Invertebrate larval distributions influenced by adult habitat distribution, larval behavior, and hydrodynamics in the retentive upwelling shadow of Monterey Bay, California, USA

One of the main goals of ecology is to quantify distributions and abundances of organisms and to elucidate the processes that influence population and community dynamics. Population dynamics are dri ven by the inputs to (birth and immigration) and outputs from (death and emigration) a population. The latter can be affected by numerous, often interacting, physical and biological processes. In marine systems, many species have a dispersing larval phase, which is a key determinant of population dynamics, espe-


INTRODUCTION
One of the main goals of ecology is to quantify distributions and abundances of organisms and to elucidate the processes that influence population and community dynamics. Population dynamics are dri -ven by the inputs to (birth and immigration) and outputs from (death and emigration) a population. The latter can be affected by numerous, often interacting, physical and biological processes. In marine systems, many species have a dispersing larval phase, which is a key determinant of population dynamics, espe-cially in recruitment-limited upwelling regions (Thor son 1950, Gaines & Roughgarden 1987. Larval phases add complexity to understanding population dynamics. They are difficult to track due to their small size (Levin 2006), are often patchily distributed, utilize a different habitat than the adult phase (i.e. pelagic for larvae and benthic for adult), and are therefore challenging to sample (Pineda et al. 2007). Thus, the suite of factors that influence marine invertebrate population dynamics is extensive and in cludes processes operating at multiple spatial and temporal scales in nearshore coastal habitats (Eckman 1996).
Since several of the windiest upwelling regions are thought to be recruitment-limited, meaning that larval supply is a key determinant of population dynamics, they provide useful study systems for elucidating patterns and processes affecting the distributions and abundance of larvae (Yoshioka 1982, Roughgarden et al. 1988). In the northern hemisphere, eastern boundary upwelling regions are characterized by strong northwesterly winds in the spring and summer, which, coupled with the Coriolis effect, transport surface water offshore (Chavez & Messié 2009). This displacement of nearshore surface waters draws cold, nutrient-rich waters to the surface into the nearshore environment to support the most productive marine ecosystems in the world (Pauly & Christensen 1995). Previous work suggests that persistent upwelling winds may rapidly transport larvae offshore, thereby limiting delivery to nearshore communities (Yoshioka 1982, Roughgarden et al. 1988), yet recent studies show that all stages of larvae from many taxa are found nearshore during upwelling conditions (Poulin et al. 2002a,b, Morgan et al. 2009a, Shanks & Shearman 2009, Bartilotti et al. 2014, Fisher et al. 2014, Bonicelli et al. 2016. Higher abundances of larvae nearshore are influenced by physical and behavioral mechanisms. For example, coastal topography in upwelling regions, such as headlands and bays, interacts with coastal circulation patterns to create nearshore retention zones, known as upwelling shadows (Graham et al. 1992, Wing et al. 1998, Mace & Morgan 2006, Morgan et al. 2011b. Upwelling shadows develop during spring to fall when cold, upwelled water is advected alongshore past an open bay or the lee of a coastal headland, generating cyclonic circulation inshore of the upwelling flow. Locally enhanced retention in the upwelling shadow results in surface warming due to solar heating and nutrient depletion due to phytoplankton uptake, leading to a region of warm, stratified, nutrient-depleted water (Graham & Lar -gier 1997, Roughan et al. 2005. Upwelling shadows are ecologically important, as they retain phytoplankton, holoplankton, and larvae of many species (Graham et al. 1992, Wing et al. 1998, Mace & Morgan 2006, Morgan & Fisher 2010, Morgan et al. 2011b, Ryan et al. 2014a,b, Harvey et al. 2018) and are associated with increased settlement (Wing et al. 1995, Morgan et al. 2009b).
Depth regulation by larvae can increase the likelihood that larvae can return to adult habitats in upwelling regimes (reviewed by Morgan 2014). Since currents vary throughout the water column, the time that larvae spend at different depths affects the direction and extent of larval transport (Queiroga & Blanton 2004). Thus, larvae employ different depthmodification 'strategies' over varying time scales. Larvae may occur at a specific depth throughout the duration of larval development or migrate vertically over the course of larval development, known as ontogenetic vertical migration. Larvae may also migrate vertically on shorter timescales, in relation to tidal or diel cycles, or in response to environmental changes, such as variation in light, temperature, salinity, or hydrostatic pressure (Queiroga & Blanton 2004, Morgan 2014, Epifanio & Cohen 2016. Nearshore species utilize one or combinations of these behaviors to mediate larval transport to suitable adult habitats. Larval migrations range from short to long distances, such that larvae may remain nearshore or in estuarine habitats, or migrate to the outer shelf, depending on the time spent in surface currents (Morgan 2014). For example, larvae may be transported away from adult habitats in seaward-flowing surface waters and return to adult habitats by descending into shoreward-flowing bottom currents (Morgan et al. 2009a, Morgan & Fisher 2010. Consequently, larval concentrations tend to increase closer to shore, even in recruitment-limited upwelling regions (Shanks & Shearman 2009, Nickols et al. 2013, Fisher et al. 2014, Morgan et al. 2009c. Thus, determining the effects of nearshore physical processes on larval supply is necessary for understanding recruitment dynamics, especially in regions characterized by strong coastal upwelling. Additionally, larval foraging behaviors can influence distributions (Woodson & McManus 2007). Phytoplankton accumulate near fronts and clines where larvae forage in high-density food patches (Woodson & McManus 2007, McManus et al. 2008, Ryan et al. 2010a. The upwelling shadow in Monterey Bay (California, USA) is an important area for the development of phytoplankton blooms (McManus et al. 2008), which are an essential re quirement for larval growth and survival (Lasker 1975). Therefore, larval presence may correspond with chlorophyll a (chl a) concentration, due to foraging behavior or passive accumulation.
Most studies of larval distributions in upwelling regions have been conducted during the peak upwelling season, but variation in seasonal oceanographic conditions (Pennington & Chavez 2000) and the persistence of the upwelling shadow in Monterey Bay have been observed (Gra ham & Largier 1997). Additionally, these studies have focused on a re latively narrow taxonomic scope (e.g. barnacles, crabs, or fishes). The goal of our study was to determine if interspecific differences in cross-shelf transport regulated by vertical positioning of larvae in the water column occur in Monterey Bay, as has been documented at other locations (Morgan et al. 2009a, Morgan & Fisher 2010, Nickols et al. 2013, Morgan 2014, and if these behaviors are consistent across differing oceanographic conditions. Variation in source waters entering Monterey Bay affected the composition of the larval assemblage, such that offshore species were more abundant in October, whereas nearshore species predominated in August (Satterthwaite et al. 2020). This study complements that of Satterthwaite et al. (2020) by (1) characterizing the spatial variation in physical characteristics of the water column, (2) assessing the cross-shore and depth distributions of larvae relative to the upwelling shadow, and (3) examining how these physical and biological patterns change with varied oceanographic conditions.

Field surveys
We conducted 2 cruises aboard the RV 'Rachel Carson' in northern Monterey Bay during late summer (12−16 August) and early fall (22−25 October) of 2013. We sampled 3 stations along an 8 km transect that extended from the inner bay, within the upwelling shadow, to the outer bay, just outside of the upwelling shadow (Fig. 1). We refer to cross-shore stations by their approximate distance from the nearest perpendicular onshore reference point (Aptos Creek: 36.969°N, 121.907°W): 4 km (inner bay), 8 km (mid-bay), and 12 km (outer bay). All sampling was conducted between early morning and midafternoon, and stations were sampled in the same order on each day (inner to mid-to outer bay). At each station, we adaptively sampled 3 depths relative to the chl a maximum layer (hereafter C max ). This layer of elevated chl a concentrations and the depth of peak concentrations were defined in real time by monitoring CTD environmental data during each cast. The depths of the C max peak ranged from 2 to 31 m over the course of the study.
At each station, plankton samples were collected above, within, and below the C max , using a gaspowered pump. We pumped 240 l of seawater per min for 10 min and sampled 2.4 m 3 of seawater per depth. Samples were filtered through a 115 µm mesh plankton net that was suspended over the side of the ship's rail, with the cod end submerged beneath the sea surface. Samples were preserved in 95% ethanol. Samples were split with a Folsom plankton splitter, and we identified the developmental stages of benthic invertebrate larvae using a dissecting microscope. Larvae initially identified to species were sub- The white contour (100 m isobath) in (b) defines the shelf break sequently grouped to class or order for data analysis due to low overall abundances. We then grouped larvae into early stages (e.g. early zoeae), late stages (e.g. late zoeae), and postlarvae (e.g. megalopae). In addition, we used high throughput DNA sequencing of the mitochondrial cytochrome-c-oxidase subunit-I gene (COI) to resolve taxa that were challenging to morphologically identify to species. Taxonomic names were assigned to operational taxonomic unit sequences by comparing them to custom reference sequence databases using the Basic Local Alignment Search Tool (BLAST). Descriptions of the complete molecular methods have been published (Harvey et al. 2018).
The hose from the plankton pump was attached to the outside of the CTD package frame. The profiling CTD package (Sea-Bird Scientific) measured temperature, salinity (Sea-Bird Scientific SBE 911Plus CTD), and chl a concentration (Wetlabs WetStar WS3S fluorometer) throughout the water column.
In addition to the CTD, the Monterey Bay Aquarium Research Institute (MBARI) autonomous underwater vehicle (AUV) 'Dorado' was deployed concurrently to measure physical, optical, and chemical oceanographic characteristics along the entire length of the transect during each day of sampling. Detailed AUV sampling methods have been previously published (Satterthwaite et al. 2020).

Data analysis
Cross-shore differences in larval assemblages were visualized using non-metric multidimensional scaling (NMDS) and tested using permutational multivariate ANOVA (PERMANOVA) ('adonis' function, 9999 permutations, R version 3.5.0). To determine which taxa contributed to cross-shore differences in larval assemblages, similarity percentages were calculated using SIMPER analyses (Clarke & Warwick 2001). NMDS, PERMANOVA, and SIMPER were run in R with the package 'vegan' (R version 3.5.0; Oksanen et al. 2018).
Relationships between abundances of larval taxa and cross-shore distance and environmental parameters (salinity, water temperature, depth, and chl a) were analyzed using general linear zero-inflated negative binomial models (with log link) or negative binomial models (when larval abundance did not have excess zeros), because larval abundances of each taxon were heavily skewed toward zero. In the case where there was one very high abundance value (e.g. Cirripedia postlarvae), we tested model re sults with and without the potential outlier and included the value if the model results were similar. Model selection was conducted using the Vuong test in R (Vuong 1989). Individual models were run for all environmental predictors within each month using the 'pscl' (Jackman 2020) and 'MASS' (Venables & Ripley 2002) packages in R. Since cross-shore distance was an ordered categorical variable, the model with cross-shore distance was compared to the null model and if significant, post hoc Tukey HSD tests were run to determine which cross-shore distances were significantly different from each other (package 'multcomp, ' Hothorn et al. 2008). Since depth, water temperature, salinity, and chl a concentration were continuous variables, significance results are reported from model output. Two metrics for depth preference were used, namely distance from the seafloor and distance from the surface, because larvae that associate with the bottom may show a strong relationship with distance from the seafloor, whereas larvae that associate with the surface may show a strong relationship with depth or distance from the surface. Nearly all larvae were associated with distance from the seafloor, so results and figures display distance from the seafloor as the depth metric.

Cross-shore variation in ocean conditions
The temperature profile of the August sampling period was characterized by strong thermal stratification along the entire transect, with warmer water occurring nearshore (Fig. 2). In August, the water temperature range was 11−15°C, and the thermocline occurred at about 10 m depth. In October, warmer water still occurred nearshore, but temperatures were cooler and spanned a smaller range (11− 13°C), and the water column was less stratified than in August (Fig. 2). In August, salinity was vertically stratified along the length of the entire transect, and values were higher than in October. October was characterized by low salinity water intruding from offshore, as evidenced by vertical isohalines extending from the bottom to the surface near the offshore station (Fig. 2). In August, chl a concentrations peaked in the middle (outer bay and mid bay) to bottom (inner bay) of the water column (~5−15 m depths) throughout the transect. In October, the area of high chl a concentration was near the surface (~0−8 m depth) and had lower maximum concentrations (Fig. 2).

Benthic invertebrate larval abundance
Eleven broad taxa were present in samples from northern Monterey Bay in August and October 2013 (Table 1). Of those, Bivalvia, Gastropoda, Echinoidea, Polychaeta, Cirripedia, Anomura, Gymnolaemata, Gebiidea, and Brachyura consisted of species generally found in the nearshore, intertidal zone, or were widespread, and Ophiuroidea and Brachiopoda consisted of species generally found deep and offshore or were widespread (Table 1).
Developmental stages of 5 nearshore taxa were more abundant nearshore (4 km) in August: Polychaeta late larvae and postlarvae, Bivalvia postlarvae, Gymnolaemata early larvae, Cirripedia early and postlarvae, and Brachyura early larvae (Table 2, Fig. 4a). A few of these taxa, namely Polychaeta late larvae, Polychaeta postlarvae, and Cirripedia postlarvae, were also found in higher abundances nearshore in October (Table 2, Fig. 4a). Offshore taxa, namely Brachiopoda and Ophiuroidea, occurred in higher abundances in the outer bay (12 km) in October, and Ophiuroidea early larvae displayed a similar pattern in August (Table 2, Fig. 4b).

Depth
Most developmental stages of nearshore taxa occurred in higher abundances close to the bottom in August (Table 2;  chiopoda and Ophiuroidea, showed a similar, although non-significant, pattern related to depth (Table 2, Fig. S1b).

Water types and chl a concentration
Water types were characterized by variation in temperature and salinity in August and October (Fig. 2). Of the nearshore taxa, Cirripedia postlarvae were found in higher abundances in warmer water in August, and the early larvae of Cirripedia and Ano mura trended toward this pattern (Table 2, Fig. S2a). Polychaeta late larvae and postlarvae, Cirripedia postlarvae, Brachyura early larvae, and Echinoidea postlarvae were found in higher abundances in relatively saltier water in October (Table 2, Fig. S3a). Larval sta ges of offshore taxa were related to water characteristics in both months. Specifically, larval stages of Brachio poda and Ophiuroidea were more likely to occur in cooler water temperatures in August ( Table 2, Fig. S2b), and in fresher water in October (Table 2, Fig. S3b).
Both nearshore and offshore taxa occurred in higher abundances in areas of high chl a concentration in August (Table 2, Fig. S4a,b). No taxa were positively related to chl a concentration in October (Table 2).

Larval supply and retention in the upwelling shadow
Larval abundances in Monterey Bay differed considerably cross-shore and with depth. Higher larval abundances of many nearshore taxa occurred in the upwelling shadow compared to farther offshore in August, consistent with previous studies in both Mon terey Bay (Graham et al. 1992, Graham & Largier 1997) and other upwelling shadows in the lee of coastal headlands (Wing et al. 1998, Roughan et al. 2005, Mace & Morgan 2006, Morgan & Fisher 2010, Morgan et al. 2011b, Hameed et al. 2018). This trend persisted in October despite different oceanic conditions. Conversely, offshore taxa, such as Brachiopoda and Ophiuroidea, occurred in the outer bay in both months, likely because adults released larvae from deeper shelf water that subsequently entered Monterey Bay. In addition, both nearshore and offshore taxa tended to occur close to the bottom in both months, yet this relationship was significant for more taxa in August than in October.
Depth distributions are likely explained by vertical positioning of larvae in the water column and food availability. Many larvae regulate depth in stratified currents, reducing offshore and alongshore transport by avoiding the surface Ekman layer and staying in slower bottom currents (Morgan et al. 2009a, Shanks & Shearman 2009, Morgan & Fisher 2010. Larvae of many abundant nearshore taxa collected during our study (i.e. Bivalvia, Polychaeta, Cirripedia, and Brachyura) tended to be uncommon in the surface layer and more abundant deeper in the water column during both months, a condition that may be instrumental in their nearshore retention.
In addition, larvae of many taxa were positively related to chl a concentration in August but not in October. Phytoplankton accumulation occurred at the thermocline in August, as has been previously described for northern Monterey Bay (McManus et al. 2008, Ryan et al. 2010b. Although weakly swimming larvae can passively concentrate at thermoclines, they typically swim across realistic thermoclines in the laboratory (Young 1995, Morgan 2020. Larvae, especially late stages of strongly-swimming taxa, may have aggregated to forage on high concentrations of phytoplankton (Lasker 1975, Woodson & McManus 2007. For example, early larval stages of Cirripedia, Polychaeta, Gastropoda, and Bivalvia eat phytoplankton (Fauchald & Jumars 1979, Turner et al. 2001, Vargas et al. 2006) and may have foraged in areas of high chl a concentration during August. Cirripedia postlarvae were also related to elevated chl a in August, but they are non-feeding (Strathmann 1985).
Since elevated chl a concentrations can be associated with convergence zones (Ryan et al. 2014b), and Cirripedia postlarvae (cyprids) swim upward to coun ter downwelling velocities characteristic of convergence zones (DiBacco et al. 2011), the combination of sinking and swimming behavior exhibited by Cirripedia postlarvae may have enhanced their concentration in the nearshore convergence zone in August. Barnacle (Shanks & Wright 1987, Pineda 1991 and fish postlarvae (Phelan et al. 2018) are transported shoreward in the convergence zone by internal waves, which commonly occur in Monterey Bay (Walter et al. 2012).
More taxa were abundant closer to the bottom in August, which could have been, in part, due to the subsurface location of high chl a concentrations. In August, chl a concentrations were higher and closer to the bottom (~8−15 m depth) than in October. This was especially true nearshore where larvae tended to be concentrated. Conversely, chl a values were much lower and chl a was concentrated near the surface (~0− 7 m depth) in October. Larvae may have occurred below surface chl a layers in October either because the phytoplankton concentrations were too low to stimulate larval feeding, as has been observed for anchovy larvae (Lasker 1975), or because larvae  Table 2. Larval abundances for each taxon and stage collected in northern Monterey Bay relative to depth, cross-shore distance, water type as best characterized by temperature in August and salinity in October, and chlorophyll a concentration. The number of samples (n) with non-0 abundance values for a given taxon within each sampling period are provided. Parentheses denote predominant species collected for each taxon, except in the case of Cirripedia and Brachyura (for full list of species, see Table 1). Significant relationships are denoted by bold font and asterisks (*p = 0.05−0.01, **p = 0.01−0.001, ***p < 0.001), trends are marked with a dagger ( †p = 0.11−0.5), and ns is not significant. Superscript '+' denotes a nonsignificant relationship for abundance but significant effect of the environmental variable on the presence or absence of taxa. 'Bottom' denotes a relationship to distance from the bottom, whereas 'near surface' denotes a relationship to depth (distance from surface). A positive relationship with chlorophyll a concentration is indicated by a 'Y'. 'Rare' indicates taxa that were present in only 3 or fewer samples, and 'absent' indicates taxa that were not detected of many species avoid the neuston (Morgan & Fisher 2010). Larvae of offshore taxa appeared to be associated with different water types in both August and October, as indicated by relationships to seawater temperature and salinity. In August, offshore taxa were associated with cold, upwelled, offshore water entering the study area. In October, offshore taxa were associated with the lower salinity water that was influenced by influx from the California Current. For example, Brachiopoda and Ophiuroidea larvae tended to be associated with cold offshore water Plots are grouped by (a) taxa that tended to be more abundant in August and are predominately nearshore taxa and (b) those that tended to be more abundant in October and are predominately offshore taxa. Taxa that were absent or rare across both sampling periods (3 or fewer samples with non-0 abundances within a sampling period) are not displayed. Refer to Table 2 for significance values and for absent and rare taxa  Table 3. Spearman rank correlation coefficients between total abundance of developmental stages for 10 larval taxa collected from northern Monterey Bay in 2013. Significant relationships are denoted by bold font and asterisks (*p = 0.05−0.01, **p = 0.01−0.001, ***p < 0.001), trends are denoted by non-bold font, and ns: not significant.
Taxa that were absent or rare (3 or fewer samples in a sampling period with non-0 abundances) are not displayed (see Table 2 Fig. S3b). Although larvae of a few nearshore taxa were associated with relatively warm water in August and relatively saline water in October, the pattern was not widespread because nearshore species may be retained in water types that become increasingly mixed over time, thereby obscuring their water type history. Most taxa with similar adult habitat (nearshore versus offshore) co-occurred in larval samples (Table 3, Fig. 5). Therefore, larvae that are released from similar habitats and have similar depth preferences may end up in water types with similar water type histories, as characterized by cross-shore distance, temperature, and salinity. For example, most nearshore taxa were correlated with one another. The nearshore taxa that were the most highly correlated were Polychaeta and Cirripedia, both of which were significantly associated with depth and occurred in significantly higher abundances inshore during both months. All offshore taxa were correlated with each other, found in higher abundances offshore, and significantly associated with depth. All larval stages (early larvae, late larvae, and postlarvae) of Cirripedia, Polychaeta, and Ophiuroidea were highly correlated with each other, suggesting that the various larval stages within a given taxon remained in similar water types throughout larval de velopment. Anomura early and late stages did not co-occur, which suggests that they may have undertaken ontogenetic vertical migrations be tween surface and bottom waters, al though it is still unknown whether they undergo ontogenetic vertical mi gration (Morgan & Fisher 2010).

Conclusions
The distribution of benthic invertebrate larvae within northern Monterey Bay appears to be strongly influenced by adult habitat distribution (nearshore versus offshore), vertical positioning of larvae in the water column (depth regulation and foraging), and changing oceanographic conditions (persistence of the upwelling shadow and the intrusion of offshore water). All of these factors interact to create associations between the composition of larval assemblages and water type characteristics, such that larvae of all stages from similar adult habitats cooccur in similar plankton samples. These results are consistent with other studies demonstrating that oceanographic processes coupled with larval behavior are important in ex plaining patterns of cross-shelf distributions and larval supply in upwelling regions (Poulin et al. 2002a,b, Shanks & Brink 2005, dos Santos et al. 2007, Morgan et al. 2009a,b, 2018, Shanks & Shearman 2009, Morgan & Fisher 2010, Nickols et al. 2013. Overall, larvae of nearshore taxa tended to be concentrated in the inner bay and near the bottom, throughout development. By remaining near the bottom, shoreward-flowing bottom currents may have retained most larvae nearshore (Morgan et al. 2009a, Morgan & Fisher 2010. Late stage larvae of offshore species, such as Ophiuroidea, were not present during upwelling conditions (August), but were observed in the offshore, less saline, bottom water during weakened upwelling conditions (October). Hence, stronger onshore currents during weakened upwelling may have kept the late stage larvae in the sampling domain (Satterthwaite et al. 2020). Offshore postlarvae occurred near the bottom in both months, suggesting that they descended to deeper, soft sediment habitats prior to settlement.
Most studies in upwelling systems have focused on specific nearshore species of barnacles and crabs. Our study is novel in that we sampled larvae originating from across the continental shelf and identi- Nodes represent larval taxa and stage and edges represent significant Spearman rank correlation coefficients between total abundances (dark purple indicates significant positive correlations and red indicates significant negative correlations). Variables that are more highly correlated appear closer together. Spearman rank correlation coefficients and associated significance values are reported in Table 3. Taxa that were absent or rare across both sampling periods (3 or fewer samples with non-0 abundances within a sampling period) are not displayed fied a greater array of taxa than has previously been studied. Larvae from adults living in similar habitats exhibited similar spatial distributions, and these patterns persisted among different ocean conditions. Therefore, it is important to consider not only physical oceanographic factors coupled with larval behavior, but also habitat characteristics of the benthic adults, such as the depth and geographic range, when defining factors that govern larval distributions in upwelling regions. These aspects of invertebrate life histories must be integrated when attempting to elucidate drivers of larval transport, supply, and population dynamics of benthic marine organisms. ment of crab postlarvae in a region of strong upwelling.