Session 1: Theory Behind Evolution I

Transcript of Part 3: Local Interactions Determine Collective Behavior

00:00:08.01	Hello.
00:00:09.10	I'm Deborah Gordon.
00:00:10.25	I'm a professor at Stanford,
00:00:12.02	and I'd like to talk to you today
00:00:13.23	about the evolution of collective behavior.
00:00:16.15	We see collective behavior all around us.
00:00:19.15	Here's an example of collective behavior:
00:00:22.00	it's a group of starlings turning.
00:00:24.18	They have an amazingly fluid way
00:00:27.02	of moving a flock collectively.
00:00:30.02	But, of course, there's lots of collective behavior
00:00:32.25	going on around us that we don't see.
00:00:35.08	Gene transcription networks
00:00:37.15	are a form of collective behavior.
00:00:40.05	Cells work collectively, for example,
00:00:42.15	in an embryo,
00:00:45.04	the development of an embryo and differentiation
00:00:47.25	is the result of collective behavior among cells.
00:00:51.01	Cancer cells work collectively
00:00:53.09	to establish tumors.
00:00:56.00	In a brain,
00:00:57.27	neurons work collectively to produce perception,
00:01:01.05	and memory,
00:01:02.18	and all of the functions of brains.
00:01:05.12	What all these systems have in common
00:01:07.08	is that there's no central control.
00:01:09.11	There's nobody in charge,
00:01:11.00	nobody telling anybody what to do.
00:01:13.01	I study collective behavior in ants.
00:01:15.21	An ant colony consists of sterile female workers,
00:01:19.27	those are the ants you see walking around,
00:01:22.01	and although there are reproductive females called queens,
00:01:26.22	they don't give any instructions
00:01:28.11	or tell anybody what to do.
00:01:30.17	So, instead,
00:01:32.07	ant colonies work through local interactions.
00:01:35.06	Although this is the way
00:01:37.16	that most people think about ant colonies
00:01:39.26	-- this is a silly picture that's been staged --
00:01:42.03	in fact this never happens.
00:01:44.01	There's no foremen,
00:01:45.28	there's no bureaucrats,
00:01:47.14	there are no managers...
00:01:49.19	somehow the behavior of the colony,
00:01:51.21	the way that it can respond to its environment,
00:01:54.23	arises through interactions among ants.
00:01:57.26	Systems without central control
00:01:59.27	always use networks of local interactions.
00:02:03.22	In ants, those are networks
00:02:06.07	of antennal contact
00:02:08.16	and chemical interactions.
00:02:10.24	In cells also,
00:02:12.27	those are networks of chemical interactions between cells,
00:02:15.13	and between cells and their environments.
00:02:18.09	And so, all of these interactions together
00:02:20.25	create a network.
00:02:25.14	The history of biology,
00:02:27.22	especially in the last hundred years,
00:02:30.11	has been to try to understand
00:02:33.00	the function and dynamics of networks,
00:02:36.09	and it began with trying to associate
00:02:39.13	function with type.
00:02:42.11	So, illustrated here, for example,
00:02:45.29	is the idea of one gene - one protein.
00:02:50.18	Then, in studies of neuroscience,
00:02:54.11	early on, we hoped to find particular parts of the brain,
00:02:59.25	each of which would do a certain function,
00:03:02.11	and the study of social insects
00:03:04.02	proceeded in the same way,
00:03:06.25	by looking at the minority of species
00:03:09.25	in which workers come in different sizes,
00:03:12.29	and trying to assign a function
00:03:15.05	to each type of worker.
00:03:18.12	But over time,
00:03:20.05	we've understood that, instead,
00:03:22.02	function and dynamics are produced by interactions.
00:03:24.29	In genes, there are very complex regulatory processes
00:03:28.25	that determine the relationship
00:03:30.24	between genotype and phenotype.
00:03:32.29	The function of brains arises
00:03:35.13	from interactions among many different groups of neurons
00:03:39.06	in the brain
00:03:40.22	that form circuits that interact with each other,
00:03:42.26	and in the same way,
00:03:44.20	in ant colonies,
00:03:46.13	we can see how local interactions
00:03:48.12	produce the behavior of the system.
00:03:52.06	So, ants operate mostly by smell.
00:03:55.01	Most ants can't see.
00:03:57.19	And, they smell with their antennae,
00:04:00.10	so one very important interaction among ants
00:04:03.07	is when one ant touches another with its antennae,
00:04:06.02	and when one ant touches another with its antennae,
00:04:08.23	it can tell by the odor
00:04:10.26	whether the other ant belongs to the same colony
00:04:13.17	and what task it's been doing.
00:04:16.28	So here, we see a laboratory arena.
00:04:20.10	The ants are moving around and interacting.
00:04:22.22	In this arena,
00:04:24.22	there are two tubes connecting to other arenas.
00:04:28.25	When one ant meets another,
00:04:30.28	it doesn't matter which ant it's meeting,
00:04:33.02	they're not exchanging any complicated signals or messages,
00:04:37.11	all that matters to the ant
00:04:38.29	is the rate at which it meets other ants.
00:04:44.19	Taken together,
00:04:46.11	all these interactions produce a network.
00:04:48.17	This illustrates the network and the paths
00:04:51.20	of all the ants that you saw in the film
00:04:54.04	in the previous slide.
00:04:56.01	And it's this constantly shifting
00:04:58.08	network of interactions
00:05:00.01	that produces the behavior of the system.
00:05:02.25	A brain works the same way,
00:05:04.15	but the great thing about ants
00:05:06.04	is that we can see all of the interactions
00:05:08.03	as they're happening,
00:05:09.26	and so we can see how this network of interactions
00:05:12.19	is related to the function of the system.
00:05:17.16	4 I study ants in a desert in Arizona
00:05:21.13	and I'm going to be telling you about
00:05:23.16	some of the work that I've done with harvester ants in the south...
00:05:27.24	I'm gonna be telling you about some of the work
00:05:29.21	that I've done with harvester ants
00:05:31.25	at a study site in southeast Arizona.
00:05:34.20	This is what the nest of a mature colony looks like.
00:05:39.21	You can see the nest entrance,
00:05:41.18	and then there's a trail leading away from the nest entrance,
00:05:44.15	sometimes cleared and sometimes not,
00:05:47.01	that goes about 20 meters,
00:05:49.03	and these ants are called harvester ants
00:05:51.01	because they eat seeds.
00:05:52.16	So, they travel along this trail,
00:05:54.00	collect seeds, and bring them back to the nest.
00:05:57.06	And, I divide all the behavior that I see outside the nest
00:06:00.00	into these four categories:
00:06:02.14	foraging, that's going out and collecting seeds
00:06:05.02	and bringing it back,
00:06:06.16	then the patrollers, shown here with a magnifying glass,
00:06:09.22	are an interesting group of workers
00:06:11.20	that go out early in the morning,
00:06:13.19	they move around the foraging area,
00:06:15.17	they meet the neighbors...
00:06:18.11	they meet the ants of the other neighboring colonies,
00:06:21.21	and it's their safe return
00:06:24.02	that signals the foragers that it's time to go out.
00:06:26.26	The nest maintenance workers work inside the nest.
00:06:29.20	They line the walls of the chambers
00:06:31.16	with moist soil that dries to a kind of adobe finish,
00:06:34.28	and then they carry out the dry soil.
00:06:37.07	So, you see nest maintenance workers
00:06:39.11	coming out, putting down soil, and going back in.
00:06:42.15	And finally, the midden workers
00:06:44.03	work on the refuse pile, or midden,
00:06:46.02	where they put a colony-specific odor
00:06:48.03	that helps guide foragers back into the nest.
00:06:52.24	It's only about 25% of the colony
00:06:55.02	that works outside the nest,
00:06:57.07	so these four task groups that I told you about
00:06:59.07	are only 25% of the colony.
00:07:01.09	Deep inside the nest,
00:07:03.01	which goes down a meter, sometimes two,
00:07:06.10	there are ants that are storing the seeds
00:07:11.00	and processing the seeds.
00:07:13.02	The queen is down somewhere,
00:07:15.05	she just lays the eggs.
00:07:17.07	Then, there are ants
00:07:19.07	that are feeing the larvae and brood.
00:07:20.27	It's actually the larvae that consume most of the food.
00:07:23.21	And, despite what it says in the...
00:07:26.25	and, despite what it says in the bible
00:07:28.15	about how hard-working ants are,
00:07:30.08	there are a lot of ants
00:07:32.12	that are just hanging around doing nothing,
00:07:34.02	and it's a very interesting question about the function of the network,
00:07:36.18	why the colonies...
00:07:39.28	it's a very interesting question
00:07:41.17	about how that group of reserve,
00:07:44.19	or inactive colonies
00:07:46.20	might function to contribute to regulating
00:07:48.20	the network of interactions.
00:07:52.00	In this species, as in most ant species,
00:07:55.05	all of the ants are the same size,
00:07:57.18	so you can't identify the task of an ant by its size,
00:08:00.29	but you can identify the task of an ant
00:08:02.26	by what it's doing.
00:08:04.25	And it turns out that ants change tasks.
00:08:07.14	So, this shows the results of experiments
00:08:10.07	in which I created a need
00:08:12.13	for more ants to do a certain task.
00:08:14.18	So, the arrows point to the...
00:08:18.11	the arrows show the outcome
00:08:20.08	of experiments where I created a need for more ants
00:08:23.01	doing that task.
00:08:24.03	So, for example,
00:08:26.04	if more ants are needed to forage,
00:08:28.01	then the patrollers will change to forage,
00:08:29.23	the midden workers will change to forage,
00:08:31.25	and the nest maintenance workers will change to forage.
00:08:34.22	In response to a lot of really exciting new food
00:08:38.09	that I put out there,
00:08:40.03	everybody will switch to forage.
00:08:42.15	If there's a need for more patrollers,
00:08:44.05	so for example, if I create a disturbance,
00:08:46.10	then the nest maintenance workers
00:08:48.13	will switch to do patrolling,
00:08:50.08	but if more ants are needed to do nest maintenance,
00:08:53.07	so for example, if I create a mess that the nest maintenance workers
00:08:55.19	have to clean up,
00:08:57.23	then none of the others will switch back
00:08:59.15	to do nest maintenance work,
00:09:01.07	and they have to go new nest maintenance workers
00:09:03.09	from the younger ants inside the nest.
00:09:05.13	So, there's a one-way flow of ants
00:09:08.06	from the younger ants inside the nest,
00:09:09.29	through nest maintenance,
00:09:11.20	some of them become patrollers,
00:09:13.10	and everybody eventually ends up as a forager,
00:09:15.27	and once an ant is a forager, it doesn't come back.
00:09:20.13	All of this is regulated
00:09:22.12	by the interactions of the ants
00:09:24.10	as they come in and out of the nest,
00:09:26.22	and so this process
00:09:28.22	that I call task allocation
00:09:30.24	is the process at the level of the colony
00:09:33.01	that gets the right numbers of ants
00:09:35.12	to each task in a given situation,
00:09:38.19	and so now we understand
00:09:40.07	that it's this regulatory process of task allocation
00:09:42.28	that determines how the system functions,
00:09:45.29	not the inherent characteristics of any particular ant,
00:09:50.17	but the regulation of the whole system
00:09:53.05	that shifts ants around
00:09:55.11	into different tasks
00:09:57.02	as they're needed in changing conditions.
00:10:01.23	So, this raises the question,
00:10:03.14	how do these interaction networks evolve?
00:10:06.00	How does evolution shape the regulation
00:10:08.11	of a system with no central control?
00:10:12.12	I'd like to begin with a quote from Dobzhansky,
00:10:15.03	that "nothing in biology makes sense
00:10:16.23	except in the light of evolution,"
00:10:18.22	and to modify that in my own way
00:10:20.09	by saying that nothing in evolution makes sense
00:10:23.03	except in the light of ecology,
00:10:24.26	and what I'd like to do now
00:10:26.13	is to explain that in more detail.
00:10:29.27	So, ecology comes from the words for village.
00:10:33.13	Ecology is about how different parts of a system interact,
00:10:38.03	and we think of ecology
00:10:40.04	in terms of interactions at different levels.
00:10:42.18	First, interactions among individuals
00:10:44.24	like the trees shown here,
00:10:47.00	and then interactions within a population
00:10:50.01	and between populations,
00:10:51.21	where a population is the set of all individuals
00:10:54.28	that may reproduce with each other.
00:10:58.15	So, populations are defined in terms of reproduction.
00:11:02.00	And then a community
00:11:04.07	is all of the populations that are living together
00:11:06.11	and interacting in a certain place,
00:11:09.00	so here, there are some birds in the forest.
00:11:11.25	Of course there are many, many kinds of organisms
00:11:15.12	in the community that that forest is part of.
00:11:18.14	And we could go even one level up
00:11:20.13	and think about the interactions
00:11:22.19	between all of those organisms
00:11:24.14	and all of the other factors that affect them,
00:11:27.01	like the air, and the water, and the wind,
00:11:29.29	and all of the chemicals that are circulating
00:11:32.05	through the system.
00:11:34.04	Ecology is the science
00:11:36.06	that helps us to understand
00:11:38.12	how all of the interactions lead to changes in a system.
00:11:42.22	And so when we want to ask,
00:11:44.07	how do interaction networks evolve,
00:11:45.28	we really have to ask,
00:11:48.12	how does evolution
00:11:50.25	react to the interactions within the system?
00:11:56.13	So, we have to think about ecology to understand evolution.
00:12:01.19	Another way to think about the relationship
00:12:03.21	between ecology and evolution
00:12:05.23	is to think about what natural selection really is.
00:12:09.01	Natural selection requires three conditions.
00:12:14.14	First, there has to be variation in a trait,
00:12:17.23	and we can think of a trait as anything.
00:12:19.09	It could be eye color,
00:12:20.23	or the height of a tree,
00:12:23.04	or the time of year that the tree flowers,
00:12:25.03	or how the thick the polar bear's fur is.
00:12:27.10	A trait could be anything,
00:12:29.09	including how individuals within a system
00:12:32.13	interact to regulate that system.
00:12:34.25	So, first of all, there has to be variation.
00:12:37.17	Then, that trait has to be heritable,
00:12:39.28	because natural selection
00:12:41.18	acts over many generations,
00:12:43.19	and if the trait isn't heritable,
00:12:45.19	if the offspring don't resemble their parents,
00:12:49.00	then over many generations
00:12:50.28	things might reshuffle
00:12:52.25	but there will be no trends.
00:12:54.25	And finally,
00:12:56.09	and here's where the ecology comes in,
00:12:58.04	there has to be...
00:12:59.22	and finally, here's where the ecology comes in,
00:13:01.20	there have to be differences
00:13:03.14	in reproductive success
00:13:05.12	related to the trait.
00:13:06.27	The trait has to make a difference ecologically
00:13:08.27	to how the organisms survive and reproduce.
00:13:14.26	So, here's a diagram to illustrate that.
00:13:17.18	So here,
00:13:20.01	let's imagine that we have lots of individuals in a population,
00:13:22.27	and each row in this diagram
00:13:24.23	is gonna be a different generation,
00:13:27.11	and the different colors represent the trait,
00:13:29.29	so the trait here is color,
00:13:31.19	and some of them are blue,
00:13:33.02	some are red,
00:13:34.12	some are orange.
00:13:35.23	That's the variation in the trait,
00:13:37.06	so there's variation to start out with.
00:13:39.14	And then, that trait is heritable,
00:13:43.00	so blue circles make more blue circles,
00:13:45.13	red circles make more red circles,
00:13:47.25	orange circles make more orange circles...
00:13:50.10	the offspring resemble their parents,
00:13:52.20	so it's heritable.
00:13:54.07	And now, here's where the ecology comes in.
00:13:56.08	In order for their to be any change over generations,
00:14:01.00	some of these individuals have to reproduce more than others,
00:14:05.13	and in this story let's say that it's really great,
00:14:08.05	ecologically, to be orange.
00:14:10.04	And so, over generations,
00:14:11.23	there will be more orange individuals,
00:14:13.19	because their ecology is such
00:14:16.22	that they can reproduce more.
00:14:18.28	And here we get natural selection,
00:14:20.24	a change over many generations,
00:14:22.17	in the frequency of individuals that are orange.
00:14:25.01	That's how natural selection works,
00:14:27.12	and it always requires an ecological process
00:14:30.21	that affects the reproductive success
00:14:33.10	of the trait
00:14:35.07	in such a way that some individuals
00:14:37.15	reproduce more than others.
00:14:41.00	So, when we ask,
00:14:42.20	how do interaction networks evolve,
00:14:44.14	really we're asking an ecological question
00:14:46.19	about why networks function differently
00:14:50.04	in a given environment
00:14:52.13	so that some forms of an interaction network
00:14:55.10	allow more reproduction than others.
00:14:59.25	So, we could ask this question about cancer cells.
00:15:02.27	We could ask,
00:15:05.11	why is it that certain ecological conditions
00:15:07.14	within a body
00:15:10.02	allow the evolution of particular types of cancer,
00:15:13.24	and allow the cancer cells
00:15:16.29	to change over generations of cells?
00:15:20.07	That's a difficult question, although it's one that's really important,
00:15:23.23	and we can also ask the same kind of question
00:15:26.09	about ant colonies.
00:15:28.00	And there, because we can see the interactions
00:15:30.09	among ants
00:15:32.07	and we can look at them in their environments,
00:15:34.02	we have the opportunity to learn
00:15:35.26	how interaction networks
00:15:38.01	are evolving in certain environments.
00:15:40.04	So, there are more than 12,000 species of ants.
00:15:43.02	They live in every conceivable habitat on Earth,
00:15:45.26	and they all have to solve ecological problems
00:15:48.01	because they have to explore their environments,
00:15:50.12	they have to get resources,
00:15:52.03	and then they have to reproduce.
00:15:54.07	And when we look at different species,
00:15:56.02	we can see how interaction networks
00:15:58.11	are evolving to work differently in different environments.
00:16:01.09	So, one species that we study in my lab
00:16:03.13	is the Argentine ant.
00:16:05.03	It's an invasive species.
00:16:06.29	They came from Argentina,
00:16:08.16	they have spread around the world,
00:16:10.04	and everywhere that there's a Mediterranean climate,
00:16:12.22	there now are Argentine ants,
00:16:14.21	on the coast of California,
00:16:16.22	the coast of South Africa, Australia,
00:16:19.08	the whole Mediterranean coastline,
00:16:21.03	Japan, Hawaii...
00:16:24.22	And one of the interesting things
00:16:26.05	about how their interaction networks operate
00:16:28.01	is how different they are from other ant species.
00:16:32.24	Many ant species use interaction networks
00:16:35.23	to create what is called central place foraging.
00:16:39.12	So, you can think of it like a broom.
00:16:41.11	The ants all live in one nest,
00:16:43.09	they go out to forage, and they bring their resources
00:16:45.20	all back to the central nest.
00:16:48.07	But, Argentine ants, like other species,
00:16:50.16	are really good at a different kind of strategy.
00:16:53.11	They make circuits,
00:16:55.09	it's more like a vacuum cleaner.
00:16:57.11	They move from nest to nest, they have many different nests,
00:17:00.00	they move from nest to nest
00:17:01.28	and they sweep up resources as they go.
00:17:04.20	So, they create a different kind of interaction network
00:17:06.29	that functions differently
00:17:09.11	from the species that use central place foraging.
00:17:12.15	And one of the things that we've been studying
00:17:14.17	is how new paths form,
00:17:17.03	and we find that Argentine ants
00:17:18.28	actually recruit from the trail and not from the nest.
00:17:21.18	That is, they create a large network like the vacuum cleaner
00:17:23.17	going around.
00:17:25.05	If you're operating a vacuum cleaner,
00:17:27.03	instead of going all the way back
00:17:29.19	to a corner of a room all the time,
00:17:31.13	you keep moving the vacuum cleaner from wherever you are,
00:17:33.13	and that's the way that Argentine ants work also.
00:17:37.26	In the tropics, we see different kinds of interaction networks.
00:17:41.12	One of the species that we're studying
00:17:43.05	is the turtle ant,
00:17:45.10	and here you can see ants marked...
00:17:49.18	those green ants
00:17:51.20	and the pink ants
00:17:53.21	are marked with nail polish by us,
00:17:55.14	they're not really that color,
00:17:57.02	and we did that in order to see how ants
00:17:59.09	are allocated on different trails.
00:18:01.06	And what we find is that ants create circuits in the trees,
00:18:05.24	ongoing circuits, again, like a vacuum cleaner,
00:18:08.10	from a nest to another nest,
00:18:10.13	fort to a food source,
00:18:12.07	so the ants create a circuit in the trees
00:18:14.06	from nest to food sources and back again,
00:18:16.28	around and around,
00:18:18.22	and those circuits are shaped by negative interactions
00:18:21.13	with the ants of another colony.
00:18:24.17	So for example, here you can see
00:18:27.03	another Cephalotes species
00:18:29.12	plugging up its nest entrance with its head
00:18:32.07	in response to negative interactions
00:18:35.11	with a different species,
00:18:37.17	those smaller ones running around with their abdomens in the air,
00:18:44.06	and so this species has a form of security
00:18:47.14	that's based on interactions with ants of other species,
00:18:51.15	and they regulate the flow of ants
00:18:53.15	in and out of the nest based on those interactions.
00:18:58.17	So, ants are using interactions differently
00:19:01.23	in a huge range of environments,
00:19:03.18	and we can study the evolution of collective behavior
00:19:06.19	by understanding, ecologically,
00:19:09.28	how the function of a network in one environment
00:19:13.07	affects the reproductive success of the colonies
00:19:16.10	using that network.
00:19:20.23	Interaction networks
00:19:22.20	evolve in response to environmental challenges,
00:19:24.28	and so the way to study the evolution of collective behavior
00:19:27.27	is to try to understand
00:19:30.16	how interaction networks
00:19:32.17	are operating in particular environments,
00:19:35.02	and what that means for the survival
00:19:37.00	and reproductive success
00:19:38.29	of the system using those networks.
00:19:41.20	Thank you.