Desert ants achieve reliable recruitment across noisy interactions

Abstract

We study how desert ants, Cataglyphis niger, a species that lacks pheromone-based recruitment mechanisms, inform each other about the presence of food. Our results are based on automated tracking that allows us to collect a large database of ant trajectories and interactions. We find that interactions affect an ant's speed within the nest. Fast ants tend to slow down, whereas slow ones increase their speed when encountering a faster ant. Faster ants tend to exit the nest more frequently than slower ones. So, if an ant gains enough speed through encounters with others, then she tends to leave the nest and look for food. On the other hand, we find that the probability for her to leave the nest depends only on her speed, but not on whether she had recently interacted with a recruiter that has found the food. This suggests a recruitment system in which ants communicate their state by very simple interactions. Based on this assumption, we estimate the information-theoretical channel capacity of the ants' pairwise interactions. We find that the response to the speed of an interacting nest-mate is very noisy. The question is then how random interactions with ants within the nest can be distinguished from those interactions with a recruiter who has found food. Our measurements and model suggest that this distinction does not depend on reliable communication but on behavioural differences between ants that have found the food and those that have not. Recruiters retain high speeds throughout the experiment, regardless of the ants they interact with; non-recruiters communicate with a limited number of nest-mates and adjust their speed following these interactions. These simple rules lead to the formation of a bistable switch on the level of the group that allows the distinction between recruitment and random noise in the nest. A consequence of the mechanism we propose is a negative effect of ant density on exit rates and recruitment success. This is, indeed, confirmed by our measurements.

Figure 1.

Recruitment behaviour. (a) A snapshot of the experimental arena. The entrance chamber of the nest (d = 9 cm) is at the left and the fixed food item at the bottom right. The red line shows the recent trajectory of one ant. (b) The number of different ants outside the nest during 1 min periods as averaged over (n = 36) experiments. Time is aligned such that t = 0 is the entry of the first recruiter to the nest. The recruiter herself is excluded from the count. The nonlinear rise at t = 0 indicates recruitment. (c) The effect of a recruiter on the number of ants that exit the nest is compared with the effect of a spontaneously (i.e. not immediately following an interaction) moving ant. To isolate the effects of these ants, we take into account events where only all other ants were initially immobile. Results shown are for nests with four to eight ants (n = 229 events). The number of initially immobile ants that left the nest in the minute following recruiter entry is sevenfold higher than the number of ants that exit following a spontaneous movement.

Figure 2.

Pairwise interactions and speed. (a) Rates of events experienced by an ant in the nest are a function of her speed. (b) An example of a contact-dependent, pairwise interaction: a recruiter entering the nest contacts the abdomen of an immobile ant, causing her to commence movement. Panels (i–iii) in (b) are 8 s apart and are overlaid with the trajectory of each ant in the preceding 8 s. (c) The probability per time unit (of approx. 5 s) that an ant starts moving as a function of the speed at which she will move. Movement is more likely to happen directly following an interaction than not (n = 52 330, 5 s periods). (d) Interactions have an averaging effect on the speeds of non-recruiter ants. Solid lines specify the speed distributions of fast and slow ants just before interaction, whereas dashed lines specify those just after interaction. Vertical lines denote the corresponding distribution's mean. (Distributions were calculated over all interactions in which at least one of the ants changed her speed, n = 1722 interactions.)

Figure 3.

Information content of interactions. (a) Reactions of immobile non-recruiter ants to interactions with another non-recruiter (blue circles) or recruiter (green circles) ant moving at a given speed. (i) Fraction of immobile ants that start moving after the interaction. (ii) The speed of the immobile ants that became mobile after the interaction. The speed change depends on the speed of the ant contacted but not on whether this ant is a recruiter or not (n = 636 interactions). (b) Smoothed histograms of speed responses of immobile ants to an interaction with either a fast (more than 8 cm s⁻¹; blue) or slow (less than 1 cm s⁻¹; cyan) moving ant. Distribution means are specified by vertical lines and correspond to the data given in (ii) of (a). The two distributions are statistically different (Kolmogorov–Smirnov test, p < 0.05). However, the large overlap makes the attribution of a single sample to one histogram or the other ambiguous.

Figure 4.

Differentiating recruitment from random nest activity. (a) Speed distributions of recruiters (green) and non-recruiters (blue) as they approach an interaction within the nest. Recruiters tend to be faster than non-recruiters (n = 2693 interactions). (b) The mean linear speed shift owing to interaction (speed after interaction minus the speed before interaction), as a function of the speed before the interaction. While recruiter (green) speed changes only marginally following an interaction, the speed of a non-recruiter (blue) typically decreases (n = 4912 interactions). (c) The mean linear speed shift of ants within the nest following an interaction (blue), an interaction with a relatively fast ant (cyan) (speeds over 11 cm s⁻¹, typical of a fast recruiter), and no interaction at all (red), i.e. spontaneous shift (n = 59 940 events).

Figure 5.

Group-level phenomena. (a) The fraction of interactions that a moving ant has with immobile ants increases with the number of ants in the nest (blue symbols, n = 2497 interactions). Although the total number of interactions a recruiter has within the nest in a single bout slightly increases with group size, this effect is marginal. In fact, the number of interactions between the recruiter and a given ant decreases as group size increases (red symbols, n = 108 recruiting bouts). (b) The number of different ants that have been outside the nest during the 10 min after the return of the first recruiter given as a function of the total number of ants in the experiment (n = 34 experiments). The three points with 0 s.e.m. signify single measurements. (c) The probability that the first ant's exit from the nest had already occurred as a function of the time from the beginning of the experiment (n = 96 experiments). Each trace signifies a different number of ants in the nest as specified in the legend. (d) A schematic summarizing the feedback loops that regulate collective behaviour and suppress noise as suggested by our model. The diagram reduces the ants’ state from continuous values of speed into three discrete states: ‘slow’, ‘fast’, and ‘recruiter’. A fourth, ‘forager’, state relating to ants that have left the nest but did not yet find the food is ignored for simplicity. Non-filled arrows depict the possible transitions between states, and thin arrows signify how interactions between ants change the rate at which ants transition between these states. For example, a larger number of ants in the slow state induce an increased suppression of the rate at which slow ants transition to the fast state. The feedback loops between the slow and fast states constitute a bistable switch that is initially skewed towards the slow state. It is this dissipation that prevents fast movers from retaining their speed and protects the switch from erroneous activation. The appearance of ‘recruiter’ ants in the system reduces the dissipative skew of the bistable switch and allows for a recruitment process. Finally, recruited ants become recruiters and allow for a positive feedback loop as long as the food item is present.