Material belonging to six species of the genus, A. aidae, A. artankeri Palacios Theil and Felder, 2020, A. felipensis, A. marianae (via photographs), A. leptodactyla, A. patagoniensis (Rathbun, 1918), and A. roblesi Palacios Theil and Felder, 2020, was obtained from the following collections: Coleção de Crustáceos do Departamento de Biologia (CCDB), Faculdade de Filosofia, Ciências e Letras de Ribeirão Preto (FFCLRP), Universidade de São Paulo (USP), Ribeirão Preto, São Paulo, Brazil; Coleção Carcinológica, Museu de Oceanografia Prof. Petrônio Alves Coelho (MOUFPE), Universidade Federal de Pernambuco (UFPE), Recife, Pernambuco, Brazil; Coleção Biológica (ColBio), Instituto Oceanográfico (IO), Universidade de São Paulo (USP), São Paulo, São Paulo, Brazil; Invertebrate Zoology Collection, Museum of Comparative Zoology (MCZ), Harvard University, Cambridge, Massachussets, USA; Regional Collection of Marine Invertebrates (EMU), Instituto de Ciencias del Mar y Limnología (ICML), and Universidad Nacional Autónoma de México (UNAM), Mazatlán, Sinaloa, Mexico. Data on the morphology of the remaining species was obtained from recent literature and robust taxonomic revisions (Manning and Felder 1989; Palacios Theil and Felder 2020a; Salgado-Barragán et al. 2021; Almeida et al. 2023).

The specimens were analyzed under a Leica^© M205 C stereomicroscope equipped with a digital camera Leica^© DFC295 and the image capturing software Leica Application Suite v. 3.8.0. The identification of the individuals was done according to recent literature (Coelho 1997; Palacios Theil and Felder 2020a; Almeida et al. 2023). The overall morphology of several structures was observed and then compared in the context of the molecular phylogenetic hypothesis proposed herein.

For DNA extraction, we selected three species: A. aidae, A. leptodactyla and A. patagoniensis. Austinixa leptodactyla has not been included in any previous phylogenetic analyses, and specimens of A. patagoniensis from Argentina were included for the first time as well. Individuals from several localities that cover most of the distribution of each species were included when possible. Total genomic DNA was extracted from pereopods, eggs or, although rarely, gills of the individuals, using Chelating Ion Exchange Resin (ChelexVR 100) (Estoup et al. 1996).

The target genes were the mitochondrial 16S rRNA, 12S rRNA, and the nuclear H3, which were selected given their relevance and previous use in studies of pinnotherid systematics (Palacios-Theil et al. 2009; Palacios Theil et al. 2016; Palacios Theil and Felder 2020a, b). To amplify such genes, the following primer combinations were used: 16SL2 (TGC CTG TTT ATC AAA AAC AT)/16S-pH2 (CCT GGC TCA CGC CGG TCT GAA) (Schubart et al. 2001; Palacios-Theil et al. 2009) and 16S-pL1 (AAC TTT TAA GTG AAA AGG CTT)/16S-pH2 (Palacios-Theil et al. 2009) for 16S rRNA; 12Sf (GAA ACC AGG ATT AGA TAC CC)/12S1R (AGC GAC GGG CGA TAT GTA C) (Buhay et al. 2007) for 12S rRNA; and H3af (ATG GCT CGT ACC AAG CAG ACV GC)/H3ar (ATA TCC TTR GGC ATR ATR GTG AC) (Colgan et al. 1998) for H3.

Amplification was done using Polymerase Chain Reactions (PCRs) following the procedures proposed by Ocampo et al. (2013) and Mantelatto et al. (2018) including appropriate modifications (adjustments in the concentration of DNA samples and temperature of primer annealing). The following cycle was used: initial denaturation at 95°C for 5 minutes, followed by 40 cycles of denaturation at 94°C for 1 minute, primer annealing at 52–55.5°C, depending on the primer, for 1 minute, and extension at 72°C for 2 minutes, followed by a final extension at 72°C for 10 minutes. Positive PCR results were purified and sent for Sanger sequencing at the Departamento de Tecnologia FCAV-UNESP in Jaboticabal, São Paulo, Brazil, using the kit ABI Big Dye Terminator Mix (Applied Biosystems) and at the private company BPI Biotecnologia Pesquisa e Inovação, in Botucatu, São Paulo, Brazil, following their own methodology. Both forward and reverse strands were sequenced.

Sequences were visualized and edited in Geneious Prime^® v.2021.2 (Kearse et al. 2012), where the consensuses were obtained. The consensus sequences, along with others obtained from GenBank (Table SS1) were aligned with the program MAFFT v.7 (Katoh et al. 2019). The program Gblocks v.0.91b (Castresana 2000; Talavera and Castresana 2007) was then used to detect and eliminate poorly or ambiguously aligned regions.

The alignments were concatenated using Geneious Prime^® v.2021.2. Maximum Likelihood (ML) and Bayesian Inference (BI) criteria were applied in the reconstruction of our phylogenetic hypotheses. The BI analysis was also used for the estimation of time of divergence. The sequences of the specimens included in the analyses are listed in Table SS1.

The best fit substitution models (HKY+F+I+G4 for 16S rRNA and T92+G for 12S rRNA and H3) were obtained with the program MEGA 11 (Tamura et al. 2021) using the Bayesian Information Criterion (BIC).

The ML analysis was carried out on the online W-IQ-TREE platform (http://iqtree.cibiv.univie.ac.at/) (Trifinopoulos et al. 2016). Branch consistencies were assessed using ultrafast bootstrap values of 1000 pseudoreplicates. Values of 70 and above were considered sufficient for further discussion.

The BI analyses were carried out in BEAST v.2.6.2 (Drummond and Rambaut 2007). The parameters and priors for the analyses were selected in BEAUti v2.7.7 (Bouckaert et al. 2014). The best fit model available in BEAST2 was HKY+G+I for all three genes. The alignments were separately uploaded into the BEAST2 program (Drummond and Rambaut 2007), but the trees were linked to form a single tree. The parameters used for the three different runs were: Birth Death model, sampling one tree every 1000 generations for 50 million generations, sampling frequencies equal to 1000. The results of the runs were visualized in Tracer 1.7 (Rambaut et al. 2018), where we observed whether the values of statistics of each ESS (Estimated Sample Size) parameter were above 200. The final results were obtained from an analysis of all trees in the TreeAnnotator v.2.4.4 program of the BEAST2 package (Drummond and Rambaut 2007). This analysis was performed to synthesize the results of three different BI runs into a single tree. The three different analyses were added together in the LogCombiner v2.6.7 program of the BEAST2 package (Drummond and Rambaut 2007), generating a consensus tree. The final tree was viewed and edited using the FigTree v.1.4.4 program (Rambaut 2018). Values of posterior probability above 0.9 were considered for further discussion.

Divergence time estimation was performed by calibrating a relaxed molecular clock under a Bayesian Inference framework. For this, while selecting the priors and specifications in BEAUTi v2.7.7, Relaxed Clock Log Normal was selected as the model for all three genes. To improve topological stability for the divergence time estimation, monophyly constraints were applied at the family and subfamily level, since the monophyly of these groups has been consistently supported by previous phylogenetic studies (Palacios Theil et al. 2016; Chow et al. 2023; Tsang and Naruse 2023) and by the ML analyses conducted herein. Since no fossils confidently assignable to the genus Austinixa are currently known and the fossil record for Pinnotheridae is fragmentary (Luque et al. 2024), we selected the oldest fossil confidently assigned to the crown-group of Pinnotheridae De Haan, 1833, Pinnixa sensu lato (Minimum age = 18.7 Ma), based on recent literature regarding fossil calibrations in brachyuran phylogeny (Luque et al. 2024). Furthermore, we chose Ocypodidae Rafinesque, 1815 as the outgroup for the analysis based on recent phylogenetic studies of Thoracotremata (Tsang et al. 2022) and given that fossils of this group are known and can be used for time calibration (Luque et al. 2024). Thus, one fossil belonging to the crown-group of Ocypodidae, Afruca miocenica (Artal, 2008) (Minimum age = 13.8 Ma), was selected. Both calibrations were implemented as MRCA priors with Log Normal distribution, with the fossil minimum age used as the offset and the standard deviation parameter (S) set to 1.5 to accommodate uncertainty associated with the sparse fossil record. The final closure of the Isthmus of Panama (3 mya) (O’Dea et al. 2016; Bourke et al. 2025) was used as a minimum constraint for the divergence between Atlantic and Pacific lineages in the subclades of Austinixa obtained either in the ML analysis and/or other studies (see Palacios Theil and Felder 2020a). This calibration was also included as a MRCA prior with Log Normal distribution, with the age of the final closure of the isthmus (3 mya) used as the offset and S set to 1.5 to account for the possibility that the divergence between these lineages predated this geological event. Lastly, given the extremely sparse fossil record of Pinnotheroidea and the focus of the present study on a single genus, a secondary calibration was made based on the results provided by Chow et al. (2023), which included a broader taxonomic sampling, eight genes, and several fossil calibrations in outgroup lineages. This secondary calibration was applied to the crown group of Pinnotheroidea and was implemented as a MRCA prior with Normal distribution with a mean of 72.8 Ma and a standard deviation (σ) of 8.0 Ma, reflecting the confidence interval reported in that study. Mean node ages along with 95% Highest Posterior Density (HPD) intervals are shown in the tree (Fig. 1).

Ancestral distributions were reconstructed in RASP v.4.4 (Yu et al. 2020) using BioGeoBEARS v.1.1.1 (Matzke 2013a, b, 2014). The BI tree was imported into RASP and distribution states were assigned to each terminal, which in this case were the individuals. The marine provinces proposed by Spalding et al. (2007) (MEOW) were used as the distribution states for the focal group. The following provinces (with their respective original classification by numbers) were included: Tropical Southwestern Atlantic (14), Warm Temperate Southwestern Atlantic (47), Tropical Northwestern Atlantic (12), Warm Temperate Northwestern Atlantic (6), Tropical Eastern Pacific (43), Warm Temperate Northeastern Pacific (11), Warm Temperate Southeastern Pacific (45), and Lusitanian (3). To reduce the number of geographic states and avoid unnecessary model complexity, adjacent MEOW provinces represented only by outgroup taxa were grouped into broader geographic regions: Central Western Pacific (grouping provinces 9, 25, 26, 30 and 32), and Cold Temperate Northeastern Pacific + Northwestern Pacific (8 and 10). Outgroup taxa belonging to Ocypodidae, Parapinnixidae Števčić, 2005 and Tetriasidae Tsang and Naruse, 2023 were removed from this analysis to simplify reconstruction and avoid estimating biogeographic events that are unrelated to the diversification of the ingroup.

With the states assigned to all terminal taxa of the tree, a model test was carried out in RASP using the function “Model Test” with the option “Compare Models Using BioGeoBEARS”. The ancestral distribution reconstruction model with the lowest Akaike Information Criterion (AIC) value, and thus with the highest AIC weight (AIC_wt) was selected: Dispersal-Extinction-Cladogenesis (DEC) (0.69 AIC_wt). The maximum number of ancestral areas was restricted to two, reflecting the pattern observed for most extant species within Pinnotheridae, which generally occupy one or two adjacent marine provinces, thus avoiding the inference of unrealistically widespread ancestral distributions. Furthermore, pairs of provinces that are not adjacent or historically connected were excluded from the possible ancestral areas. The ancestral distribution reconstruction was then carried out using the function “Reconstruction”, with the option “DIVALIKE/DEC/BAYAREALIKE on BioGeoBEARS”, with the DEC model being selected.

To simplify the character state reconstruction, a new ML tree was constructed using a single representative sequence per species. This did not change the overall topology of the tree (see Figs 1, 3, 4). Eleven morphological characters (Table 1) were selected for character state reconstruction based on importance to taxonomy, suspected phylogenetic signal and for each state being discrete and easily discernible, without overlapping. The ancestral state reconstruction was carried out for each character independently. Most characters only had two states and were thus coded with binary states (0 and 1). Characters that had three independent states were coded with three states (0, 1 and 2).

Aspects of the same structure were divided into separate characters when they reflected different qualitative properties that can vary independently of each other among taxa. Herein, these characters include: presence/absence of a branchial ridge, the texture of the branchial ridge when present, and the orientation of the branchial ridge when present; G1 apex curvature and G1 apex flattening (Table 1). Character states were identified based on the analysis of physical specimens of A. aidae, A. artankeri, A. felipensis, A. leptodactyla, A. patagoniensis, A. roblesi, Tubicolixa chaetopterana (Stimpson, 1860), Rathbunixa sayana (Stimpson, 1860), Rathbunixa occidentalis (Rathbun, 1894), Glassella faxoni (Rathbun, 1918), and Dissodactylus crinitichelis Moreira, 1901. For the remaining species, illustrations and descriptions published in recent literature were used (Manning and Felder 1989; Zmarzly 1992; Campos 1996; Campos and Wicksten 1997; Thoma et al. 2009; Rahayu and Ng 2010; Naruse and Maenosono 2012; Campos and Vargas-Castillo 2014; Palacios Theil and Felder 2020a, b; Almeida et al. 2023; Balbino et al. 2026). Character states were then coded into a taxon-character matrix (Table 2), with all being treated as unordered. Characters with no available information in literature were treated as missing data and coded as “?”. Furthermore, when a character was not applicable to a taxon, it was coded as “—”. The character state reconstruction analyses were carried out individually using Likelihood Reconstruction with the default model Mk1 in Mesquite v.4.01 (Maddison and Maddison 2025). The final trees were visualized and edited in Mesquite.

The following abbreviations were used for morphological characters: Third maxilliped (Mxp3), pereopods 2–5 (P2–P5), first male gonopod (G1). Additionally, abbreviations for biogeographic regions and marine provinces include: Tropical Southwestern Atlantic (TSWA), Warm Temperate Southwestern Atlantic (WTSWA), Tropical Northwestern Atlantic (TNWA), Warm Temperate Northwestern Atlantic (WTNWA), Tropical Eastern Pacific (TEP), Warm Temperate Northeastern Pacific (WTNEP), Cold Temperate Northeastern Pacific + Northwestern Pacific (CTNEP + NWP), Central Western Pacific (CWP), Warm Temperate Southeastern Pacific (WTSEP), and Lusitanian (LUS).

The ML and BI phylogenetic trees recovered similar overall topologies and were compiled into one tree (Fig. 1). Pinnotheroidea was recovered as monophyletic with maximum support in ML and in BI (100/1). All three pinnotheroid families (Pinnotheridae, Parapinnixidae, and Tetriasidae) were also recovered as monophyletic groups with maximum support in both analyses (100/1). However, the family-level relationships were not concordant between the analyses, with Parapinnixidae being the sister group of Tetriasidae with moderate support (87) in the ML tree, whereas Parapinnixidae was the sister group of (Tetriasidae + Pinnotheridae) in the BI tree, although with no support. Within Pinnotheridae, Pinnixulalinae Palacios Theil, Cuesta and Felder, 2016 was recovered with maximum support (100/1) and formed the sister group of (Pinnotherinae De Haan, 1833 + Pinnixinae Števčić, 2005), although the latter clade was only well-supported in the ML analysis (95). Pinnotherinae received moderate and maximum support in the ML and BI analyses respectively (70/1), whereas Pinnixinae was recovered as monophyletic with maximum support in both (100/1).

Within Pinnixinae, the clade (Glassella Campos and Wicksten, 1997 + Indopinnixa Manning and Morton, 1987) was recovered with high and maximum support in the ML and BI analyses respectively (98/1) and formed the sister group of the clade containing all other genera. The clade containing all other pinnixine genera was recovered with high support in the ML tree (98) and maximum support in BI (1). Within this clade, all intergeneric level relationships were recovered with very high to maximum support in both analyses (see Fig. 1). The genus Austinixa was recovered as monophyletic with high support in ML (98) and maximum support in BI (1).

Within Austinixa, A. patagoniensis was recovered as the sister group of all other species of the genus, with the latter clade being recovered with maximum support (100/1). In the ML analysis, within this subclade, A. cuestai was recovered as the sister group of the clade containing all other species, although the latter clade received no support. In the BI analysis, this clade is divided into two subclades, (A. cuestai + (A. cristata + (A. gorei + (A. artankeri + A. aidae sensu lato)))) and (A. felipensis + ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla))), both recovered with maximum support (1). The clade (A. felipensis + ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla))) was also recovered in ML, although with low support. The clade (A. cristata + (A. gorei + (A. artankeri + A. aidae sensu lato))) was recovered with moderate support in ML (85) and BI (0.93). The subclade (A. gorei + (A. artankeri + A. aidae sensu lato)) received moderate supported in ML (85), but no support in BI. The clade (A. artankeri + A. aidae sensu lato) was recovered with maximum support in both trees (100/1). Within the other subclade, ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)) was recovered with high support in ML (97) and maximum support in BI (1). Additionally, (A. chacei + A. behreae) was recovered with maximum support in both analyses (100/1). Austinixa leptodactyla, included in a molecular phylogenetic hypothesis for the first time, was recovered as the sister group of the morphologically similar A. roblesi with maximum support in ML and BI (100/1).

All of the traditionally recognized species of Austinixa were recovered as monophyletic groups, and with maximum support (100/1), with the exception of A. aidae, A. chacei, A. leptodactyla and A. roblesi, which only had maximum support in the BI analysis (74, 98, 99, and 97 support in the ML analysis, respectively). There appeared to be little to no intraspecific structuring in these taxa, except for A. aidae, in which well separated groups are observed. The first, recovered with high and maximum support in ML and BI respectively (98/1), contained specimens from Brazil, whereas the second, recovered with low support in both trees, contained specimens from Venezuela and Panama. Within the latter clade, a subclade containing only specimens from Panama was recovered with maximum support in both analyses (100/1). For this reason, and the divergence time estimates (see below), we have opted to treat A. aidae from Panama and Venezuela as a separate entity, A. aff. aidae.

The mean estimated time for the divergence between the outgroup, Ocypodidae Rafinesque, 1815, and Pinnotheroidea was 62.57 mya (95% highest posterior density (HPD) interval: 39.85–85.08), during the early Paleocene. Parapinnixidae diverged from (Tetriasidae + Pinnotheridae) 57.7 mya (HPD 37.76–76.43) during the late Paleocene, and Tetriasidae diverged from Pinnotheridae 52.72 mya (HPD 32.24–72.69) during the early Eocene. Within Pinnotheridae, Pinnixulalinae diverged from (Pinnotherinae + Pinnixinae) 35.81 mya (HPD 20.78–49.79), and Pinnotherinae diverged from Pinnixinae 34.18 mya (HPD 20.19–47.82), both in the late Eocene. Within Pinnixinae, (Indopinnixa + Glassella) diverged from the common ancestor of the remaining pinnixine genera 27.81 mya (HPD 16.12–38.98) in the mid Oligocene and diverged from each other 17.35 mya (HPD 9.31–25.71), in the early Miocene. Sayixa diverged from (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax)))) 22.67 mya (HPD 12.95–31.87) and Pinnixa diverged from (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))) 19.24 mya (HPD 11.28–27.27), both in the early Miocene. Austinixa diverged from (Tubicolixa + (Rathbunixa + Scleroplax)) 17.08 mya (HPD 9.97–24.06), also in the early Miocene. Tubicolixa diverged from (Rathbunixa + Scleroplax) 13.78 mya (HPD 7.81–19.9) in the mid Miocene, whereas Rathbunixa and Scleroplax diverged 9.2 mya (HPD 4.95–13.75).

Within Austinixa, A. patagoniensis diverged from the common ancestor of the remaining species 15.4 mya (HPD 8.93–21.71) in the mid Miocene. The mean estimated time of divergence between the remaining two subclades of Austinixa is 12.94 mya (HPD 7.53–18.4) in the mid Miocene. A. cuestai is estimated to have diverged from (A. cristata + (A. gorei + (A. artankeri + A. aidae sensu lato))) 12.16 mya (HPD 7.06–17.43) in the mid Miocene, and A. cristata from (A. gorei + (A. artankeri + A. aidae sensu lato)) 10.71 mya (HPD 6.08–15.58) in the late Miocene. Furthermore, A. gorei diverged from (A. artankeri + A. aidae sensu lato) 9.8 mya (HPD 5.22–14.43) in the late Miocene, and A. artankeri diverged from A. aidae sensu lato 5.29 mya (HPD 2.68–8.09) in the early Pliocene. Within A. aidae sensu lato, A. aidae diverged from A. aff. aidae 3.16 myaa (HPD 1.43–4.98) in the late Pliocene, whereas within A. aff. aidae, the lineage from Venezuela diverged from the lineage from Panama 2.5 mya (HPD 0.94–4.32). In the other subclade, A. felipensis is estimated to have diverged from ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)) 11.99 mya (HPD 6.8–17.08) in the mid Miocene. The latter two clades are estimated to have diverged from each other 9.59 mya (HPD 5.31–13.95) in the late Miocene. A. chacei and A. behreae diverged 3.95 mya (HPD 1.85–6.2) in the Pliocene, and A. roblesi and A. leptodactyla diverged 2.48 mya (HPD 1.07–4.1) in the Pleistocene. These two species pairs have the youngest times of divergence between the currently described and valid species. The two lineages within A. aidae sensu lato are estimated to have diverged before A. leptodactyla and A. roblesi.

A tree with most likely ancestral distribution areas is shown in Figure 2. No ancestral area with over 50% probability was recovered for the common ancestor of Pinnixinae, with (TNWA + TEP) having the highest probability (38%), followed by (TEP + WTNEP) with 33.3%. Likewise, no area with over 50% probability was recovered for the common ancestor of (Glassella + Indopinnixa), with (TEP + WTNEP) having the highest probability (49.3%), followed by (WTNEP + (CTNEP+NWP)), with 36.7%. No distribution area was recovered with over 50% probability for the ancestor of (Sayixa + (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))))) with (TNWA + TEP) having the highest probability (48.6%), followed by the TNWA (29%). No ancestral area with more than 50% probability was recovered for (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax)))), with the TNWA having the highest probability (47.8%), followed by (TNWA + TEP) with 28%. For (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))), the most probable ancestral distribution of the common ancestor was recovered as the TNWA (52.5% probability), followed by (TNWA + TEP) (34.1%). No ancestral area for the genus Austinixa was recovered with over 50% probability, with the highest being the TNWA (49.9%), followed by (TSWA + TNWA) (40.5%). Thus, the probability of the ancestral area of the genus being in the tropical western Atlantic was recovered as 90.4%.

For A. patagoniensis, the ancestral distribution was recovered as the WTSWA with very high probability (96.7%). The distribution of the common ancestor of the clade containing all other species of Austinixa was recovered as the TNWA with moderate probability (72%). The common ancestor of the subclade (A. cuestai + (A. cristata + (A. gorei + (A. artankeri + A. aidae sensu lato)))) was likely distributed in the TNWA (80.4% probability), with A. cuestai having subsequently dispersed into the TEP (90.3% probability). The common ancestor of (A. cristata + (A. gorei + (A. artankeri + A. aidae sensu lato))) originated in the TNWA (85.9% probability). The ancestor of Austinixa cristata later colonized the WTNWA, with its distribution including both the TNWA and WTNWA (98% probability), which remains as its current distribution (Manning and Felder 1989; Palacios Theil and Felder 2020a). The common ancestor of (A. gorei + (A. artankeri + A. aidae sensu lato)) also originated in the TNWA (90% probability), and the ancestor of A. gorei remained in the TNWA (96.5%). The most likely ancestral distribution of the common ancestor of (A. artankeri + A. aidae sensu lato) was the TNWA with moderate probability (56.3%) followed by (TSWA + TNWA) (39.8%), with a very high probability of having originated in the tropical western Atlantic (96.1%). The ancestor of Austinixa artankeri remained in the TNWA (98.9% probability), whereas the ancestor of A. aidae sensu lato expanded its distribution into the TSWA, occupying both the TNWA and TSWA (94.6% probability). A vicariant event then likely led to the separation of A. aff. aidae and A. aidae, with the former becoming restricted to the TNWA (93.9%). The ancestor of A. aidae remained in the TSWA and also colonized the WTSWA, occupying both areas (99.1%), which is still the case for its modern distribution (Balbino et al. 2026).

The distribution of the common ancestor of the clade (A. felipensis + ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)) was recovered as the TNWA with moderate probability (60.1%), with the ancestor of A. felipensis subsequently colonizing and becoming restricted to (TEP + WTNEP), where it is still currently distributed (Palacios Theil and Felder 2020a). The most likely distribution of the common ancestor of ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)) was recovered as the TNWA with moderate probability (50.2%), followed by (TNWA + WTNWA) (37.7% probability), and thus, a high probability of being in the Northwestern Atlantic (87.9%). The common ancestor of (A. roblesi + A. leptodactyla) extended its distribution to (TNWA + TSWA) (95.2% probability), with a subsequent vicariant event leading to the separation of A. roblesi in the TNWA (99.3% probability) and A. roblesi in the TSWA (97.3% probability). The distribution of the common ancestor of (A. chacei + A. behreae) was recovered as the WTNWA with moderate probability (74.7%), which might have resulted from a dispersal event from the TNWA. The ancestors of each species remained in the WTNWA, recovered with high probability (98.6% and 98.9% for A. behreae and A. chacei respectively).

The trees showing the history of state changes for the eleven characters (Tables 1, 2) are depicted in Figure 3. For Character 1 (branchial region; 0 = without branchial ridge, 1 = with branchial ridge) (Fig. 3A), the absence of the branchial ridge was recovered as the most likely state for the common ancestor of Pinnixinae (65%). The branchial ridge appears to have convergently evolved two separate times in species of Glassella (G. abbotti and G. miamiensis), Indopinnixa moosai, and in the common ancestor of (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax)))) (84.1% likelihood for the latter). The common ancestor of the genus Austinixa likely had branchial ridges (92.9%), with most of its descendants retaining it. A process of reversal was subsequently observed convergently in A. felipensis and A. cuestai. In the common ancestor of (A. gorei + (A. artankeri + A. aidae)) the presence of a branchial ridge was the most likely state, although with moderate probability (64.3%). A. gorei also convergently lost the branchial ridge, whereas the common ancestor of (A. artankeri + A. aidae) showed a slightly higher likelihood of also lacking it (53.1%). If this is the case, A. artankeri would have evolved branchial ridges convergently. However, if the ancestor of (A. artankeri + A. aidae) had a ridge, A. aidae would have lost the ridge convergently. In either case, the branchial ridge was convergently lost in four lineages of the genus Austinixa.

The reconstruction for Character 2 (surface texture of branchial ridge; 0 = tuberculated, 1 = smooth) (Fig. 3B) is dependent on the presence of the branchial ridge and therefore only applies to lineages in which it is present. For Character 2, the most likely ancestral state for (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax)))) was recovered as a tuberculated ridge (94.7%). The common ancestor of Austinixa showed a slightly higher likelihood of having a tuberculated branchial ridge (54.3%) and A. patagoniensis would have acquired a smooth branchial ridge. The common ancestor of the remaining species of the genus also showed a slightly higher likelihood of having a tuberculated ridge (56.2%), with the same happening in the ancestor of ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)) (57.9%), with the ancestor of (A. chacei + A. behreae) retaining this state (98.5% likelihood), whereas the ancestor of (A. roblesi + A. leptodactyla) acquired a smooth ridge (99.6% likelihood). The ancestor of (A. cristata + (A. gorei + (A. artankeri + A. aidae))) was slightly more likely to have had a tuberculated ridge (56%), with A. cristata retaining this state and A. artankeri acquiring a smooth branchial ridge.

For Character 3 (orientation of branchial ridge; 0 = almost parallel to anterolateral margin, 1 = posteriorly bent towards posterolateral margin) (Fig. 3C), the same methodological limitation as Character 2 applies, with the character being dependent on the presence of the ridge and, thus, not applicable in cases where the ridge is absent. The common ancestor of (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax)))) very likely had a parallel ridge (99.6%), which is also the case for the common ancestor of Austinixa (~100%), with A. patagoniensis and the common ancestor of the remaining clade retaining this state (99% for the latter). The common ancestor of (A. cristata + (A. gorei + (A. artankeri + A. aidae))) very likely had a parallel ridge (99.7%), with A. cristata and A. artankeri) retaining the state. The sharply bent branchial ridge likely arose in the common ancestor of ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)) (76.1%), with the ancestors of the two subclades (99.7 and 99.9% likelihood, respectively) and the descendent species having retained a bent ridge.

The most likely state for Character 4 (cardiac ridge; 0 = absent, 1 = not crossing entire cardiac region, 2 = crossing entire cardiac region) (Fig. 3D) in the common ancestor of Pinnixinae was an absent cardiac ridge (99%). Likewise, the lack of a cardiac ridge was recovered as the most likely state for the common ancestor of most clades, including (Indopinnixa + Glassella) (95.2%), (Sayixa + (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))))) (99.4%), (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax)))) (94.8%), (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))) (70.6%), (Tubicolixa + (Rathbunixa + Scleroplax)) (65.5%), and (Rathbunixa + Scleroplax) (69.5%). A cardiac ridge not crossing the entire cardiac region probably arose separately in Tubicolixa and Rathbunixa, with the common ancestor of R. sayana and R. occidentalis very likely having such ridge (98.4%). The cardiac ridge crossing the entire width of the carapace appeared in the common ancestor of Austinixa (96%), with all subclades and species retaining this state.

For Character 5 (cardiac tuft of setae in males; 0 = absent, 1 = present) (Fig. 3E), the most likely state for the common ancestor of Pinnixinae was lacking a cardiac tuft of setae (99.1%), with the ancestors of all analyzed genera inheriting the same state, including Austinixa (~100% probability). Most subclades and species of Austinixa inherited the same state, with only A. cuestai, A. aidae and the ancestor of (A. roblesi + A. leptodactyla) acquiring a tuft of setae in the cardiac region (the latter with 99.8% probability).

For Character 6 (cheliped symmetry; 0 = symmetrical, 1 = asymmetrical) (Fig. 3F), symmetrical chelipeds were recovered for all nodes with very high probability (~100%), with only one species, A. felipensis, acquiring asymmetrical chelipeds.

The most likely state for the common ancestor of Pinnixinae for Character 7 (length of fixed finger and dactylus deflection in adult males; 0 = fixed finger long, dactylus not deflected, 1 = fixed finger short, dactylus deflected, 2 = fixed finger extremely short, dactylus extremely deflected) (Fig. 3G) is a long fixed finger and non-deflected dactylus in male chelipeds (88.4%), with the ancestor of the subclade (Indopinnixa + Glassella) inheriting this state (98.7% probability), along with all of its descendants. The males of the ancestor of the subclade (Sayixa + (Pinnixa + (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))))) likely had an extremely reduced fixed finger with an extremely deflected dactylus (71.7% probability). The common ancestor of Pinnixa and (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))) still likely presented this state (74%), although a process of reversal happened in Pinnixa. The ancestor of (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))) retained an extremely reduced fixed finger and deflected dactylus (98.1%). The same applies for the ancestor of Austinixa and the ancestors of all its subclades (probabilities always 95.1% or higher), except for the ancestor of (A. roblesi + A. leptodactyla), for which the fixed finger is reduced and the dactylus deflected, but not extremely (99.7% probability). Two additional convergent appearances of this state were shown in A. cuestai and A. aidae.

For Character 8 (P4 merus; 0 = without depression, 1 = with depression) (Fig. 3H), the lack of a depression in the P4 merus was present in almost all nodes with very high likelihood (~100%), with the depression originating in the common ancestor of (A. roblesi + A. leptodactyla) (99.9%), thus representing a well-supported synapomorphy of this clade.

For Character 9 (P4 propodus opposable margin; 0 = unicarinate, 1 = bicarinate, 2 = without carinas) (Fig. 3I), a bicarinate P4 opposable margin was recovered as the most likely state for the ancestor of Pinnixinae (96.9% probability) being inherited by most of its descendants, including the common ancestor of Austinixa (~100%). A P4 propodus margin without carinas appears in the genus Rathbunixa (98.3%), whereas P4 propodus with a single carina only appears in two species of Austinixa, A. gorei and A. chacei, convergently.

The reconstruction of Character 10 (G1 apex curvature; 0 = not curved outwards, 1 = sharply curved outwards) (Fig. 3J) showed that a sharply outwards curved G1 apex was the most likely state for the ancestor of Pinnixinae (66.1%). The same state was recovered for the common ancestor of (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))), (77.2%), which was retained by the common ancestor of Austinixa (86.1%) and inherited by A. patagoniensis, by the common ancestor of all other species (77.5%), and by the ancestor of (A. cuestai + (A. cristata + (A. gorei + (A. artankeri + A. aidae)))) (78.8%), being retained by all of the descendants within the latter. However, although all extant members of the clade (A. felipensis + ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)) present a not-curved G1 apex, the most likely state recovered for its common ancestor is a curved G1 apex (71.5%). Furthermore, the most likely state for the ancestor of ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)), although only slightly (50.5%), is an outwards curved G1, even though it is not curved in all extant members of the clade. A not-curved G1 apex was recovered for the common ancestors of (A. chacei + A. behreae) and of (A. roblesi + A. leptodactyla) (93 and 98.1% respectively).

Lastly, for Character 11 (G1 apex flattening; 0 = not flattened, 1 = flattened horizontally, 2 = flattened laterally) (Fig. 3K), the ancestral state for Pinnixinae was recovered as not flattened (83.7% probability), with the same happening for the common ancestor of the clade (Austinixa + (Tubicolixa + (Rathbunixa + Scleroplax))) (71.9%). The ancestor of Austinixa had a slightly higher probability of also having a not-flattened G1 apex (53.6%), with A. patagoniensis acquiring a laterally flattened apex. No state with over 50% likelihood was recovered for the ancestor of the clade containing the remaining species, with very similar probabilities for states 0 and 1 (49.6% and 47.8%). The ancestor of the subclade (A. cuestai + (A. cristata + (A. gorei + (A. artankeri + A. aidae)))) showed a moderate probability of having a horizontally flattened G1 apex (67.7%), with of all the descendant species retaining this state. The ancestor of the other subclade (A. felipensis + ((A. chacei + A. behreae) + (A. roblesi + A. leptodactyla)) likely presented a not-flattened G1 apex (63.2%), with all of the descendants retaining this state.

The phylogenetic relationships among species of Austinixa have been the target of two previous studies, Harrison (2004) and Palacios Theil and Felder (2020a). Harrison (2004) used sequences of two mitochondrial genes, 16S rRNA and COI, separately and combined, to reconstruct phylogenetic hypotheses using Maximum Likelihood and Maximum Parsimony. All eight recognized species classified under Austinixa at the time were included in the analyses. Although tree topologies varied greatly between the different genes and criteria, A. patagoniensis was always recovered as the sister taxon of the clade containing the remaining species. Within this clade, A. felipensis was recovered as the sister taxon of the clade containing all other species. Close relationships between A. aidae + A. hardyi and A. gorei, and between A. chacei and A. behreae were also recovered in the analyses proposed by Harrison (2004). The phylogenetic relationships of A. cristata remained unresolved, as its position varied greatly among the therein proposed trees. There are some similarities between the phylogenetic tree proposed herein (Fig. 1) and the one proposed by Harrison (2004), such as the position of A. patagoniensis as the sister taxon of the clade containing all other species and the close relationships between (A. chacei + A. behreae) and (A. aidae + A. gorei). However, a major difference is the low-supported position of A. felipensis as the sister group of the clade containing all other species (except for A. patagoniensis), whereas herein it was found to be closer to (A. chacei + A. behreae). These differences might be the result of a different outgroup choice, a larger number of ingroup taxa included herein, and the use of different genes.

This initial topology proposed by Harrison (2004) also contrasts with the phylogenetic tree presented by Palacios Theil et al. (2016), which shares a very similar topology with our analyses. The only major difference between our tree (Fig. 1) and the hypothesis proposed therein (fig. 1) and developed further by Palacios Theil and Felder (2020a: fig. 2) is the position of A. cristata and A. cuestai, with A. cristata being the sister group of the clade (A. cuestai + (A. gorei + (A. aidae + A. artankeri))) with no support, whereas in our BI analysis, A. cuestai is the sister taxon of (A. cristata + (A. gorei + (A. aidae + A. artankeri))) with maximum support (1), although this relationship is not supported in our ML tree. Nonetheless, the position of A. patagoniensis as the sister group of all other species is shared in all three studies. These similarities between the phylogenetic hypotheses proposed herein and the one proposed by Palacios Theil et al. 2016 and Palacios Theil and Felder (2020a) could be due to the use of the same sequences and genes (16S rRNA, 12S rRNA and H3), which are deemed useful in phylogenetic studies of pinnotherid crabs. Harrison (2004) even mentioned that the COI gene was not as useful for phylogenetic studies in Austinixa as 16S rRNA, and should rather be used for population studies. Despite these differences, it is reliable to assume that the phylogeny of Austinixa is well-understood compared to many other pinnotheroids, such as Glassella and Parapinnixa Holmes, 1895.

Even though these two studies focused on the phylogeny of Austinixa as thoroughly as possible, little was inferred regarding morphological evolution within the genus. Harrison (2004) mentioned that the species of Austinixa have strikingly similar morphologies given their genetic distance, and that this could be related to stabilizing selective pressures regarding their lifestyle of being commensal with ghost shrimps. The author also mentioned that A. cristata is morphologically similar to A. behreae, A. chacei, A. gorei and A. aidae, taxa with which it formed a polytomy in the phylogenetic tree presented therein. Harrison (2004) also mentions that A. cristata, A. gorei and A. aidae are similar in having small patches of setae on the dorsal surface of fourth pereopod and the lateral surface of the carapace, which the author deemed to be plesiomorphic given the configuration of the tree. Other mentioned similarities include the shape of the G1 and overall P4 morphology. However, there are no further mentions of character evolution. Palacios Theil and Felder (2020a) advance in this discussion by arguing that, based on the tree topologies observed therein, characters used in species identification have little phylogenetic signal and do not represent synapomorphies of subgroups within Austinixa. The present study is, however, the first to further explore the evolution of morphological characters within Austinixa using character state reconstruction based on a baseline multigene phylogenetic tree. For our analysis, we picked characters that are useful in distinguishing between species of Austinixa or between pinnixine genera to search for possible phylogenetic signal (Manning and Felder 1989; Heard and Manning 1997; Harrison and Hanley 2005; Palacios Theil and Felder 2020a, b; Salgado-Barragán et al. 2021).

The presence of a cardiac ridge, either crossing the entire cardiac region of the carapace or not, appears to be a convergent characteristic between several different groups (Fig. 3D). This trait was likely absent in the ancestor of Pinnixinae and most of its descendants, having evolved convergently in a clade containing some species of Glassella, in one species of Indopinnixa, in Tubicolixa, Rathbunixa, and Austinixa. Thus, this character is not homologous between these groups, and the cardiac ridge transversally crossing the entire cardiac region of the carapace probably did not appear from a pre-existing ridge that did not cross the carapace and rather evolved independently. Furthermore, the presence of a transverse ridge crossing the entire posterior surface of the carapace, in addition to being a useful character in distinguishing this genus from other pinnixines (see Manning and Felder 1989; Harrison 2004; Palacios Theil and Felder 2020a), is proposed as a synapomorphy of Austinixa. Pinnixa transversalis (H. Milne Edwards & Lucas, 1842) shares a similar transverse cardiac ridge, but Palacios Theil and Felder (2020a) stated that this species does not share other diagnostic characters with species of Austinixa and did not include it in the genus. However, recently, Poore and Ahyong (2023) transferred P. transversalis to Austinixa. After analysis of the illustrations provided by Garth (1957: fig. 4), it appears that the ridge does not cross the entire width of the carapace and falls short of the lateral edges on both sides. Furthermore, the pleon and cheliped morphology of P. transversalis are not similar to those of species of Austinixa. Thus, we conclude that the cardiac ridge of P. transversalis is not identical to the species of Austinixa and thus should be removed from the genus. However, the phylogenetic position of this taxon is still not resolved, and understanding the relationship between this taxon and Austinixa could provide insights into the evolution of this trait.

For characters that are important for diagnosis of species within the genus, we have observed that most of the shared characteristics are actually homoplasies, and convergently evolved in different clades within Austinixa. For example, the presence of the branchial ridge is plesiomorphic for the genus (Fig. 3A), and the loss of this trait has happened multiple times, being a homoplasy shared by A. aidae, A. gorei, A. cuestai and A. felipensis. Our analysis also indicated that it is slightly more likely that the ancestor of A. artankeri and A. aidae lost the branchial ridge, and it appeared once again in A. artankeri, in a process of reversal. The same is true for the texture of the ridge (Fig. 3D), which can be either smooth or tuberculated. The plesiomorphic state for Austinixa is a tuberculated ridge, and thus the change to a smooth ridge happened three times (in A. patagoniensis, A. artankeri and A. leptodactyla + A. roblesi). Furthermore, the presence of setal patches on the cardiac region of the carapace of males (Fig. 3E) evolved convergently three times, once in the common ancestor of A. leptodactyla and A. roblesi, in A. cuestai and in A. aidae. The number of carinas in the P4 propodus opposable margin (Fig. 4C), also used as a diagnostic character for species of Austinixa, shows a similar pattern, with the plesiomorphic state being two carinas (bicarinate), and the unicarinate state evolving twice, in A. chacei and A. gorei separately.

Additionally, several aspects of the cheliped morphology are also used to distinguish between species of Austinixa (see Manning and Felder 1989; Palacios Theil and Felder 2020a; Balbino et al. 2026). Some of these aspects include the length of the fixed finger and dactylus deflection, which we have found herein to be correlated. In cases where the fixed finger is long, the dactylus is not deflected, whereas an increased degree of dactylus deflection is observed when the fixed finger is shorter. Thus, we treated these characters as one in the analysis. We recovered that the most likely state for the common ancestor of Austinixa is an extremely reduced fixed finger with an extremely reduced dactylus for the males, which was retained in most of the descendant species. The appearance of a less reduced fixed finger and less deflected dactylus is convergent in Austinixa aidae and in (A. leptodactyla + A. roblesi). All of these possible convergences observed herein are in accordance with the discussion provided by Palacios Theil and Felder (2020a), in which the authors state that many of the characters used for species diagnoses do not reflect synapomorphies or the evolutionary history of these groups.

Some degree of phylogenetic signal can be observed in smaller clades for these characters with several convergences. For example, for the branchial ridge texture, cardiac tuft of setae in males, and fixed finger shortening and dactylus deflection, the common ancestor of A. roblesi and A. leptodactyla acquired state changes that were retained by both taxa, and thus, even though convergent with other species, these are synapomorphies of this clade in a more restricted context.

The presence of asymmetrical chelipeds evolved only in A. felipensis, being an autapomorphy of this species, as symmetrical chelipeds are the ancestral state for the genus Austinixa, and for other broader clades within Pinnixinae and Pinnotheridae. Given that asymmetrical chelipeds is the most common and widely distributed state within Thoracotremata and Eubrachyura as a whole (e. g. Abby-Kalio and Warner 1989; Spani et al. 2020; Rozaimi et al. 2023; Guinot and Rodríguez Moreno 2026), it is possible that the symmetrical chelipeds are a synapomorphy of Pinnotheroidea and a process of reversal happened in A. felipensis.

Some characters analyzed herein represent synapomorphies for smaller clades of Austinixa with no other cases of convergence. For example, for the orientation of the branchial ridge, the sharply bent ridge near the lateral margin originated in the ancestor of the clade ((A. behreae + A. chacei) + (A. leptodactyla + A. roblesi)), and was inherited by all of its descendants, thus representing a synapomorphy. Likewise, the depression on the P4 merus was likely acquired by the ancestor of (A. leptodactyla + A. roblesi) and retained in the descendant species. These character states have been pointed out as useful in species diagnoses (see Manning and Felder 1989; Palacios Theil and Felder 2020a; Balbino et al. 2026). The morphologically similar A. marianae, which was not included in the analyses due to lack of tissue samples, also shares these characteristics (see Salgado-Barragán et al. 2021). It is likely that this species is contained within the clade ((A. behreae + A. chacei) + (A. leptodactyla + A. roblesi)) and is likely more closely related to A. leptodactyla and A. roblesi, since it also shares the P4 merus depression. If this were the case, this would support these proposed synapomorphies.

The morphology of the G1 appears to reflect some phylogenetic signal at a larger scale within Austinixa. The plesiomorphic state for the G1 apex curvature for Austinixa is recovered as sharply curved outwards. Both states, sharply curved outwards and not curved outwards, are relatively common and widespread in Pinnotheridae (e. g. Zmarzly 1992; Campos 1996; Thoma et al. 2009; Rahayu and Ng 2010; Naruse and Maenosono 2012; Felder and Palacios Theil 2020; Palacios Theil and Felder 2020a), and have likely converged in many different groups. However, in our analysis, the ancestor of Austinixa and (Tubicolixa + (Rathbunixa + Scleroplax)) likely had a curved G1, although only Scleroplax presents this state in the latter clade. The curved G1 of the ancestor of Austinixa was inherited by A. patagoniensis and the common ancestor of the clade that contained all other species. The clade (A. cuestai + (A. cristata + (A. gorei + (A. artankeri + A. aidae)))) also retained this character state. Similarly, and somewhat surprisingly, the state for the common ancestor of (A. felipensis + ((A. behreae + A. chacei) + (A. roblesi + A. leptodactyla))) was also recovered as a curved G1 apex, although all of its descendant species present straight G1 apexes. It is possible that this happened due to limitations in our analyses, such as the inability to code the G1 morphology for certain outgroup taxa, which might have added some bias. Nonetheless, the curved G1 apex appears to be a retained plesiomorphy in at least two major groups of Austinixa (A. patagoniensis and the clade containing A. cuestai). If the common ancestor of the clade (A. felipensis + ((A. behreae + A. chacei) + (A. roblesi + A. leptodactyla))) had a straight G1 apex, this would be a reversal and would represent a synapomorphy of this subclade in a more restricted context. The scenario for the flattening of the G1 apex, however, is slightly different. The most likely ancestral state is “not flattened” for Austinixa as a whole, which was retained by the clade (A. felipensis + ((A. behreae + A. chacei) + (A. roblesi + A. leptodactyla))), whereas an autapomorphic state was acquired by A. patagoniensis (flattened laterally) and a horizontally flattened G1 apex evolved in (A. cuestai + (A. cristata + (A. gorei + (A. artankeri + A. aidae)))). Combining these characters, the most likely G1 morphology for the ancestor of Austinixa was a G1 with a sharply bent and non-flattened apex, which is a state combination not observed in any extant species of the genus. The presence of a laterally flattened G1 apex seems to be an autapomorphy of A. patagoniensis, the absence of curvature a synapomorphy of (A. felipensis + ((A. behreae + A. chacei) + (A. roblesi + A. leptodactyla))), and the horizontal flattening a synapomorphy of (A. cuestai + (A. cristata + (A. gorei + (A. artankeri + A. aidae)))). Thus, the G1 morphology reflected evolutionary history within Austinixa, which further supports the larger subclades obtained herein and by Palacios Theil and Felder (2020a).

In summary, the common ancestor of Austinixa most likely had a parallel and tuberculated branchial ridge, a cardiac ridge crossing the entire width of the carapace, no tufts of setae on the cardiac region, symmetrical chelipeds, extremely reduced cheliped fixed fingers with extremely deflected dactyli, no depression on the P4 merus, a bicarinate P4 propodus opposable margin, and a sharply curved and non-flattened G1 apex. Thus, apart from the G1 morphology and orientation of the branchial ridge, no other diagnostic characters analyzed herein have phylogenetic importance for major clades. Although some of these characters present some degree of phylogenetic signal for smaller clades, at a larger scale plenty of convergence and reversions are observed. Only some morphological support is observed for the clades recovered in the phylogenetic hypothesis proposed herein.

All that was known so far about the evolutionary history and biogeography of Austinixa was based on inferences made from phylogenetic trees (see Harrison 2004; Palacios Theil and Felder 2020a) and one broader study addressing Pinnotheroidea that used divergence time estimates and ancestral distribution reconstruction (Chow et al. 2023). Harrison (2004) stipulated that A. felipensis, the only species of Austinixa known for the Pacific at the time, diverged from other species at least 3 mya, with the closing of the Isthmus of Panama. This is also mentioned by Palacios Theil and Felder (2020a), which noted that both Pacific species known at the time, A. felipensis and A. cuestai, were the sister groups of the Atlantic species of their respective clades, which indicated that they might have diverged longer ago, with populations becoming completely disconnected with the closing of the isthmus. Thus, vicariant events caused by the formation of the Isthmus of Panama, a clear geographic barrier, were suggested to have potentially caused cladogenesis events in Austinixa. However, Chow et al. (2023) proposed much earlier divergence times for the eight species of Austinixa used therein, with A. felipensis (the only Pacific species included, although erroneously assigned to “Western Atlantic” in the ancestral state reconstruction analysis) having diverged from the Atlantic counterparts of its subclade about 20 mya (see Chow et al. 2023: fig. 3), which is before even deep-water connections between the Atlantic and Pacific were terminated (O’Dea et al. 2016), thus indicating that the formation of the Isthmus of Panama might not have led to vicariant events, since the separation of Atlantic and Pacific lineages was much older. According to our analysis, A. felipensis diverged from ((A. behreae + A. chacei) + (A. leptodactyla + A. roblesi)) and A. cuestai diverged from (A. cristata + (A. gorei + (A. artankeri + A. aidae))) about 12 mya. Our divergence time estimates, although younger than those proposed by Chow et al. (2023), also indicate that the Atlantic and Pacific lineages within Austinixa diverged before the deep-water connections between the Atlantic and Pacific ceased about 9.2 mya (O’Dea et al. 2016) and long before the final closure of the Isthmus of Panama about 3 mya (Fig. 1). Thus, it is probable that the Isthmus of Panama was not a geographic barrier responsible for vicariant events that separated the Pacific and Atlantic lineages of Austinixa included in the present study. Furthermore, the ancestral distribution reconstruction shows two clear dispersal events into the Pacific by A. felipensis and A. cuestai, whose ancestors were most likely present in the Atlantic, and more specifically in the TNWA. It is possible that more species of Austinixa might be present in the Eastern Pacific, given that this area remains largely unexplored compared to the Western Atlantic (see Palacios Theil and Felder 2020a), and if so, including such species in these analyses, along with the recently described Pacific taxon A. marianae, could further elucidate biogeographic aspects and the evolutionary history of the genus.

The third Pacific species, A. marianae, shares several similarities with the Atlantic A. roblesi and A. leptodactyla, including a patch of setae on the cardiac region of the carapace, posteriorly bent branchial ridges, and cheliped morphology. It is possible that this taxon is more closely related to these two species than to the other species in the genus. The divergence between (A. chacei + A. behreae) and (A. leptodactyla + A. roblesi) happened about 10 mya, also before the ending of the deep-water connections between the Atlantic and Pacific (O’Dea et al. 2016). If A. marianae is indeed more closely related to (A. leptodactyla + A. roblesi) than to (A. chacei + A. behreae), the divergence between the forementioned taxa must be younger than 10 Ma. Furthermore, since A. leptodactyla and A. roblesi are estimated to have diverged from each other about 2.5 mya, after the final closure of the Isthmus of Panama, the divergence between A. marianae and the common ancestor of the other two species must be 3 Ma old or older. Thus, it is possible that the earlier or later stages of the formation of the Isthmus of Panama affected cladogenesis in this group. However, it is still unclear whether the possible common ancestor of A. marianae and (A. leptodactyla + A. roblesi) inhabited the TEP, with a vicariant event leading to speciation, or if the ancestor of A. marianae dispersed into the TEP and the formation of the isthmus reinforced the lack of gene flow.

Since species of Austinixa occur only in shallow waters, it is likely that they began to be affected by the closing of the Isthmus of Panama more recently, and not when deep-water connections were severed, or that cladogenesis events were already taking place within these subclades and that the closing of the isthmus only further isolated these groups. Our estimations of time of divergence support this idea, given that most Atlantic species and subclades had already diverged from each other prior to the ceasing of deep-water connections. For example, A. cristata diverged from (A. gorei + (A. artankeri + A. aidae)) 10.7 mya, and A. gorei diverged from (A. artankeri + A. aidae) 9.8 mya. Additionally, as previously mentioned, (A. leptodactyla + A. roblesi) diverged from (A. chacei + A. behreae) approximately 10 mya. Thus, several cladogenetic events within Austinixa were happening between Western Atlantic and Pacific lineages, and within Western Atlantic groups before the formation of the Isthmus of Panama started having an effect on shallow-water taxa, which were the last to be affected (Knowlton et al. 1993; Coates and Obando 1996).

Harrison (2004), given his phylogenetic hypothesis, mentioned that there might have been a widespread ancestral population of Austinixa in the Atlantic that later separated into A. patagoniensis and a “northern clade” (as mentioned by the author) prior to the closure of the Isthmus of Panama, given that A. felipensis belongs to the latter group. Our ancestral distribution reconstruction points to an alternative situation as the most likely scenario (49.9% probability), where the ancestor of the entire genus inhabited the Tropical Northwestern Atlantic (TNWA), with the lineage leading to A. patagoniensis having dispersed into the Warm Temperate Southwestern Atlantic (WTSWA). However, the second most likely scenario (40.5% probability) recovered the distribution of the ancestor of Austinixa as the (TNWA + TSWA), which supports the hypothesis proposed by Harrison (2004). The first and most likely alternative is more consistent with the degree of endemism observed in most species of Austinixa, which are usually restricted to smaller areas (Palacios Theil and Felder 2020a), in most cases in only one marine province, as observed herein. Furthermore, Chow et al. (2023) also recovered the Western Atlantic as the most likely distribution area for the common ancestor of Austinixa. However, the areas used by Chow et al. (2023), such as the “Western Atlantic” or “Eastern Pacific”, were much broader than the marine provinces used herein, and thus, we were also able to investigate biogeographic aspects of the genus within the Western Atlantic and Eastern Pacific.

In accordance with Harrison (2004), the cladogenesis event that led to the separation of A. patagoniensis from other species happened before the closing of the Isthmus of Panama and is herein estimated to have occurred 15.4 mya. This is about 10 Ma later than what was recovered by Chow et al. (2023), but both estimations are much older than the ceasing of the deep-water connection between the Atlantic and Pacific. These younger divergence time estimates obtained herein in comparison to Chow et al. (2023) may have happened due to differences in taxon sampling, with the latter using several more outgroups and other pinnotherid genera, whereas the present study focused on obtaining a good coverage of the species within Austinixa, and differences in which fossils were used for calibration, with the use of a crown-group pinnotherid herein.

The northwestern Atlantic was recovered as the distribution area for the ancestor of the clade containing all species of Austinixa except for A. patagoniensis. The closure of the Isthmus of Panama was not only important in separating taxa in the Atlantic and Pacific oceans, but also dramatically affected the Northwestern Atlantic, including the Caribbean and Gulf of Mexico, altering ocean currents, nutrient distribution, temperature, and salinity, which influenced the evolution of marine fauna (Coates and Obando 1996; Allmon 2001). This, together with the glaciation events that happened in North America, could have greatly influenced the many cladogenetic events that occurred in the Northwestern Atlantic, whereas cladogenesis events, as far as we know, did not happen within the lineages from the Pacific or Southwestern Atlantic. This region was affected by changes in ocean currents, such as the intensification of the Gulf Stream, as the shoaling of the Central American Seaway progressed at about 4.6 to 4.2 mya (Auderset et al. 2019), which could have affected speciation events between A. behreae and A. chacei, which happened shortly after (3.95 mya) according to our analysis. Additionally, an increase in global dryness and decrease in overall temperature during the late Miocene (between 7 and 5.4 mya) could have also affected marine ecosystems (Herbert et al. 2016), which could be related to the separation between A. artankeri and A. aidae sensu lato. The distribution of the ancestor of (A. cristata + (A. gorei + (A. artankeri + (A. aff. aidae + A. aidae)))) was also recovered as the TNWA, with A. cristata and the ancestor of (A. aff. aidae + A. aidae) extending their distributions, colonizing the WTNWA and TSWA respectively. A vicariant event probably led to the separation of the latter into two species about 3.16 mya, with A. aff. aidae becoming restricted to the TNWA and A. aidae to the Southwestern Atlantic. In the other Atlantic clade, ((A. behreae + A. chacei) + (A. leptodactyla + A. roblesi)), the ancestor likely occupied either the TNWA or (TNWA + WTNWA). In the first case, the ancestor of (A. behreae + A. chacei) likely dispersed into the WTNWA, where the later cladogenetic event occurred and separated the two species. However, in the second case, a vicariant event could have led to the separation of (A. behreae + A. chacei) from (A. leptodactyla + A. roblesi), whose ancestor likely occupied the (TNWA + TSWA).

The speciation event separating A. leptodactyla and A. roblesi is the most recent for any species pairs in Austinixa and happened about 2.5 mya likely due to a vicariant event. At around this time (2–5 mya), there was an intensification of the Amazon-Orinoco Plume (AOP), a freshwater discharge that deposits sediment into the Atlantic, reducing salinity and increasing turbidity of the ocean in northern South America (Figueiredo et al. 2009; Hoorn et al. 2010). The Amazon-Orinoco plume is also a barrier that is known to affect the dispersal capabilities of many different groups of animals, including decapods (Peres and Mantelatto 2020; Peres et al. 2022; Mantelatto et al. 2023). Two of the three southwestern Atlantic species of Austinixa are known to occur in northern Brazil, near the region affected by the plume. As mentioned above, it is most likely that one dispersal event led to the colonization of the Southwestern Atlantic by A. patagoniensis, whereas the ancestors of the other two species likely expanded their distributions into the area, followed by vicariant events that separated them from their northern Atlantic counterparts, which is the case for A. leptodactyla and A. aidae sensu stricto. The first event was estimated to have happened prior to the intensification and full establishment of the plume. The AOP, although probably not responsible for the original reduction of gene flow and speciation, likely maintained this species’ isolation and stopped it from dispersing back into the Caribbean. For the other two species, the vicariant event likely happened during its establishment, however, given that both taxa are found around the plume area, it is unlikely that this was the main factor that influenced speciation in these cases.

Other important factors that might have affected the evolution of Austinixa and speciation events are ocean currents, larval characteristics, and host association, as previously mentioned by Palacios Theil and Felder (2020a). Modern ocean currents may be involved in limiting and maintaining modern distributions of the species (Fig. 4). Such currents are affected by geological processes and have changed over time, as mentioned above, which may have led to cladogenetic events (Coates and Obando 1996; Allmon 2001). For decapods, larval characteristics are also associated with species distribution, including the length of larval development, physiological tolerances, settlement preferences and even swimming patterns (Palacios Theil and Felder 2020a). However, as also mentioned by Palacios Theil and Felder (2020a), these traits are understudied in Austinixa. Host association might also be a determining factor in the biogeography of Austinixa. Palacios Theil and Felder (2020a) reported that intraspecific genetic variability within A. cristata and A. gorei mirrors the genetic structure of host populations to a certain extent, as other studies have observed for other pinnotherids (Stevens 1990). However, to thoroughly understand how host association has affected the evolution of the genus, more studies are necessary.

Austinixa aidae was originally described by Righi (1967) (as Pinnixa aidae) based on material collected from the mouth of the Juqueriquerê river, in Caraguatatuba, São Paulo, southeastern Brazil. The only provided illustration of the carapace depicts the female holotype (see Righi 1967: fig. 21). Decades later, Heard and Manning (1997) described A. hardyi arguing that it was distinct from all other species of Austinixa since males have dense patches of setae anterior to cardiac ridge. However, the authors mention that it is similar to A. aidae and A. gorei in their lack of branchial ridges. Additional differences mentioned by the authors include the relative width of the carapace and body size. The authors were probably unaware that Righi (1967) had described the gastro-cardiac sulcus as strongly pubescent in males, which likely led to the use of these patches of setae as a diagnostic character for A. hardyi. Harrison (2004) raised suspicions regarding the validity of A. hardyi, given that 16S rRNA sequences of this taxon and A. aidae were identical and COI sequences differed by only two base pairs. The specimens used for the phylogenetic and pairwise distance analyses in the study were collected from areas near the type locality of both taxa. Harrison and Hanley (2005) then classified A. hardyi as a junior synonym of A. aidae based on molecular, morphological, and morphometric data. The authors also detected the patches of setae on the carapace of males of A. aidae. Palacios-Theil et al. (2009) then suggested that A. aidae and A. hardyi might be distinct and should be regarded as separate species until further analyses are carried out. However, Palacios Theil and Felder (2020a) regard A. hardyi as a junior synonym of A. aidae without further discussion. Recently, Almeida et al. (2023), using molecular and morphological data, including material from the type locality of A. aidae, argued that A. hardyi is actually a junior synonym of A. aidae, also arguing that the carapace size and proportions mentioned by Heard and Manning (1997) were not useful in distinguishing between the two supposed groups.

The ML molecular phylogenetic hypothesis using the 16S rRNA/tRNA-Leu/NADH1 gene complex proposed by Palacios Theil and Felder (2020a: fig. 1) recovers a clade composed of specimens of A. aidae from Panama and a clade composed of individuals from Brazil and Venezuela, both with high to very high support (86 and 99, respectively). Within the South American subclade, a clade composed of specimens from Brazil, including A. bragantina which was later synonymized with A. aidae by Almeida et al. (2023), was recovered with very high support (99). Similarly, Almeida et al. (2023) in their phylogenetic hypothesis using 16S rRNA recovered two large clades within A. aidae, one consisting of individuals from Panama (97 support), and another with specimens from South America and Tobago (89 support). Within the South American subgroup, two large clades were also recovered, one including one sample from Venezuela, the specimen previously identified as A. hardyi by Harrison (2004) from Tobago, and the specimen from São Paulo used by Harrison (2004), and another with specimens from several localities in Brazil. This led the authors to conclude that the populations in South America and Tobago are connected, and that A. hardyi should indeed be regarded as a junior synonym of A. aidae. However, in the concatenated phylogenetic hypothesis using multiple genes (16S rRNA/tRNA-Leu/NADH1 complex, 12S rRNA and H3) generated by Palacios Theil and Felder (2020a), the specimen from Venezuela is recovered as the sister group of the clade from Panama with moderate to high support. This was also the case for our analyses, although with low support.

The Bayesian analysis with a molecular clock also shed light on the relationship between these subgroups of A. aidae sensu lato. The divergence between A. aff. aidae from Panama and Venezuela and A. aidae from Brazil (3.16 Ma) is longer than that observed for the traditionally recognized species A. leptodactyla and A. roblesi (2.48 Ma). Furthermore, another group of morphologically similar species diverged at similar times, A. chacei and A. behreae (3.95 Ma). Thus, it is very likely that A. aff. aidae from Panama, and possibly Venezuela, is a separate species from A. aidae sensu stricto from Brazil, even though they are likely morphologically very similar, which led to their identification as A. aidae until now. Furthermore, this raises questions regarding the taxonomic status of the specimens from Venezuela and Tobago, which are recovered as the sister group of A. aidae sensu stricto in some analyses (see Palacios Theil and Felder 2020a: fig. 1), and as the sister clade of A. aff. aidae in others (Palacios Theil and Felder 2020a: fig. 2; present study: Fig. 1). In our estimation of time of divergence, it was suggested that the Venezuelan group diverged from the Panamanian subclade 2.5 mya, which falls within the range of what is observed for other species-level separations. Additionally, in the 16S rRNA tree proposed by Almeida et al. (2023: fig. 10), the specimen used by Harrison (2004) from near the type locality of A. aidae clusters with A. aidae from Venezuela and A. “hardyi” from Tobago, even though all other individuals from Brazil obtained by Palacios Theil and Felder (2020a) and therein form a separate clade, which is similar to what we have recovered in the present study. If this is correct, there appears to be a certain degree of connection between Brazilian and Venezuelan/Tobagonian populations. However, if this happened due to contamination, or a poor-quality sequence, it is also possible that the population found in Venezuela and Tobago represents a separate species, which is likely A. hardyi. Thus, the taxonomic status of A. hardyi remains unresolved and requires future analyses with more specimens and access to the type material, which we were not able to examine herein.

Austinixa leptodactyla and A. roblesi are extremely similar morphologically, with even their G1 being indistinguishable, which is a characteristic that can be used to identify most species of Austinixa. However, differences in the male telson shape, being quasi-semiellipsoid and as long as the sixth pleonite in A. roblesi (Palacios Theil and Felder 2020a) and subtriangular with rounded margins and longer than the sixth pleonite in A. leptodactyla (Balbino et al. 2026), can be used to differentiate these species. It is possible that A. aidae sensu stricto and A. aff. aidae from Panama and Venezuela are just as similar to each other, and that only minor differences are going to be helpful in distinguishing them. This could also be the reason why they were identified as A. aidae in the first place. The same might be true for the specimen of A. aff. aidae from Venezuela.

Austinixa leptodactyla was herein included in molecular phylogenetic analyses for the first time. This species, endemic to Brazil, was recovered as the sister species of A. roblesi, found in the Caribbean, to which it is extremely similar morphologically (see Palacios Theil and Felder 2020a; Balbino et al. 2026). Additionally, these species also share several morphological similarities with the recently described A. marianae from the Eastern Pacific, for which we have not obtained any sequences. Similarities between these three species include the presence of a bilobed tuft of setae in the cardiac region of males, the cheliped morphology, including the level of fixed finger reduction and dactylus deflection and dentition of the cutting edge of both fingers, and the presence of a dorsodistal depression on the P4 merus (see Palacios Theil and Felder 2020a; Salgado-Barragán et al. 2021). The phylogenetic proximity of A. roblesi and A. leptodactyla indicates that these characters are likely synapomorphies of this group and thus allows us to infer that A. marianae is phylogenetically close to these species, although its exact position within the tree is impossible to be determined without further molecular data.

Austinixa leptodactyla is likely more closely related to A. roblesi, having diverged from each other 2.48 mya, representing the most recent divergence event between any described species of Austinixa. It is very likely that A. marianae diverged from (A. leptodactyla + A. roblesi) much earlier, before the total closure of the Isthmus of Panama, which occurred 3 mya (O’Dea et al. 2016; Bourke et al. 2025). This very close proximity between the latter two taxa is also evidenced by the species’ morphology, which is almost identical. For these two species, the overall carapace, Mxp3, cheliped, pereopod and G1 morphology is indistinguishable (see Palacios Theil and Felder 2020a: figs. 8, 9; Balbino et al. 2026). The male telson shape, however, can be used to separate these taxa. In A. leptodactyla, the telson is subtriangular with rounded margins and longer than the sixth pleonal somite, whereas in A. roblesi, the telson is quasi semiellipsoid and as long as the sixth somite (see Palacios Theil and Felder 2020a: fig. 8G; Balbino et al. 2026: fig. 6A, C). Palacios Theil and Felder (2020a), when describing A. roblesi, did not have access to specimens of A. leptodactyla for comparison, but assumed that the P4 merus depression in the latter was proximal based on the description provided by Coelho (1997). However, the original description does not adequately represent the morphology of the species, and the P4 merus depression is similar in both taxa. The description of A. leptodactyla provided by Coelho (1997), as is the case for other species addressed in the same study, was vague and does not thoroughly discuss important aspects (see Balbino and Mantelatto 2024; Balbino et al. 2025; Balbino et al. 2026; Oliveira et al. 2026).

Although not as similar to A. leptodactyla as A. roblesi, A. marianae shares important similarities with the former. The telson morphology of A. marianae (see Salgado-Barragán et al. 2021: fig. 3E) is very similar to that of A. leptodactyla (see Balbino et al. 2026: fig. 6A, C), and both differ from that of A. roblesi (see Palacios Theil and Felder 2020a: fig. 8G). However, the pleon of A. marianae tapers more abruptly towards the distal end than that of A. leptodactyla. Additionally, the G1 of A. marianae (see Salgado-Barragán et al. 2021: fig. 3J, K) has a bifurcated apex, as opposed to a straight and simple apex, with a very slight indentation in A. leptodactyla (see Balbino et al. 2026: fig. 6D). The male cheliped morphology of the three species appears to be very similar, however, the female cheliped morphology of A. marianae (see Salgado-Barragán et al. 2021: fig. 3H, I) differs slightly from A. leptodactyla and A. roblesi in having a much less pronounced median tooth on the cutting edge of the fixed finger.