me....

The first thing to think about would be what you want to learn from your model - there are two approaches to modeling as far as i can tell - First, to create complex models that incorporate every interaction within an ecosystem. These are realistic but too complicated to be able to see the mechanisms that cause the reactions. Second, you could strip this model down to its key components - not so realistic but easier to understand how they produce the outcomes and therefore understand the underlying mechanisms at work - i.e why your populations go extinct.

Thing to incorporate (depending on how complex a model you want)
- Growth rate - Maybe the recovery rate of the plants / herbivores is not fast enough? Preditor-prey cycles don't usually mean extinction because while the preditor numbers are reduced, the prey has time to recover.
- Demographics (ie. Survival, dispersal, fecundity) - these will give you your growth rate. They will be different for each organism at leach stage of its life. You can add this in by creating an 'age-structured model' - (a matrix of how much each life stage contributes to the next stage, given the survival rates etc.)
- Threshold numbers - Depending on the demographics / interactions of each population, there will be a limit on the number of survivors, below which the population will go extinct. (there won't be enough reproductive ability to increase numbers fast enough)
- Genetics - You've already incorporated mututions which lead to natural selection - at very low populations, 'inbreeding' would have a big effect on which mutations increased in abundance - it wold increase deleterious mutations and reduce the fitness of the population - not sure if this is worth putting in?
- Weather - Abiotic factors also have a big effect on ecosystems. Environmental variation affect the whole population, albeit in different ways, but could add another element of density dependance, therefore reducing the rate at which the animal populations grow, giving the plants time to recover.
This does mean adding an element of randomness to you model, which would make it more complicated but more realistic!

I've only done this on paper - so not sure how you would program it!

A really good book if you can find it is 'An introduction to ecological modeling' by Michael Gillman - it goes through each interaction and how to analyse it.

Hope that's relevant, i'm revising models at the moment so you'll have to excuse the essay!!
Treefrog
Garter
Garter
 
Posts: 6
Joined: Mon Dec 15, 2008 5:14 pm

Re: Conditions for a stable ecosystem?

Postby JorgeLobo » Tue Jan 06, 2009 1:35 am

Overly ambitious for someone who admits no knowledge of biology.
JorgeLobo
Coral
Coral
 
Posts: 424
Joined: Mon Nov 17, 2008 12:12 am

Postby AstusAleator » Wed Feb 11, 2009 12:44 am

Sounds like a fun project, one I've daydreamed of undertaking.

From reading your initial post, it seems to me that there is little friction in your model. No that's not an ecological term, but I'm using it because it makes sense. It sounds like every trophic level is able to operate with zero resistance (ie competition, disease, environmental hazards, etc).
I know that every factor added to your application will greatly increase its complexity, but if you want to decrease the severity of these boom and bust cycles, you need to add some resistance to the organisms' abilities to carry out their functions.

Probably the simplest way to achieve this is to add an "environmental" factor to each troph. So for the veg layer, program x% chance that a plant will die due to causes other than herbivory. For the herbivore layer, program x% chance that the animal will die or not reproduce due to causes other than predation. And for the predator, program x% chance that the predator will die or not reproduce for some reason other than lack of food.

And finally, as someone else has suggested, make sure age is a factor in your program. These organisms need to have lifecycles in which they mature, breed, and die - or the model fails. Keep in mind that the lifecycle of an herbivore will be typically longer than the plants it eats, and the carnivore's lifecycle will be comparable to the herbivore's. Also fecundity was mentioned (number of offspring). That's important. Plant lifecycle may be shorter, but it's capable of producing many many more offspring if conditions are right. Herbivores and predators may only produce a couple of offspring in a lifetime.

What you're trying to do is incredibly complicated. I'd recommend trying to find the simplest real-life example you can, and try to emulate it.
What did the parasitic Candiru fish say when it finally found a host? - - "Urethra!!"
User avatar
AstusAleator
King Cobra
King Cobra
 
Posts: 1039
Joined: Tue Jan 24, 2006 8:51 pm
Location: Oregon, USA

Postby JorgeLobo » Mon Feb 16, 2009 2:13 pm

I'm sure this is mere speculation but recommend learning biology if you ever attempt anything ike it.
JorgeLobo
Coral
Coral
 
Posts: 424
Joined: Mon Nov 17, 2008 12:12 am

Postby Smig » Thu Mar 26, 2009 4:25 pm

Thanks for your replies, I haven't realized that my thread was actually answered until a few days ago so I'm still digesting all input!!

JorgeLobo, I agree with your comments but unfortunately, an academic education on biology is out of the question because my professional career is already headed in another direction but I have a profound interest on the subject and I can learn a lot from this project. Besides, and most importantly, it's a fun project to think about in my spare time :)

I thought of replying to each post individually but since some subjects are similar I'll go through each subject instead:

Demographics:
Thank you for your suggestions, what I have is very simplistic but I think it's realistic enough at this point. There's a maximum age after which the organism dies, and there's an age when an element will become an adult. Before this age each element will follow the parent that gave birth to it as long as it's alive, they'll grow gradually in size and speed until the adult age which makes them vulnerable until that point, and they're unable to reproduce before that.

Growth rate:
Right now, this is what governs reproduction. Fitness checks are almost nonexistent, partners just need to be adults from the same species and the opposite gender. They need to be above a certain threshold of energy, otherwise they'll prioritize getting food instead. Newborns are born instantly just to simplify and the female (or asexual) element will need time to recover. This time is what I've been using to tune the growth rate.

Resistance to growth:
I actually thought about disease as a great way to control growth but I'm still trying an implementation that makes sense and is useful for the simulation at the same time. I've been trying this concept that each time an animal interacts with another, it's chances of dying from disease increases. This would make social animals that group together more vulnerable to disease than more loner animals. I don't know if this can translate in realistic results but this has been the only way I've been able to control the herbivore population. I've got this statistic that 3 in 5 animals in the African Savanna die from sexually transmitted diseases so I'm using that as point of reference.

Genetics:
I never thought of introducing something as detailed as you've suggested. I'm merely merging all properties of both parents and adding a small chance that they might increase or decrease by small amounts. The main properties are speed and weight, both influence the energy consumption needed and the range of prey/predators so a very fast and very big creature would have a hard time getting all the food needed to survive. The detail you're suggesting could be very interesting but for now I wouldn't meddle with something as detailed as that. I was actually going in the opposite direction, in a much less realistic but more entertaining way. I've been adding "special" traits that could be anatomical or behavioral adaptations like "climber", eyes on front, eyes on side, social, loner, etc. Of course this is highly unrealistic as each of the things I've mentioned represent many different genes that wouldn't just pop up like that but it could make sense if seen in a macro perspective of the passage of time and it would be fun to be able to customize new creatures that would behave much differently from one another. It might still be interesting to set the appropriate chances of certain types of mutations (deleterious) in certain types of reproduction events (inbreeding, mitosis?) but I don't know much about that.

Weather:
I also thought about this as a fun extra. When I thought of plants I thought of seasons, but I decided to think about that later on.


Lately, I've been following AstusAleator advice and I've been trying to simulate a well known ecosystem like the one in a savanna, consisting only of grass, a type of ruminant and a type of feline predator. Looking back at the time I made that first post I think I'm now in the right track but even though I have a more stable ecosystem, I'm still facing that challenge. I'll upload it to the web shortly to get more concrete feedback but feel free to mention anything that I might be doing wrong or things that are missing and could be useful.

Treefrog, thanks for the book you've suggested, I'll look it up!
Smig
Garter
Garter
 
Posts: 19
Joined: Wed Dec 10, 2008 11:52 am
Location: Portugal

Postby Smig » Sun Apr 19, 2009 12:31 am

I finally have something worth uploading so you can see exactly what I'm talking about but before that, I'd just like to say that I'm not sure if posting links is against some rule as it could be considered advertising. I couldn't find any rule against it and as you can see, there's nothing to advertise through that link since it's a non-profit and unfinished software. I'm truly seeking advice.

Here's the link: http://evolv.alturl.com

It's still buggy but you can see what I meant in my first post. Depending on what constants I change, I always get one of two results, either the carnivores deplete the herbivore layer or the carnivores can't get enough food to sustain themselves and all that remains is the herbivore and plant layer. To see this happening faster click on "settings" and then on "faster".

The results are rather variable because the starting placement of every element and trees is totally random. There's also some problems with path finding around trees and getting out of corners when cornered by a predator so these things also play in but I feel that regardless of some problems, I should still be able to get something more stable than this.

Any other ideas and suggestions?
Last edited by Smig on Thu Sep 03, 2009 10:33 am, edited 1 time in total.
Smig
Garter
Garter
 
Posts: 19
Joined: Wed Dec 10, 2008 11:52 am
Location: Portugal

Postby AstusAleator » Mon Apr 27, 2009 9:07 am

Smig
Very entertaining. I love the cartoonish feel. This is certainly something that - once refined - could be a useful educational tool.

Here are some suggestions:

Predator adjustments:
-decrease the rate of reproduction. This can be accomplished in a couple of ways. a) require a higher amount of energy before reproduction becomes an option. b) simulate population dispersal (predators don't really hang out with eachother - with a few notable exceptions). c) raise minimum reproductive age, and increase time between births.
-Prevent kin-mating
-Less predators, better pathing. Think pac-man. But prioritize targeting according to an herbivores relative distance from other herbivores (ie the outlier - the straggler). You might not want a predator to be able to bee-line across the map to a prey, but once it is within a certain radius of a prey it should get a pretty good idea of which way to go.
-Decrease chance of success. I know this is a factor you're trying to work into the evolution of the herbivores, but the initial value for success rate could be lower.
-Implement a general selection factor like I suggested before. If you use the suggestions above, I'd recommend keeping the frequency of this factor pretty low in predators.


I have more comments and suggestions but I'm really tired and need to go to sleep. I'll try to post more tomorrow.
What did the parasitic Candiru fish say when it finally found a host? - - "Urethra!!"
User avatar
AstusAleator
King Cobra
King Cobra
 
Posts: 1039
Joined: Tue Jan 24, 2006 8:51 pm
Location: Oregon, USA

Postby mith » Mon Apr 27, 2009 5:47 pm

I'm assuming you have a series of differential equations to simulate growth/death? Surely there's a steady state solution if you play it right.
Living one day at a time;
Enjoying one moment at a time;
Accepting hardships as the pathway to peace;
~Niebuhr
User avatar
mith
Inland Taipan
Inland Taipan
 
Posts: 5345
Joined: Thu Jan 20, 2005 8:14 pm
Location: Nashville, TN

Re: Conditions for a stable ecosystem? (for a simulation)

Postby AstusAleator » Sun May 03, 2009 6:44 am

You make it sound so simple Mith ;)

Smig I haven't forgotten about this and will post more suggestions when I get a chance. I've been very busy lately though.
What did the parasitic Candiru fish say when it finally found a host? - - "Urethra!!"
User avatar
AstusAleator
King Cobra
King Cobra
 
Posts: 1039
Joined: Tue Jan 24, 2006 8:51 pm
Location: Oregon, USA


Return to Ecology

Who is online

Users browsing this forum: No registered users and 0 guests

cron