This blog post is provided by Vinícius Caldart. Vinícius is a shortlisted candidate for the 2023 Elton Prize, for work on Function of a multimodal signal: a multiple hypothesis test using a robot frog.

Animals communicate with each other through signals. Signals can be visual, auditory, olfactory, tactile, seismic, or electric and play a crucial role in the behavior and social interactions of animals. Signals of different modalities can be emitted individually (unimodal signals) or associated with each other (multimodal signals). Imagine these three situations: a handshake, a smile, and a handshake accompanied by a smile. The first two are examples of unimodal signals and the last one is an example of a multimodal signal. When we compare them, they seem to have very different meanings, don’t they? In fact, these examples illustrate the idea that unimodal and multimodal signals can have different functions in the same communication system. However, although there is a great diversity of unimodal and multimodal signals in nature, this idea is not easy to test in practice.

Why do complex signals like multimodal ones evolve? There are three paths that help to answer this question. Multimodal signals can evolve if they reveal information about the sender (content-based selection). They can encapsulate a range of information such as quality (“I am a good mate”), condition (“I am full of energy for fighting”), or specific recognition (“I am of the same species as you”). Multimodal signals can also evolve if they improve transmission and detection in the environment (efficacy-based selection). In some cases, one component of the signal travels long distances (e.g., a low-pitched call) while the other facilitates identification at short distances (e.g., vocal sac movement). Finally, multimodal signals can evolve if the interaction between signal components causes one to improve or modify the other (inter-signal interaction selection). For example, one component of the signal can serve only to draw the receiver’s attention (“hey, you there”) to the other component (subsequent waving).

In an attempt to unravel the function of multimodal signals, researchers generally focus on one or another type of selection, testing hypotheses about content-based selection, efficacy-based selection, or inter-signal interaction selection. An alternative approach involves testing multiple hypotheses that incorporate the different types of selective pressures to which a multimodal signal may be subject. With this purpose in mind, we took advantage of extensive existing knowledge about the ecology and behavior of the diurnal stream frog Crossodactylus schmidti to test functional hypotheses postulated in the literature about complex communication. We selected from the literature four hypotheses compatible with the ecology of our study species to understand the function of a multimodal display emitted by males during territorial contests.

Crossodactylus schmidti is a diurnal frog species that inhabits streams with varying levels of background noise and sunlight incidence (Figure 1). Male frogs compete for territories containing rocks, which serve as signaling sites, and subaquatic chambers, which serve as oviposition sites. When an intruder approaches a resident male and emits aggressive calls, the resident responds with aggressive calls. If the intruder does not retreat, both males exchange long-lasting aggressive calls and various visual signals. Before escalating to physical confrontations, males exchange for a long time the most common multimodal signal: aggressive calls accompanied by toe flags. Toe flags are repeated up-and-down movements of the toes in both feet. They are often seen during agonistic interactions among diurnal frogs in lotic habitats.

Figure 1. Above, a male Crossodactylus schmidti defending its territory and two males involved in a territorial contest. Below, a male interacting vocally and physically with the robot frog during the experiment.

We considered that the multimodal signal used by males during territorial contests could be subject to different types of selection. This is because males need to exchange information about fighting ability and motivation for aggressiveness during such disputes, making the signal subject to content-based selection. The disputes occur in an environment with varying levels of background noise and sunlight incidence, which could make the signal subject to efficacy-based selection. Additionally, the visual component of the signal (toe flag) is rarely emitted in isolation. Therefore, the addition of toe flags to the acoustic display could serve to provide a new context for the receiver to modify its response to the aggressive call. This makes the signal subject to inter-signal interaction selection.

How to test multiple functional hypotheses? This is where the robot frog comes in. In a natural setting, we exposed thirty-eight resident males to a robot frog that simulated a male emitting aggressive calls (acoustic stimulus) and toe flags (visual stimulus), combined and isolated. In each exposure, we measured characteristics related to the quality of the resident males (size and body condition) and local levels of background noise and light intensity. The results demonstrate that toe flags are insufficient to elicit a response from a receiver on their own, while aggressive calls are sufficient. However, when toe flags are emitted along with aggressive calls, they evoke longer aggressive responses in the receiver than those evoked by aggressive calls alone. Moreover, aggressive calls and toe flags do not provide complementary or redundant information about male quality, and the multimodal display did not increase the receiver’s response along natural gradients of light and background noise.

This combination of results is consistent with the predictions of the context hypothesis. According to this hypothesis, the function of toe flags is to provide a new context that is used by the receiver to interpret aggressive calls and modify the response to them. Since the emission of toe flags often increases as fights progress, we suggest that the contextual information provided by toe flags accompanying aggressive calls is motivation for escalating the aggressive interaction. In the example at the beginning of the text, a smile accompanying a handshake could well indicate a more friendly social context than just a handshake. In the case of C. schmidti males, in turn, toe flags accompanying aggressive calls seem to indicate a context of higher aggressiveness than when calls are emitted alone.

A multimodal contextual signal can be advantageous in situations where the information transmitted by a signal changes over time or is difficult to discern due to less stereotyped signal properties. By adding a contextual signal to a display, the ambiguity of the other signal can be reduced, which benefits receivers who acquire less ambiguous information about the signaler. This is especially important in contests, as individuals who are unable to acquire adequate information about the aggressiveness of their rival may incur injury costs. Examples of context-dependent multimodal signals are rare in the literature, likely because most studies focus on single hypotheses based on content or efficacy-based selection. Therefore, this study provides one of the few pieces of evidence for a multimodal signal with a context function in animal contests and emphasizes the importance of considering multiple selective pressures when testing the function of multimodal signals.

Literature hyperlinked in the text:

Bro-Jørgensen, J., & Dabelsteen, T. (2008). Knee-clicks and visual traits indicate fighting ability in eland antelopes: Multiple messages and back-up signals. BMC Biology, 6, 47.

Grafe, T. U., & Wanger, T. C. (2007). Multimodal signaling in male and female foot‐flagging frogs Staurois guttatus (Ranidae): An alerting function of calling. Ethology, 113(8), 772-781.

Hebets, E. A., & Papaj, D. R. (2005). Complex signal function: Developing a framework of testable hypotheses. Behavioral Ecology and Sociobiology, 57, 197–214.

Rosenthal, G.G. Rand, A.S. Ryan, M.J. 2004. The vocal sac as a visual cue in anuran communication: an experimental analysis using video playback. Animal Behaviour, 68:55–58.

Videos hyperlinked in the text:

Supp_file_Appendix S1 (Male signals): https://besjournals.onlinelibrary.wiley.com/action/downloadSupplement?doi=10.1111%2F1365-2656.13620&file=jane13620-sup-0001-AppendixS1.mp4

Supp_file_Appendix S4 (Sequence of videos of male responses): https://besjournals.onlinelibrary.wiley.com/action/downloadSupplement?doi=10.1111%2F1365-2656.13620&file=jane13620-sup-0004-AppendixS4.mp4

About the author