taxonID	type	description	language	source
BA767A6D320EFFC0FEFFFD9F7DF3579A.taxon	discussion	Divergence of the MRCA of Floridichthys – J. pulchra 42.3 – 21.6 Mya (Fig. 3, node 8; Table 2) is consistent with our hypothesis that Cyprinodontidae originated in the northwestern Gulf (above). This divergence estimate spans a dramatic cooling event at the Eocene – Oligocene boundary 33.9 Mya, which initiated the Oligocene icehouse climate that lasted until ~ 26 Mya (Zachos et al. 2001, Miller et al. 2020 b). Sustained icehouse conditions possibly facilitated southward invasion of Nearctic Cyprinodontidae (Fig. 7), consistent with predictions of the intothe-tropics paradigm (Vrba 1995, Eronen et al. 2012, Meseguer and Condamine 2020) and with our ancestral-areas analysis that included the potential for jump dispersal between realms (Fig. 5). The point estimate for this divergence (31.8 Mya) is aligned with Oligocene isotope zone 1 b (abbreviated Oi 1 b), dated 31.8 Mya (Boulila et al. 2011). Oxygen isotope zones are periods of increased global glaciation (at which times, sea level is lowered), indicated by maximum relative abundance of δ 18 O (Boulila et al. 2011). Concurrent timing of sea-level fall with divergence of the Floridichthys – J. a pulchra lineage agrees with the hypothesis that sea-level fall isolates coastal fishes among disjunct estuaries (Dolby et al. 2016, 2018). Hence, we hypothesize that climatic cooling and periodic sea-level rise during Oligocene oscillations (Boulila et al. 2011) allowed southward expansion of ancestral cyprinodontids, with one or more periods of sea-level fall (potentially Oi 1 b) subsequently isolating populations that had settled on the Yucatán platform. Because living Floridichthys have a disjunct distribution across the Gulf of México between Yucatán and Florida, the MRCA of Floridichthys – Jordanella could be from Florida, Yucatán, or points between. Presence of the sister taxon J. pulchra in Yucatán favours Yucatán as the ancestral area for Floridichthys and is compatible with the hypothesis (above) that Cyprinodontidae originated in the western Gulf of México (closer to Yucatán than to Florida). Furthermore, the Caribbean Loop Current could have facilitated later oceanic dispersal of Floridichthys from Yucatán to Florida (as seen in several cyprinodontiform taxa; MassipVeloso et al. 2024), whereas the current would have hampered immigration in the reverse direction. Hence, we suggest that the Oligocene ancestor of Floridichthys – J. pulchra dispersed along the western Gulf Coast from the Río Grande delta to the Bay of Campeche, where presence of coastal and shallow-marine areas (( Villagómez et al. 2022) could have provided suitable habitats (Fig. 7). However, Floridichthys – J. pulchra has uncertain phylogenetic placement (compare Fig. 3 with Piller et al. 2022), hence further study is needed. Our estimate for vicariance of Cualac tessellatus Miller, 1956 in the Río Pánuco 29.9 – 14.6 Mya (Fig. 3, node 9) provides further support to the hypothesis that ancestral cyprinodontids inhabited the western Gulf Coast in the Oligocene. Presence of a nascent Río Pánuco (Beltrán-Triviño et al. 2021) suggests that ancestral Cualac tessellatus ranged inland along this route (Fig. 8). Late Oligocene – Early Miocene uplift of the Chicontepec Basin (Roure et al. 2009) could have stranded this population inland (Table 2). As already mentioned, our phylogeny indicates that Jordanella is polyphyletic (Fig. 3). Separation of J. floridae from Megupsilon – Cyprinodon 23.3 – 10.8 Mya (Fig. 3, node 10; Table 2) suggests that an ancestral cyprinodontid dispersed eastwards from the northwestern Gulf of México during the Early Miocene, when the nascent Florida Peninsula was an island (i. e. Ocala High). Our 16.8 Mya point estimate for time of divergence closely follows the onset of the MMCO, a period of high temperatures, reduced global glaciation, and elevated sea levels spanning 17.0 – 13.8 Mya (Miller et al. 2020 a). This timing suggests that an ancestral cyprinodontid was distributed along the northeastern Gulf Coast, ranging between the Ocala High and the mainland while the Gulf Trough, which separated these lands, was narrowed (Fig. 8). Sea-level rise during the MMCO greatly broadened and deepened the Gulf Trough, shrinking the Ocala High to a small island (Popenoe 1990). We hypothesize that J. floridae descends from a peripheral cyprinodontid population persisting around this island. Notably, the genus Jordanella has never been resolved as monophyletic with molecular data (Miller et al. 2005, Echelle and Echelle 2020). This, along with the biogeographical evidence provided here, could justify resurrection of the name Garmanella pulchra for the Yucatán species. In this case, morphological similarities between species presently recognized as Jordanella might reflect ancestral traits that persisted in ancient peripheral-isolate taxa sequestered in ancestral cyprinodontid niches on either side of the Gulf of México. The taxonomy of these species needs further study. Late Miocene divergence of Megupsilon (14.9 – 7.4 Mya; Fig. 3, node 11) might represent Late Miocene emergence of ancestral Río San Fernando (Río Bravo in the work of Snedden and Galloway 2019) as a corridor for inland invasion (Table 2; Fig. 8). This species (extinct in the wild) inhabited a spring-fed habitat (Miller et al. 2005) and might have speciated as a spring endemic. It is possible that episodes of aridity, tectonism, or volcanism isolated the spring system, but this needs further study. The rate of Cyprinodon speciation increased upon separation from Megupsilon (node 11; Figs 3, 4). Several factors are likely to have contributed to this trend. Initially, inland invasions into five Late Miocene rivers subdivided Cyprinodon into as many upland lineages, four of which dispersed far across the desert region (Hoagstrom and Osborne 2021). A relative of this western inland radiation also dispersed to Yucatán, founding an endemic lineage there (Cyprinodon artifrons species group; Figs 2, 5). Once inland invasions were underway, barrier displacement via drainage rearrangements, climate fluctuations, tectonism, and volcanism caused widespread allopatric diversification. Meanwhile, a maritime lineage of Cyprinodon remained along the Gulf Coast (Echelle et al. 2005, 2006). This geography is unclear in our reconstruction of ancestral habitats (Fig. 2), because the same widespread ancestor produced sequential upland invasions from the coast, as already described. Furthermore, during the Pleistocene, maritime Cyprinodon made new invasions into the desert region (Lozano-Vilano and Contreras-Balderas 1999, Hoagstrom and Osborne 2021) and immigrated to Caribbean islands and South America (Haney et al. 2007, 2009). The Yucatán lineage produced a species flock (Strecker 2006; Fig. 3). Beyond this, our analysis might underestimate the Cyprinodon speciation rate, because a species flock from San Salvador, Bahamas (Martin and Wainwright 2013) and the subspecies Cyprinodon variegatus hubbsi Carr, 1936 (a potential distinct species; Brix and Grosell 2013, Jung et al. 2019) are absent from our phylogeny, which also leaves out seven Cyprinodon species for which we had no genetic data. Finally, six of the recognized species included in the analysis are likely to be polyphyletic (Echelle and Echelle 2020) but are represented as one taxon here.	en	Hernández-Ávila, Sonia Gabriela, Hoagstrom, Christopher W., Matamoros, Wilfredo A. (2024): Historical biogeography of North American killifishes (Cyprinodontiformes) recapitulates geographical history in the Gulf of México watershed. Zoological Journal of the Linnean Society 202 (2), DOI: 10.1093/zoolinnean/zlae105, URL: https://doi.org/10.1093/zoolinnean/zlae105
BA767A6D3207FFC6FBD9FEBB78A85484.taxon	discussion	Phylogenetic analyses The MRBAYES and BEAST produced similar topologies. Bayesian inference recovered almost all genera with posterior probabilities> 0.95 (Fig. 3). Most genera were monophyletic except the following: (i) Xenotoca variata (Bean, 1887) was sister to Ameca splendens Miller & Fitzsimons, 1971 and placed well outside genus Xenotoca Hubbs & Turner, 1939; (ii) Jordanella pulchra (Hubbs, 1936) was sister to Floridichthys Hubbs, 1926, while Jordanella floridae Goode & Bean, 1879 was sister to Megupsilon Miller and Walters, 1972 – Cyprinodon Lacépède, 1803; and (iii) Lucania Girard, 1860 was sister to Wileyichthys Ghedotti & Davis, 2013 (Fundulus lima Vaillant, 1894 – Fundulus parvipinnis Girard, 1856). In our phylogeny, the families Cubanichthyidae, Cyprinodontidae, Fundulidae, Goodeidae, and Profundulidae were recovered as composing a monophyletic group, in agreement with recent studies (Ghezelayagh et al. 2022, Piller et al. 2022). The branching pattern among families in this clade was in agreement with genomic results provided by Ghezelayagh et al. (2022) and Piller et al. (2022). However, in contrast to their and our findings, which place Cubanichthyidae sister to Goodeidae – Profundulidae, Piller et al. (2022) found that Cubanichthyidae was paraphyletic, with Cubanichthys cubensis (Eigenmann, 1903) sister to Cyprinodontidae, and with Cubanichthys pengelleyi (Fowler, 1930) sister to C. cubensis – Cyprinodontidae. Confidence intervals for divergence estimates of family originations in our tree (nodes 1 – 4; Table 2) overlapped broadly with confidence intervals for the same nodes in the study by Ghezelayagh et al. (2022). The greatest difference was for node 4 (separation of Goodeidae and Profundulidae), wherein our confidence intervals overlapped with theirs, but their range also extended into more recent time, with their point estimate of ~ 22 Mya being substantially younger than ours of ~ 30.0 Mya. Diversification through time The lineage-through-time plots (Fig. 4) indicate that Cyprinodontidae is the only family that experienced a diversification-rate shift, with acceleration initiating ~ 10.9 Mya and continuing to the present (λ = 0.23 mean; range = 0.14 – 0.38). The initial rate increase occurred when Megupsilon diverged from Cyprinodon. No shifts were detected in Fundulidae (λ =. 12 mean; range =. 07 –. 21), Goodeidae (λ = mean. 13; range =. 09 –. 29), or Profundulidae (λ = mean. 12; range =. 07 –. 24). Reconstruction of ancestral areas The best-fitting model for reconstruction of ancestral areas was the BAYAREALIKE + J (Supporting Information, Table S 3), consistent with the coarse level of analysis, which recognized only the Nearctic and Neotropical realms. Most divergence events occurred within one realm or the other, as required by the BAYAREALIKE model (Garcia-R and Matzke 2021). Inclusion of the + J portion of the model implies that immigrations between realms occurred via long-distance dispersal (Matzke 2014). The ancestral distribution for this clade of North American killifishes is optimized as Nearctic – Neotropical, suggesting a widespread most recent common ancestor (MRCA) (Fig. 5). This clade first split into a northern fork with primarily Nearctic affinities (Fundulidae – Cyprinodontidae) and a southern fork with Nearctic – Neotropical affiliation (Cubanichthyidae, Profundulidae, and Goodeidae). Appearance of fundulids in the Neotropics is relatively recent (Late Neogene – Quaternary), limited to the Fundulus grandis Baird and Girard, 1853 species group. Nearctic Cyprinodontidae made several Neotropical invasions (Floridichthys, Yucatán Cyprinodon, Caribbean Cyprinodon, and Cyprinodon variegatus Lacépède, 1803 species group). The MRCAs of Cubanichthyidae and Profundulidae were Neotropical, and both families are restricted to that realm. Goodeidae is most likely to have had a Nearctic origin, with representatives of Girardinichthys, Allodontichthys Hubbs & Turner, 1939, Xenotaenia Turner, 1946, and Ilyodon reaching Neotropical drainages in the Late Neogene – Quaternary. Reconstruction of ancestral habitats Ancestral-habitats reconstruction (Fig. 2 B) indicates that the MRCA of North American killifishes was coastal. Inland invasions within Fundulidae include Lucania interioris Hubbs & Miller, 1965, the Fundulus sciadicus Cope, 1865 – Plancterus Garman, 1895 group, and later-branching lineages within subgenera Fundulus Lacépède, 1803 and Zygonectes Agassiz, 1854. Fundulidae are the only family with species broadly distributed between coastal and upland habitats. In Cyprinodontidae, Floridichthys and Jordanella Goode & Bean, 1879 retained coastal affinities, while the maritime branch of Cyprinodon gave rise to upland and coastal lineages. Other branches within Cyprinodon diverged to become upland lineages. Cubanichthyidae remained coastal, whereas Profundulidae and Goodeidae are likely to have originated and diversified entirely in uplands. Overall, 58.6 % of species had coastal distributions, 33.5 % had upland distributions, and 7.9 % had distributions spanning coastal and upland habitats (Fig. 2 B). Biogeographical synthesis Early branching of major lineages Evidence suggests that the MRCA of North American killifishes arose in the Early Eocene Gulf of México (Figs 3, 4). Accordingly, the MRCA is likely to have had high salinity tolerance (Ghedotti and Davis 2013). This timing corresponds precisely with a period of increased taxonomic and morphological diversification (Ghezelayagh et al. 2022). The distribution of the MRCA potentially spanned the present-day Nearctic and Neotropical realms (Fig. 5). However, it was likely to be tropical because at this time; the tropics extended to 27 ° N, 3 ° north of the present-day boundary of the tropics (Zhang et al. 2019). The primary phylogenetic split in North American killifishes separated the northern sister families Cyprinodontidae – Fundulidae from a southern subclade of Cubanichthyidae – Goodeidae – Profundulidae 58.6 – 45.8 Mya (Fig. 3, node 1; Table 2). The relative distributions of these families (Fig. 1), our ancestral-areas reconstruction (Fig. 5), and our ancestral-habitats reconstruction (Fig. 2) combine to suggest that separation occurred as a north – south division within the Gulf of México. The Tropic of Cancer was gradually retracting southwards through this period (Zhang et al. 2019), which could imply that distinct subtropical ecosystems developed in the northern Gulf. We hypothesize that the northern MRCA of Fundulidae – Cyprinodontidae adapted to inhabit subtropical estuaries, including those of the Mississippi, Brazos, and Río Grande (Fig. 6). Concurrently, the southern MRCA of Cubanichthyidae – Goodeidae – Profundulidae potentially occupied a tropical estuary in the southwestern Gulf (Bejuco, Chicontepec, and Nautla) (Fig. 6). Geological evidence suggests the Gulf of México became isolated from the world ocean 55.8 – 55.0 Mya, accompanied by sea-level recession of 900 – 1300 m (Cossey et al. 2016, 2021). Because sea-level fall isolates estuaries (Dolby et al. 2020), this extreme drawdown could have caused strong separation of northern and southern clades. We hypothesize divergence of Cyprinodontidae from Fundulidae, 54.8 – 37.6 Mya (Fig. 3, node 2; Table 2) was via east – west vicariance along the northern Gulf Coast. Drainage reconfiguration occurred in the northern Gulf of México 48.2 Mya (Early – Middle Eocene transition), when reduced sediment supply caused alluvial deltas to retract and separate (Snedden and Galloway 2019). Shifting delta locations (Snedden and Galloway 2019) might have initially dispersed and then subdivided killifish populations. To the northeast, the MRCA of Fundulidae was potentially isolated in the Middle Eocene Brazos River delta. From there, it is likely to have ranged east into the adjacent Mississippi Embayment, which flooded at this time (Fig. 6). The embayment refilled gradually through the Middle Eocene (Snedden and Galloway 2019), suggesting that it contained extensive estuarine habitats that could have been the cradle for Fundulidae. To the southwest, the MRCA of Cyprinodontidae putatively originated in a separate delta region associated with the Río Grande and Río Bravo deltas (Fig. 6) that were comparatively sediment rich and wave swept (Snedden and Galloway 2019). Such habitat is seemingly well suited for the ancestor of this group, which is hypothesized to have resembled Cyprinodon variegatus (Echelle & Echelle, 2020), a species well adapted for open coasts with fluctuating temperatures and salinities (Simpson and Gunter 1956, Martin 1968, Nordlie 2006). This period also included ~ 40 m sea-level fall between 48.0 and 47.5 Mya (Miller et al. 2020 a, 2024), which is likely to have isolated each delta region (Dolby et al. 2020). Subsequent sea-level recessions 46.9 and 45.6 Mya (Miller et al. 2020 a) might have amplified barriers to dispersal. Divergence of Cubanichthyidae from Goodeidae-Profundulidae, 54.8 – 35.8 Mya (Fig. 3, node 3; Table 2), could also be linked to sea-level fluctuations. We hypothesize that maximum sea levels associated with ice-free conditions in the Early Eocene, including a peak at 48.0 Mya (Miller et al. 2024), allowed an ancestral taxon to range across the Yucatán platform, into the Caribbean Sea. If so, then subsequent sea-level falls in the Middle Eocene, along with failure to return to ice-free conditions until the end of the Eocene (Miller et al. 2024), could have made the Yucatán platform a more formidable barrier, isolating the ancestor of Cubanichthyidae in the Caribbean (Fig. 6). However, as already noted, phylogenetic placement of Cubanichthyidae is inconsistent among recent studies (e. g. Ghezelayagh et al. 2022, Piller et al. 2022), indicating that this family requires additional study. Divergence of Goodeidae from Profundulidae, 39.2 – 21.4 Mya (Fig. 3, node 4), presumably reflects upland isolation of lineages in separate river drainages (Table 2; Fig. 3). We hypothesize that Goodeidae descend from a population that ranged inland into a Madrean River (Fig. 6), which joined the Bejuco delta system in the Early Eocene, after uplift of the Tamaulipas Arch diverted it southwards (Snedden and Galloway 2019). As the Madrean River drainage coalesced among emerging tectonic basins and colliding terranes of southwestern North America (Lawton et al. 2009), not necessarily incorporating any pre-existing rivers, this drainage might have been depauperate in freshwater fishes or lacked them entirely. If so, the Madrean River system might have provided open niches for coastal killifishes transitioning into freshwater habitats. In the Middle – Late Eocene, the outlet of the Madrean River shifted between the Tamaulipas Diversion and Río Bravo delta, but at the end of the Eocene (33.9 Mya) the Río Grande captured the inland drainage (Galloway et al. 2011) (Fig. 7). A resulting influx of freshwater fishes might have constrained the distributions of goodeids to more marginal habitats isolated by distance upstream, barriers, or habitat conditions (i. e. salinity), presumably isolating upland populations from coastal relatives. For Profundulidae, the timing of divergence corresponds with uplift of the Chiapas Massif (40 – 25 Mya) (Villagómez and Pindell 2020). Drainage from the Massif to the Bay of Campeche was localized at this time, not reaching the Gulf of México basin (Beltrán-Triviño et al. 2021, Villagómez et al. 2022), isolating inland populations (Fig. 7). Given that cyprinodontids evidently inhabited the southwestern Gulf by ~ 31.8 Mya (see below; Table 2; Fig. 7), we cannot rule out the possibility that a southward cyprinodontid invasion contributed to the disappearance of coastal Goodeidae – Profundulidae.	en	Hernández-Ávila, Sonia Gabriela, Hoagstrom, Christopher W., Matamoros, Wilfredo A. (2024): Historical biogeography of North American killifishes (Cyprinodontiformes) recapitulates geographical history in the Gulf of México watershed. Zoological Journal of the Linnean Society 202 (2), DOI: 10.1093/zoolinnean/zlae105, URL: https://doi.org/10.1093/zoolinnean/zlae105
BA767A6D320CFFDFFE18FD1C7CC455C0.taxon	discussion	As described above (Early branching of major lineages), Middle Eocene fundulids theoretically inhabited the Brazos River delta and flooded Mississippi Embayment (Fig. 6). Late Eocene – Oligocene divergence of Leptolucania Myers, 1924 39.4 – 22.5 Mya (Fig. 3, node 12) suggests eastward immigration of an ancestral fundulid across the Suwannee Channel to the Ocala High (Table 2; Fig. 7), which developed peritidal landforms by this time (Avon Park and Ocala formations; Randazzo and Jones 1997, Maliva et al. 2011). During the Oligocene, any of several sea-level fluctuations could have facilitated lineage separation if a sea-level fall (detected as an oxygen isotope event) facilitated dispersal across the Suwannee Channel to the Ocala High, with subsequent sea-level rise causing vicariance of the Ocala population owing to widening and deepening of the ocean gap (as proposed above for J. floridae). The Oi 2 event ~ 30.0 Mya (Boulila et al. 2011) appears best aligned in time with our divergence estimate for Leptolucania. Although modern Leptolucania is a freshwater genus, the ancestor is likely to have had high salinity tolerance (Ghedotti and Davis 2013), consistent with this hypothesis. Our phylogeny places Lucania within Fundulus, on an earlydiverging branch that includes Fundulus subgenus Wileyichthys (Fig. 3). This is contrary to the work of Ghedotti and Davis (2017), who recovered Lucania outside Fundulus, but in partial agreement with the fish tree of life (Rabosky et al. 2018), which places Lucania inside Fundulus, albeit on a separate branch from Wileyichthys (this tree did not include Leptolucania). Given that Wileyichthys includes only the Pacific Coast taxa Fundulus lima and Fundulus parvipinnis and given that the Coahuilan endemic Lucania interioris is the earliest diverging lineage of Lucania, we propose (following Fig. 3) that the MRCA of Lucania – Wileyichthys originated from range expansion of fundulids westwards from the Brazos delta to the Río Grande-Río Bravo delta 31.3 – 19.1 Mya (Fig. 3, node 13; Fig. 7). During this time, there were sea-level falls of 30 – 40 m below modern sea level approximately every 1.2 Mya (Miller et al. 2020 b). Following Dolby et al. (2016, 2018), we postulate that sea-level fluctuations promoted cycles of dispersal and vicariance. Any one or a combination of recessions might account for vicariance of fundulids between the Brazos and Río Grande-Río Bravo deltas, but the Oi 2 c event dated ~ 25.1 Mya (Boulila et al. 2011) best aligns in time with divergence of Lucania – Wileyichthys (Table 2). In the Oligocene, the common ancestor of Lucania – Wileyichthys presumably immigrated into the ancestral Río Grande drainage. This massive river system extended far across southern North America to what would become the Basin and Range geomorphic province (Snedden and Galloway 2019) and appears to have been the only Cenozoic route to the southern Basin and Range from the western Gulf of México (Fig. 7). This geographical scenario comes with relatively strict time constraints. Divergence of Wileyichthys from Lucania must have occurred when Basin and Range drainages became disconnected from the Río Grande, dated ~ 23.0 Mya (Snedden and Galloway 2019). This timing is compatible with our 30.7 – 7.5 Mya estimate for divergence of Wileyichthys (Fig. 3, node 14; Table 2). Fossil Fundulus in the Great Basin (Smith et al. 2002) also support this hypothesis. Finally, distribution of Wileyichthys on the Pacific Coast of southern California and south along the Baja Peninsula agrees with the Río Grande as the dispersal corridor, because the headwaters of the Oligocene Río Grande extended to the Continental Divide in what would become southern California (Karlstrom et al. 2020). Tectonism in this region potentially resulted in stream transfers that introduced Wileyichthys to the Pacific slope. Likewise, Empetrichthyinae must have been living in Basin and Range drainages along the continental divide, after break-up of the Madrean River (above). Fossil evidence from Empetrichthyinae (Uyeno and Miller 1962) indicates that this group was also transferred to the Pacific slope, although no representatives survived to the present time. The major split within the Fundulus crown group separated subgenus Fundulus as a clade 24.2 – 17.1 Mya (Fig. 3, node 15; Table 2). An eastward shift of the Oligocene Brazos River at the Oligocene – Miocene transition (23 Mya) from its existing delta to form a new Red River delta relatively close to the Mississippi River delta (present Sabine River delta; Galloway et al. 2011) might have provided a catalyst for this event by providing an uninhabited delta open to immigrants. During this time, there were sea-level falls of 30 – 40 m below modern sea level approximately every 1.2 Mya (Miller et al. 2020 b). Any one or a combination of these fluctuations might account for a dispersal – vicariance sequence in fundulids between the Mississippi River and Red River deltas (Dolby et al. 2016, 2018). However, the Mi 1 a isotope zone (sea-level recession) dated ~ 20.8 Mya (Boulila et al. 2011) best aligns in time with our divergence estimate for subgenus Fundulus. We hypothesize that the Zygonectes – F. sciadicus – Plancterus clade descends from a lineage that originated in the Mississippi River delta, because Plancterus (node 18) has a Mississippian distribution and is the westernmost lineage on this branch (Fig. 8). By default, this implies that subgenus Fundulus descends from fundulids that immigrated into the Red River delta, then diverged into a separate lineage during sea-level recession. Later abandonment of the Red River delta (Fig. 8), which merged with the Mississippi River delta 15 Mya (Snedden and Galloway 2019), mixed Red River and Mississippi River fishes (Hoagstrom and Echelle 2022). This would have brought subgenus Fundulus into sympatry with Zygonectes. Thereafter, sea-level rise 19 – 17 Mya leading up to the MMCO 17.0 – 13.8 Mya, during which high sea levels were sustained (Miller et al. 2020 a), theoretically aided range expansion of all coastal fundulids (Dolby et al. 2016, 2018).	en	Hernández-Ávila, Sonia Gabriela, Hoagstrom, Christopher W., Matamoros, Wilfredo A. (2024): Historical biogeography of North American killifishes (Cyprinodontiformes) recapitulates geographical history in the Gulf of México watershed. Zoological Journal of the Linnean Society 202 (2), DOI: 10.1093/zoolinnean/zlae105, URL: https://doi.org/10.1093/zoolinnean/zlae105
BA767A6D3213FFDFFEFFFEC67D7C56D1.taxon	discussion	First branching within subgenus Fundulus might reflect both ecological and geographical speciation. Fundulus majalis (Walbaum, 1792) – Fundulus persimilis Miller, 1955 – Fundulus similis (Baird & Girard, 1853) inhabit unvegetated coastal habitats where they dive into soft sediments for cover rather than retreating to vegetation like typical Fundulus (Martin & Finucane, 1968; Harvey, 1998; Miller et al. 2005). They are adapted for continuous swimming in the surf zone (Yetsko and Sancho 2015) and spawn in unvegetated habitats (Greeley et al. 1986). Fundulus majalis segregates from congeners in unvegetated, high-salinity waters (Weisberg 1986, Wagner and Austin 1999). We propose that this ecologically divergent lineage descends from an ancestor adapted for open, wave-swept coasts. Divergence 21.0 – 15.2 Mya (Fig. 3, node 16) was concurrent with sea-level rise that culminated in the MMCO (Table 2) and could have facilitated range expansion (Dolby et al. 2016, 2018). Given that coastlines adjacent to Early Miocene deltas were wave swept and sand dominated (Snedden and Galloway 2019), this could account for ecological specialization in this species group. In the earliest Middle Miocene (15.6 Mya), a new Guadalupe River delta emerged to the west (Fig. 8). Wave-dominated conditions there (Snedden and Galloway 2019) suggest that the MRCA of the F. similis species group evolved there.	en	Hernández-Ávila, Sonia Gabriela, Hoagstrom, Christopher W., Matamoros, Wilfredo A. (2024): Historical biogeography of North American killifishes (Cyprinodontiformes) recapitulates geographical history in the Gulf of México watershed. Zoological Journal of the Linnean Society 202 (2), DOI: 10.1093/zoolinnean/zlae105, URL: https://doi.org/10.1093/zoolinnean/zlae105
BA767A6D320AFFC2FBA5FA8F7CB75718.taxon	discussion	Through the Late Eocene and Oligocene, reorganization and drainage expansion within the Madrean River (Galloway et al. 2011, Snedden and Galloway 2019) could have allowed goodeids to disperse along the front of the Sierra Madre Occidental to the southern Great Basin (Fig. 7). Divergence of Great Basin Empetrichthyinae from Goodeinae of the Trans-Mexican Volcanic Belt and Sierra Madre 23.3 – 12.9 Mya (Fig. 3, node 6; Table 2) agrees with break-up of this drainage at the end of the Oligocene (Snedden and Galloway 2019). The great distance between northern (Empetrichthyinae) and southern (Goodeinae) goodeids is a biogeographical anomaly (Webb 2020). Parenti (1981) proposed that regional desiccation eliminated intervening populations (Grant and Riddle 1995, Webb et al. 2004, Miller et al. 2005, Webb 2020), but living Empetrichthyinae occupy the driest region of North America (i. e. aridity is associated with survival). Fragmentation of the Madrean River might have been an additional cause of extirpations (Fagan et al. 2002), and climatic cooling might have reduced habitat suitability at higher elevations and latitudes, perhaps explaining why Empetrichthyinae are associated with warm-water springs. Loss of goodeids from the Gulf of México drainage could reflect Late Miocene integration of the ancestral Río Nazas with the Río Grande, which facilitated an influx of fishes (Hoagstrom and Osborne 2021). Characodon Günther, 1866 persisted only in remote basins († Characodon garmani, Mayran-Parras Basin; Characodon audax Smith & Miller, 1986 – Characodon lateralis Günther, 1866, Río Tunal) (Beltrán-López et al. 2021), potentially protected from invading fishes. It is also possible that remnant goodeid populations were present in the uninhabited region before European settlement, but disappeared owing to habitat destruction, dewatering, or invasive species associated with early European contact. Because this scenario agrees with reconstructed hydrography, it is unnecessary to invoke the popular hypothesis (Miller and Smith 1986, Minckley et al. 1986, Grant and Riddle 1995, Miller et al. 2005, Webb 2020) that northward drift of the Pacific Plate created the gap between Empetrichthyinae and Goodeidae. Furthermore, thehypothesisthatGoodeidaeusedawesternroute from the Great Basin into central México (Pérez-Rodríguez et al. 2015) is unnecessary because goodeids dispersing inland from the Gulf of México via the Madrean River would have reached central México en route to the Great Basin (Fig. 7), not from the Great Basin. Separation of Empetrichthyinae from Goodeinae is attributable to vicariance during fragmentation of the Madrean River drainage, with Empetrichthyinae representing a relict lineage that has persisted in the Mojave Desert after reaching the region when the Madrean River extended from the Gulf Coast to the Basin and Range geomorphic province. Goodeinae represent lineages descending from populations further downstream that persisted within the Sierra Madre Occidental and Trans-Mexican Volcanic Belt after separation from the Gulf of México drainage. Within Goodeinae, divergence of CharacodontiniIllyodontini from crown Goodeinae 16.8 – 10.3 Mya (Fig. 3, node 7; Table 2) appears to have been an east – west divergence, possibly between ancestral ríos Nazas and Aguanaval (Fig. 8). Reactivation of the San Marcos Fault 14.0 – 5.0 Mya and contemporary volcanism on the borders of the Mesa Central (Aranda-Gómez et al. 2005, 2007, Chávez-Cabello et al. 2007, Nieto-Samaniego et al. 2007) are likely to have contributed to drainage fragmentation. Positioning of crown Goodeinae on the Trans-Mexican Volcanic Belt throughout its tectonic evolution, which initiated ~ 20 Mya and lasted 17 Myr (Ferrari et al. 2012), ensured that goodeids experienced frequent drainage reorganizations and shifting hydrographic barriers (Barbour 1973, Domínguez‐Domínguez et al. 2010, Pérez-Rodríguez et al. 2015). Barrier displacement via stream capture was probably the main mode of range expansion and diversification (Webb et al. 2004, Domínguez‐Domínguez et al. 2010, Beltrán-López et al. 2021).	en	Hernández-Ávila, Sonia Gabriela, Hoagstrom, Christopher W., Matamoros, Wilfredo A. (2024): Historical biogeography of North American killifishes (Cyprinodontiformes) recapitulates geographical history in the Gulf of México watershed. Zoological Journal of the Linnean Society 202 (2), DOI: 10.1093/zoolinnean/zlae105, URL: https://doi.org/10.1093/zoolinnean/zlae105
BA767A6D320AFFC6FBB5FE027FF25008.taxon	discussion	The major node within Profundulidae 30.7 – 14.2 Mya (Fig. 3, node 5) separates Profundulus Hubbs, 1924 from Tlaloc Álvarez and Carranza, 1951. Tlaloc occurs across the Chiapas Massif and adjacent highlands within the Río Grijalva drainage (Miller 1955, Cashner and Echelle 2020). The Upper Río Grijalva flows northwest until making an abrupt turn northeast, suggesting capture from a north-flowing river, possibly the ancestral Río Uxpanapa or Tonalá. This capture could have been caused by uplift of the northwestern Chiapas Massif 30 – 25 Mya (Witt et al. 2012) or head-cutting of the lower Río Grijalva during a period of extensive erosion 35 – 25 Mya (Abdullin et al. 2016) and potentially explains the distinctiveness of the Upper Grijalva fish community (Elías et al. 2020). Divergence of Tlaloc from Profundulus, which probably arose in an ancestral Río Coatzacoalcos (Fig. 8), provides an estimate for timing of this capture (Table 2). Notably, Profundulus is now largely restricted to Pacific slope drainages (Domínguez-Cisneros et al. 2023). Presuming that Miocene, upland Profundulus no longer tolerated high-salinity or coastal environments, rising sea levels could explain the rarity of Profundulus on the Gulf slope, as proposed for Herichthys Baird & Girard, 1854 (Pérez-Miranda et al. 2020). During the Middle Miocene Climatic Optimum (MMCO) 17.0 – 13.8 Mya (Miller et al. 2020 b), seas inundated broad areas north and east of the Sierra Madre del Sur and Chiapas Massif (Blakey and Ranney 2018). Our chronogram indicates that diversification within Profundulus occurred thereafter (Fig. 3), potentially from one refugial population.	en	Hernández-Ávila, Sonia Gabriela, Hoagstrom, Christopher W., Matamoros, Wilfredo A. (2024): Historical biogeography of North American killifishes (Cyprinodontiformes) recapitulates geographical history in the Gulf of México watershed. Zoological Journal of the Linnean Society 202 (2), DOI: 10.1093/zoolinnean/zlae105, URL: https://doi.org/10.1093/zoolinnean/zlae105
BA767A6D3213FFDFFF46FBD57F0B53C1.taxon	discussion	– F. sciadicus – Plancterus Our tree uniquely groups F. sciadicus as sister to Plancterus (Fundulus kansae Girard, 1859 – Fundulus zebrinus Jordan & Gilbert, 1883). Divergence of this group from Zygonectes 20.4 – 13.0 Mya (Fig. 3, node 17; Table 2) and a transition from coastal to upland habitat (Fig. 2) suggest that the MRCA of F. sciadicus – Plancterus immigrated up the Mississippi River (Fig. 8), which extended to the nascent Great Plains (Snedden and Galloway 2019). Fundulus sciadicus and Plancterus species are grassland associates (Cross and Moss 1987, Fausch and Bestgen 1997). We hypothesize that the MRCA diverged from eastern relatives via adaptation to grassland habitats on the northern Great Plains. Accordingly, the 16.6 Mya point estimate for divergence of this group corresponds to a time when the Great Plains region was already ~ 68 % open habitat (Edwards et al. 2010, Andermann et al. 2022). Also, the Ogallala Formation, which forms the substrate of the Great Plains, was forming at this time (Chapin 2008, Galloway et al. 2011). The 5.0 – 2.6 Mya fossil † Fundulus detillae Hibbard & Dunkle, 1942, recovered from the Ogallala formation in Kansas (Ghedotti and Davis 2017, Cashner and Echelle 2020), is consistent with the hypothesis that this lineage originated on the northern plains. Altogether, this evidence suggests that the F. sciadicus group might be a case of ecological isolation.	en	Hernández-Ávila, Sonia Gabriela, Hoagstrom, Christopher W., Matamoros, Wilfredo A. (2024): Historical biogeography of North American killifishes (Cyprinodontiformes) recapitulates geographical history in the Gulf of México watershed. Zoological Journal of the Linnean Society 202 (2), DOI: 10.1093/zoolinnean/zlae105, URL: https://doi.org/10.1093/zoolinnean/zlae105
