Collecting permits were issued by the following agencies: Brazil, Ministério do Meio Ambiente, Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio #17974-3), and Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis (IBAMA #10148-1); Bolivia, Ministerio de Medio Ambiente y Agua (MMAyA VMA-DGBAP #1319); Perú, Direccion General de Forestal y Fauna Silvestre (#002-2008-INRENA-IFFS-DCB); Ecuador, Ministerio de Ambiente (scientific research authorization #007-14 IC-FAU-DNB/MA); Argentina, Secretaría de Ambiente de la Provincia de Córdoba (#SECA01-524433053-908). Specimens from Colombia belong to registered biological collections and did not require of specific permissions for this study. Specimens from Venezuela are part of old collections for which no permissions could be traced. This study did not involve any endangered or protected species.

Ananteris specimens were collected in Argentina, Bolivia, Brazil, Colombia, Peru and Venezuela (details of collection localities in S1 Appendix) by turning stones during the day, or by ultraviolet (UV) light detection at night [27], using portable UV lamps, comprising mercury vapor tubes attached to a chromium reflector, and powered by a 12V, 7 Amp/hour battery, or Maglite flashlights modified with UV LED attachments. Autotomy was recorded when part of the metasoma was shed by an individual during collection. When autotomy occurred, the sex, age (adult or juvenile), and site of detachment (metasomal segments on either side of the cleavage plane, Fig. 1A), were noted.

The scorpion collections of several museums were searched for Ananteris specimens that had undergone autotomy prior to collection (complete list of material examined in S1 Appendix). Autotomy was recorded as present when part of the metasoma (including telson) was missing, and a brown scar was evident at its severed stump (as Ananteris specimens become brittle after ethanol fixation, we could not be confident that autotomy had occurred if the metasoma was broken but no scar was evident). The species, sex, age, site of detachment, and presence of a scar, were noted for each specimen. The incidence of autotomy, i.e., the percentage of scorpions with part of the metasoma (including telson) missing and a scar present at the severed stump, in a single population was calculated for those species with more than ten specimens from a single collection locality. The results for different populations of the same species were not combined because it was assumed that the incidence of autotomy could vary among localities due to ecological factors (e.g., presence and abundance of predators). Differences among sexes and stages (adult males vs adult females vs immatures) were analyzed as 2×3 contingency tables with Fisher’s exact test in R v. 3.1.1 statistical package [28].

Twenty-five adult males and five adult females of Ananteris solimariae Botero-Trujillo & Flórez, 2011 were collected for experimentation from a population at Girón, Santander Department, Colombia. One of the females, gravid when captured, gave birth, and two second instar juveniles from her litter were also used in the experiments.

Scorpions were housed separately in plastic containers (7 cm diameter; 10 cm height) with moistened cotton as water supply and fed every two weeks with crickets, Gryllodes sigillatus Walker, 1869. After 7 to 10 days of acclimation, each adult scorpion was placed separately in a plastic container with a rough surface (humid cotton), and subjected to the following treatment, designed to demonstrate if detachment was provoked by external stimuli and achieved by an intrinsic mechanism. The metasoma was held with forceps on segments III, IV or V and gently pulled backwards to simulate capture by a predator, for no more than 30 seconds (sufficiently short duration to categorize the phenomenon as autotomy). The time to detachment of the metasomal segments was noted, if applicable. In order to avoid damaging the small second instar juveniles with the forceps, the posterior part of the metasoma was instead attached to sticky tape, which was pulled backwards with the forceps. Ten adult males were subjected to the same treatment, but using glass Petri dishes as a substrate, and the same specimens were held in the air for 30 seconds, without being allowed to contact any substrate. Experiments were filmed with a SONY Cyber-shot DSC-W35 camera and photographs taken with a Canon EOS Rebel T2i camera fitted with a 50 mm macro lens. The healing process of the wound was also documented, by photographing the development of scars on the severed stump of the metasoma during successive time intervals. The effect of anesthesia on autotomy was also investigated, to confirm whether autotomy is mediated by the nervous system, by placing another five adult male individuals of A. solimariae in a styrofoam box with ice, prior to manipulation of the metasoma. Differences between the experiments (adult males on rough substrate vs adult females on rough substrate; adult males on rough substrate vs adult males in Petri dishes; adult males on rough substrate vs adult males held in the air; adult males on rough substrate vs anesthetized adult males on rough substrate) were analyzed as 2×2 contingency tables with Fisher’s exact test in R v. 3.1.1 [28].

Additional experiments, identical to those performed on A. solimariae, were conducted with 92 live individuals (32 adult males, 32 adult females, 28 immatures) of ten other scorpion species in seven genera and three families: Bothriuridae Simon, 1880: Bothriurus cordubensis Acosta, 1995: 3 adult males, 2 adult females; Bothriurus flavidus Kraepelin, 1911: 2 adult males, 1 adult female, 3 immatures; Brachistosternus ferrugineus Thorell, 1876: 4 adult males, 5 adult females, 5 immatures; Timogenes elegans Mello-Leitão, 1931: 4 adult males, 1 adult female; Timogenes dorbignyi Guérin Méneville, 1843: 3 adult males, 1 adult female; Urophonius brachycentrus Thorell, 1876: 2 adult males, 4 adult females, 3 immatures; Buthidae C.L. Koch, 1837: Tityus trivittatus Kraepelin, 1898: 5 adult females; Zabius fuscus Thorell, 1876: 8 adult males, 6 adult females, 4 immatures; Zabius birabeni Mello-Leitão 1938: 4 adult males, 2 adult females; Hormuridae Laurie, 1896: Opisthacanthus elatus Gervais, 1844: 2 adult males, 5 adult females, 13 immatures. The bothriurid and buthid specimens were collected at several localities in Córdoba Province, Argentina, whereas the O. elatus specimens were collected in Santander Department, Colombia, in 2012.

In order to determine whether a defined cleavage plane exists in the metasoma, six ethanol-preserved specimens of three buthid species (two adult specimens per species), Ananteris balzani Thorell, 1891, Tityus uruguayensis Borelli, 1901 and Z. fuscus, the last two species included for comparison with A. balzani, were manipulated with forceps to induce detachment of the metasoma between segments II and III, and III and IV. Scanning electron micrographs (SEM) were taken of the sites of detachment in these specimens, as well as of the scarred, post-autotomy metasomal segments of two A. solimariae specimens, with a Philips XL30 TMP SEM. Samples for SEM were dehydrated and coated with gold-palladium in a Thermo VG Scientific SC 7620 sputter coater.

The behavior of A. solimariae specimens post-autotomy was recorded in the laboratory and, when possible, compared with the behavior of intact (i.e., pre-autotomy) specimens. In order to assess the effect of losing part of the metasoma on male mating success, four mating trials were conducted with two adult females and four adult males, two with the metasoma intact and two without the last three metasomal segments. Each pair was placed in a mating arena (20 × 40 × 30 cm) with a substrate comprising soil, stones and pieces of tree bark from the collection locality. Mating behavior was observed and filmed under a 40 W red lamp. Two mating trials were conducted per female. The first two trials, involving males with an intact metasoma, were conducted when the females were gravid. The second and third trials, involving post-autotomy males, were conducted several months later, after both females had given birth and the juveniles had left their mothers.

Evidence of autotomy was found in fourteen species of Ananteris (Table 1, S1 Appendix). In several species, part of the metasoma readily detached during capture in the field: holding the metasoma for a few seconds was sufficient to induce the scorpion to shed part of the metasoma, consistent with autotomy. In one case, involving an adult male A. balzani from Mato Grosso, Brazil, the detached metasoma writhed intensely, as if attempting to sting, for about one minute. Autotomy was not observed in A. solimariae specimens anesthetized with ice prior to manipulation, or when A. balzani and A. solimariae specimens were held by body parts other than the metasoma (i.e., the prosoma, mesosoma, pedipalps or legs). Other defense behaviors exhibited by some species at the time of collection included fleeing and tanatosis.

Most cases of autotomy involved adult males (n = 23). The phenomenon was observed in only six adult females and not in immatures. The metasoma detached most frequently between segments III and IV (Table 1, n = 33), less often between II and III (Table 1, n = 15), and only twice between I and II (Table 1).

Incidence of autotomy in the field for five populations of four species was low, from 5.26 to 8.33% (Table 2). Only adult males were observed with part of the metasoma missing (10 out of 96). The metasoma was intact in all adult females (46) and immatures (12) observed. The frequency of autotomy was significantly greater among adult males than adult females and immatures across all species (Fisher’s exact test, p = 0.04605), but the difference was not significant within each species (p > 0.05). Autotomy affected 8.33 to 14.29% of the adult males in the population.

Under laboratory conditions, 22 of 25 adult males and one of five adult females of A. solimariae shed part of the metasoma after being held with forceps for 30 seconds or less, whereas three adult males, four adult females and both immatures did not. The frequency of autotomy was significantly greater among adult males than adult females (Fisher’s exact test, p = 0.005796).

Autotomy only occurred when scorpions were in contact with a rough substrate with their pedipalps and/or legs. Autotomy did not occur when scorpions were in contact with smooth, slippery surfaces (glass Petri dishes, n = 10), when held in the air (n = 10), or when anesthetized with ice prior to manipulation (n = 5). The frequency of autotomy was significantly greater among adult males on rough surfaces (22 out of 25) than adult males on Petri dishes (0 of 10), adult males held in the air (0 of 10) and anesthetized adult males (0 of 5) (Fisher’s exact test, p < 0.0001 in all cases).

Time to detachment of the metasomal segments, after contact with the forceps (Fig. 1B), varied from 0.29 to 12.99 seconds (median, 4.68 seconds). The metasoma detached most frequently between segments III and IV (n = 16) (Fig. 1C), less often between II and III (n = 4), and rarely between I and II (n = 2). Cleavage occurred between metasomal segments III and IV if the specimen was held with the forceps on segments IV, V, or on the telson, between segments II and III if held on segments III or IV (see S1 Movie), and between segments I and II if held on segment III. A lateral twisting motion of the metasomal segments anterior to the cleavage plane, immediately preceding detachment, was interpreted as part of the process of metasomal autotomy (see S2 Movie). Autotomy occurred rapidly, requiring minimal stimulus with the forceps, in some individuals whereas others only shed the metasoma after several attempts to escape (see S3 Movie).

The detached metasoma of some individuals carried part of the digestive tract of the preceding metasomal segments along with it (see Movies S1–S4).

After autotomy, the detached metasoma writhed vigorously, as if attempting to sting, for some time (8.8–169.95 seconds; median, 49.32 seconds). If touched, the telson reacted to stimuli, attempting to sting in 47% (n = 19) of cases (see Movies S1 and S4). Immediately after autotomy, a small drop of hemolymph was evident at the site of detachment (Fig. 2A). Some loss of hemolymph occurred from the wound during subsequent days, but reduced with time, with no further loss after five days (Figs. 2B-F). The process of scar formation was rapid. One day after autotomy, a brown spot appeared medially, apparently produced by hemolymph coagulation, eventually darkening to form a blackish brown scar (Figs. 2G-H) which completely blocked the digestive system, preventing defecation.

Post-autotomy survival rate was high. All individuals were alive three weeks afterward and only one male (5%; n = 22) was dead by day 25. The remainder survived up to 8 months in the laboratory. No evidence of autotomy was observed in the other scorpion taxa tested.

Scorpions offered prey every two weeks after autotomy only fed on small crickets (up to 5 mm, adult males of A. solimariae range from 25–27 mm in total length, whereas adult females reach up to 34.32 mm in total length) using the pedipalps and chelicerae to grab and consume them alive (Fig. 3A). After 20–25 days, the opisthosoma of these individuals had become swollen due to the accumulation of excrement visible ventrally through the sternites in the posterior part of the mesosoma (Figs. 3B, C). In two cases, a second autotomy was observed. The first case was apparently spontaneous (no prey was present when it occurred, and the round, plastic container contained nothing which could have caused the specimen inside to become stuck) and occurred between metasomal segments I and II, both of which had remained (along with segment III) following detachment of segments IV and V, eight months earlier. In this case, autotomy was apparently induced by the internal pressure of the excrement because, without having been subjected to manipulation, metasomal segments II and III were found to have detached and a new scar developed at the severed stump on segment I. The detached segment was empty, and did not contain living tissue. The second case of autotomy, also between segments I and II, was induced by holding segment III (the segment that remained after the first autotomy) with forceps. In both cases, several drops of white excrement were expelled and a new scar developed at the end of the segment, several days after detachment.

After autotomy, scorpions attempted to use the remainder of the metasoma to sting prey and for defense, when captured, as if the metasoma and telson were fully intact (Figs. 3A, B). Attempts to attack prey of moderate to large size met without success and these individuals could only capture small prey with the pedipalps and chelicerae. Other behavior was similar to that observed prior to autotomy, except for a grooming behavior performed with the telson and posterior segments of the metasoma (lubricating these segments with fluid from the mouthparts, and using them to clean the body), which was no longer possible. The single adult female that underwent autotomy, was apparently not gravid.

The first two mating trials with post-autotomy males and gravid females terminated rapidly after female aggression led to stinging and cannibalization of the males. In contrast, two complete courtship rituals, both ending in successful sperm transfer, were observed with post-autotomy males and post-parturition females (see S5 Movie). At the onset of courtship, after making contact with the female, the male curved what remained of his metasoma, performing balanced movements from side to side. Afterward, the male grasped the female’s pedipalps with his, and guided the courtship ‘dance’, walking backwards. On several occasions, the couple interrupted the dance, and the male grasped the female chelicerae with his (cheliceral ‘kiss’; [29]), without releasing her pedipalps. After locating a suitable substrate (rock or dry leaf), the male deposited his spermatophore and moved slightly backwards, guiding the female over it. During the course of guiding the female over the spermatophore, which lasted 70 to 85 seconds, and while continuing to grasp her pedipalps, the male moved rhythmically, by rapidly vibrating his pedipalps and gently pushing the female every three seconds, touched the ventral surface of the female’s prosoma and genital operculum with his first legs (‘rubbing with legs’; [29]), and touched her chelicerae and anterior carapace margin with his chelicerae. When sperm transfer was complete, the male disengaged the pedipalps and ran away, after limited aggression by the female. Neither female ate the spermatophore. Except for the first contact with the female, neither male made any further attempt to use the severed stump of the metasoma. As a complete courtship ritual with the intact males was not observed, and courtship behavior has not been previously described in any Ananteris species, it is impossible to know whether male Ananteris perform ‘clubbing’ (striking the partner with the metasoma while the sting is tucked away [29]) or the ‘sexual sting’ (male punctures the female’s body with his aculeus [29]).

Scanning electron micrographs of the metasomal segments of ethanol-preserved specimens of A. balzani revealed a cleavage plane between segments II and III and between segments III and IV (Figs. 4A, B). The severed surfaces exhibited sharp edges (indicative of a clean break) in the tegument of the intersegmental membrane, almost without a trace of the severed metasomal muscle or digestive system. In contrast, similar tests on other buthid scorpion species revealed ragged edges in the tegument of the intersegmental membrane, with part of the muscles and digestive system protruding (Figs. 4C, D). The wound on the severed stump of a post-autotomy A. solimariae displayed a large star-shaped scar closing the posterior end of the segment (Fig. 4E), whereas the severed surfaces exhibited sharp edges, without any structures protruding (Fig. 4F).