by Matthew Cobb
One of the big problems that worried Darwin in his theory of evolution by natural selection was what he called ‘the example of neuter insects’ – social insects such as ants, wasps and bees—groups in which most of the members of a colony are female, and yet do not reproduce and are sterile. As Darwin wrote in On the Origin of Species:
Darwin’s answer was that selection operated at the level of ‘the family’ – sterile workers shared characters with those individuals who were reproducing, and thus those characters could be ‘seen’ by natural selection.
We now interpret worker sterility in terms of genes shared by highly related groups of insects, which in the case of Hymenoptera (bees, wasps and ants) show a bizarre form of sex determination called haplodiploidy, whereby males are haploid, produced from unfertilised eggs, while females are diploid, produced by mating.
This means that if a queen has mated with only a single male, her offspring (the workers) are more closely related to each other (sharing 75% of their gene copies) than they would be to their own offspring (50%). In other words, it is in the genetic interest of the workers to be sterile and rear their sisters rather than mate themselves – they will pass on more copies of their genes to the next generation this way!
Life is (of course) more complicated than this – many queens mate with more than one male, so those ideal relatedness levels aren’t always seen. Furthermore, while haplodiploidy encourages the evolution of sociality, it isn’t necessary. Many Hymenoptera are not social, despite being haplodiploid, while some of the most successful social insects – termites – have XY sex determination like you and me.
However, for the last few decades the basic view of those studying sociality has been that it’s in the workers’ genetic interest to be sterile – they are passing more of their genes on to the next generation this way. They haven’t been ‘sterilised’ by the queen, they are ‘choosing’ to be sterile (or at least, choosing not to mated, for in some species workers can produce males from unfertilised eggs).
This would mean that for a worker to turn off her ovaries, she would need two bits of information – she is surrounded by closely-related individuals (that is, there is a colony-specific signal) and that there is a reproductive individual present who is churning out eggs (that is, there is a fertility signal provided by the queen or reproductive).
There were two reasons why we came to this view. Firstly, in 1993 there was an excellent article by Laurent Keller and Peter Nonacs which looked at what we would expect if the queen was controlling the workers, or if she was merely signalling her fertility and thereby enabling the workers to ‘decide’ to turn their ovaries off. Keller & Nonacs pointed out that if the workers were being forced into sterility, then there would be an ‘arms race’ and in some groups we would expect to see that workers had started mating. So you might expect, for example, to see some species of solitary ants. There is no such evidence, so on the basis of a tight argument, most people accepted that there was no evidence for queen ‘control’. That doesn’t mean to say it doesn’t exist, merely that there is no evidence of this.
The second reason is that there was a mass of evidence showing that chemical signals are involved in affecting the growth of worker ovaries. If you removed the reproductive individual from the colony, in some species worker ovaries would soon start to grow. Furthermore, the chemical signature of the reproductive is correlated with her ovarian function – these ‘queen pheromones’ are in fact a chemical signal indicating her fertility. Further, similar effects were seen in solitary insects, like flies, providing an insight into how the sensory systems of social insects could have evolved, on the basis of the pre-existing sensory systems of solitary insects.
As Michel Chapuisat puts it:
The hypothesis that queen pheromones evolved from a preexisting communication system in solitary ancestors has interesting implications for the evolution of eusociality. In the early stage of sociality, daughters may respond to maternal fertility signals by helping the mother if she is highly fertile, and reproducing if she is not. By allowing this conditional response, a preexisting pheromonal communication of fertility may have facilitated the transition to eusociality.
In the latest issue of Science there is a stunning proof of this suggestion (Chapuisat’s article is commenting on this), but frustratingly it is partly framed – and is certainly being reported – in the wrong way.
To summarise an awful lot of very impressive work, the scientists looked at three very distantly related species – a wasp, a bumblebee and an ant, in each of which sociality evolved independently (in other words, their most recent common ancestor was a solitary insect).
They looked at the chemicals on the cuticles of the various castes within these species (queens, workers, males), and identified a set of compounds that appeared to be common to the queens in these species. They then removed the queen from colonies of these insects, and introduced synthetic versions of the compounds, and observed what happened to the workers’ ovaries. If the ovaries remained regressed, then this would indicate that the compound was perceived as indicating the presence of a queen.
Amazingly, they found that some of these compounds were indeed common to these three groups (note the effect is clearer in the wasp and the ant – which are more closely related – than it is in the bee):
The legend says: “The results demonstrate that long-chain cuticular hydrocarbons act as a conserved class of sterility-inducing queen pheromones in three independently evolved social insect lineages, represented by the wasp V. vulgaris (A), the bumblebee B. terrestris (B), and the ant C. iberica (C). Treatment of queenless worker groups with the linear alkanes n-C27 and n-C29 and the methyl alkane 3-MeC29caused a two- to sevenfold reduction in the odds of workers having fully developed ovaries in the common wasp and the Iberian ant (bar charts, red bars) relative to a pentane-treated control (left, stacked bar charts)”
So this suggests that there may be common chemical signals relating to ovarian function that are used in all these species. The authors then looked at a large number of other studies and plotted the putative fertility signals onto a phylogenetic tree, with fascinating results:
The legend says: “Fig. 2 The evolutionary history of queen and fertility signals across major clades of social hymenopteran insects. Each alternately shaded clade indicates an independent origin of eusociality. The pie charts show the likelihoods of different compound classes being used as queen or fertility signals (…). Saturated hydrocarbons (linear and methyl-branched alkanes) receive very high support for being used as conserved queen or fertility signals across several independent origins of eusociality.”
The authors conclude with this great summary of a marvellous piece of work:
“our ancestral state reconstruction shows that saturated hydrocarbons were most likely used as fertility cues in the common solitary ancestor of all ants, bees, and wasps, which lived ~145 million years ago”.
Note that they aren’t suggesting that exactly the same molecules are used today – “saturated hydrocarbons” is a pretty broad class of substances, and insect physiology keeps on popping up with the same molecules (they’re all related to fatty acid biosynthesis), so this is an entirely legitimate suggestion.
So what’s my beef? The data are fantastic and change the way we think about the evolution of queen signals, suggesting the same signals may exist in different lineages. The problem comes with the way the findings are being presented. Here are two screenshots from the Science magazine website. See if you can spot the problem.
Both these presentations – especially the second one – suggest that the authors have proved that queens control their ‘underlings’. My guess is that this view will be repeated in the media over the next few days. In fact, the study does nothing of the sort – it doesn’t show how these chemicals exert their effect. If anything, as the authors argue, and emphasised by Chapuisat, it supports the view that these are fertility signals, not methods of ‘control’ used by the queens ‘to prevent worker reproduction’. The idea of ‘underlings’ being ‘controlled’ might seem sexy to some; to anyone who knows, it’s just plain wrong.
So where does this view come from? It’s not simply sub-editors at Science who don’t ‘get’ what is, I accept, quite a subtle point. I fear the authors have inadvertently contributed to the confusion.
The title of the paper is ‘Conserved class of queen pheromones stops social insect workers from reproducing’. To which I would answer, ‘well yes and no’. The ‘stops’ is very affirmative and suggests strongly this is a manipulation of the workers’ fertility by the queen. Elsewhere in the paper, they suggest ‘pheromones emitted by the queen are thought to play a key role in suppressing worker reproduction’, ‘we searched for sterility-inducing queen pheromones’, ‘the more volatile queen pheromone blend not only stops workers from reproducing’.
These minor terms create the impression that these compounds act against the interests of the workers. This might be the opinion of some of the authors, or they might have felt that this was a sexier way of presenting the findings, or it might just be sloppy writing. Whatever the case, it’s unfortunate as it lessens the impact of a fantastic piece of work.
Why do I care? Well apart from being punctilious/cranky, I have 150 final year students sitting a Chemical Communication in Animals exam on Monday, and I don’t want them to go off writing about this new study showing queen control!
A final intriguing point. About 15 years ago, Jean-François Ferveur and myself, together with two of his students, suggested that closely-related Drosophila species (the flies studied by Jerry) might use common sex pheromones, which are related to ovarian activity. Among the compounds we suggested could be involved were saturated hydrocarbons such as 2-methyl hexacosane, 7-heptacosene and n-heptacosane – chemicals with structures similar to those identified as common fertility signals in social insects by this study. The deep history of chemical communication in insects might go even further than anyone had suspected.
[UPDATE: I told you so. Here’s New Scientist’s coverage of the article. *Sigh*. “Minions”. They should know better, but maybe it’s me who should not expect so much from them… h/t @dunbarrover]
References (mostly $$$ I’m afraid, but the link will get you to the abstract):