The surface morph of the Mexican tetra fish (Astyanax mexicanus) exhibits strong territoriality behavior and high levels of aggression. In contrast, the eyeless cave-adapted morph from Sierra de El Abra, México, rarely are aggressive and have totally lost the territorial behavior. These behaviors are part of what has been called the cavefish behavioral syndrome. Here, we report that several Astyanax cave populations of Sierra de Guatemala, unlike those reported for the Sierra de El Abra cave populations, display significant territoriality and aggression when confined into a reduced space. We discuss divergent evolutionary trajectories in terms of agonistic behavior for cavefish populations inhabiting different mountain ranges.

Abra, behavior, stygobyte, troglobite

Organisms in which variation in behavior can be linked to the environment and their evolutionary history are key to understanding how behavior evolve. The Mexican tetra Astyanax mexicanus has both an eyed surface morph and a blind cave-adapted morph. The cave-dwelling Astyanax are characterized by conspicuous morphological traits that have evolved in response to their subterranean environment. Among these traits, the most obvious are degenerate eyes and reduced pigmentation (Jeffery 2001; Krishnan et al. 2017). Behavioral differences are also found between the morphs, such as a reduction of intraspecific aggression in some cave-adapted Astyanax (Burchards et al. 1985; Parzefall 1985).

Astyanax genus has emerged as a powerful model system for genotype-phenotype analysis because surface fish and cavefish morphs are interfertile and high-quality genome information is available (Gross 2012; Warren et al. 2021). Furthermore, the molecular basis of repeated trait evolution can be examined in this species, as multiple cave populations have independently evolved the same traits (Jeffery 2001; Jeffery 2009). Thirty-four caves are currently known to be inhabited by Astyanax (Espinasa et al. 2020). They are found in three mountain ranges in Northern México (Sierra de Guatemala, Sierra de El Abra, and Micos), and one mountain range in Southern México (Guerrero; Granadas and La Joya). Multiple authors (e.g. Bradic et al. 2013; Gross 2012; Herman et al. 2018) have suggested that each mountain range has independent evolutionary histories, although colonization independence has been disputed for the Sierra de Guatemala and Sierra de El Abra by Espinasa et al. (2014). All cave localities (Molino, Caballo Moro, Jineo, Vásquez, and Escondido) involved in this study come from Sierra de Guatemala, with the exception of the Pachón and Tinaja cave populations, which are from Sierra de El Abra (Fig. 1).

Figure 1. 

Localities of cave populations A satellite photo of Sierra de Guatemala (Red) and Sierra de El Abra (Blue). Inset shows the area of México. Map data 2021 (C) Google B topographic map modified from Elliot (2018). Sierra de Guatemala caves are shown in red while Sierra de El Abra caves in blue

Burchards et al. (1985) and Parzefall (1985) showed that epigean specimens are intensely aggressive and develop territoriality in the laboratory. On the contrary, in the Sierra de El Abra cave populations, aggressive behavior is highly reduced, biting seldom occurs, and fish have lost territorial behavior (Langecker et al. 1995). Even the few attacks performed by cavefish during the behavioral tests have been linked to food-seeking behavior (Elipot et al. 2013). In the natural hybrid populations of Micos and Guerrero, aggression levels are also reduced (Espinasa et al. 2001). The virtual absence of aggressive behavior in these cavefishes has a genetic basis. F1 and F2 hybrids resulting from crosses between epigean and Sierra de El Abra cave specimens exhibit a very high to intermediate level of aggression, respectively (Parzefall and Hausberg 2001; Elipot et al 2013). In backcrosses with the cave morph, the aggressive behavior disappears (Wilkens 1988). Of note, F1 hybrids from crosses between Molino cavefish (Sierra de Guatemala) with Surface fish are exceptionally aggressive (Elipot et al. 2013).

Breder (1943) first described the aggressive behavior characteristic of surface Mexican tetra fish as “erratic viciousness”. Among the observable agonistic behavioral patterns are “fin spreading”, “snake-swimming”, and “tail striking”, with the most prominent being “ramming” and “biting” (Wilkens 1988). Subordinate individuals with no opportunity to hide or escape may eventually be killed. The intensity of the aggression depends on many factors, particularly an individual’s size (John 1964), the amount of space available (John 1964), and the overall state of nutrition (Wilkens 1988; Elipot et al. 2013). Vision is not essential to trigger or release aggressive behavior (Espinasa et al. 2005). The agonistic behavior establishes a hierarchical order among the fish (Elipot et al. 2013).

Aggressive interactions and territoriality are correlated, with fish defending an individual space. Astyanax surface fish show territoriality that is dependent upon the aquarium space available. In small tanks less than 250 L, surface fish stop schooling and start defending territories (Burchards et al. 1985; Parzefall 2000). Hierarchy and territoriality are linked, with the dominant and most aggressive fish defending the largest area of the tank (Burchards et al. 1985). However, no territoriality behavior has been reported in the cave populations (Parzefall 1983).

The intense dominance-related aggressiveness in surface Astyanax is inversely correlated with the serotonin (5HT) levels in the hindbrain raphe nucleus (Elipot et al. 2013; Rétaux and Elipot 2013). A mutation has been identified in the monoamine oxidase (MAO) coding sequence of several cave populations from Sierra de El Abra (Elipot et al. 2014). The P106L MAO mutation would appear as an ideal candidate to serve as the genetic basis of reduced aggression. However, experiments using surface fish and Pachón cavefish lines with or without the P106L MAO mutation have demonstrated that even though MAO P106L does increase 5-HT levels in fish, it does not affect aggression and cannot be considered as the genetic basis for the loss of aggressive behavior in cavefish (Pierre et al. 2020). The P106L mutation is present in Sierra de El Abra cavefish but absent in Sierra de Guatemala (Molino, Caballo Moro, Jineo, and Escondido cavefish) as well as in surface fish (Pierre et al. 2020).

Since populations in three mountain ranges (El Abra, Micos, and Guerrero) have independently converged in a reduction of aggression and territoriality, it has been suggested that the loss may be an adaptive change for cave Astyanax (Hinaux et al. 2015). The purpose of this study is to resolve whether troglomorphic Astyanax can retain aggression and territoriality by studying the Sierra de Guatemala cave populations.

Cavefish and surface fish localities used in this study can be seen in figure 1. Quantitative data was gathered from fish within four days after being collected from the field in 2016 (Pachón cave collected on August 6^th, Caballo Moro cave on May 18^th; and Rascón surface fish on August 5^th). Data was also gathered from Molino, Vásquez, Caballo Moro, Jineo, and Escondido cavefish, kept in P. Ornelas’ laboratory at IBUNAM, México, which had been collected on February 2016, four months prior to the study. Choy surface river and Tinaja cavefish originally bred in Borowsky’s laboratory and now kept in Espinasa’s laboratory were also analyzed. Fish kept in the laboratory have been under standard husbandry with a water filtration and aeration system. Fish have been fed about 3% of the fish’s bodyweight daily. Fish used in experiments were not fed on the 24 hrs before the experiments. Since the Caballo Moro cave holds a mixed population of eyed and eyeless fish, only completely eyeless Caballo Moro specimens were used (Fig. 2). All other cave populations studied only have eyeless fish. Collecting permit # SGPA/DGVS/02438/16 from Secretaría del Medio Ambiente y Recursos Naturales, México, was issued to Patricia Ornelas García. To unify criteria of what constitutes an attack, one of us (LE) went to Rétaux (SR) laboratory to personally observe the pattern of behavior of the specimens described by Elipot et al. (2013) and kept at the French research facility. Afterwards both LE and SR observed the fish kept at Ornelas’ Laboratory in México. All counts were performed by at least two observers at the same time to minimize bias. Counts were always very similar and were averaged to present the data.

While most authors agree that aggression in Astyanax cavefish populations from Sierra de El Abra, Micos, (reviewed in Hinaux et al. 2015) and Guerrero (Espinasa et al. 2001) is highly reduced when compared to surface fish, there have been conflicting accounts for the populations in Sierra de Guatemala. Elipot et al. (2013) reported that both Molino and Pachón fish showed ten times less aggression than surface fish during an hour-long resident-intruder assay. In contrast, an earlier study (Espinasa et al. 2005) reported that Molino cavefish had the “erratic viciousness” equal to surface fish and were clearly more aggressive than the Pachón cavefish.

In the present study, cavefish were directly observed by authors of both the Elipot et al. (2013) and Espinasa et al. (2005) studies to standardize criteria. Five populations were examined (Molino, Caballo Moro, Jineo, Vásquez, and Escondido). Assays used were for two to four individuals. Fish were observed with or without acclimation (like in resident-intruder assays), and with individuals that had just been collected from the field or that had been kept in the laboratory for extended times. In all these tests, 4 out of 5 Sierra de Guatemala cave populations (except Escondido cavefish) have displayed levels of aggression significantly higher than Sierra de El Abra cavefish. The Molino fish have similar histological characteristics of eye reduction to El Abra cavefish (Wilkens 1988). So, their higher levels of aggression are not due to having significantly better vision.

Espinasa et al. (2005) reported that in a “three fish assay”, Molino cavefish had aggression levels over a 10 min period equivalent to surface fish from Río Boquillas. Similar results were obtained by us when comparing against the surface population of Choy. Yet, in our study, surface fish from Rascón were 2 to 5 times more aggressive than both the surface Choy population and the aggressive Sierra de Guatemala cave populations. Two interpretations seem plausible. The first is an ongoing process to loss the aggression, in which most Guatemala fish display “intermediate” aggressiveness between surface Rascón and cave Sierra de El Abra fish. This corresponds to a plastic behavior whose environmental conditions, phylogenetic inertia, or genetic architecture have not promoted the genetic fixation in comparison with El Abra system, while in Escondido the lack of aggressiveness has been favored.

But there is a second alternative that focuses on the variability in aggression levels among surface populations. Sierra de Guatemala cavefish are as aggressive as surface fish from Río Boquillas and Choy, but less aggressive than those from Rascón. Qualitative and quantitative observations in the fish kept at the UNAM laboratory suggest that Rascón fish are more aggressive than other surface populations from streams closer to Sierra de El Abra. While few live specimens are available, tanks with Rascón fish typically need to have reduced number of individuals so as to prevent them from killing each other. Rascón population is isolated from the rest of the surface populations by the 100 m high Tamul waterfall and its known to be genetically distinct by harboring a different mitochondrial haplotype from other surface populations (Ornelas-García et al. 2008) as well as with nuclear data (Herman et al. 2018). So, it could be that Rascón is a particularly aggressive population of surface Astyanax and the Sierra de Guatemala cavefish are simply as aggressive as their neighboring surface populations of Choy and Boquillas.

Our study also confirmed that the absence of aggression in the three Molino individuals kept in France used in the Elipot et al. (2013) study was “real” and not an artifact of subjective evaluation or different data collection techniques. We propose that apart from variability among populations, there is variability among individuals within a population regarding their aggression levels. For stochastic reasons, the handful of specimens that were shipped to France displayed little aggression and represented a divergent subset of the natural population with regards to average aggression. In fact, while the norm is that levels of aggression in Sierra de Guatemala cavefish are higher than in Sierra de El Abra, the absence of aggression displayed in the Escondido cavefish housed in Ornelas’ laboratory in México may be the result of this same natural variability. Escondido cave is part of the same hydrologic system as Molino and Jineo Cave (Espinasa and Espinasa 2015). It remains to be seen if for stochastic reasons the pool from which Escondido cavefish were collected happened to have a high number of non-aggressive fish, or if it is the norm throughout that particular cave.

It has been suggested that agonistic behaviors in surface Astyanax rely on visual cues and that aggressiveness is no longer performed in darkness at all (Burchards et al. 1985; Parzefall and Hausberg 2001; Langecker et al. 1995). In contradiction, Elipot et al. (2013) reported that surface fish remain aggressive in the dark. These differing findings come about by the fact that Elipot et al. (2013) observed the fish in darkness by using infrared lamps. This method does not consider that many fish (including characids) are able to perceive near-infrared (Meuthen et al. 2012). It is interesting to note that surface fish that have been blinded by surgical lens removal at larval stages are aggressive (Espinasa et al. 2005).

Our analyses of the swimming paths indicate that unlike Sierra de El Abra cavefish, most Sierra de Guatemala cavefish develop territories. Individuals appear to try to minimize contact with each other and the available space is subdivided. The borders of these territories are where most attacks are concentrated. Again, it is worth noting that the cave specimens in these assays were completely eyeless individuals and thus they could not use visual cues to generate a spatial mapping of the tank and their territories. Nonetheless, individuals adjusted their swimming paths to form new and stable territories without the help of visual cues.

Swimming paths of most Sierra de Guatemala cavefish kept inside the laboratory are drastically different from Sierra de El Abra cavefish. Pachón and Tinaja fish typically swim near the edges of the tank, following the entirety of the side and turning at the corners, thus going in circles around the whole tank. Aggressive Sierra de Guatemala fish that hold territories have swimming paths that do not follow the entirety of the side of the tank. Instead, they swim a few strokes and turn around before encountering any physical obstruction. Non-visual spatial mapping must allow them to turn at the frontier of their territories, even when the neighboring individual is distant. At high densities, swimming paths can even be restricted to reduced circling around in areas only slightly larger than the fish itself. Teyke (1985) has shown that active flow sensing, using locomotion and the lateral line are utilized during cavefish navigation. Blind cavefish use a higher frequency of swim cycles than eyed surface fish, consisting of burst and glide in straight trajectories. Santacruz et al. (2015) have suggested that three-dimensional space is learned and remembered differently in Astyanax surface fish and cavefish. It is likely that the modifications to the sensory, processing, and motor neural system of the Sierra de Guatemala cavefish must have occurred such that they are equipped for navigation in perpetual darkness to generate territoriality without the need of visual triggering stimuli.

Based on studies of the Pachón and Micos cave, the surface fish, and F1-hybrids between surface and Pachón fish, Langecker et al. (1995) revealed two releasers for aggressive behavior, ‘natural shape’ and ‘locomotion’. Both would differentially evolve in A. mexicanus during the transition from a surface to a subterranean life style. The releaser ‘natural shape’, on the one hand, cannot be manifested in blind cave specimens and became a subject to regressive evolution because of the lack of stabilizing selection. The releaser ‘locomotion’, on the other hand, may be progressively selected because it can still be stimulated by the lateral line system. It is likely to play a role in the new aggressive behavioral patterns of cave fishes (Langecker et al. 1995).

Elipot et al. (2013 and 2014) proposed that the loss of aggressiveness in the Sierra de El Abra cavefish is a “side effect” and not something to be directly selected by living in caves. Raphe serotonergic neurons control aggressive/dominance behavior while at the same time, hypothalamic serotonergic paraventricular neurons regulate food-seeking behavior. They suggested that evolution of the Sierra de El Abra cavefish, in seeking to enhance feeding, lost aggression as a pleiotropic side effect due to modifications of their shared serotonergic neuronal network. We propose that the Sierra de Guatemala cavefish may have achieved the regulation of adaptive feeding behaviors or the control of serotonergic networks through different evolutionary, genetic, and physiologic pathways (e.g., mc4R mutation in Molino cavefish; Aspiras et al. 2015). Thus, the pleiotropic and side effects are different such that aggression and territoriality were maintained.

Espinasa et al. (2021) have found that variability of food resources among different pools within a single cave can be higher than between different caves. Thus, we can assume that it is unlikely that caves in Sierra de Guatemala are environmentally significantly different from those in Sierra de El Abra with regards to selective conditions for loss of aggression and territoriality. We suggest that the behavioral differences are rather due to their independent evolutionary histories, where adaptation occurred through modification(s) of different genes with different pleiotropic effects. Future studies may resolve if aggression and territoriality, or their absence, evolves differently among karstic zones.

Several Astyanax cave populations of Sierra de Guatemala, unlike those reported for the Sierra de El Abra cave populations, display significant territoriality and aggression when confined into a reduced space. The Sierra de Guatemala cavefish may have achieved the regulation of adaptive feeding behaviors or the control of serotonergic networks through different evolutionary, genetic, and physiologic pathways than at Sierra de El Abra populations. Thus, the pleiotropic and side effects may be different such that aggression and territoriality were maintained.

Jae Wong helped process the videos. Support for a short sabbatical stay at CNRS in France to LE was provided by a VPAA grant from Marist College and by an ANR (Agence Nationale pour la Recherche) grant ASTYCO to SR. Work supported by an CONACYT, ECOS-NORD, exchange grant between SR and POG. We thank the support from the CONACYT project N191986 to POG. Zsuzsanna Sidlo provided comments on the manuscript. The authors have declared that no competing interests exist.