Rise of modern marine fishes captured in an early Paleocene Lagerstätte Authors: Sanaa El-Sayed, Matt Friedman, Belal S. Salem, Abdullah S. Gohar, Shorouq F. Al-Ashqar, Mohamed Amin, Hossam El-Saka, Hadeel Saad, Robert P. Speijer, Hesham M. Sallam Published: 2026 ABSTRACT The Cretaceous-Paleogene (K-Pg) extinction reshaped Earth’s biodiversity, yet its impact on marine fishes remains debated due to gaps in the Paleocene record. Here, we report a paleotropical assemblage from the early Paleocene (Danian) of Egypt that provides a window into this transition. The Qreiya 3 Lagerstätte [62.2 million years ago (Ma)] reveals an offshore marine ecosystem with at least 21 actinopterygian taxa across nine orders, exceeding the diversity of all other Danian skeletal assemblages combined. Most fishes are percomorphs and include the oldest skeleton-based records for at least six ecologically divergent extant groups. These findings reinforce inferences of fish extinction linked to the K-Pg and the rapid establishment of compositionally modern communities, marked by the first occurrences of new lineages no later than ~4 million years (Myr) after the event. Comparisons across sites indicate that percomorphs appear more common at lower paleolatitudes in the Paleocene, expanding into higher paleolatitudes by the Eocene. INTRODUCTION The Cretaceous-Paleogene (K-Pg) mass extinction resulted in the loss of an estimated 75% of Earth’s species (1) but with varying degrees of severity among different biological groups, environmental settings, and geographical regions (2). On land, recovery from the K-Pg mass extinction includes apparent bursts of taxonomic diversity and morphological disparity in placental mammals (3), neoavian birds (4), and squamates (5), as well as establishment of modern plant communities (6) and localized environmental changes across the boundary (7). In marine settings, complementary ichthyolith (8) and otolith (9) records of actinopterygian fishes reveal turnover and subsequent diversification. Skeletal fossils add more anatomical and taxonomic detail to this picture, pointing to the disappearance of several long-lived fish groups typical of Cretaceous marine assemblages (10) followed by the appearance of diverse, compositionally more modern assemblages no later than the earliest Eocene (11, 12). Percomorph fishes, the teleost analogs of placental mammals or neoavian birds in terms of both taxonomic richness and phenotypic variety, represent the principal drivers of this conspicuous Paleogene faunal shift. Numbering over 17,000 extant species, percomorphs are exceedingly rare in Late Cretaceous marine faunas but abundant from the earliest Eocene onward, with clustered first appearances of dozens of extant families captured by multiple exceptional fossil sites dating between 56 and 49 million years ago (Ma) (13). Although past paleontological (12, 14) and biological (11, 15, 16) work often implicates the K-Pg in the rise of these compositionally modern marine fish faunas, a detailed picture of how this major shift transpired remains elusive. The scarcity of well-preserved, articulated actinopterygian skeletons from the first 10 million years (Myr) following the K-Pg extinction—part of a paleontological hiatus recently termed Patterson’s Gap (13)—represents a major barrier. Meaningful skeletal evidence from the Danian, the critical first stage of the Paleogene encompassing the initial ~4.34 Myr after the extinction (17), is essential but currently restricted to two relatively shallow marine localities: Palenque in Mexico and Limhamn in Sweden (Fig. 1A). However, both assemblages are taxonomically sparse and lack precise age control within the Danian. The Limhamn fishes are tentatively regarded as late Danian in age, based on their assumed collection horizon within the Copenhagen Limestone (18). The reported estimate for the Palenque assemblage, based on Sr isotope stratigraphy dating, shows substantial uncertainty for a marine deposit (reported as 63 ± 1.5 Ma) (19). Paleogeographic context and taxonomic composition of early Paleocene marine fish assemblages. (A) Paleogeographic map of the Danian age (~62 Ma) showing three fossil localities that have yielded articulated skeletons of multiple marine fish taxa. The map is adapted from (74). The circle diameter reflects total actinopterygian taxonomic richness at each site. (B) Taxonomic composition of each assemblage, divided into percomorph (green) and non-percomorph (orange) taxa. Qreiya 3 (Egypt) preserves a compositionally modern assemblage, with high percomorph diversity, whereas Palenque (Mexico) and Limhamn Quarry (Sweden) are less taxonomically rich. Only taxa with secure taxonomic identifications were included in site-level counts. Taxa designated as tentative, uncertain (e.g., preceded by “?”), or assigned only to higher-level groups with ambiguous affinity were excluded from counts in (B) [Palenque (19, 37, 75–78); Limhamn (13, 52); Qreiya 3; table S1]. Paleogeographically, these two Danian sites differ by over 30 degrees of paleolatitude, with Palenque located only a few hundred kilometers away from the Chicxulub impact crater. Despite this geographic contrast, these modest faunas show important similarities. Each contains ten or fewer taxa, of which less than half are percomorphs (Fig. 1B). On both counts, these Danian faunas differ from the much more diverse, percomorph-dominated early Eocene communities revealed by localities concentrated at middle latitudes in Europe and central Asia (20). Danian otolith data reinforce the compositional pattern suggested by the sparse skeletal record, with percomorphs representing at most around one-third of taxa at any particular locality (9) (table S2). Taken at face value, the known Danian record suggests depauperate marine actinopterygian faunas and a protracted ascent of percomorphs to taxonomic dominance in the wake of the K-Pg event. Here, we report an exceptional previously unidentified Danian fossil Lagerstätte that adds to the current picture of postextinction marine fish diversity: Qreiya 3. This locality in the Eastern Desert of Egypt records the negative carbon isotopic excursion diagnostic of the Latest Danian Event (LDE) (figs. S1 and S2). Correlation with the top of magnetochron C27n provides a robust age constraint of 62.2 Ma (21, 22) with uncertainty on the order of ~100 thousand years (kyr), in contrast to larger uncertainties in age for other Danian sites [63 ± 1.5 Ma for Palenque (19) and late Danian for Limhamn (18)]. The LDE horizon at Qreiya 3 represents an outer neritic to upper bathyal setting at a paleodepth of 150 to 250 m (22), providing an environmental contrast to these localities. To date, it has yielded ~500 specimens including numerous intact, well-preserved fish skeletons. With more than 21 distinct actinopterygian taxa represented by the ~180 specimens studied in more detail here, material from Qreiya 3 matches or exceeds the entirety of the known Danian skeletal record in terms of richness and specimen numbers (13) (tables S1 and S3). Taxonomically, this diverse majority-percomorph assemblage resembles Eocene and younger communities and includes the oldest representatives of multiple extant families. Temporally, Qreiya 3 postdates the K-Pg boundary by less than 4 Myr and predates the oldest comparably diverse Paleogene marine fish assemblages by 6 Myr or more. Paleogeographically, it provides the earliest post-Cretaceous window on tropical (paleolatitude 18°N) diversity in the Western Tethys, a region widely considered to be a Paleogene biodiversity hotspot (23, 24), bearing on competing hypotheses of Cenozoic marine fish biogeography (25). As the only offshore Danian skeletal assemblage, it captures an ecosystem comparable to that of some Cretaceous sites while offering a critical perspective on recovery in tropical regions far removed from the Chicxulub impact site. Association of the Qreiya 3 assemblage with the LDE, a minor hyperthermal marked by a rise of ~2° to 3°C in temperature, provides an opportunity to explore the impact of changing climate on early Paleocene marine ecosystems. In combination, these attributes make Qreiya 3 an important source of new constraints on the timing and dynamics of the rise of modern marine fish communities. RESULTS Fishes from Qreiya 3 are abundant and preserved as laterally compressed, articulated skeletons (Figs. 2 to 4). Preservation was likely enhanced by anoxic bottom-water conditions during the onset of the Late Danian Event, representing a taphonomic window (22, 26). Comparable preservation characterizes Egyptian fish-bearing deposits associated with the Paleocene-Eocene Thermal Maximum (PETM) (27) and other PETM horizons in the Tethys region (20), which likewise reflect oxygen-stressed conditions. Fossils also include isolated chondrichthyan teeth (Hexanchidae and Odontaspididae) and squamate (†Palaeophiidae) material (Supplementary Text and figs. S3 and S4). Here, we provide only a brief description and systematic treatment of the actinopterygian assemblage at the level of family and order in most cases, with comprehensive descriptive and phylogenetic analyses to follow in subsequent taxon-specific studies. Further details and figures for the taxa reported below are presented in the Supplementary Materials (Supplementary Text, figs. S5 to S44, and table S1). Non-acanthomorphs and non-percomorph acanthomorphs from Qreiya 3. (A to D) Non-acanthomorph fishes. (A), †Pycnodontidae (cf. †Nursallia), specimen missing caudal fin (MUVP 642). [(B) and (C)] Osteoglossidae (bony tongues) (†Phareodontinae) undet. (B) maxillary fragment (MUVP 652). (C) Partial skeleton (MUVP 650). (D) Clupeiformes (Clupeoidei) undet. (MUVP 661). (E to G) Non-percomorph acanthomorph fishes. (E) Veliferidae (sailfin velifers) undet, partial skeleton (MUVP 666). (F) Zeiformes (dories) undet. Morphotype I, partial skeleton (MUVP 670). (G) Acanthomorpha undet., Morphotype I, complete skeleton (MUVP 671). Percomorphs from Qreiya 3. (A to C) Syngnathiformes. (A) Syngnathidae (pipefishes) undet., articulated skeleton missing caudal region (MUVP 698). (B) Aulostomoidea undet., skeleton missing caudal region (MUVP 705). (C) Aulostomoidea undet., skull (MUVP 706). (D and E) Scombriformes. (D) Scombridae (†Eocoeolopomini) (mackerel, tuna, and kin) undet., articulated skeleton missing the neurocranium and the caudal fin (MUVP 708). (E) Trichiuroidea undet. (MUVP 829). (F and G) Carangiformes. (F) Menidae (moonfish) (Mene sp.), complete skeleton (MUVP 804). (G) Carangidae (jacks) undet., complete skeleton (MUVP 805). (H and I) Distinct percomorph morphotypes. (H) Percomorpha undet. Morphotype I, skull and anterior region of the body (MUVP 709). (I) Percomorpha undet. Morphotype II, skull and the anterior abdominal region (MUVP 812). Key taxa and ecosystem reconstruction of the Qreiya 3 assemblage. (A) Summary of the new earliest skeletal occurrences of several acanthomorph lineages represented in this study. Previously documented stratigraphic ranges are shown in white (see table S4), and stratigraphic range extensions based on Qreiya 3 are shown in orange; open-ended arrows indicate lineages that extend beyond the plotted interval to the present. An asterisk in Aulostomoidea indicates that this may represent the earliest occurrence of the group, contingent on resolving the phylogenetic position of the Mexican syngnathiform †Eekaulostomus (38). (B) Artistic reconstruction of the Qreiya 3 fauna. Individual taxa shown approximately at scale to one another, with anatomy based on specimens from Qreiya 3 augmented by close living and extinct relatives. Image credit: I. Baylatry. Abbreviations: Ca, Carangidae; Cl, Clupeoidei; He, Hexanchidae; Me, Menidae (Mene sp.); Od, Odontaspididae; Pa, †Palaeophiidae; Ph, †Phareodontinae; Py, †Pycnodontidae (cf. †Nursallia); Sc, Scombridae; Sy, Syngnathidae; Tr, Trichiuroidea; up, undetermined percomorph; ua, undetermined acanthomorph; Ve, Veliferidae; Ze, Zeiformes. Non-acanthomorphs The assemblage includes at least five different actinopterygian taxa that fall outside of Acanthomorpha (spiny-rayed fishes), the principal group of fishes in modern marine settings (11). †Pycnodontids, an early-diverging group of neopterygians, are represented by several specimens (Fig. 2A and figs. S5 to S7) attributable to the long-ranging (Late Cretaceous–late Eocene) genus †Nursallia based on the presence of curved frontals contributing to a rounded anterior dorsal profile of the head and elaborate intervertebral sutures (28). We interpret a single elongated specimen as a probable anguilliform (eel) (fig. S8), based on the overall body geometry and the presence of greatly elongated branchiostegal rays (29). Osteoglossiformes (bony tongues) (Fig. 2, B and C, and figs. S9 to S11) are represented by two partial skeletons, including the largest specimen in the assemblage, which has a preserved length of ~50 cm and estimated total length of 75 cm. This fusiform taxon (Fig. 2C) exhibits a long, robust first pectoral-fin ray and a broad frontal bone extending above the orbit, indicating affinities with †phareodontine osteoglossids. An isolated osteoglossid maxilla (Fig. 2B) bears a single row of teeth with their bases encased by bony collars (30). Isolated scales with contrasting ornamentation and different vertebral proportions between the two partial skeletons suggest the presence of at least two osteoglossid taxa. Clupeiformes are represented by several specimens (Fig. 2D and figs. S12 to S15) that show ventral scutes, fusion of the first uroneural with the first preural centrum, and a reduced first ural centrum that indicate placement within Clupeoidei (31). Non-percomorph Acanthomorpha Only two spiny-rayed taxa definitively belonging to groups other than percomorphs are present. Three lampriform specimens representing a single species of Veliferidae (sailfin velifers) show hypurostegy, insertion of the first dorsal-fin radial anterior to the first neural spine, scaly bases of median fins, and greatly elongated dorsal-fin rays (32, 33) (Fig. 2E and fig. S16). They closely resemble other extinct and extant velifers, and are the oldest lampriforms clearly belonging to a living family (34). Zeiformes (dories) are represented by one definitive representative and potentially two additional morphotypes (Fig. 2F and figs. S17 to S21). These minute, deep-bodied forms display a first neural spine fused posteriorly to the skull, vertically oriented neural spines on the posterior abdominal and anterior caudal vertebrae, and vacant interneural spaces (35). The definitive zeiform bears a ventral body scute (35). Percomorpha Perch-like fishes constitute the primary taxonomic component of the Qreiya 3 assemblage, with at least eight different taxa. Syngnathiformes (pipefishes, seahorses, and kin) are represented by multiple specimens, showcasing elongated skulls and tubelike mouths and belonging to at least two different taxa (Fig. 3, A to C, and figs. S22 and S23). One syngnathiform taxon is encased in bony plates and exhibits neural arches that are expanded anteroposteriorly and perforated by numerous foramina along the length of the preserved vertebral column (36). As in syngnathids (pipefishes), the bony plates are arranged in rows and are unlike either the stellate ones of solenostomids (ghost pipefishes) or the irregularly arranged scutes of the coeval Mexican †Eekaulostomus, which has recently been interpreted as a stem syngnathoid (37, 38). A diminutive, long-bodied syngnathiform lacking large scutes joins this larger armored form and likely represents the earliest clear member of Aulostomoidea (coronetfishes and trumpetfishes) (Fig. 3, B and C, and fig. S24). Qreiya 3 provides the oldest definitive skeletal records of Scombriformes and Carangiformes, two anatomically diverse groups hypothesized to have radiated in the aftermath of the K-Pg (15, 16, 39). Those include first occurrences of the scombriform clades Scombridae (mackerel, tuna, and kin) and Trichiuroidea (snake mackerels and cutlassfishes) and the carangiform families Menidae (moonfish) and Carangidae (jacks). Scombrids (Fig. 3D and figs. S25 and S26) are represented by three specimens: first, an articulated partial skeleton that shows a nonprotrusible upper jaw, a single row of widely spaced teeth, horizontally oriented neural and hemal spines in the caudal region, and the presence of finlets; second, an articulated neurocranium, jaws, suspensorium, and branchiostegal rays, with a frontal bone featuring a postero-superior fossa bordered anteriorly by a substantial thickening (fig. S26, A and B); and third, a hypural plate, overlapped by the caudal fin rays lacking a caudal notch (fig. S26, E and F). This suite of features suggests placement in †Eocoeolopomini, an extinct scombrid tribe ranging from the latest Paleocene-Eocene (40). Trichiuroids are represented by an incomplete, elongated skull (Fig. 3E and fig. S27). As in other trichiuroids, the mandible bears a hook-shaped anterior fang followed by larger, broadly separated teeth, and the apparently nonprotrusible premaxilla bears similarly enlarged dentition (41). The anatomically distinctive menid Mene (42) (Fig. 3F and fig. S28) is the most abundant identifiable fish from Qreiya 3, represented by ~80 individuals. Carangids are represented by three fusiform specimens (Fig. 3G and fig. S29) displaying the principal characteristic of the family: a wide gap between the second and third anal-fin spines (43). The Qreiya 3 assemblage includes several small- to medium-sized fish specimens that are incomplete or require additional preparation and study, complicating taxonomic placement (Fig. 3, H and I, and figs. S30 to S35). Nevertheless, they display suites of traits consistent with assignment to Percomorpha, including ~10 abdominal vertebrae, pelvic fins with a spine and five or fewer rays, transformed ctenoid scales, the absence of a second ural centrum, and an anterior portion of the pelvic bones in contact with the pectoral girdle (29). One morphotype shows relatively large hooklike teeth whereas the other bears small denticles, indicating trophic diversity among these otherwise morphologically conservative taxa. Acanthomorpha incertae sedis and indeterminate Teleostei In addition to the well-diagnosed taxa described above, the Qreiya 3 assemblage includes several specimens whose taxonomic placement remains less securely resolved. These include two distinct acanthomorph morphotypes. The first is a uniformly small sized and morphologically distinctive spiny form with highly ambiguous systematic affinities that could possibly represent a juvenile stage (44). Characterized by greatly elongated, barbed dorsal-fin and pelvic-fin spines and an anteriorly positioned dorsal-fin insertion, it is the second most abundant taxon from Qreiya 3 (Fig. 2G and figs. S36 to S38). The second is a single slender fish with greatly elongated pelvic fins (figs. S39 and S40). In addition, the collection includes various partial fish specimens with distinctive traits; however, their incompleteness poses challenges for taxonomic classification (figs. S41 to S44). Quantitative comparisons of marine fish assemblages around the K-Pg Quantitative comparisons across Campanian-Eocene marine actinopterygian assemblages yield conspicuous patterns. An ordination of skeletal assemblages (n = 28) based on taxonomic similarity shows the principal axis of variation in faunal composition aligns roughly with stratigraphic age, with conspicuous separation between Late Cretaceous and Paleogene assemblages. Qreiya 3 plots within the convex hull of Eocene assemblages, closer to these localities than any other Paleocene site (Fig. 5A). A coarser examination of faunal composition for a more inclusive set of sites (n = 97) that integrates parallel skeletal and otolith taxonomies shows temporal and geographic patterns in the establishment of percomorph-dominated marine assemblages. No Cretaceous assemblage is percomorph dominated, but most Eocene ones are. The Paleocene shows a mixture of these patterns. Higher proportions of percomorphs are restricted to low paleotropical latitudes, whereas higher-latitude sites generally show a greater share of non-percomorphs (Fig. 5B). Quantitative comparisons of marine actinopterygian assemblages in the late Late Cretaceous and early Paleogene (Campanian-Priabonian). (A) Metric multidimensional scaling ordination of faunal similarity for actinopterygian faunas represented by articulated skeletons. Pairwise similarities calculated using the Forbes index for the presence/absence of families (or more inclusive groups). The plot rotated so that the first ordination axis, which strongly covaries with stratigraphic age, is oriented vertically. (B) Proportion of percomorph species in marine actinopterygian skeletal (filled symbols) and otolith (open symbols) faunas in the late Late Cretaceous and early Paleogene (for sources, see table S2). Error bars represent 50% binomial confidence intervals. Majority percomorph faunas are restricted to low latitudes in the Late Cretaceous and Paleocene but occupy all latitudes by the Eocene. DISCUSSION Implications for extinction and recovery Qreiya 3 punctuates Patterson’s Gap (13) in the marine fish fossil record and provides an unprecedented view of community structure after the K-Pg mass extinction in an offshore, tropical setting in the Western Tethys (Figs. 2 to 4). It is unquestionably the oldest majority percomorph locality (either skeletal or otolith) known, yielding twice as many extant percomorph families as the roughly coeval Palenque with a relative abundance of percomorphs over twice as high (67% of individuals versus 29%; Fig. 1B and table S3). Despite the high yield of fossils from Qreiya 3, traces of major Cretaceous lineages hypothesized to have gone extinct at the K-Pg boundary are conspicuously absent (12). Key groups missing from the assemblage include †pachycormids, †pachyrhizodontids, †ichthyodectiforms, †enchodontids, and †cimolichthyids. These large, widely distributed pelagic predators feature robust and readily identifiable skeletal elements and are generally represented in Cretaceous sites capturing offshore environments like Qreiya 3 (45). Their absence bolsters earlier interpretations, based on limited data from other Danian sites, that these groups were victims of the extinction (10, 12, 46). In the place of these emblematic Cretaceous lineages, Qreiya 3 reveals a cast of Danian predatory pelagic fishes like those of younger deposits: osteoglossids, carangids, scombrids, and trichiuroids. Osteoglossids have a Cretaceous record in freshwater environments, with their widespread appearance in marine settings in the Paleocene and Eocene interpreted as a short-lived opportunistic expansion into ecological roles cleared by the K-Pg extinction (10). Carangids, scombrids, and trichiuroids are key living predatory groups within larger percomorph clades inferred to have diversified around the K-Pg boundary based on fossil-calibrated molecular phylogenies (15, 16). Until now, they have lacked a Danian skeletal record, perhaps because other sites of this age represent shallower, more coastal settings or these groups had restricted geographic distributions at that time (15, 16). Unambiguous examples from Qreiya 3 represent the oldest articulated remains for these lineages, establishing robust new minimum age constraints on the timing of their early diversification (Fig. 4A and table S4). Notably, fossils of these groups are widely used in molecular clock studies (table S3), making articulated material from Qreiya 3 uniquely valuable for future refinement of the evolutionary timescale of Percomorpha. Within Scombriformes and Carangiformes alone, fossils from Qreiya 3 shift minimum divergence times to the Danian for roughly a dozen lineages classically recognized as taxonomic families (15, 16). This is consistent with both the rapid model of post K-Pg radiation suggested by past molecular and morphometric studies (15, 16, 47), otoliths (9), ichthyoliths (8), and patterns of functional replacement hypothesized from skeletal fossils (12, 46). Some of these percomorph lineages appear in the otolith record later in the Paleocene, with only trichiuroids (gempylids) reported as early as the Danian (9). However, articulated skeletons offer substantially more information than otoliths, including character evidence for refined phylogenetic placements and ecologically important aspects of anatomy like jaw geometry (10) and body shape (12). Moving beyond this highest trophic level, other fishes from Qreiya 3 add to the picture of a taxonomically and ecologically diverse community closely comparable to those of the Eocene. They include a mixture of deeply diverging lineages with extensive Mesozoic records like †pycnodontids, anguilliforms, and clupeiforms, coupled with a large variety of spiny-rayed fishes that constitute most of the assemblage in terms of taxonomic richness as well as specimen counts. The acanthomorphs include members of lineages previously known from the Late Cretaceous (e.g., zeiforms, lampriforms, and syngnathiforms) and those only recovered thus far from Cenozoic strata (e.g., carangiformes and scombriforms). However, the Qreiya 3 acanthomorph assemblage shows some notable differences from both Late Cretaceous and post-Paleocene faunas. Among non-percomorphs, we have not recognized any polymixiiforms, beryciforms, or trachichthyiforms, all of which are prominent members of many Late Cretaceous marine acanthomorph communities but relatively uncommon in most Cenozoic skeletal assemblages (13, 48, 49). Although several percomorphs are present, those that can be clearly assigned to specific clades all belong to relatively early-branching percomorph groups (Fig. 4A). We have not identified any unambiguous representatives of the highly nested percomorph radiation Eupercaria from Qreiya 3. Modern examples of this clade include major marine lineages like tetraodontiforms (pufferfishes and kin) and labrids (wrasses) that represent emerging macroevolutionary model systems (50, 51). Evidence of eupercarians is also sparse in the handful of other skeleton-yielding Paleocene sites (13). By contrast, Eupercaria is well represented in most Eocene and younger marine localities, including environments ranging from offshore pelagic settings (20) to shallower, reef-adjacent ones (48). Their absence from Qreiya 3 may reflect limited ecological prominence in offshore Danian settings, but additional collecting will be essential to determining whether this pattern is real or only apparent. Faunal heterogeneity in the Danian Higher taxonomic richness and a preponderance of percomorphs give Qreiya 3 a notably modern aspect compared to contemporaneous skeletal assemblages, highlighting compositional contrasts among marine fish communities in the critical first few million years after the K-Pg event. Qreiya 3 corresponds more closely to Eocene faunas than it does to any Paleocene ones (Fig. 5B). Notably, the Qreiya 3 fauna shows little resemblance to that from the higher-paleolatitude (54.4°N) Limhamn site that provided the only meaningful skeleton-based data for Danian marine fishes for over 100 years. Among Limhamn acanthomorphs, two generalized percomorphs of uncertain relationships to extant lineages (18, 52) are joined by polymixiiforms and trachichthyiforms, groups common in Late Cretaceous skeletal assemblages but not recovered to date from Qreiya 3. The position of the Palenque assemblage at the periphery of the Eocene cluster may indicate a transitional fauna that is less compositionally modern than Qreiya 3 but more so than most other Paleocene assemblages. This pattern is consistent with the possibility that Palenque samples an earlier recovery interval (63 ± 1.5 Ma) than Qreiya 3. These two important paleotropical localities share multiple non-acanthomorph lineages in common (e.g., †pycnodontids and osteoglossids) but display no overlap between their percomorph components although both include extant families. Despite the assumed association of the Palenque assemblage with reefs, a locus of extant fish diversity, its richness of fishes generally—and percomorphs specifically—is lower than that of the offshore Qreiya 3 community. Taken at face value, this could be seen as additional evidence of geographically heterogeneous fish recovery from the K-Pg crisis, a pattern previously inferred from ichthyoliths (53). However, whether the contrast in diversity between Palenque and Qreiya 3 reflects age differences between the assemblages, protracted disruption to ecosystems proximal to the K-Pg impact site, offset timing of diversification between environmental settings, or an artifact of differential sampling effort or preservation potential is unclear at present. The percomorph radiation: Out of the tropics? The distinctly modern composition of the Qreiya 3 fauna, combined with a paleotropical latitude, throw geographical differences in faunal patterns after the K-Pg extinction into sharp contrast. Past works suggest possible spatial heterogeneity in the diversification of modern marine fish faunas, with the proposal that paleotropical settings might have been key to the early radiation of percomorphs (13, 25). Our survey shows that percomorph-rich assemblages are absent from the Late Cretaceous, restricted to paleotropical to paleosubtropical latitudes in the Paleocene, and encompass higher paleolatitudes only in the Eocene (Fig. 5B). These findings suggest that Paleocene marine fish faunas at higher latitudes may have retained a structure more reminiscent of pre–K-Pg assemblages. By contrast, the paleotropical Qreiya 3 site is the only Danian locality known to preserve a majority percomorph assemblage closely resembling the taxonomic structure characteristic of Eocene and younger marine faunas. At these coarse levels of comparison, it is unclear whether this indicates range expansions of tropical lineages as a function of warm early Eocene climates with low pole-to-equator thermal gradients (27, 54), delayed in situ diversification of percomorphs at mid-to-high paleolatitudes (55), an artifact reflecting environmental or spatial distribution of available fossil sites (55), or some combination of these. Additional localities and more detailed analysis will be necessary to determine the relative contributions of these factors to observed patterns. The location of Qreiya 3 within the Western Tethys—preceding the region’s well-established status as a marine biodiversity hotspot in the Eocene (23)—might invite interpretation of its elevated diversity as an early expression of this pattern. We treat such an inference with caution. The Danian skeletal fish record remains extremely sparse, with limited overlap among localities in age, geography, and depositional environment, compounded by substantial age uncertainties (13, 18, 19). This makes precise biogeographic inferences highly sensitive to the addition of new data (25, 56). To date, evidence bearing on a possible Paleocene Western Tethyan biodiversity hotspot remains incompletely synthesized (23), although there are indications the region may have acted as a center of elevated diversity for reef-associated ecosystems (24). Because Qreiya 3 is currently the only well-documented offshore paleotropical Danian locality, this raises the possibility that its composition reflects environment (offshore pelagic versus more nearshore settings) rather than a uniquely Western Tethyan geographic signal within the broader paleotropics. Resolving the dynamics of fish diversification in the tropics will require expanded sampling across multiple paleotropical regions and depositional settings, together with refined age constraints for existing Danian localities. In sum, the Qreiya 3 site brings shifting aspects of marine fish biodiversity around the K-Pg boundary into clearer focus. First, despite collecting hundreds of fossil specimens, we have found no evidence of several anatomically distinctive predatory fish lineages that are well represented in geographically widespread Late Cretaceous marine deposits. The conspicuous absence of these groups suggests that their extinction before the late Danian is genuine—and likely associated with the K-Pg crisis itself—rather than an artifact of an especially poor Paleocene fossil record (10, 46). Second, the Qreiya 3 assemblage yields the earliest majority percomorph marine assemblage with numerous, unambiguous representatives from anatomically specialized clades (Figs. 3 and 4). The presence of extant lineages of Scombriformes (Scombridae and Trichiuroidea), Carangiformes (Menidae and Carangidae), and Syngnathiformes (Syngnathidae and Aulostomoidea) provides unambiguous evidence that some of the most ecologically divergent living lineages of the marine percomorph radiation were established no later than a few million years after the K-Pg extinction. This invites comparisons to patterns in other vertebrate groups. Neoaves—the avian counterpart to percomorph fishes—are sparsely represented in the Paleocene, with the Danian bird record limited to a handful of taxa that clearly belong to extant clades alongside others of uncertain phylogenetic affinity (57). Molecular clock studies of birds likewise recover early Cenozoic ages for major neoavian lineages (4). Paleontological patterns are well established for eutherian mammals, the classic vertebrate model for evolutionary radiation in the Cenozoic (58). The Danian mammalian fossil record yields an assortment of problematic “archaic” placentals, with only a few taxa that can be confidently assigned to extant lineages. It is only in the Eocene or after that the mammalian assemblages become dominated by anatomically distinctive modern orders (3). Despite differences in details, all three groups show a consistent pattern of significant taxonomic diversification and ecological expansion around the K-Pg, achieving substantial diversity in the fossil record by the Eocene. Third, and perhaps most relevant to understanding patterns in other major lineages during a critical time in Earth history, Qreiya 3 and the handful of coeval Danian sites suggest substantial geographic or environmental heterogeneity in the response of marine fishes to the K-Pg extinction and subsequent recovery. Similar patterns have been reported in a host of marine groups including chondrichthyans (59), macroinvertebrates (60), and microinvertebrates (61), as well as plants (62) and mammals (63) in terrestrial settings. Available evidence points to paleotropical regions as being important to understanding the origin of modern marine fish faunas. Poor sampling of the paleotropics in the Cenozoic represents a well-recognized anthropogenic bias in the fossil record, and marine fishes are no exception to this general pattern (64). Apart from revealing biotic responses at low latitudes during intervals of extreme climate (27), our results imply that sites from low paleolatitudes could have outsized impact on our understanding of early events in the evolution of today’s diverse marine fish communities. As a geographically clustered set of low-latitude marine faunas, continued work on Egyptian fish fossil assemblages from the Danian (Qreiya 3), PETM (Ras Gharib A) (27), and Lutetian (Mokattam) hold particular promise for addressing these questions. MATERIALS AND METHODS Fossil collection and preparation Fossils were collected during three field expeditions of the Mansoura University Vertebrate Paleontology Center (MUVP) in 2020, 2021, and 2023 to the Qreiya 3 section at Gebel Qreiya in the Eastern Desert of Egypt. Paleontological quarrying of the LDE horizon yielded ~500 individual fish specimens, multiple rock slabs, and six large plaster jackets. Two of these large jackets have been prepared, and their fauna was included in this study. Where necessary, specimens were mechanically prepared following standard techniques at the University of Michigan Museum of Paleontology (UMMP) and MUVP. All specimens are deposited in the MUVP. Computed tomography scanning Computed tomography (CT) scanning for some of the specimens (table S5) was conducted at the CTEES facility of the University of Michigan using a Nikon XT H 225 ST scanner. CT scans were segmented using the commercial biomedical software Materialise Mimics (Leuven, Belgium). Figured bones were exported as surface files (.ply format) and rendered in Blender (Blender Documentation Team, blender.org). In addition, CT scanning was performed to survey the fossil content of five of the plaster jackets using a GE (General Electric) Revolution EVO CT Scanner in Mansoura, Egypt. Specimen photography Most of the specimens were photographed using a Keyence Microscope (VHX-7000/7020) located in the Department of Materials Science and Engineering at the University of Michigan. In addition, Reflectance Transformation Imaging (RTI) was used for certain specimens. Some photographs were taken at the MUVP and UMMP using a Nikon D850 digital SLR. Ordination analysis To place the Qreiya 3 assemblage in the broader context of Late Cretaceous and early Paleogene fish assemblages, we conducted a quantitative comparison of the composition of marine fish faunas represented by articulated skeletons (rather than isolated bones, teeth, or otoliths). The skeletal faunal list, which records the presence or absence of taxa mostly at the family level, is adapted from El-Sayed et al. (27) with the addition of the following literature and/or localities (the complete list with references is in table S2): Campanian: Coesfeld Member, Sendenhorst, Germany. Maastrichtian: Pindos Unit, Gavdos Island, Greece; Liburnica Formation, Trebiciano, Italy; Muwaqqar Chalk Marl Formation, Jordan. Danian: Dakhla Formation, LDE horizon, Qreiya 3 Lagerstätte (this study). Ypresian: “Nummulitic Limestone,” Bolca, Italy (the faunal list was updated with information from publications published after 2021); Solteri Lagerstätte, Italy (65); Fur Formation, Fur, Denmark (the faunal list was updated with information from publications published after 2021). Lutetian: Lillebaelt Clay, Denmark (66); Mokattam Formation, Jebel Turah, Egypt. Priabonian: Pabdeh Formation, Istehbanât, Iran. We updated the taxonomic conventions used by El-Sayed et al. (27). We updated the syngnathiform classification to include the following higher-level groupings (38): Syngnathoidea (Syngnathidae, Solenostomidae, and †Eekaulostomidae), Centriscoidea (Centriscidae, Macroramphosidae, †Gasterorhamphosus, †Gerpegezhidae, and †Paraeoliscidae), and Aulostomoidea (Fistulariidae, Aulostomidae, †Urosphenidae, and †Parasynarcualidae). The Eocene †Aulorhamphidae and †Rhamphosidae remain incertae sedis within Syngnathiformes, and we therefore treat them as separate clades in our faunal analysis. For the so-called “veliferoid” lampridiforms, here we regard †Turkmene, †Whitephippus, †Danatina, and various †Analectis-like taxa as members of an unnamed group we informally term “Veliferoidei.” Fossil fishes of uncertain taxonomic placement were excluded (e.g., indeterminate teleosts in this study). To minimize problems associated with heterogeneous sampling (e.g., unequal sample size and varying collecting intensity) that could affect the presence/absence of data in our faunal list, we used the Forbes index (67) with the modification suggested by Alroy (68, 69) and implemented code provided by Brocklehurst et al. (70) to determine pairwise distances between all of our assemblages. We used the R script from El-Sayed et al. (27) to perform a classic multidimensional scaling (i.e., principal coordinates analysis) using the function pcoa() in the ape package using the correction = “lingoes” option to implement the procedure for yielding results without negative eigenvalues (71). Articulated fish skeletal fossil record sites containing three or fewer taxa were excluded from this study. The R script is provided as part of the Supplementary Materials as an R file “El-Sayed_et_al_ordination.” Percomorph fish species proportions in the Late Cretaceous and early Paleogene To complement our ordination analysis of skeletal faunas, we developed a parallel dataset to examine the proportion of percomorph fish species within Late Cretaceous and early Paleogene faunas. This coarse taxonomic binning allowed us to make use of both skeletal and otolith assemblages, as well as sites for which relative abundances are not available. Faunas represented by skeletons are those from the previous section, and otolith data were compiled from the available literature (table S2). Latitude data were sourced directly from publications where available. In cases where specific latitude information was not provided, location details from the respective papers and maps were used to determine the latitude for each locality. To estimate paleolatitude and paleolongitude, we used the R package rgplates with the MULLER2016 plate reconstruction model (72, 73). For locality ages, point estimates were used when available. If a range was provided (e.g., a stage), the midpoint was selected to represent the faunal age. For site-level taxonomic counts, taxa whose identifications are described as “tentative,” “possible,” or otherwise uncertain were excluded. Taxa preceded by a question mark (“?”) were likewise omitted. In addition, taxa assigned only to higher-level groups whose phylogenetic placement could ultimately fall either within percomorph or non-percomorph lineages (e.g., Acanthomorpha) were excluded from the total counts to avoid inflating diversity estimates. Only taxa with secure and stable placements were included in the analysis and incorporated into the dataset. For clarity of interpretation, skeletal fossil assemblages with fewer than three species and otolith-bearing sites with fewer than 20 species were excluded from the analysis. The complete dataset is available in the Supplementary Materials “El-Sayed_et_al_percomorph_proportion_without_plate_reconstruction_DS2.” In addition, a separate file is provided that includes the reconstructed paleocoordinates using the MULLER2016 plate reconstruction model “El-Sayed_et_al_percomorph_proportion_with_plate_reconstruction,” facilitating ease of analysis. The accompanying R script offers two options: Users can either run the analysis to generate the paleocoordinates within the code or directly use the dataset with the reconstructed paleocoordinates. The R script is provided as part of the Supplementary Materials as an R file “El-Sayed_et_al_percomorph_proportion.” All datasets and codes are available at Zenodo (https://doi.org/10.5281/zenodo.19054170).