Open ocean ecosystems are dominated by mesopelagic fishes, which occupy depths between 200 and 1,000 m and collectively comprise the largest proportion of global fish biomass (Irigoien et al., 2014). As both major consumers of microzooplankton and key prey for pelagic predators, including commercially important species, mesopelagic fishes provide an important intermediate link in the pelagic food web (Choy et al., 2013; Young et al., 2015; Drazen and Sutton, 2017). Additionally, the diel vertical migration behavior characteristic of many mesopelagic fish taxa contributes to the global carbon cycle, actively transporting carbon consumed in the epipelagic to the mesopelagic, where it is deposited via respiration, fecal material, or mortality (Davison et al., 2013). While the importance of mesopelagic fishes to ecosystem function is now well recognized, the early life history stages of these taxa remain understudied.

Larval mesopelagic fishes are a large, often numerically dominant, component of oceanic ichthyoplankton catch in regions worldwide (e.g., eastern North Pacific, Moser and Smith, 1993; western North Pacific, Sassa et al., 2004; northwest Mediterranean, Olivar et al., 2010; equatorial and tropical Atlantic, Olivar et al., 2018; southern Gulf of Mexico, Daudén-Bengoa et al., 2020). Mesopelagic larvae are thought to be largely constrained to productive epipelagic regions of the water column (<200 m) with limited or small-scale diel vertical movements relative to adults (Ahlstrom, 1959; Sassa et al., 2002). However, broad-scale surveys of ichthyoplankton rarely sample at depths exceeding 200 m (exceptions include Moser and Smith, 1993; Sassa and Kawaguchi, 2006; Olivar et al., 2018; Dove et al., 2021), so mesopelagic larval dynamics throughout the water column remain largely unknown.

Understanding the spatial and temporal distributions of larval fish assemblages and the physical processes that shape them can provide insights into the patterns that structure adult populations (Fuiman and Werner, 2002). Indeed, the geographic distributions of adult mesopelagic fishes have been found to reflect larval fish distributions accurately, relative to the dispersal potential of the larval stage (Olivar et al., 2017, 2018). Because of their limited mobility and high vulnerability (Houde, 1989), larval fish assemblage dynamics are heavily influenced by oceanographic features and often reflect changes in the marine environment (Kingsford, 1993; Brodeur et al., 2008; Hsieh et al., 2009). This makes the early life stages of fishes uniquely susceptible to environmental perturbations, which can have impacts on adult recruitment, food web dynamics, and broader ecosystem services.

In the northern Gulf of Mexico (GOM), the Deepwater Horizon Oil Spill (DWHOS) occurred at ∼1,500 m depth in April 2010 and exposed a deficiency in baseline data for deep-sea ecosystems in the GOM and beyond (Ramirez-Llodra et al., 2011; Fisher et al., 2016). The adult stages of mesopelagic fishes have long been studied in the GOM, albeit on a limited spatial scale (see Hopkins et al., 1996) or without depth-discrete sampling (Murdy et al., 1983); focus on quantifying the ecology of the mesopelagic GOM, including fishes, has increased dramatically in the wake of DWHOS (Cook et al., 2020). Mesopelagic fish larvae, however, receive little targeted attention in the GOM. Studies of broad-scale ichthyoplankton assemblages in the GOM that include (but are not focused on) mesopelagic fish taxa are constrained to sampling of the epipelagic region (Muhling et al., 2012; Lyczkowski-Shultz et al., 2013; Meinert et al., 2020) and to coastal waters (Rakocinski et al., 1996; Carassou et al., 2012). High abundances of mesopelagic larvae in the mixed layer (0–100 m) of the open ocean GOM suggest that epipelagic regions represent important habitat for these taxa during early development (Meinert et al., 2020), but questions remain about larval distribution patterns and usage of mesopelagic portions of the water column.

Adult mesopelagic fish assemblages have been found to lack geographic and temporal patterns of structure in the GOM, suggesting that these assemblages are well-mixed and highly dispersed over large distances despite variation in local oceanographic features (Ross et al., 2010; Milligan and Sutton, 2020). On the continental shelf, distributions of larval mesopelagic fishes in the GOM were largely explained by sampling depth and were not associated with environmental factors, despite strong correlation of epipelagic-associated larval fishes to sea surface temperature (Muhling et al., 2012). These previous findings suggest that mesopelagic larval fishes in the oceanic GOM may have limited spatial assemblage structure, although the response of vertical assemblage distribution to environmental drivers remains unknown.

Here, we characterize the dynamics of larval fish assemblages across epipelagic, mesopelagic, and bathypelagic (0–1,500 m) regions of the GOM, encompassing both epipelagic- and mesopelagic-associated fish families. We examine indices of abundance and diversity, as well as environmental drivers of larval fish assemblage structure in the different pelagic zones. Insights into the spatial and temporal scale of variability in pelagic ichthyoplankton communities in the GOM will provide a reference point for a previously unstudied ecosystem that may guide response efforts to future environmental perturbations.

Mesopelagic-associated taxa comprised the majority of ichthyoplankton in both the epipelagic and mesopelagic regions of the northern Gulf of Mexico. Four dominant mesopelagic families (Myctophidae, Gonostomatidae, Sternoptychidae, Phosichthyidae) were particularly common, with few other families making significant contributions to larval assemblage groups within mesopelagic waters. Mesopelagic-associated families outside of the four dominant taxa (e.g., Chauliodontidae, Melamphaidae, Paralepididae) were largely constrained to larval assemblage groups within epipelagic waters, and while these families were occasionally found in the mesopelagic, catches were minimal in comparison to the epipelagic and associations with mesopelagic assemblage groups were weak. Overall, larval abundance declined steeply with increasing depth and even the dominant mesopelagic families were most abundant in epipelagic regions. In both epipelagic and mesopelagic regions, the distinct larval assemblage groups identified were predominantly structured by depth and season, with surface conditions (temperature, salinity, mixed layer depth) and geographic locale (longitude) providing some finer-scale distinctions among assemblages. The most common (largest) epipelagic assemblage groups consisted of samples distributed widely across the GOM, as did all mesopelagic assemblage groups (southeastern gap in sampling coverage during the fall notwithstanding). Meaningful assemblage structure was not detected in the bathypelagic region, which is not surprising given the environmental stability of the pelagic realm beyond 1,000 m depth.

The ubiquity of the dominant mesopelagic ichthyoplankton families detected aligns with prior surveys of adult mesopelagic fishes in the GOM, where Gonostomatidae, Myctophidae, Phosichthyidae, and Sternoptychidae accounted for 96% of all specimens collected (Ross et al., 2010). Ichthyoplankton surveys encompassing outer shelf and oceanic regions of the GOM also found high abundances of these dominant mesopelagic families when sampling epipelagic waters (Muhling et al., 2012; Meinert et al., 2020). Among the few ichthyoplankton surveys sampling to depths (800–1,000 m) comparable to this study, similar broad patterns of distribution were detected in the equatorial and tropical Atlantic when accounting for both epipelagic and mesopelagic waters, with prevalence of the same four dominant families and decreasing abundance and richness with increasing depth (Olivar et al., 2018; Dove et al., 2021).

We found that larval fish assemblages in the oceanic GOM were largely structured by depth and season, which is consistent with patterns from other regions and often reflect the spawning habitat preferences of adults (Doyle et al., 1993; Duffy-Anderson et al., 2006; Muhling et al., 2012). Adult mesopelagic fish assemblages in the GOM were found to lack spatial structure despite the presence of distinct physical oceanographic features among widely separated sampling sites (Ross et al., 2010; Milligan and Sutton, 2020), so it is not unexpected that the epipelagic and mesopelagic larval assemblage groups identified here were broadly distributed across the GOM with little horizontal differentiation. Adult distributions have been found to be important in structuring patterns of larval distribution in mesopelagic fishes in other regions (tropical and equatorial Atlantic, Olivar et al., 2017, 2018; southern Gulf of Mexico, Daudén-Bengoa et al., 2020). While the extent of larval distribution often mirrors patterns of adult distribution, larval assemblages are subsequently subject to strong influence by oceanographic conditions (Grothues and Cowen, 1999). Surface salinity was identified by MRT as the primary environmental variable responsible for differentiating larval fish assemblages within upper and lower epipelagic regions. In the GOM, freshwater and high nutrient discharge from the Mississippi River interact with the Loop Current system to influence salinity variability in the GOM (Dagg and Breed, 2003; Brokaw et al., 2019), resulting in salinity-driven variability in ichthyoplankton diversity and taxonomic richness in the GOM, both of which were highest during low salinity conditions (Meinert et al., 2020). We did not examine Loop Current and Mississippi River influences directly, but variability in salinity was high in the GOM both within and between seasons in surface and upper epipelagic waters, and strongly influenced epipelagic larval fish assemblage structure.

While surface features (surface salinity and temperature) were found to define ichthyoplankton assemblages, mean salinity and temperature for depth-discrete nets were not identified by MRT as structuring environmental variables. The MRT technique is robust to collinearity (De’ath, 2002), and the best fit trees typically remove variables that are redundant. Temperature and depth are not independent, with temperature decreasing as depth increases. Similarly, salinity varies with season in upper epipelagic waters. Both temperature and salinity have limited variability within depth bins in mesopelagic waters. Depth and season are strong drivers of ichthyoplankton assemblage structure in the selected trees, so temperature and salinity may not contribute to additional definition of assemblage groups. Indeed, substituting temperature as a proxy for depth (or including temperature in addition to depth) in our final analyses did not result in appreciably different tree structures. The importance of surface feature variability in defining assemblage structure in both epipelagic and mesopelagic regions here is not surprising, considering that primary productivity is tightly linked to mesopelagic fish biomass in the open ocean (Irigoien et al., 2014). Primary productivity at the surface can impact the mesopelagic food web via vertical migration of fishes and zooplankton prey, as well as sinking particles (Kelly et al., 2019). Larval mesopelagic fishes are weak vertical migrators relative to adults (Sassa et al., 2002), so larvae distributed in mesopelagic waters are likely to benefit from the downward export of surface productivity rather than active migration to feed at the surface. The larvae of common mesopelagic fish species have been found to display different diel patterns of vertical distribution, but migration in the water column was constrained within ∼20–200 m (Sabatés, 2004). Diel periodicity was not identified by MRT as a structuring factor for GOM larval fish assemblages, likely due to our broad epipelagic depth bins precluding detection of any fine scale differences in diel distribution patterns.

Outside of the GOM, the structure of mesopelagic larval fish assemblages has been influenced by currents (Sassa et al., 2004; Olivar et al., 2010), availability of planktonic prey (Olivar et al., 2010; Koslow et al., 2014), oxygen minimum zones (Koslow et al., 2011; Dove et al., 2021), as well as bathymetry (Olivar et al., 2010). Distinct larval fish assemblages have been identified between oceanic and continental shelf waters in the GOM (Richards et al., 1993), and bottom topography can influence the abundance and distributions of mesopelagic fauna by modifying local hydrography (Fock et al., 2004; Sutton et al., 2008; De Forest and Drazen, 2009). Significantly higher abundance and diversity were identified in the offshore GOM when compared to the slope environment in other mesopelagic taxa (Frank et al., 2020). Here, MRT analyses did not distinguish between larval fish assemblages in slope and offshore sampling locales in the GOM. Ichthyoplankton in the GOM were largely concentrated in epipelagic waters, so it is possible that beyond the continental shelf, bathymetric features have minimal influence on the surface conditions that appear to drive assemblage structure for the early life history stages of fishes.

Epipelagic-associated larval fishes were not the primary focus of this study, but the broad distribution patterns found here were generally in agreement with previous studies of common epipelagic larval fishes in the GOM. In particular, we found that larval Engraulidae in oceanic GOM waters are largely associated with fall assemblage groups (September–October). Analyses of long time series of GOM ichthyoplankton on the continental shelf found that Engraulidae were strongly associated with late summer and fall sampling periods (August–October), suggesting that spawning peaks during summer in shallow inshore waters (Muhling et al., 2012; Lyczkowski-Shultz et al., 2013) before dispersing to slope and offshore GOM environments. This timing is further supported by the lack of detection of larval Engraulidae in offshore GOM ichthyoplankton samples during early summer (June–July) (Meinert et al., 2020), although we did occasionally detect low concentrations of larval Engraulidae in spring samples. Engraulidae have previously been associated with western GOM locales (Muhling et al., 2012; Lyczkowski-Shultz et al., 2013). While a weak western pattern can be seen in the epipelagic assemblage groups strongly associated with Engraulidae in this study, these assemblage groups were not defined by longitude in the MRT analysis. Additionally, a lack of sampling effort during fall in the eastern portion of the GOM sampling range precludes us from drawing any conclusions about spatial distributions of larval Engraulidae here.

While our analyses generally indicate broad GOM distribution for most larval assemblages, MRT analysis identified two small assemblage groups (Epi-B and Epi-H) in upper epipelagic waters that were defined by their geographic location (encompassing the westernmost sampling stations, delineated at approximately 90°W). In a study encompassing both shelf and slope ichthyoplankton samples in the GOM, a faunal discontinuity was identified at 87°W with breaks in species distribution and distinct eastern and western patterns of abundance between common species (Lyczkowski-Shultz et al., 2013). This longitudinal differentiation in distribution patterns was attributed to distinct bottom topography (i.e., DeSoto Canyon) driving local circulation, as well as differences in riverine influence (i.e., Mississippi River outflow) (Lyczkowski-Shultz et al., 2013). Single-species evaluations of the larvae of commercially important fisheries species in the GOM have found higher larval abundance in the western GOM (delineated at 89.25°W) when assessing regional patterns of distribution (Hanisko et al., 2017a,b). Eastern and western differentiation in larval fish distribution and abundance in the GOM is well documented, so our finding of a pair of small western assemblage groups certainly warrant further evaluation beyond our family-level analyses.

Assemblage analyses in this study were conducted at the family level, which is not uncommon for studies of ichthyoplankton communities (e.g., in the GOM alone: Richards et al., 1993; Carassou et al., 2012; Muhling et al., 2012; Meinert et al., 2020). It is possible that with improved taxonomic resolution, additional structure could have been detected among GOM larval fish communities. The most common assemblage groups in both epipelagic and mesopelagic regions were largely structured by depth and season, suggesting that ichthyoplankton within families may respond similarly to broad-scale habitat changes. However, many families were associated with multiple assemblage groups, both large and small and across seasons and depth regions, so it is certainly feasible that the finer scale assemblage structure we detected may be attributable to varying environmental tolerances among species within families. It is important to note that MRT assemblage groupings result from evaluations of multivariate abundance and are not defined by the presence or absence of prominent indicator families, but by variation in the relative proportions of abundance of all families in the assemblage. Families that occur across assemblages with disparate environmental characters may differ in proportion of abundance or in coincidence with other families, in addition to potentially reflecting species-specific differences in environmental response that are masked by poor taxonomic resolution. The oceanic GOM is known to support one of the most speciose ichthyofaunas in the world (Sutton et al., 2017), and the ubiquitous family Myctophidae is known to be particularly diverse, with the early life stages of approximately 50 species from 18 genera identified in the GOM (Richards, 2005). Outside of the GOM, both adult and larval myctophids have demonstrated species-specific or subfamily-specific patterns of vertical distribution (Watanabe et al., 1999; Sassa et al., 2002; Collins et al., 2008; Olivar et al., 2018). Within the GOM, it is possible that improved taxonomic resolution could reveal that larval myctophid species are occupying different portions of the water column and contributing to vertical assemblage differentiation. Horizontally, assemblages of adult myctophid species in the GOM were found to be well-mixed, with limited spatial differentiation despite variation in major hydrographic features (Milligan and Sutton, 2020). Adult myctophids are presumed to be less susceptible to local oceanographic conditions due to their large capacity for vertical migration, which may allow for passive advection at depth with varying currents, promoting more horizontal assemblage homogenization than would be expected with environment-driven structuring (Heino et al., 2015; Milligan and Sutton, 2020). Larval myctophids display limited vertical migration behavior compared to adults (Sassa et al., 2002) and ichthyoplankton are highly sensitive to the local environment (Kingsford, 1993; Brodeur et al., 2008; Hsieh et al., 2009), so it’s possible that larval myctophid distribution patterns are quite distinct from those of their more mobile adult counterparts. It remains unknown whether increased taxonomic resolution (i.e., species-specific assemblage analyses of Myctophidae and other speciose families such as Carangidae and Scombridae) would have resulted in the detection of additional partitioning of GOM larval assemblages or aided in the ecological interpretation of defined assemblages. Regardless, when the aim is to describe overall ichthyoplankton structure or assess changes in ichthyoplankton assemblages resulting from environmental perturbations, family-level analyses have been found to be sufficient (Hernandez et al., 2013).

One goal of this study was to fill a gap in knowledge about larval fish distributions beyond the epipelagic waters of the GOM. A lack of baseline data about the early life stages of mesopelagic fishes in the GOM was apparent following the DWHOS, limiting the efficacy of subsequent damage assessment efforts. Effects of the DWHOS on epipelagic-associated larval fish populations in the GOM have been equivocal, but studies to date largely reflect resilient populations. On the continental shelf, red snapper showed declines in larval condition in the years following DWHOS (Hernandez et al., 2016), while detrimental effects on Spanish mackerel larval condition were not detected (Ransom et al., 2016); declines in abundance were not detected in either species in relation to the timing of DWHOS. In oceanic waters, declines were detected in the larval abundance of common pelagic fish species (blackfin tuna, blue marlin, dolphinfish, sailfish) in the months following DWHOS, although the declines were within the scope of interannual variability resulting from shifts in oceanographic conditions that may be unrelated to DWHOS (Rooker et al., 2013). Mesozooplankton communities were found to recover rapidly after initial DWHOS impacts (Carassou et al., 2014), perhaps contributing to food web continuity and minimizing DWHOS declines for the largely zooplanktivorous ichthyoplankton communities in the GOM.

Larval fish in the epipelagic GOM are surveyed annually (spring and fall) as part of a long-term monitoring program (Southeast Area Monitoring and Assessment Program [SEAMAP], Lyczkowski-Shultz et al., 2013), but no such survey exists for the water column beyond 200 m. This study provides the first large-scale characterization of larval fish assemblages in the GOM that encompasses epipelagic, mesopelagic, and bathypelagic portions of the water column, providing a vital description of mesopelagic-associated larval fish distributions within the full scope of their vertical habitat distribution. While the brief sampling timeframe of this study provides only a snapshot of two seasons of larval fish distributions in the GOM, our findings will provide a useful reference point for GOM ichthyoplankton communities nonetheless. We found that the vast majority of ichthyoplankton (including both epipelagic- and mesopelagic-associated families) were located in the epipelagic portion of the water column. Mesopelagic waters were dominated by larvae from four dominant mesopelagic families, which made up distinct assemblage groups that were widespread and largely structured by depth and season, not by spatial distribution. The bathypelagic GOM consisted of very few larval fish overall with limited diversity. Together, these results suggest that long-term epipelagic-focused surveys such as SEAMAP may be sufficient for characterizing the distributions of most mesopelagic-associated larval fishes and evaluating the responses of these communities to environmental perturbations. To capture variation in abundances of dominant mesopelagic-associated larval fish families associated with mesopelagic waters, the spatial scope of mesopelagic sampling for future assessments can likely be reduced considering the horizontal ubiquity of most larval assemblage groups. Tremendous resources and effort are required to assess damages to the marine environment, particularly when accounting for the logistical challenges of sampling the deep pelagic, so concentrating future evaluation efforts on epipelagic waters and reduced mesopelagic spatial extent can aid in speeding assessment, restoration, and mitigation.

The data are publicly available through the Gulf of Mexico Research Initiative Information and Data Cooperative (GRIIDC) at https://data.gulfresearchinitiative.org (10.7266/n7-z8ry-6j20, 10.7266/n7-30jwaq82, 10.7266/n7-n2t3-aa19, 10.7266/n7-s18hwr95, 10.7266/n7-9vcj-nf90, 10.7266/n7-dp76-tz89, 10.7266/n7-j6rb-dt32, 10.7266/n7-7fxg-2980, 10.7266/n7-qecb-7x31, 10.7266/n7-rjqw-tf97, 10.7266/n7-60cgk731, and 10.7266/n7-1v14-fc70).

FH, VW, and CZ contributed to the conception and design of the study. VW and CZ organized the database. VW conducted the data analyses, prepared the figures, and wrote the first manuscript draft. All authors reviewed and edited the final draft.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.