For most of the Miocene, the Pebas system experienced a connection to the Caribbean Sea, which may have offered an ideal transitional environment for adaptations and colonization of proto-marine lineages in freshwater environments, especially in the region that now comprises the Amazon [14, 15]. Subsequently, the diversification of taxa of marine origin depends on speciation processes common to continental freshwaters. In the case of the Amazon, periods of glaciation during the Pleistocene and tectonic processes associated with the uplift of the Andes resulted in changes in abiotic conditions and drainage connections in the region [4, 16, 17, 18]. However, for other families that have both marine and freshwater species (e.g., Achiridae), there is little information about the influence of these past processes of speciation [19]. 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101

The flatfish family Achiridae is a member of the superfamily Soleoidea, suborder Pleuronectoidei, Order Pleuronectiformes [ 9, 20, 21 ]. Although some molecular studies have questioned the monophyly and ordinal status of Pleuronectiformes (see references in Atta et al. [22]. Achiridae is composed of six genera: Achirus Lacépède, 1802, Apionichthys Kaup, 1858, Gymnachirus Kaup, 1858, Catathyridium Chabanaud, 1928, Trinectes Rafinesque, 1832, and Hypoclinemus Chabanaud, 1928. Some achirids are exclusive to marine environments (Gymnachirus), while others are entirely restricted to freshwater (Hypoclinemus). Other achirids are predominantly freshwater (Apionichthys and Catathyridium) and some (Achirus and Trinectes) inhabit multiple environments across a range of salinities [ 9 ]. The phylogenetic relationships of Achiridae have been debated in recent years, with some studies placing it near Citharidae [ 19, 23 ] or Rhombosoleidae [24, 25]. The placement of Achiridae within Soleoidea is uncertain due to disagreement over the composition of this superfamily [22]. Betancur-R et al. [24] and Harrington et al. [26] support Achiridae as the first lineage to diverge within a clade that also includes Achiropsettidae, Paralichthodidae, and two clades of Rhombosoleidae (I and II); that larger clade is sister to one composed of the remaining members of Soleoidea (Cynoglossidae, Poecilopsettidae, Samaridae, Soleidae). Atta et al. [22], however, transferred Achiropsettidae, Paralichthodidae, and Rhombosoleidae to a separate superfamily (Pleuronectoidea), and supported Achiridae as sister to a clade composed of Cynoglossidae, Poecilopsettidae, Samaridae, and Soleidae within a restricted Soleoidea. Recently, Bitencourt et al. [27] showed that Achiridae is indeed a monophyletic family that arose during the Oligocene-Miocene transition.

Bitencourt et al. [27] also point out that some genera may harbor cryptic lineages and that Hypoclinemus may have a closer relationship with the samples of Achirus achirus (Linnaeus, 1758) used in their study. Chabanaud [28] proposed Hypoclinemus for two 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 species, H. paraguayensis which he described from the Paraguay river, and the nominal Solea mentalis described by Günther (1862) from the Capim river near Belém, Brazil. Ramos [29] synonymized the former species with Catathyridium lorentzii (Weyenbergh, 1877). Described as Achirus hasemani (Steindachner, 1915) from Rio Branco, Brazil, which was also synonymized with H. mentalis [28]. Thus, Hypoclinemus is monotypic with only H. mentalis considered valid. Hypoclinemus is distinguished by having teeth on both rami of the dentary vs. restricted to the left ramus (blind side) in other achirids [21, 30]. This characteristic is considered the most plesiomorphic among achirids, providing osteological support for H. mentalis as the sister species to the rest of the group. Hypoclinemus mentalis is widely distributed throughout Greater Amazonia, which includes the Amazon, Essequibo, and Orinoco basins [31]. Its large distribution raises questions as to whether Hypoclinemus is in fact monotypic, since other widespread neotropical fishes have been recently split into multiple species (e.g., De Santana et al. [32]; Garavello et al. [33]) or shown to include hidden diversity represented by multiple species-level lineages (e.g., Mendes et al. [34]).

Given the complex topology of Greater Amazonia (sensu Van der Sleen and Albert, [31]), Hypoclinemus is also a good candidate to investigate how speciation processes may affect the evolutionary history of Neotropical fish species, which in many cases include the divergence of cryptic lineages [18, 35, 36, 37]. In freshwater environments, fishes found in smaller streams and headwaters tend to be more strongly fragmented and isolated, even on small spatial scales, providing ample opportunity for speciation [18, 38], while fishes common to major rivers with accumulated sedimentary substrates tend to speciate based on large-scale vicariance events between basins and ecological differences across basins [32]. The geographic range of H. mentalis spans major basins (Amazon, Orinoco, Essequibo) that present spatial and environmental 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 heterogeneity, factors that can influence speciation [39]. Therefore, this study aims to test the monophyly of the genus Hypoclinemus, exploring whether H. mentalis is composed of a single evolutionary lineage across its distribution in the Neotropics, or whether multiple lineages exist that can create a biogeographic profile for understanding the history of the region.

The present study also incorporated public sequences from GenBank [42] for the three amplified markers for all six achirid genera (Achirus, Apionichthys, Gymnachirus, Catathyridium, Trinectes, and Hypoclinemus) as well as the outgroup genera Bothus Rafinesque, 1810, and Symphurus Rafinesque, 1810 (S1 Table). For our study, we used three different datasets. First, the concatenated dataset for all the three markers (rrln+cox1+RHO - dataset 1). This dataset only lacked the cox1 sequences from Bitencourt et al. [27] that were not available in Genbank. With dataset 1, we generated phylogenies (Maximum likelihood-ML and species tree) and estimated the Time to Most Common Ancestor (TMRCA) and Reconstruction of Ancestral States (RASP 4). We also generated phylogenies using only the cox1 data (ML and Bayesian trees - dataset 2, S2 Figure 2A) and using a concatenation of the rrln and RHO sequences (ML and species trees - dataset 3, S2 Figure B).

For all phylogenetic analyses performed in this study, we used concatenated datasets; however, evolutionary models were independently estimated for each partition using ModelFinder 2 [43], implementing the AIC criterion to determine the best evolutionary model for both Maximum Likelihood (ML) and Bayesian Inference (BI) analyses. The models GTR+I+G (rrln), HKY+I+G (cox1), and HKY+I (RHO) were selected for both ML and BI analyses for dataset 1. For dataset 2 (cox1), TIM2+I+G (ML) and HKY (BI) were selected. Finally, for dataset 3 (rrln+RHO), TIM2+I+G (ML and BI) and HKY+I+G (ML and BI) were selected, respectively.

Initially, an ML tree was generated in IQ-TREE 2 [44] using the 1000 ultrafast bootstrap replicates [45]. A species tree was then produced using *BEAST [46] which is part of the BEAST v.1.7.4 package [47]. In BEAUTi, the evolutionary models were “unlinked” to use the previously estimated models for each separate partition. Only 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 species with sequence data for more than two markers were analyzed. For this, we used an uncorrelated relaxed clock with a lognormal distribution prior on rates and a Yule speciation prior [46, 48]. MCMC (Markov Chain Monte Carlo) samples were estimated through four simultaneous runs containing four chains (one cold and three warm) with 50 million generations. All posterior probabilities were defined using the 80% consensus rule, with samples taken every 1000 generations, and 10% of the initial trees discarded as burn-in. Log-likelihood files generated in each run were viewed in Tracer v.1.4 [47], where only runs with ESS values equal to or greater than 200 were considered. The consensus tree was then generated using TreeAnnotator v.1.4 [47].

The time to The Most Recent Common Ancestor (TMRCA) was estimated with BEAST 2 [49], using the three previous independent partitions. The TN93+G model was used with base frequencies determined empirically. For tree prior parameters, the Yule model was selected, using the uncorrelated relaxed clock and a lognormal distribution for rates without enforcing monophyly for any genera [48]. We used two dating priors in the BEAST2 analysis, both of which were estimated in Shi et al. [ 23 ], based on fossil calibrations from Near et al [50]. The first one is the separation between Samaris cristatus Gray, 1831, and Poecilopsetta beanii (Goode, 1881) at 52.9 million years ago (mya). We assigned a log-normal distribution prior and set this prior to have an offset of 51.0, standard deviation (S) of 0.5, and mean (M) of 17.78. For the second calibration, we used the separation between Trinectes and Achirus at 18 mya, where we assigned a log-normal distributed prior and set this prior to have an offset of 15.0, standard deviation (S) of 0.5, and mean (M) of 16.7. The MCMC method was used to infer the divergence times with four independent runs with 120 million generations through four simultaneous runs containing four chains (one cold and three heated) with sampling performed every 1000 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 generations. Only runs with ESS values equal to or greater than 200 for all marginal parameters were used. Log-likelihood files generated in each run were viewed in Tracer v.1.4 [47], where only runs with ESS values equal to or greater than 200 were considered. The consensus tree was then generated using the TreeAnnotator v. 1.4 [47].

RASP 4 [51] was used to reconstruct the ancestral biogeographic area, using the Bayesian Binary Method (BBM). The TMRCA tree generated in BEAST 2 was used as the guide tree. We used the ecoregion level areas defined for marine [52] and freshwater [53] species. The BBM analyses were run for eight million cycles, using ten chains, with sampling every 100 cycles. The temperature was set to 0.1, and a fixed JC model was used. The maximum number of areas for all nodes was set at four. Subsequently, the information of each node was plotted on pie charts.

A total of 30 individuals were sequenced for each of the three genes (rrln, cox1, and RHO) in the present study. All GenBank sequences utilized for each dataset are listed in S1 Table. Together these resulted in a total of 187 individuals for analyses.

Based on the phylogenetic reconstructions simulated here (based on two mitochondrial regions and one nuclear gene) Hypoclinemus was recovered as a monophyletic genus, and as sister taxon to Apionichthys with a high posterior probability (Fig. 2). 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283

Fig 2. Species tree generated using *BEAST. Only support values above 0.95 for Bayesian Inference are shown, for maximum likelihood, only supports above 85% are indicated. Color of bars represent ecoregion associations inferred in RASP 4. Tip nodes are numbered according to location numbers in Fig. 1.

We also recovered seven distinct lineages within Hypoclinemus. Lineage I is from the Lower Xingu (Amazonas Estuary and Coastal Drainages ecoregion). Lineage II is composed of two specimens from the Orinoco basin (Orinoco Guiana Shield ecoregion) with high support values (Fig. 2). Lineage III is composed of two specimens from the Iriri River (Xingu ecoregion). Lineage IV is composed of two specimens from the Upper Amazon (Iquitos, Amazonas Lowlands ecoregion). Lineage V is represented by a single specimen from the Lower Tocantins (Amazonas Estuary and Coastal Drainages ecoregion). Lineage VI is composed of six specimens, including three from the Upper Amazon (Iquitos and Amazonas) and three from the middle Amazon (Amazonas) (Amazonas Lowlands ecoregion). Lineage VII is composed of a large clade containing the rest of the specimens utilized in the present study from many ecoregions including Essequibo, Lower Xingu, Middle Amazon and Delta Amazon. (Figure 2). Some differences in the relationships between, and support for, lineages can be found when analyzing the Dataset 2 (cox1) or Dataset 3 (rrnl+RHO), but are generally concordant (S2 Figure A and B). Using only Dataset 2 weakens support for most recent nodes, resulting in poor support for monophyly of the clade Apionichthys+Hypoclinemus, since A. cf. menezesi was a recovered sister to Hypoclinemus lineage I previously estimated here (S2 Figure A). But we also recovered significant support for four of the lineages within H. mentalis. A similar trend occurs when using Dataset 3 recovering also four lineages of H. 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 mentalis, and for this dataset Hypoclinemus was recovered as a monophyletic clade (S2

Three ecoregions predominated the analyses of speciation events in Hypoclinemus: Amazonas Lowlands, Xingu, and Amazonas Estuary and Coastal Drainages, where the latter concentrates most of the cases (Fig. 3, Table 2).

Fig 3. Historical biogeographical reconstruction. Inferred by BBM method in RASP 4 [51] combined with Time of Most Recent Common Ancestor tree (TMRCA) generated in BEAST 2 utilizing the log-normal relaxed clock. Blue horizontal bars on the nodes correspond to 95% High Posterior Density (HPD). Numbers indicate the split between the nodes as presented in Table 2. The different color ellipses combine the ecoregion's description from Spalding et al. [52] and Abel et al. [53]. Lineages are numbered I-VII on the right side of the phylogeny. Tip nodes are numbered according to location numbers in Fig. 1. 307 Ancestral Area

Dispersion

Vicariance

Extinction

Route (Probability) I II III IV V VI VII VIII

Emergence of Hypoclinemus ancestral Separation of Hypoclinemus lineage I (Xingu/Iriri)/Hypoclinemus species complex Separation of the ancestral of Hypoclinemus lineages II, III, IV, V and VI from Hypoclinemus lineage VII Separation of the ancestral of Hypoclinemus lineages II (Lower Xingu) and III (Orinoco) and IV (Peru 1) from lineages V (Lower Tocantins) and VI (Peru 2) Emergence of Hypoclinemus lineage VII Separation of Hypoclinemus lineages II (Lower Xingu) and III (Orinoco) and from lineage IV (Peru 1) Separation of the ancestral of Hypoclinemus lineages II (Lower Xingu) from ancestral of Hypoclinemus lineage III (Orinoco) Separation of Hypoclinemus lineages V (Lower Tocantins) from lineage VI (Peru 2) ~9.20 mya (6.064 - 12.364 mya.

Posterior=1)

F – 79.72%: E – 6.70%; * - 13.58 % ~5.55 mya (3.130 - 8.264 mya.

Posterior=0.99) ~3.81 mya (2.337 - 5.848 mya.

Posterior=0.99) ~2.84 mya (1.627-4.567 mya.

Posterior=0.99) ~2.60 mya (1.387-4.327 mya.

Posterior=0.99) ~2.10 mya (990ka-3.603 mya.

Posterior=0.93) ~1.43 mya (538ka-2.662 mya.

Posterior=1)

F – 68.28%; E – 21.44%; EF – 7.24%; * - 6.04% F – 85.14%; * - 14.84% F – 87.58%; * - 12.42% F – 85.30%; * - 14.70% G – 44.80%; F – 34.38%; EF – 18.67%; * - 2.15%

F – 89.35%; * - 10.65% ~1.18 mya (303ka-2.429 mya.

Posterior=0.98)

F – 52.43%; G – 28.79%; FG – 16.41%; * - 2.37% 2 2 1 2 2 2 1 1 1 1 1 1

F->F^F->F|F (0.3467) F->EF->E|F (0.5392) F->F^F->F|F (0.6360) F->FG->F|G (0.3506) F->HF->H|F (0.4442) G->GF->G|F (0.2277) F->FJ->F|J (0.8709) F->FG->F|G (0.4971)

The following symbols represent dispersion (-> -between areas, ^ - within a geographic area) and vicariance ( | ). Collection localities associated with Freshwater and Marine Ecoregions of the World: E – Xingu; F – Amazonas Estuary and Coastal Drainages; G – Amazonas Lowlands; *-not recovered. 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333

Our TMRCA shows that Hypoclinemus split from its common ancestor with Apionichthys in the Miocene (~9.20 mya), most probably in the Amazon Estuary and Coastal Drainages (72.72%) and Xingu (6.70%) ecoregions (Fig. 3, Table 2, Node I). The end of the Miocene/beginning of the Pleistocene marked the rise of the first distinct lineage (Lineage I) inside Hypoclinemus, diverging from the common ancestor of all Hypoclinemus at ~5.55 mya, probably in the same ecoregions (Amazon Estuary and Coastal Drainages and Xingu) but resulting from… two events of dispersion and one vicariance (Fig. 3, Table 2, Node II). The subsequent divergence of Lineage VII from its shared common ancestor (the ancestor of lineages II, III, IV, V, and VI) is estimated to have taken place at ~3.81 mya. The common ancestor of lineages II and III is then estimated to have divergence from its shared common ancestor (the common ancestor of lineages IV, V, and VI) at ~2.84 mya (nodes III and IV, respectively, in Fig.3, Table 2).

In a short interval of time, the TMRCA and RASP analyses (Fig. 3, Table 2) estimated divergence within lineage VII at ~2.60 mya (node V) and the separation of lineage IV from its common ancestor with the common ancestor of lineages V and VI at ~2.10 mya (node VI). Finally, in the Pliocene, these analyses estimated the divergence between lineage I from Lower Xingu and lineage II from Orinoco at ~1.43 mya and the divergence of lineage V from the Tocantins and VI from the Upper Amazon at ~1.18 mya. (Nodes VII and VIII, Fig. 3, Table 2).

Our results shed new light on hidden diversity inside the monotypic genus Hypoclinemus. However, the results generated here are different from Bitencourt et al. 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 [27], most likely because of the distinct composition of our datasets in comparison with those of Bitencourt et al. [27], primarily because sequences of cox1 from their work were unavailable. Additionally, our study employed a smaller number of molecular markers, and we used different sequences and samples compared to those utilized by those authors.

The presence of A. achirus within Hypoclinemus (rendering it paraphyletic) described in Bitencourt et al. [27] was not observed in any of the phylogenetic reconstructions generated in the present study. Instead, A. achirus was only recovered as a sister lineage to the Hypoclinemus species complex in reduced datasets and this relationship was not always strongly supported. This was observed in the topologies derived from dataset 2 (only cox1 sequences - S2 Figure A) and dataset 3 (only rrln + RHO sequences - S2 Figure B). However, with dataset 1 (the most complete dataset formed by the three markers) and in our TMRCA analysis (dataset 1), Apionichthys was consistently recovered as the sister genus. It is worth noting that Byrne et al. [19] conducted a phylogenetic study with nine molecular markers (four mitochondrial and five nuclear) within the order Pleuronectiformes, including representatives of 13 of the 14 families. Their results, similar to ours, confirmed the validity of the genus Hypoclinemus at the genus level. However, there were differences in the placement of Hypoclinemus as a sister of Achirus, as opposed to our findings where the sister genus was Apionichthys.

Ramos [21] diagnosed Hypoclinemus based on the presence of teeth on both rami of the dentary, as opposed to them being restricted to the left ramus (blind side) as in other achirids. This led Ramos [21] to suggest that Hypoclinemus could be the earliest lineage, possibly originating in freshwater. However, our results present a contrasting scenario. Hypoclinemus, according to all phylogenetic reconstructions in our study, does form a monophyletic group, but it is not monotypic. Our findings indicate that it is sister to the genus Apionichthys. We also discovered the existence of at least seven lineages (putative 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 cryptic species) within Hypoclinemus. To further clarify and confirm the presence of these species-level groups within the genus Hypoclinemus, additional taxonomic and systematic investigations are required.

Several authors have discussed the origin and diversification of freshwater fish lineages with recent evolutionary ties to marine species. Among the most used hypotheses is marine invasion through river mouths [54]. However, according to many authors, marine incursions on the South American continent began in the Cretaceous period, but it was during the Miocene that the greatest number of neotropical freshwater taxa diverged from their marine relatives [ 4, 7, 8 ]. During the transition between Oligocene to Miocene, the proto-Amazon was draining into Caribbean region which promotes several marine intrusions associated with Pebas system generating colonization in western Amazon region, which in turn could promote biotic interchanges of faunas between marine and freshwater environments [4, 11].

The formation of the Pebas mega-wetlands, stretching from the Caribbean to southern South America [11, 14], played a significant role in driving the speciation of neotropical fish species [18]. This extensive swampy system covered the western Amazon during most of the Miocene (around 10 to 20 mya) [15] and experienced marine incursions from the Caribbean, which likely influenced the adaptations of species from estuarine to freshwater conditions The Pebas mega-wetlands

This supports the hypothesis that the colonization of inland waters originated from the north and the Caribbean Sea, resulting in the mixed sister clade position of Apionichthys relative to Hypoclinemus in our biogeographic reconstruction (Fig. 4). The formation of the Andes and Cordillera de Mérida disrupted the connection between Lake 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406

Pebas and the Caribbean. Additionally, the emergence of the Vaupes Arc, associated with global cooling and lower sea levels, increased sediment discharge from the Andes. The formation of the Acre system also contributed to this process [55, 56]. This pattern of colonizing freshwater environments in the Amazon region has been observed in other aquatic species, such as Amazonian pufferfish [57] and potamotrygonin stingrays [4].

Before 10 million years ago (mya), the region now recognized as the lower Amazon was geographically separated from the upper Amazon by the Purus Arch. This geographical barrier created distinct eastern and western (Pebas) faunas [4, 58, 55] (Fig. 4a). The rise of the Vaupes Arch, approximately 10 mya (Fig. 4b), led to a substantial influx of sediments from the sub-Andean foreland and eventually caused the rupture of the Purus Arch [59].

Fig 4. Graphical summary of overall changes in distribution of Hypoclinemus lineages over the time based on BBM analysis of RASP 4 proposed by our study. Red Arrows show the direction of colonization. Colors of ellipses correspond to the same ecoregions shown in Fig. 3. (a) Early/middle Miocene (Pre ~10 mya); (b) Late Miocene/Pliocene (~10-5 mya); and (c) Pliocene (5 mya – Present) distribution.

The subsequent formation of the modern transcontinental Amazon during the midPliocene, around 5.11 mya (Fig. 4c), allowed uninterrupted sediment flow from the Andes to the Atlantic [56]. This connection between the western and eastern parts of the Amazon basin facilitated the mixing of lineages in the lower Amazon. This aggregation was favored by the river capture system [60], which enabled the dispersal of species from different origins, including the Orinoco (within lineage 3), the Upper Amazon (from 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 Pleuronectoidei: Molecular and morphological analysis of living and fossil taxa.

Zoologica Scripta. 2019; 48, 540. https://doi.org/10.1111/zsc.12372 in Amazonia: Evidence from the Geological Record. In: Hoorn C and Wesselingh FP, editors. Amazonia: Landscape and Species Evolution; 2009. 11. Hoorn C, Wesselingh FP, Hovikoski J, Guerrero J. The development of the amazonian mega-wetland (Miocene; Brazil, Colombia, Peru, Bolivia). In: C. Hoorn & F. P. Wesselingh, editors. Amazonia, landscape and species evolution: A look into the past; 2010. pp. 123–142. https://doi.org/10.1002/9781444306408.ch8 12. Bamber RN, Henderson PA. Pre-adaptative plasticity in atherinids and the estuarine seat of teleost evolution. Journal of Fish Biology. 1988; 10, 17– 23. 13. Hughes LC, Cardoso YP, Sommer JA, Cifuentes R, Cuello N, Somoza GM, González-Castro M, Malabarba LR, Cussac V, Habit EM, Betancur-R R, Ortí R. Biogeography, habitat transitions and hybridization in a radiation of South American silverside fishes revealed by mitochondrial and genomic RAD data. Mol 14. Bernal R, Bacon CD, Balslev H, Hoorn C, Bourlat SJ, Tuomisto H, Salamanca S, van Manen MT, Romero I, Sepulchre P, Antonelli A. Could coastal plants in 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 western Amazonia be relicts of past marine incursions? Journal of Biogeography. 2019; 46(8), 1749–1759. https://doi.org/10.1111/jbi.13560 western Amazonia and its role in Neotropical biogeography, Botanical Journal of the Linnean Society. 2022; v199, Issue 1, Pages 25–35. 16. Albert JS, Reis R. Historical Biogeography of Neotropical Freshwater Fishes. 1st ed. California: University of California Press; 2011. 17. Machado C, Galetti JP, Carnaval A. Bayesian analyses detect a history of both vicariance and geodispersal in Neotropical freshwater fishes. Journal of Biogeography. 2018; 45(6), 1313-1325. https://doi.org/10.1111/jbi.13207 18. Gales SM, Ready JS, Sabaj MH, Bernt MJ, Silva DJF, Oliveira C, Oliveira G, Sales JBL. Molecular diversity and historical phylogeography of the widespread genus Mastiglanis (Siluriformes: Heptapteridae) based on palaeogeographic events in

Samples sequenced in the present study are indicated with an asterisk (*) with their respective markers. The sequence samples from Genbank are shown according to each gene and representative of the family Achiridae.

Species

Code

Location

Ecorregion 16S

COI

INPA-40685 2 - L. Amazon/Xingu – PA, Brazil

Amazonas Estuary and Coastal Drainages

OM066810 1 - D. Amazon/Portel - PA, Brazil 1 - D. Amazon/Portel - PA, Brazil 1 - D. Amazon/Portel - PA, Brazil 1 - D. Amazon/Portel - PA, Brazil 1 - D. Amazon/Portel - PA, Brazil 3 - M.Amazon/Alenquer - PA, Brazil 3 - M.Amazon/Alenquer - PA, Brazil 3 - M.Amazon/Alenquer - PA, Brazil 8 - Tocantins/Mocajuba - PA, Brazil 6 –Negro/Takutu, Guiana 6 - Negro/Takutu, Guiana 5 - Essequibo/Rupununi, Guiana 9 - Orinoco, Venezuela 9 - Orinoco/Ventuari, Venezuela 7 - Iquitos - Peru 11 - Apure, Venezuela 2 - L. Amazon/Xingu - PA, Brazil 10 - Essequibo, Guiana 37 - Mishahua River

Amazonas Estuary and Coastal Drainages Amazonas Estuary and Coastal Drainages Amazonas Estuary and Coastal Drainages Amazonas Estuary and Coastal Drainages Amazonas Estuary and Coastal Drainages Amazonas Guiana Shield Amazonas Guiana Shield Amazonas Guiana Shield Amazonas Estuary and Coastal Drainages Amazonas Guiana Shield Amazonas Guiana Shield Essequibo Orinoco Guiana Shield Orinoco Guiana Shield Amazonas Lowlands Orinoco/Llanhos Amazonas Estuary and Coastal Drainages Essequibo Ucayali–Urubamba Piedmont

OM066830 OM066831 OM066832 OM066837 OM066838 OM066834 OM066835 OM066836 OM066833 OM066827 OM066828 OM066839 OM066811 OM066812 OM066817 OM066840 OM066843 OM066844 ON617866.1

Afin6017 Adum Adum5543 HmP1 HmP2 HmP3 HmP4 Hmen3207 Hmen5548 Hmen3209 Aachi01 Aachi02 Aachi2301 Aachi3053 Aachi4509 Adecl01 Adec5554 Adec5562

Amazonas Lowlands Amazonas Estuary and Coastal Drainages Amazonas Estuary and Coastal Drainages Amazonas Lowlands Amazonas Lowlands Amazonas Lowlands Amazonas Lowlands Amazonas Lowlands Amazonas Lowlands Amazonas Lowlands Amazonas Estuary and Coastal Drainages Amazonas Estuary and Coastal Drainages Amazonas Estuary and Coastal Drainages Northeastern Caatinga and Coastal D.

Northeastern Brazil Southeastern Brazil Amazonia Eastern Brazil KP213863.1 KP213862.1 ON617839.1 ON617840.1 ON617840.1 KP213865.1 ON617852.1 ON617854.1 KT310050.1 MH883040.1 MH898947.1 ON562462.1 ON562463.1 ON562463.1 ON562475.1 ON562477.1 Achirus lineatus Achirus lineatus Achirus lineatus Achirus lineatus Achirus lineatus Achirus lineatus Achirus mucuri Achirus mucuri Gymnachirus texae Gymnachirus texae Gymnachirus texae Gymnachirus nudus Gymnachirus nudus Gymnachirus nudus Gymnachirus nudus Gymnachirus nudus Gymnachirus melas Catathyridium jenynsii Catathyridium jenynsii

Aline_SP2 Aline_SP3 Aline18 Aline2300 Aline2285 Aline3917 Amucu5294 Amucu6514 Gtex01 Gtex_US1 Gtex04 Gnud_SP1 Gnud_SP2 Gnud_BR1 Gnud_AL1 Gnud6531 Gme5123 Catje01 CatjePR1 13 - Cananéia, SP, Brazil 13 - Cananéia, SP, Brazil 30 - Delta do Parnaíba 25 - Barra Grande 17 - Itaipú Lagoon - RJ 22 - Salsa River/Canavieiras 23 - Cachoeira River 38 - USA, Texas 13 - Cananéia - SP, Brazil 13 - Cananéia - SP, Brazil Brazil Alagoas, Brazil 27 - Todos os Santos Bay 38 - North America 15 - Paraná, Brazil Paraná, Brazil Northeastern Brazil Northeastern Brazil Southeastern Brazil Northeastern Mata Atlantica Northeastern Mata Atlantica Northern Gulf of Mexico Eastern Brazil Northern Gulf of Mexico Iguassu

Cjen5998 Cjen5999 Cga6511 Tpaul01 TpaulBR1 Tpaul3064 Tpaul3073 Tpaul4502 Tpaul4881 Tpaul6221 Tmacu01 Tmacu03 Tmacu_US1 Tmacu_US3 Tmacu10 Tmacu11 Tmicr_SP1 Tmicr_SP2 Tmicr_BR1 18 - Taquarassu Mato grosso 18 - Taquarassu Mato grosso 21 - Jequitinhonha River 13 - Cananéia - SP, Brazil Brazil 28 - Manaíra Beach 29 - Paraíba River 30 - Delta do Parnaíba 20 - Mucuri River 16 - Beach Barra do Sul - SC 39 - Lousiana, EUA 39 - Lousiana, EUA USA USA 13 - Cananéia - SP, Brazil 13 - Cananéia - SP, Brazil Brazil Northeastern Brazil Northeastern Caatinga and Coastal D.

Northeastern Brazil Northeastern Mata Atlantica Southeastern Brazil Sabine Galveston Sabine Galveston Southeastern Brazil Southeastern Brazil

Tmicr_BR2 Tmicr3450 Tmicr3799 Tinsc2237 Tinsc5578 Sleuco Bpodas Brobinsi Scris Pbea

Eastern Brazil Northeastern Mata Atlantica Southeastern Brazil Eastern Brazil East China Sea Azores Canaries Madeira Virginian Southern China Virginian

S1 Fig. Bayesian inference (Bi) trees of cox1 (A) and rrLn+RHO (B) of H. mentalis cryptic lineages recovered in the present study. Only support values above 0.95 for Bi are shown and for maximum likelihood, only supports above 75% are indicated. Grey filled circles indicate a posteriori probabilities n 0.8 (80%). Colors represent ecoregion associations inferred in RASP 4.