How do you build a new behavior? First, you need changes in the neural circuit to express the new behavior, and under the right context. Then, you also need the behavior to be meaningful to the intended recipient, so that it can be selected for and passed on to the next generation.

But let's take a step back. Why do you want to build a new behavior?

In animals, behaviors can be crucial for every facet of survival and remarkably diverse in complexity and form. Behaviors are also often used to distinguish between self and other, for example in the selection of a mate of the correct species. Mating with the wrong species can be costly - for this and other reasons, mating related behaviors are often rapidly evolving and can be quite different between closely related species. This is where new social behaviors can come in.

In our lab, we use fruit fly (genus Drosophila) courtship behaviors to help us uncover the mysteries of behavior evolution. Drosophila courtship behaviors are complex (involving many sensory modalities such as songs, pheromones, and visual signals), and can be quantitatively and qualitatively different between even closely related species. The diversity of male courtship songs among Drosophila species has fascinated researchers for decades, and a tremendous body of research has collectively discovered much of the organization and evolution of song neural circuitry. 

Female courtship-related behaviors, on the other hand, are less well characterized until recently. We now know that females perform different abdomen behaviors when they accept or reject a male's courtship advances, respectively, and we know the neural circuits controlling each behavior. Interestingly, although female perception and preference in the species-specific courtship songs must be as diverse as the songs themselves, the acceptance and rejection behaviors are very conserved throughout Drosophila. 

Therefore, it was a big surprise when we saw a new female courtship-related behavior in D. santomea.

In the traditional model species D. melanogaster, when a female is receptive to a male's courtship, she slightly extends her abdomen to open her vaginal plates (vaginal plate opening; VPO). In D. santomea, in addition to VPO, females also spread their wings. Wing spreading always co-occurs with VPO (but not necessarily vice versa). As a male typically performs multiple rounds of singing before attempting to copulate, the female being courted can decide whether to respond to each song with VPO and/or wing spreading. Wing spreading encourages males to devote more effort into singing longer songs, and pairs that successfully adopt this social feedback loop of "male singing - female wing spreading - extended male singing" are more likely to copulate. These observations tell us that wing spreading is a meaningful social signal, and we have satisfied the second requirement of building a new behavior.

Moving on (or back?) to the first requirement. To understand how the neural circuit evolves to encode a new behavior, we first need to identify the neurons that control it. Neurons that express the sex determination gene doublesex (dsx) is involved in a lot of mating and reproductive behaviors in Drosophila, which makes them an attractive starting point. We generated parallel genetic reagents in D. santomea, its closest relative D. yakuba, and D. melanogaster to target brain dsx neurons for labeling and neuron activity manipulation.

Activating brain dsx neurons in all 3 species triggered VPO, consistent with our knowledge of the evolutionary conservation of this acceptance behavior. The same treatment also drives robust wing spreading only in D. santomea, which is also consistent with wing spreading being a species-specific behavior. Furthermore, restricting activation only to the descending neurites of 2 pairs of brain dsx neurons in the ventral nerve cord triggers the same behaviors in each species. This means evolutionary change must have happened at or downstream of one of these 2 pairs of brain dsx neurons in D. santomea. This also elegantly solves the "under the right context" aspect of the first requirement by layering a new behavior, wing spreading, on top of an existing neural circuit that integrates external cues (male courtship songs) and internal states (female receptivity state) to direct the appropriate motor behavior (VPO).

To our delight, one of the 2 pairs of brain dsx neurons controls VPO (called vpoDN), and the other pair controls the rejection behavior. Because both VPO and wing spreading are receptive behaviors, and because wing spreading co-occurs with VPO but not the rejection behavior, we believe that vpoDN might also control wing spreading. The perfect experiment here would be to activate vpoDN in the 3 species and show that, similar to activating all brain dsx neurons, while all 3 species perform VPO, only D. santomea performs wing spreading. Unfortunately, making genetic reagents that target very specific subset of neurons in non-melanogaster Drosophila species is still pretty challenging.

We did the next best thing, to try and see if we could find out more about vpoDN in D. melanogaster. When we activated vpoDN neurons in D. melanogaster females that were raised at a high developmental temperature, we noticed that quite a few females also performed wing spreading in addition to VPO. High developmental temperature is known to be an environmental stressor that can unmask some of the buffering built in to developmental programs to safeguard the correct development of organisms under different natural conditions. Normally D. melanogaster females should not perform wing spreading under vpoDN activation, but with the developmental buffering reduced or gone, some actually did. This tells us that the wing spreading circuit is latent in D. melanogaster females, and that they can be activated under certain conditions to express wing spreading.

As we have established, D. santomea is the only species among its close relatives (within the melanogaster subgroup) that performs wing spreading, but a few other species across the broader Drosophila phylogeny also perform this behavior. Phylogenetic reconstruction of wing spreading across 46 species revealed that this behavior has evolved multiple times throughout the genus, but none of the ancestors of D. santomea and D. melanogaster were likely to have expressed wing spreading (i.e., wing spreading is unlikely to have been a re-emergence of a lost ancestral behavior). How can the same behavior evolve repeatedly without an ancestral blueprint?

We think that the latent circuit potential for wing spreading, but not the wing spreading behavior itself, is ancestral. This latent potential might have been maintained over long evolutionary time to support another behavior or for some other reasons that we do not yet know. But crucially, in certain lineages, evolutionary changes take place to actualize the circuit potential and drive stable expression of wing spreading. Similar observations have been made about latent developmental potentials in evo-devo studies. This emerging theme in phenotypic evolution research seems to blur the line that traditionally demarcates "novelty" - when the potential is ancestral and can be repeatedly tweaked to express the "novel" phenotype, where does novelty begin?