(a) “Hitchhiking” for dispersal
Remarkably, despite tens of thousands of known insect fossil inclusions, our report corresponds to about 6% of all the previous occurrences of springtails ever reported in Dominican amber and consists of two taxa out of twelve described in total [32–34]. As this association is unique among many comparable specimens of termites from La Cumbre, it cannot represent the general abundance of these springtails in termites’ environments but must correspond to a more specific biotic explanation.
Other symphypleonan fossil associations are known from amber, beginning with two cases of Eocene harvestmen legs supporting up to five individuals arranged in a row and clasped by the antennae (Cholewinsky pers. comm. in ; Fig. 4). While the springtails were differently positioned (either facing the leg or facing away from it), they all show antennae distinctly bent toward this appendage, suggesting antennae were initially secured to it. The resulting position was interpreted as their immediate detachment following resin entrapment . Associations with winged insects are described from attachments on the forewing base of a mayfly (Miocene ) and the leg of an false blister beetle (Eocene ; Fig. 4C). In the case of the beetle association, antennae were implicated, as were the mouthparts which may have grasped onto the leg surface (or perhaps tibial setae, ). Thus, all previous cases of preserved attachment of springtails (excluding that of the mayfly association) were described from smooth cylindrical legs. The associations here reveal springtails with a positioning attachment preserved on novel structures consisting of insect antennae and wing margins. In those cases, we observe that the attachment is also achieved by rolling up of the fourth antennomere around termite antennae/sclerotized wing margins, implicating this structure in the general phoretic ability of the springtails. Possible variations in grasping mechanisms have been briefly addressed, stressing the attachments of both a long-antennomere species (Sminthurus longicornis†) to a 25 µm-thin leg, and that of a medium-antennomere species (S. sp.) to a 100 µm-thick leg [22, 23]. The herein case confirms these possible variations revealing the attachment of a much smaller species with quite short antennomeres grasped onto 20-40 µm thick structures. We stress that this behavior must have been even more dependent on antenna/body lengths. Indeed, in Poduromorpha and even more in Neelipleona morphotypes the antenna/body length ratio is less than 1:3 (Fig. 3). And if many Entomobryomorpha possess longer antennae – reaching a third to half of their total length – their limited antennal subdivision would preclude any antennal attachment that could support their total body weight (see , Fig. 3). Attachment efficiency may be restricted to symphypleonan morphotypes that correspond to a short bulbous body with (1) antennae reaching a third to half of the body length and a (2) fourth prehensile antennomere with subdivisions consisting at least 1/3 of the antenna length (Fig. 3). As an example, in E. helibionta sp. nov. the fourth antennomere is divided in eight sub-segments representing in total 60% of the total antennal length. Variations in length and subdivision of fourth antennomere among Symphypleona might have driven specificities in host selections/attachments, their increase probably contributing to some more generalist behaviors. Our observations finally confirm the involvement of mouthparts as a secondary important element to the primary antennal attachment, grasping surfaces or acting as an abutment.
Antennae appear to be key for phoretic attachment, however they are subject to significant sexual dimorphism. Unlike other Symphypleona, the second and third antennomeres of male Sminthuridoidea are modified into a clasping organ; this structure is preserved in E. helibionta nov. sp. However, among 25 individuals, only three males (AMNH DR-NJIT001_ss, si; Fig. 2E-F) could be identified in the inclusion. Could this disparity in sex representation have a biological underpinning? Phoretic behaviors restricted to females have been reported in other phoretic wingless arthropods. Aggregations of female mites and pseudoscorpions, sometime identified as gravids, were reported in extant fauna competing for various animal hitchhiking hosts (harvestmen, beetle, diptera, frog) [36–39]. From laboratory experiments, Zeh & Zeh  even demonstrated a female bias in phoresy by pseudoscorpions that increases over time, and with mated females exhibiting a slightly higher rate of phoresy than unmated ones. Given pseudoscorpions’ ecological proximity and ametabolian life cycle, this – rather unexplained bias – could relate to that of phoresy among springtails observed here. The amber piece contains four distinct flying insects, some caught with wings still open, which is suggestive of an arboreal capture in the tree resin for which all trapped sminthuridids would have detached from the ant/termite association. However, the presence of one Isotomida (Entomobryomorpha) in the inclusion, could advocate for a trapping in amber close to the soil, implying the possible trapping of a few non-associated sminthuridids. Given this taphonomic limit, we cannot definitively conclude on a discrepancy in sex representation in this association. Besides, we note that comparable disparity in sex representation have been suggested for the modern symphypleonan Sminthurides in general sampling context .
While migration strategies in springtails are today poorly understood, they have been observed to occur through four vectors: windborne, pedestrian, rafting, and wind propelled on water surface . Because Collembola are highly susceptible to desiccation, it is highly unlikely they are capable of moving over oceanic distances through aerial movements, favouring recently accepted water dispersal option confirmed by experiments [17, 19]. However, the evidence so far of waterborne dispersal only applies to Poduromorpha and Entomobryomorpha; as well as for large “swarm” pedestrian migrations [17, 19, 20, 41, 42]. Symphypleona have so far only been caught in air (up to 3350 m, [18, 43]) and, as strictly terrestrial and freshwater inhabitants, there is no evidence that they could maintain themselves in association with brackish water . However, Symphypleona, which extends into Spanish amber dated to the Lower Cretaceous , display many recently diverged clades (e.g. genera Sminthurinus and Sminthurides) on every continent. This present-day distribution suggests the existence of additional mechanisms for significant dispersal. The type of phoresy evidenced herein from the fossil record may have facilitated this intensive dispersal.
(b) Springtails and phoresy
Previous fossil reports all noted the absence of modern phoresy in springtails, highlighting a questionable discrepancy (lack of modern records vs extinct behavior, [22, 23]). Given the introduced biases affecting the identification of phoresy among commensals, we disagree with that first appraisal and uncover some, so far, hidden cases.
In fact, springtails have been reported to have associations with a diverse assortment of invertebrates, although the nature of the associations has not been determined. A frequently documented case corresponds to the repeated observation of termitophile and myrmecophile inquiline springtails. Different genera of cyphoderid springtails have been observed clinging to the head and back of soldiers and queen termites within nests ([46–48], Fig. 4C). In those cases, specialized sucking mouthparts, their location on soldiers’ heads and their posture – drooping toward the termite labrum – suggest that springtails obtain small meals from trophallactic soldier-worker food transfers. Cyphoderids have also been reported attached to reproductive alate ants (both females and males, ). Food-supply commensalism has been suggested from the observations of springtails (both Entomobryomorpha and Symphypleona) on Palearctic slugs, feeding on their mucus (, Fig. 4C). But springtails are also curiously reported as commensals of the shells of hermit crabs of tropical land environments ([51–57], Fig. 4C). These associations were reported from about a hundred of specimens from Mexico, New-Guinea, the Caribbean (Guadeloupe, Dominican Republic, Saint Croix). In 2005, we observed 80 individuals caught from 7 hermit-crabs and sands from different Martinican beaches (Anse Trabaud, NE of Les Salines, Grand Macabou, Presqu'ile de la Caravelle) evidencing the relative commonness of these associations within the pantropical range of these terrestrial crustaceans (Coenobitidae). The biological purpose of these associations has been addressed as a possible trophic inquilinism  linked to consumption of host food remains or feces; but the actual springtails’ (Coenolatidae) location inside shells or their feeding activity was never observed. Alternatively, these associations characterize specimens collected from beach environments into which the backshore microhabitats of springtails are discontinuously distributed, requiring alternative dispersal options. Associated individuals consist mostly of females and juveniles whereas males represent about 10% of documented samples; which could be linked to an unequal sex distribution in species or a difference in sex involvement when associations relate to transport. These reports of poorly understood commensalisms with other invertebrates argue for the existence of actual phoretic-like behaviors in modern Collembola.
No modern Symphypleona is known to perform attachment to cylindrical structures by use of antennae. It has been suggested that this behavior could have been restricted to extinct lineages of Symphypleona having especially elongate flexible antennae . From the variability of antennae exhibited here, we exclude this possibility. In addition, the herein fossil association reveals the nature of Symphypleona reactions to disturbances. The fossil inclusion preserves an alate termite that could not begin to fold its wings when trapped, implying an almost immediate capture in the resin. In that meantime, most springtails managed to get completely or partially detached of their host revealing the reactive mobility of the antennae, perhaps comparable to that of the furcula. A reflexive detachment may explain the apparent absence of modern phoretic Symphypleona based on modes of collection in the field. Insects and arachnids are collected most frequently by direct immersion in ethanol before being prepared for collections. It is very likely that given their superficial attachment and quick detachment system, phoretic springtails may detach from hosts in ethanol or even before immersion, effectively removing link between host and commensal (see Fig. 5a for comparison with other wingless phoretic arthropods). The tiny proportions of the fossil phoronts relative to all types of soil arthropod host (including ants and termites) numbers would have impeded the finding of modern representatives even when initially collected and kept in alcohol with that host (Fig. 5b). In this context, the immediate embedding of springtails onto insects by tree resin would characterize a unique situation enabling the preservation of these associations, making the fossils of high significance to the documentation of the phenomenon.
(c) Collembola evolving with social insects
Three of the four orders of modern springtails are reported from termite and ant nests [48, 58–60]. This includes cyphoderids primarily described as nests inquilines as well as direct commensals of termites and ants [46–49, 60]. Thus, despite constraints in collecting modern phoretic springtails, other reported cases suggest significant relationships with social insects. Moreover, we report paleoecological insight that further bolsters links between springtails and social insects. Based on an initial screening of more than 1300 inclusions within rough Cambay amber, from the Lower Eocene (54 Ma) of western India, we report only two Symphypleona. Those two individuals, representing different genera, are both located a few millimeters from a single alate termite.
Today, ants and termites together comprise a significant component of many terrestrial ecosystems. The biomass of termites is estimated to be approximately equal to that of humans  and in tropical localities, termites and ants together may outweigh all vertebrates and all other insects combined . It is possible to trace the “rise” of soil-dwelling social insects from their first appearance in the fossil record to their remarkable ecological impact today. The termite and ant fossil records extend to the Lower Cretaceous of Russia (Berriasian) and Charentese-Burmese ambers (Albian, Fig. 4A), respectively. In the Cretaceous, ants and termites never comprise more than 2% of all fossil hexapods by locality [63, 64]. This changes markedly by the Cenozoic. In Dominican amber, termites make up to 6% of inclusions while ants represent ~30% of all hexapods; social insects represent more than 1/3 of the total entomofauna at that time (Fig. 4A). As they constitute the vast majority of soil insects, termites and ants stand as an immediate model for widespread biological interactions in springtails.
Through various modern ecologies (soil, leaf litter, canopy), ant lineages are primarily soil or surface dwelling . For Blattodea – ie. cockroaches and termites – modern groups live in the superficial soil including litter and in wood logs, corresponding to their ancestral ecology [66, 67]. The host termite identified here, Coptotermes is also distinctly subterranean [66–68]. Advanced levels of sociality are documented in fossil stem-ants as early as the mid-Cretaceous [50,51] and must have applied to the roaches-termites lineage since the early Mesozoic [43, 44], implying that the organisation in nests was present early in the emergence of these two taxa.
Consequently, various springtail families have been confronted with the fast-increasing prevalence of eusocial insects from the mid-Cretaceous onwards, eventually leading to advantages of living close or inside their nests, as reported for most orders of springtails ([48, 58–60], Fig. 4B). Termito/myrmecophile ancient behaviors have previously been suspected for other Dominican amber springtails (cyphoderines, ).
Living in close proximity to termites and ants, as they increased in ecological impact, would provide significant benefits for springtails. An initial termito/myrmecophile behavior would then act as a catalyst, providing opportunity for the evolution of more specialized associations. This type of specialization to eusocial environment is actually also observed in other groups of early diverging hexapods like bristletails (Thysanura), which are reported as myrmecophiles  but also as direct commensals climbing on large ant larvae at colony migration . Thus, beside termito/myrmecophile behaviors, associations with eusocial insects would have appeared at least twice in the evolution of springtails: in cyphoderines (Entomobryomorpha) for apparent feeding and in Symphypleona for (primary or secondary) dispersal. From both fossil and modern examples, we infer that social insects would have acted as Symphypleona dispersal agents. The phoresy would have been enabled through hitchhiking on alate termite/ant adults, at the time they reach the leaf litter to begin their nuptial flights. Comparable to many symphypleonan genera, Coptotermes has distributed onto every continent with diversification since just the Miocene , implying its strong ability for overseas dispersal. As for an dispersal, modern subterranean termites showed real flight performances reaching a 900 m distance in a single take-off and crossing in the same time very large water mass (e.g. the Mississippi river, ). Obviously, these relatively short distances cannot relevantly explain the worldwide distribution of the termites themselves. But is stands very relevant that associations with lineages that shows comparable timing of diversification/distribution over continents could explain the dispersal process of sminthuridan clades. The grasping ability of Symphypleona would have been enhanced by the elongation/segmentation of the fourth antennomere. This derived feature may have allowed the hitchhiking of other soil surface inhabitants (toward more or less efficient dispersal), as observed in other fossil examples (Fig. 4C).
From both fossil and modern cases, soil-dwelling social insects show an obvious trend for biotic associations with smaller apterous arthropods. Indeed, modern mites (both parasitiforms and acariforms) are known for living in ant nests, with phoretic nymphs found on extant ant alates, as well as army ant workers [43, 59, 60]. These nymphs, called hypopi, correspond to a non-feeding stage of mite ontogeny, specialized for phoresy. These associations have been reported twice as early as 44 Ma (Mesostigmata; 56,57). If less described, termitophilous mites are also abundant [77, 81, 82], with not less than eight (parasitiform and acariform) families detected from the study of only three subterranean termites (including Coptotermes) in various continents [63–67]. Limits on these associations remain the size of soil social insects, restricting phoresy to mites and other millimeter-sized organisms, such as springtails (Fig. 5b). Contrary to mites, the ontogeny of springtails displays no metamorphosis (hexapods ametaboly, ) and thus, no possible specific stage to increase dispersal efficiency. It is therefore relevant that phoresy in this group appears in the adult-phenotype (through attachment structures) and primarily in lineages most pre-equipped for this kind of acquisition (Symphypleona). Springtails share this criterion with pseudoscorpions in which phoretic abilities are displayed by the adult phenotype as well, namely through adapted claws (Fig. 5b). In this regard, springtail phoresy compares with the selection constraints of the attachment strategy found in pseudoscorpions, applying within the host range found in mites, but controlled by a much more reactive behavior, probably reflective of hexapod motility speed compared to that of most arachnid orders (, Fig. 5). Through their body proportions and antennae length, Symphypleona would have checked the requirements for specialized phoretic behaviors among other springtail morphologies (Fig. 3). Limited understanding of modern counterparts impedes consideration of whether these phoretic behaviors could have shaped the Symphypleona general morphology as well.