Animals are capable of forming complex societies with distinct roles and kin relationships among individual members. Animal social structures and cooperation are primarily explored in vertebrates with developed sensor systems and cognitive abilities (Clutton-Brock, 2021) and in social insects, such as Hymenoptera and termites (da Silva, 2021). At the same time, intraspecific cooperation remains scarcely and non-systematically studied in other groups (see discussion in Elgar, 2015), potentially leading to an underestimation of their social complexity. Meanwhile, one such group, сrustaceans, exhibits significant morphological and ecological diversity and is capable, similarly to insects, of constructing a variety of individual and communal structures (Atkinson and Eastman, 2015; Moore and Eastman, 2015). Moreover, certain crustaceans have complex levels of social organization, including true eusociality among some shrimp species, exhibited by cooperative brood care, overlapping generations, and a division of labor (Ashrafi and Hultgren, 2023; Duffy et al., 2000; Duffy and Thiel, 2007; Duffy, 2007).

Amphipods are a diverse, abundant, and ecologically significant group of crustaceans (Ritter and Bourne, 2024). Two amphipod families, Dulichiidae and Protodulichiidae, construct unique vertical structures called masts (Ariyama and Hoshino, 2019; Corbari and Sorbe, 2018; Neretin et al., 2017). These masts, made of secreted silk and detritus, elevate the crustaceans far above the bottom, exceeding their body length multiple times. Constructing masts provides protection from predators and increases feeding efficiency (Mattson and Cedhagen, 1989; Thiel, 1999a).

Usually, amphipod constructions are inhabited by a single adult crustacean which protects it from representatives of its species. However, in many cases, the adult female’s structure may also be used by offspring, indicating long-term care of the offspring (Palaoro and Thiel, 2020; Thiel, 2003, 2007). In the case of corophioid amphipods, the females are usually sedentarian whereas males are less attached to the constructions and tend to wander. Often the female may share the building with the male for some time (Borowsky, 1983; Thiel, 2007). Dulichiids and their masts are no exception; Figure 1 shows such a mast inhabited by a female, a male, and the offspring. At the same time, dulichiids exhibit one of the most pronounced extended care of offspring among amphipods. Since the offspring remain on the parental mast for a long time, up to three of a mother’s successive broods can inhabit a mast simultaneously (Mattson and Cedhagen, 1989; Thiel, 1997, 1998). Furthermore, recently we found “collective” masts of the amphipod Dyopedos bispinis (Gurjanova, 1930) in the White Sea; these masts, at least in some cases, are characterized by their increased length and are inhabited by several adult female crustaceans (Neretin et al., 2017).

The discovery of collective masts raised questions about the kin relationship of their inhabitants. To test if the collective masts are inhabited by a progeny of a single founding female, we sequenced significant portions of mitochondrial genomes from 58 individuals from two such collective masts as well as two ordinary lone-female masts. Mitochondrial genomes contain regions with various levels of conservation (e.g. COX1), and in most Metazoa, mitochondrial DNA is transmitted from the mother (Birky, 2008). Therefore, mitochondrial genetic markers are convenient for pedigree reconstruction.

It should be mentioned that there are some invertebrate species that deviate from purely maternal mtDNA inheritance. For example, some bivalves exhibit doubly uniparental mtDNA inheritance (DUI), where females inherit mtDNA from mothers whereas males inherit from both parents (Breton et al., 2007). Next, some invertebrates, including arthropods, harbor mitochondrial genomes split into several circular or linear chromosomes (Lavrov and Pett, 2016; Najer et al., 2024). Finally, copepod hybrids can inherit high proportions of paternal mitochondrial DNA (Lee and Willett, 2022). Therefore, we also sequenced several mother-embryo pairs to ensure that our kinship reconstruction is not affected by paternal mtDNA inheritance.

In this study, we investigated the kin structure of collective dulichian masts. We analyzed masts containing multiple adult female individuals to determine whether they are the progeny of a single founding female or if they independently settled and cohabited the masts.

From the available set of total D.bispinis DNA, we chose one sample (ZMMU WS20509) and sequenced it using an Illumina platform. Sequencing enabled the assembly of the complete circular mitochondrial genome (Figure 2). The D. bispinis mitochondrial genome appeared to be 14,836 bp in length, had a 35% GC content, and contained all standard mitochondrial protein-coding genes (PCGs), tRNAs, and rRNAs. It also contained an AT-rich non-coding region upstream of the 12S rRNA gene. AT-richness, along with an absence of predicted open reading frames, suggests that this is the control region containing D-loop. We also sequenced, assembled, and annotated a 14,853 bp mitochondrial genome of another Dyopedos species, D. porrectus (Supplementary Data S2). Assembling the D. porrectus mitogenome allowed us to calculate dN/dS ratios for individual genes, estimating the relative strengths of purifying and positive selection effects acting after the divergence of these species. We also calculated the d0/d4F metric, which is the ratio of non-synonymous substitutions to substitutions in four-fold redundant positions, and provides greater sensitivity than dN/dS. These analyses revealed a typical metazoan pattern: a high conservation in COX1 and a relaxed selection on ATP8 and ND6 genes (Figure 3).

The order of PCGs in D. bispinis (Figure 2A) and D. porrectus (Supplementary Data S2) did not differ from that of Gammarus duebeni Liljeborg, 1853, the closest species whose complete mitochondrial genome was available in genbank (Figure 2B). The available Gammarus mitogenomes exhibited conserved architectures, with the exception of Gammarus chevreuxi Sexton, 1913 (Supplementary Data S3), and a duplicated control region seen in some species, such as Gammarus roeseli Gervais, 1835 (Cormier et al., 2018). At the same time, the position and coding strands of some tRNAs differed between the Gammarus and Dyopedos species. In available Gammarus mitogenomes, tRNA-Pro was located upstream of the ND5 gene, whereas in both Dyopedos species it was between tRNA-Ser and the ND1 gene (Figure 2B). Furthermore, a cluster of five tRNAs flanked by the putative control region and the ND2 gene in Dyopedos species was inverted compared to that observed in Gammarus species (Figure 2B, Supplementary Data S3). This suggests that the divergence of Gammarus and Dyopedos led to an architectural reorganization of the mitogenome in one of the groups, resulting in the transposition of several tRNAs.

Then, to analyze the family structure of amphipods on the collective masts, we selected two masts with multiple adult females and two typical short masts (see the full list of individuals on the masts in the Supplementary Table S1). We extracted DNA from several adult individuals on these masts, as well as some immature amphipods. In some cases, we isolated embryos from the females and processed them as separate individuals or as groups of individuals. We performed PCR with the total DNA isolated from 59 samples off these four masts and sequenced LR-PCR products using next-generation sequencing. Given that not all PCRs yielded products, we took only partial genomes in further analysis, constituting roughly half of the entire mitogenome sequence. Figure 2 illustrates the region taken for analysis and the number of polymorphism found in all samples. Phylogenetic tree of several Dyopedos and Gammarus species, generated based on partial nucleotide sequences of the COX1 gene (Supplementary Data S5), shows that intraspecies polymorphism within samples identified as D. bisminis is significantly lower than interspecies divergency, confirming their belonging to the same species.

We generated a haplotype network based on 259 variable sites in mitochondrial genome regions with a total length of 7,299 b.p. among 59 individuals or groups of embryos from four collective masts (Figure 4, see alignment in Supplementary Data S4). As expected, the small masts contained either a female with three independently sequenced embryos (mast #3), or a female with embryos and two young male individuals with identical partial mitochondrial genome sequences (mast #4). Specimens from one of the collective masts (mast #2) also shared identical mitochondrial genome partial sequences, with the exception of the two males, adult (ZMMU WS19167) and subadult or adult (ZMMU WS19173). However, one analyzed collective mast (mast #1) contained several groups of female individuals with highly diverged mitochondrial genomes. Figure 4 shows a haplotype network with seven diverged clusters (groups of more than two individuals sharing the same genotype; clusters are named A–G) and branches representing subadult females, adult males, and gravid females with mitotypes distinct from those in the clusters. Figure 5 shows pairwise distances between four clusters from mast #1, as well as three clusters of samples from the remaining three masts. The mean pairwise distance between clusters was 0.63% (45.5 substitutions per 7,299 b.p., or 93.5 substitutions per genome); the mean pairwise distance between clades from mast #1 was 0.65%, 0.61% between clades from masts #2-4, and 0.63% between clades from mast #1 compared to clades from masts #2-4.

Given that mitochondrial DNA is inherited from the female parent, our results shown in Figure 4 suggest that multiple non-kin females inhabit mast #1, and that at least four of the families inhabited the mast long enough to have offspring on it. However, as mentioned in the introduction, there are reports of paternal leakage and unusual patterns of mtDNA inheritance among various invertebrate species. To account for this, we sequenced partial mitochondrial genomes of ten mother-embryo pairs; in all cases, the embryo contained exactly the same genome as the female parent (Figure 4). Therefore, it is unlikely that the difference between clades on mast #1 can be explained by frequent paternal inheritance of mtDNA in this species.

Some animals can form societies where they share common resources and cooperate with each other. Cooperation is mainly driven by kin selection, though in some cases non-kin individuals or individuals of different species can establish evolutionarily stable cooperation (Clutton-Brock, 2009; Foster et al., 2006). The evolutionary forces shaping these societies, as well as the associated costs and benefits for their individual members, usually remain unclear because of their complexity. Therefore, animals with less complex forms of sociality can provide valuable insights into the development of more complex communities (Costa, 2018).

In this study, we reconstructed the family structures of the social amphipod D.bispinis collected from several masts including two collective masts that harbored several distinct assemblages of similar-sized individuals. On one of the masts (#2), despite the presence of multiple adult females, all the female crustaceans included in the analysis turned out to share an identical ~10 k.b. mitochondrial genome fragment (Figure 4). The presence of several adults, including pregnant females, on a mast suggests two non-mutually exclusive possibilities: (1) there are at least two generations on this mast, or (2) the pregnant females, with identical partial mitochondrial genomes, are sisters continuing to use their progenitor’s mast. At the same time, male D.bispinis are mobile and travel between masts (Neretin et al., 2017), so kinship is expected to be shared through the female line. Accordingly, male individuals from this mast contained another mitochondrial genotype and are therefore not descendants of the founding female (Figure 4).

Figure 4 shows that crustaceans from another collective mast (#1) were divided into several maternally non-kin groups (clades in which all individuals share the same genotype). These groups included (1) a group of one subadult female and two females with embryos, (2) three groups of subadult females, one of which also included small juveniles. The low level of kinship on mast #1 is a rather unexpected result because ordinary “family” masts are considered classic parent-offspring groups with strict territoriality (Mattson and Cedhagen, 1989; Neretin et al., 2017; Thiel, 1997, 1998), so it seemed natural to expect a high degree of kinship on the more populated masts that developed from them.

Shared construction and usage by non-kin individuals may result from exploitation of a founding family by later-arriving families or non-kin cooperation. By cooperation here, we mean the investment of resources, not necessarily equally, by different individuals to create a common product—namely, building and/or maintaining a mast. Although we lack the data to quantify the costs of mast construction for each genetically distinct cohort, several arguments suggest the presence of non-kin cooperation in D. bispinis. Firstly, there is evidence of cooperative behavior on ordinary family masts among dulichiids amphipods, like Dyopedos monacanthus (Metzger, 1875) males participating in mast construction as temporary guests (Mattson and Cedhagen, 1989). Secondly, we observed no differentiation in the shape and volume of the secretory apparatus among adult females inhabiting collective masts. This does not exclude the possibility of a division of labor or unequal investment in mast construction due to behavioral differences. Finally, in a minority of cases, the length of communal masts may significantly exceed the length of ordinary reproductive masts (those of females with offspring, Neretin et al., 2017). These reasonings and observations suggest that D. bispinis on collective masts may exhibit a primitive mode of cooperation, characterized by the joint construction and shared usage of the mast.

As per Hamilton’s rule, which states that the benefits of cooperative investment fade rapidly with increasing kinship distance, evolutionarily stable cooperation requires kinship (Hamilton, 1964). Uncommon exceptions to this rule occur when manipulation strategies provide individual benefits from cooperative behavior or when mechanisms ensure an immediate fitness advantage from cooperation (Clutton-Brock, 2009). However, Hymenoptera colonies often exhibit reduced relatedness within themselves due to (1) polypaternity (Delaplane et al., 2024; Duff et al., 2023), (2) colony co-founding by multiple distantly related females, and (3) acceptance of new non-sibling breeding females into the colony. The efficiency of such cooperation is likely ensured by immediate (direct) benefits for contributing participants (Ostwald et al., 2022). We suggest that all these mutually non-exclusive scenarios can contribute to the collective masts of D.bispinis.

How a mast with a non-kin population is formed remains unknown. The presence of groups of subadult siblings requires either the assumption of group migrations from other masts, or of their birth and growth on this mast. In the latter case, the formation of a communal structure on the mast should be attributed to the previous generation. Since aggression levels are lower in juveniles than in adults (Mattson and Cedhagen, 1989), one might suppose that the non-kin structure of the mast initially forms due to independent arrivals of several juveniles on the mast. One of the family groups (group A) includes two age cohorts, which may signify that their mother spent some time on the mast.

Our data indicate that the masts of D. bispinis are utilized and possibly constructed or maintained by multiple individuals (Figure 4, Neretin et al., 2017). Regardless of the specific contributions by different individuals, these masts are utilized by all adult and immature inhabitants to enhance their foraging and protective capabilities. Such social constructions are common among insects (Costa, 2018; Schwarz et al., 2007; Wcislo et al., 2009). In contrast, shared construction usage among crustaceans is usually limited to parental-offspring groups (Thiel, 2003, 2011, Thiel, 2000a). Exceptions are some burrowing species, for example gnathid isopods and talitrid amphipods form small harem groups (Iyengar and Starks, 2008; Upton, 1987). There are also occasional reports of adults sharing a common burrow for semi-terrestrial crayfish (Norrocky, 1991; Richardson, 2007) and maerid amphipods (Atkinson et al., 1982), but data on these social systems are still very incomplete. However, more structured societies including harems, overlapping generations, and true eusociality, are found in symbiotic amphipods, isopods, shrimps (Duffy, 1996, 2007; Shuster and Wade, 1991; Thiel, 1999b, 2000b), and terrestrial crabs (Diesel and Schubart, 2007). Therefore, the case of the D. bispinis communal mast #1 (Figure 4) is rather exceptional, as it does not fall into either category.

The mechanisms enabling coexistence of non-kin D. bispinis on a shared mast while preventing it in other species remain unknown. Apparently, the high level of territorial aggression that is characteristic for family masts (Mattson and Cedhagen, 1989; Thiel, 1997; Neretin et al., 2017) is decreased on communal masts and at the very least does not lead to competitors being immediately driven away. A change in aggression level towards conspecifics, depending on a set of factors (reproductive state, colony welfare, environmental factors, and others), has been observed for some social insects (Reeve and Nonacs, 1997; Sturgis and Gordon, 2012). Territorial aggression in amphipods may be influenced by selectivity based on the intruder’s size or certain chemical cues. While crustaceans are known to use chemical signals for kin recognition, this occurs very rarely (Thiel, 2007; Beermann et al., 2015). The adoption of unrelated offspring, sometimes even from closely related species, is usually independent of chemical cues (Thiel, 2007). We suggest that D. bispinis communal masts are enabled by the low ‘cost’ of mast sharing due to the type of construction. Indeed, space deficit is not as pronounced on masts as in holes or tubes exploited by other crustaceans (see the discussion in Neretin et al., 2017; Thiel, 2007).

The mitochondrial genome of D. bispinis (Figure 2) did not contain any features uncommon in other crustaceans. The protein-coding genes (PCGs) were fairly conserved, and accumulated significantly more synonymous than non-synonymous mutations since their divergence from D. porrectus (Figure 3). The relaxed purifying selection on the ATP8 gene, which we detected (Figure 3), is a common feature of mitochondrial genomes in other arthropods (da Silva et al., 2020; Hao et al., 2017; Pons et al., 2014). Furthermore, the block of five tRNAs rearranged between Gammarus sp. and D. bispinis is adjusted to a putative control region. This observation is consistent with the idea that architectural mtDNA rearrangements typically involve or occur near the control region (Macey et al., 1997).

The numerous single nucleotide polymorphisms (SNPs) in the mitogenomes of D. bispinis collected from the same location, and even the same mast, suggest a high mutation rate and/or effective population size (Ne). Indeed, mitochondrial DNA (mtDNA) is prone to mutations (Nabholz et al., 2008) but mtDNA mutation rates (μ_mito) in invertebrates are understudied, with a few exceptions in some model organisms (Allio et al., 2017; Haag-Liautard et al., 2008). Here, by sequencing mother-offspring pairs, we did not detect any de novo mutations and deviations from the strict maternal inheritance of the mitochondrial DNA. However, the data allows us to make a rough estimate of the lower limit of μ_mito for D. bispinis. Given that we sequenced approximately 7.3 kb of the mitogenome fragment from twelve embryos and found no differences from their mothers (Figure 4), we can infer that μ_mito is less than 1.1*10^–6 substitutions per base pair per generation. It should be noted, however, that this estimate is significantly higher than the known mutation rate of the fruit fly, which is 6.2 × 10⁻⁸ per site per fly generation (Haag-Liautard et al., 2008). Therefore, the actual μ_mito of D. bispinis is likely to be several orders of magnitude lower than our estimate.

To summarize, the high level of intrapopulation genetic variability in mitochondrial genomes permits for the reliable tracing of maternal kinship within D. bispinis groups. Our study provides essential resources, including an assembled mitochondrial genome and a set of primers, to facilitate kinship investigations in D. bispinis. Utilizing these tools, we have demonstrated the possibility of this marine mast-building amphipod forming non-kin societies. D. bispinis represents an unusual case among marine crustaceans, showcasing the capacity of an amphipod to share a common resource with non-sibling individuals.