The natural world is filled with networks. Predator and prey, flower and pollinator—each interacting pair forms a link in a networked community of organisms.
Now, a French research team has developed a model that explores how evolution may help shape these ecological networks over time as species interact in either antagonistic or mutually beneficial ways. The recent findings, reported in Ecology Letters, suggest that evolution could help explain common patterns that scientists find in ecological networks today. “We don’t actually have a lot of ways to observe the assembly of networks,” says ecologist and evolutionary biologist Jeremy Yoder of California State University, Northridge, who was not involved in the study. “We don’t really know what [a network’s] history is.”
As scientists explore existing networks, two major trends have emerged, as noted in a 2011 paper. Mutualistic interactions tend to be part of “nested” networks, where insects, for example, may specialize in distinct flowers, but flowers receive visits from several pollinator species.
Meanwhile, antagonistic interactions, such as between predator and prey or herbivore and plant, tend to form “modular” networks with pairs grouped into small subsets of the community—and organisms tend to interact primarily with others within their own subset. “We wanted to see whether evolutionary dynamics could lead to these kinds of [network] structures,” says evolutionary biologist Odile Maliet, study lead author and a postdoctoral fellow at the Institut de Biologie de l’École Normale Supérieure in Paris, France.
Maliet and her team simulated the formation of an ecological network. Their model simplified interactions to home in on the essential elements: Imagine an island with only two species. This island is divided into a grid of 4,000 cells, each housing one representative from each species. The species’ interactions depend on how closely their traits match.
The traits, represented by numerical values, are intentionally “super abstract,” meaning the same values could apply to any number of traits belonging to organisms in real-world networks, says Maliet. For a mutualistic network, for example, the values might represent the shapes of a flower tube and a pollinator’s beak. For an antagonistic network, the values could represent a plants’ ability to defend itself with toxic compounds and an herbivore’s capacity to overcome the defense.
The simulation begins by killing off one individual from each species at random. These deaths leave openings on the island for new individuals that represent the next generation. Initially, individuals within a species are genetically identical, so survivors all have the same probability of producing offspring. But with each new generation, the model randomly introduces small genetic mutations that may push an offspring’s traits further or closer to the other species.
In the team’s mutualistic network model, individuals have a higher probability of contributing to the next generation when their trait values more closely match those of the interacting species. In the antagonistic model, the predator species has a higher probability of reproducing when it closely matches its prey, but the prey is more successful with a mismatched trait.
The model repeats this process of random death followed by reproduction with slight mutation for thousands of generations. Over time, genetic mutations accumulate enough that species diverge into new species. When the simulation is complete, the team creates a phylogenetic tree that shows how each species is related based on family histories going back to the start of the simulation. The researchers also construct an ecological network for the resulting species that captures interactions between species pairs.
The team repeatedly ran models for antagonistic and mutualistic networks, as well as neutral networks where species interact at random. The results show patterns similar to what’s been observed in nature. “The antagonistic ones tend to be more modular and less nested than the mutualistic ones,” says Maliet.
“This paper answers what feels to me like a really longstanding question,” says Yoder. “Do patterns like nestedness or modularity tell us something about the evolution that led to the networks that we see today? I think the conclusion is that they can.”
But the research also raises other questions. Mutualistic models tended to generate networks with fewer species than antagonistic ones. This finding doesn’t match some mutualistic networks in nature, such as, for example, the hugely diverse network of figs and their specialist wasp pollinators. Maliet acknowledges that the model could be improved. “I think there is something lacking for the mutualism [model],” she says. Thinking about what’s lacking could help her team and others generate new hypotheses about what exactly is missing and, therefore, important. And they could add these parameters into future models.
The model is publicly available so others can vary its moving parts and generate their own hypotheses. “I’m looking forward to playing around with the [model] myself,” says Yoder. “And seeing what comes from this [model] in the future.”