Mutualism is very common: the classic example is the relationship pollinators and their plants. Around 70% of land plants require other species to help them reproduce via pollination. Often, the pollinators, like bees and wasps, gain food from the plant while the plant benefits by getting to mix its genes with other plants - a clear win-win for both. But both have to give up something, too, and whenever there is a cost to a relationship, both sides have good reason to cheat.
When I say cheat, I mean a species not keeping up their half of the deal. A species would gain something if they could maintain the positive benefits provided by another other species without having to expend whatever cost is associated with their side of the mutualistic bargain. A plant would benefit, for example, if it could attract its pollinators without having to make nectar or pretty flowers to attract them.
So how is mutualism maintained when there is strong evolutionary pressure to cheat? In some cases, it's by nature of the relationship. In the example above, it's simply hard for the plant to cheat because skimping on the goods directly affects how the other side acts - no nectar-laden flowers, no reason for a bee or other bug to stop and get covered in pollen.
But some mutualist relationships are easier to cheat on - take the case of fig wasps.
Fig wasps are wasps that lay their eggs in fig flowers. As these flowers turn into fruits, the wasp larvae are protected and fed by the fig, costing the tree resources. This relationship looks parasitic at first glance: the wasp gets healthy babies while the fig gets its fruit ruined. But the wasp has a promise it must keep to the tree: when it lays its eggs, it has to pollinate flowers so the tree can produce seeds.
There are actually two kinds of fig wasps: one that pollinates passively and one that pollinates actively. The passive pollinators collect pollen on their extremities and, while climbing around to deposit eggs, pollinate the trees' flowers without even thinking about it. Passively pollinating wasps do not expend extra energy to pollinate, and they cannot easily avoid carrying pollen, so there's no real way or reason for them to cheat.
David Attenborough explains their relationship rather nicely:
The active pollinators are much more deliberate about things: female wasps specifically collect pollen in specialized pouches (see R) and deposit it on another tree's flowers by choice when they lay their eggs. Active pollinators don't have to pollinate, per se - they can, and do sometimes, flit around without collecting pollen and bring it to another tree. After all, it costs the wasp time and energy to go about collecting and lugging around pollen, so why bother if they don't have to? Instead, the female wasps just infect flowers with wasp eggs, acting more like a parasite than a mutualist.
Clearly, there's an easy, good reason for the wasp to short-change the tree. But, if there's good reason for the wasps to cheat, there is equally good reason for the trees to catch them, evolutionarily speaking. Having a cheating wasps' young growing in its fruit does the tree no good whatsoever. But can the trees spot cheaters and somehow punish them for it?
That's the question that biologists K. Charlotte Jandér and Edward Allen Herre wanted to answer. To find out, they carefully watched six different species of figs, four that had active pollinating wasps and two that had passive pollinating wasps. They wanted to see if the actively-pollinated trees somehow reacted differently to loyal wasps who pollinated like they're supposed to and cheaters. Since it's hard to tell if a wasp is doing its job, instead, the researchers intentionally manipulated the wasps. For each fig tree–pollinator species-pair, they experimentally produced pollen-carrying and artificially pollen-free wasps, which, because they had no pollen, played the role of cheaters. They then waited to see how well the cheaters larvae survived.
They found that the passively pollinated figs had no system in place to protect against cheaters - which is exactly what you'd expect, since it's basically impossible for a passive-pollinating wasp to get around on the flowers without pollinating, meaning that cheating is not likely.
The actively pollinated figs, on the other hand, all punished cheaters.
First off, the figs carrying cheater offspring were aborted more frequently. When a fig aborts a larvae-containing fruit, it kills all of the larvae inside. One active species only kept around 3% of the number of figs that the passive pollinated species did. But to punish them even more, the fig also manipulated the conditions within the growing fruits which contained cheating larvae - per fruit, fewer cheater adults emerged than non-cheating ones. In one species of fig, almost no cheaters survived to adulthood - just 5% of the number that emerged from passively pollinated figs. How exactly the fig changes the condition of the fruit to harm the growing larvae isn't yet known.
This made the scientists wonder how common cheaters were in the wild, and whether the species that strongly reacted to cheating were plagued by more cheaters. As expected, they didn't find any pollen-free passive pollinating wasps, but they did find active pollinating ones that weren't carrying the goods. They also found that the species that cheated the most lived on the fig tree that punished them the least.
These data strongly support consistent coevolution between the fig wasps and their trees. If the tree doesn't catch cheaters, the wasps exploit their longtime friends, and since cheating isn't punished, cheating young grow up and continue cheating, leading to high frequencies of cheaters. This rapidly degrades their relationship from mutualism to parasite-host. However, if the trees respond by culling free-riders, they reduce the number of wasps inclined to cheat and maintain the true mutualism that the two have had for around 80 million years.
Mutualism is often portrayed as "playing nice", a beautiful harmony between species. Just listen to how the relationship between active pollinating fig wasps and their trees is portrayed in this PBS special:
How sweet. Too bad it's totally not true. Just like the arms races between predator and prey or parasite and host, mutualist species constantly adapt to try get the upper hand in their relationship. There is still a battle going on between even the best of friends to gain an evolutionary advantage, and just like other interactions, mutualists have to constantly evolve to maintain the status quo.
Jander, K., & Herre, E. (2010). Host sanctions and pollinator cheating in the fig tree-fig wasp mutualism Proceedings of the Royal Society B: Biological Sciences DOI: 10.1098/rspb.2009.2157