Nocturnal surveys of lined seahorses reveal increased densities and seasonal recruitment patterns

Abstract Although the nighttime ecology of organisms remains understudied, nocturnal surveys play an integral part in assessing fish assemblages and the selective forces shaping them. Eleuthera (Bahamas) contains an unusual population of lined seahorses (Hippocampus erectus) in an anchialine lake, possessing morphological characteristics distinct from those found in the ocean. Population surveys for seahorses and their potential predators were conducted at midnight and midday during wet and dry seasons, with belt transects perpendicular to the shoreline that increased in depth away from shore. Nocturnal surveys uncovered seahorse densities 259% higher than daytime transects on average. Sex ratios were consistently male‐biased, and the frequency of animals from different reproductive categories varied significantly by time of day, with gravid males observed around the clock but females and nongravid males observed more often at night. Spatial and seasonal recruitment was detected for the first time in this species, with an increase in juveniles detected in the shallow ends of transects during dry season surveys. Juvenile recruitment is poorly understood across syngnathid fishes, so the detection of early recruits at night has broad implications for this fish family. Seahorses from all reproductive categories were perched significantly higher in the water column during the night regardless of their depth or season. Predator densities followed a similar pattern with higher densities observed at night, indicating that elevated nocturnal perch height may be a response to predator presence. However, the selective agents driving these nocturnal behaviors have yet to be identified. Considering H. erectus is listed on the IUCN Red List as “Vulnerable,” the increase in nocturnal population size and the detection of juveniles has crucial implications for understanding their ecology, recruitment, and conservation.


| INTRODUC TI ON
As a diurnal species, humans have strongly biased our understanding of the population dynamics and behavioral patterns of many species by solely studying them during the day, which can have critical implications for their management in the wild. Until recently, we knew little about the biology of organisms at night, but with the advent of low-light trap cameras and other technology, specific behaviors, their selective factors, and the nocturnal activity cycle are coming into focus (Gaston, 2019). "Night" poses vastly different selective pressures across environments, which extend well beyond the simple lack of light (Gaynor et al., 2018). In shallow water systems, air and water temperatures fluctuate between day and night, leading to changes in thermoclines, wind, and water circulation patterns.
The nocturnal changes in climate can affect dissolved oxygen levels, cause planktonic shifts or alter broader community dynamics, and shift overall ecosystem and organism functioning. Such organismal-level effects have recently been exemplified by changes in gene activity under constant light versus intermittent light in Acroporid corals (Gaston, 2019;Rayner, 2003;Reeve, 1964;Reyes & Merino, 1991;Rosenberg et al., 2019;Sameoto, 1986).
Animals active at night display a range of unique behaviors associated with conducting the ordinary business of life in the absence of ambient light and a suite of behaviors specific to the nighttime hours (Nichols & Alexander, 2018). Most animals are clearly either nocturnal or diurnal, but in some groups, a range of activity cycles exist. Recent estimates suggest that 30% of vertebrates and 60% of invertebrates display nocturnality (Hölker et al., 2010). Alternatively, some animals are cathemeral, possessing flexible patterns of activity around the clock depending on food availability and season (Bennie et al., 2014;Colquhoun, 2006;Eppley & Donati, 2019). Others, like Merriam's kangaroo rats (Dipodomys merriami), are crepuscular and displayed their highest activity levels at twilight (Daly et al., 1992).
Lunar cycles are similarly influential upon the activity levels of animals active at night, with some mammals displaying suppressed activity during the full moon (Prugh & Golden, 2014). Among invertebrates, some taxa show predominantly nocturnal activity (like fireflies; Lewis, 2016), with others displaying a range of activity cycles.
Predation risk specifically has been shown to influence diel activity in many fish species, including minnows, guppies, carp, bluestripe grunt, gray snapper, and sea breams among others (Fraser et al., 2004;Hammerschlag et al., 2010;Metcalfe & Steele, 2001;Pettersson et al., 2001;Reebs, 2002). However, temperature variation, ontogeny, prey availability, nutritional status, lunar phase, and reproductive status have also been shown to be critical to the level of nocturnal activity in fish (Clark et al., 2009;Fraser et al., 1993;Gries et al., 1997;Metcalfe & Steele, 2001;Nagelkerken et al., 2000;Reebs, 2002;Reebs et al., 1984). The collection of research supporting a wide range of behaviors in fish throughout the day indicates the potential for many species to display diel changes in activity that have yet to be observed.
To effectively assess the conservation status and overall ecological niche of a species, it is vital to know when individuals are most active or most easily censused. Differences in the detectability of organisms between night and day play a role in accurate measures of animals' populations, not always with the expected consequence that animals are more detectable during the day. In bull trout (Salvelinus confluentus) and smallmouth bass (Micropterus dolomieu), nighttime visual counts and electrofishing census data estimated higher population sizes than during the day (Blackwell et al., 2017;Thurow & Schill, 1996). Similarly, the use of camera traps confirms that the presence of humans can reduce animal activity and detectability in multiple species (Swann & Perkins, 2014), and thus, daytime counts will often be lower than nighttime counts because animals can see researchers from further away and react accordingly.
An additional challenge in the detection of animals involves the morphological or behavioral adaptations that facilitate blending in with the landscape. In highly cryptic species like seahorses, daytime sampling can further bias against detection, leading to inappropriate management decisions (Aylesworth et al., 2017). The majority of seahorse species surveyed have been identified as diurnal, with anecdotal observations of nocturnality in the Pacific seahorse (Hippocampus ingens; Foster & Vincent, 2004), pot-bellied seahorse (H. abdominalis; Martin-Smith & Vincent, 2005;Paulin, 1992), and tiger-tail seahorse (H. comes; Foster & Vincent, 2004;Perante, 2002).
Although tiger-tail seahorses were never observed during the day across a 24-month study, seahorses were most visible at night high on top of the reef substrate (Perante, 2002). Researchers hypothesized that nocturnality in H. comes resulted from intense fishing pressures stemming from the aquaria and traditional medicine trades (Foster & Vincent, 2004). However, nocturnality or increased detectability at night may be more broadly present in this genus. If nocturnal activity is more common in seahorses, population monitoring might require surveys at night to more accurately census members of this CITES Appendix II in Appendix S1 trade-protected genus (https://www. cites.org/eng/app/appen dices.php).
Previous research has established that the Sweetings Pond marine lake (Eleuthera, The Bahamas) supports a range-restricted population of an IUCN Red List "Vulnerable" seahorse species (lined seahorse, Hippocampus erectus) of higher density than observed in nearly any other seahorse population globally (Correia et al., 2018;Rose et al., 2016). Surveys of this population have consistently indicated the presence of more males than females. This is unusual for seahorse populations, which typically display female-biased or equal sex ratios associated with their monogamous mating system (Foster & Vincent, 2004;. This high-density, male-biased population provides a unique system to assess interactions between the sexes and their behaviors over a 24-h diel cycle. Previous work included formal population sampling during the day  with animals observed engaging in cryptic behaviors in groups fewer than four seahorses (animals camouflage and tuck themselves away within the structure of the habitat) consistent with the behavior of other seahorse species (Freret-Meurer et al., 2017). Both formal nocturnal fish census using baited remote underwater video (BRUV) and informal nighttime observations in Sweetings Pond suggest that this population of H. erectus may be nocturnal, engaging in noncryptic behaviors, including congregating in large social groups of up to 14 animals, perching on the top of vegetation, and performing courtship displays (Mason, unpublished data;O'Brien et al., 2021).
This study provides the first comprehensive survey of wild seahorses at night, utilizing a paired transect design to document daytime and nighttime seahorse population counts. We investigated the selective factors that may influence their nocturnality by focusing on three main questions centered around the seahorses' behaviors and the presence of potential predators. The primary question centers on how population densities, sex ratios, and ontogenetic shifts in detectability or habitat usage, i.e., water depth and distance from shore, may vary between day and night. Secondly, we aimed to determine how seahorse behaviors might shift between midday and midnight, in terms of how high they are on vegetation in the water column (perch height), their overall body position, and holdfast preference. Finally, our study identifies the potential predator pool in this unique, closed system and how potential predator densities relate to seahorse abundance between daytime and nighttime surveys. Both octopus and crabs are occasional predators of seahorses in other systems, and we have witnessed (August 2017) one act of nocturnal predation of a male seahorse by a spider crab. As a result, we also document predator presence during our diurnal and nocturnal surveys to identify potential selective pressures acting on this population of seahorses that could cause the seahorses to shift their activity cycle from day to night.

| Experimental design
The study organism was the lined seahorse, Hippocampus erectus  (Foster & Vincent, 2004). Two deeper water transects were sampled both day and night during March sampling (identified as D transects in Figure 2). Due to sampling time constraints, we were unable to replicate these deeper water transects more broadly and excluded them from the analysis.

| Seahorse sampling
During daytime sampling, 1 m on each side of the tape was surveyed for seahorses, noting their depth in the water column, the location relative to the transect (in meters), side (right or left), and whether the animal was a juvenile, female, or male. Male reproductive status was identified in the field as either nongravid (not carrying embryos) or gravid (carrying embryos). In addition, we measured F I G U R E 1 Image taken by Shane Gross of Sweetings Pond lined seahorses at night. perch height as the distance from the sediment to the top of the seahorses' head in centimeters with a metric tape measure.
We photographed (using an Olympus TG-5) animals in situ to identify holdfast species, with a closeup of the left side of the head of the animal taken if possible, to allow the identification of individuals. Divers did not have any physical contact with animals during the day to reduce the likelihood of relocation, which has been observed in a broad range of seahorse species after interactions with divers and their cameras (De Brauwer et al., 2018Giglio et al., 2018;Harasti & Gladstone, 2013). During nighttime sampling, we took the same series of measurements for each animal and photographed them against a 1 cm grid background for size measurements.

| Potential predator sampling
To identify potential predators in the area surrounding each transect, we used two survey methods. The first included surveying the shorelines near the start of the transects during the daytime for spider crabs (Maguimithrax spinosissimus), Nassau grouper (Epinephelus

| Habitat assessment
In order to assess the benthic habitat available for use in the system, photographs of the benthic cover were taken 1 m above the tape at each meter mark with 40 cm in each frame. To assess seahorse habitat preference, we compared habitat availability from benthic photos with in situ photos of seahorses on specific holdfasts both day and night.

| Population density
Statistical analyses were performed using JMP 11.2 (JMP, 2019). All means are provided with standard error in parentheses. We estimated total seahorse density as the total number of seahorses observed in a transect divided by transect area (60 m 2 ). Densities and sex ratios were also estimated for the nearshore (the animals observed in first 0-15 m) and farshore sections (the deeper 15-30 m) of the transects to reflect patterns by depth and by distance to shoreline and thus potential predator pools. These densities were estimated by dividing seahorse counts by the area of a half transect (30 m 2 ). Seahorse density did not have a normal distribution (Shapiro-Wilk, W = 0.867, p = .0008) and had unequal variances by the time of day (Levene's, F 1,31 = 16.42, p < .0003) but not by season (Levene's, F 1,31 = 0.27, p = .605). To determine differences in total seahorse density between night and day on each transect, we used a Wilcoxon Signed Rank test because the variables were not parametric. Secondly, repeated measures MANOVAs were used to determine the relationship between density at different times of day by season and by location on transects. Adult density followed the same distribution as total density, and thus the same statistical approach was used.

| Reproduction
We estimated the sex ratio as the total number of adult males divided by the total number of adults in the population. The sex ratio was normally distributed (Shapiro-Wilk, W = 0.95998, p = .2581) and displayed equal variances (Levene's test, F 1,31 = 1.2032, p = .281, by season). We first used one-sample t-tests, with Bonferroni corrections applied for multiple tests following Holm (1979), to assess whether sex ratios deviated from a hypothesized 0.5 ratio. Paired t-tests investigated differences in sex ratios on each transect between day and night. Next, we investigated seasonal differences in adult sex ratio with repeated measures MANOVA for changes in sex ratio by the time of day (continuous variable) and season (grouping variable). Counts of animals in four reproductive categories (juvenile, female, gravid, and nongravid male) were delineated for both half and full transects and investigated for patterns in frequency shifts between night and day and between the time of day and season using a contingency table likelihood-ratio test.

| Habitat use and preference
To investigate seahorse body position relative to the substrate (n = 737 fish), we used in situ photos of seahorses on their holdfasts, categorizing them as 0 (laying flat on substrate or upside down, i.e., 90° to gravity), 0.5 (anywhere in between laying flat and vertically upwards, i.e., 90° to 180° tail to torso), and 1 (being completely upright relative to the substrate). Body posture was not normally distributed nor had equal variances and was investigated relative to the time of day and distance from shore using a general linear model. Coral Point Count (Kohler & Gill, 2006) was used to analyze the benthic cover's content at each meter along the 30-m transects from benthic photos. We identified a 30 × 30 cm region from each photo, assigned 30 random points, and then the identity of the item under the point was classified to species level if possible. The percentage cover of each habitat type per meter on each transect was then estimated from these count data. We used the Manly-Chesson Index to calculate seahorse holdfast preference, where α equals ∑ m i=1 ri pi , with m = number of benthic/holdfast categories used in the analysis, r i = the proportion of seahorses on a particular holdfast in either the night or the day, and p i = the proportion of that benthic type in the environment (Chesson, 1983;Manly et al., 1972). If α = 1/m, seahorses are using holdfasts relative to the frequency of that habitat component in the environment, but in cases where α < 1/m seahorses are avoiding that habitat component and α > 1/m, there is preferential usage.

| Potential predators
Differences between nighttime and daytime predator densities were investigated with a Wilcoxon Signed Rank test because daytime predator densities were not normal (Shapiro-Wilk, W = 0.557, p < .0001), with the relationship between predator density and both season and location on transect investigated with a repeated measures MANOVA. Finally, the relationship between the density of predators observed along the shoreline relative to the variables from perpendicular transects was estimated via linear regression.

| Population density
During night surveys, seahorses were most commonly attached to the top of the vegetation, highly visible to divers, and were reported in numbers 376% higher during the night (n = 681) compared with the same transects during the day (n = 143) when counts are pooled. Consequently, nighttime lined seahorse densities were significantly higher independent of the season (Wilcoxon Signed Rank test, S = 68.0, n = 15, p < .0001), with nighttime densities on average 259.3 (32.5)% higher than daytime ( Table 1). The highest density on a single nighttime transect was 1.67 seahorses m −2 . Overall, there was no effect of season on total seahorse density (Repeated measures MANOVA, F 1,4 = 2.39, p = .197) and no interaction effects (F 1,4 = 6.972, p = .0576). Because there was no seasonal effect detected in total density (density of adults and juveniles combined), we combined March and August samples for any further statistical analyses.
When investigated relative to distance from shore, overall seahorse density remained higher at night than during the day (Repeated measures MANOVA, F 1,28 = 5.041, p = .0328; Figure 3).
Seahorse density was higher closer to shore at night, a pattern not observed during the day (effect of the time of day: F 1,28 = 74.071, p < .0001, distance to shore: F 1,28 = 74.071, p < .0001, with significant interactions: F 1,28 = 70.27, p = .013). To investigate the relationship between adult density and location on the transect, juveniles were excluded, which eliminated both the effect of distance from shore and season (Repeated measures MANOVA, distance: F 1,28 = 2.299, p = .141; season: F 1,4 = 0.716, p = .445), but the significantly higher nocturnal fish density was maintained (Wilcoxon Signed Ranks Test, S = 68.0, n = 15, p < .0001; Table 1).
Second, we compared the sex ratio between treatments to investigate the effect of time of day and season. Sex ratio was significantly  Identifying the frequency of seahorses at different reproductive stages can provide insights into their population structure. Strong diel differences existed in the frequency of female, nongravid male, gravid male, and juvenile seahorses but only during the dry season (Contingency Table Analysis, X 2 = 30.133, df = 3, p < .0001; Figure 5). Diurnally, females, juveniles, and nongravid males each represented only 1.8%-2.5% of the total number of fish observed in the dry season, compared with 13%-28% of the population at night.
Gravid males were more consistently observed, although fewer were observed overall during the day compared with at night (8.7% of the population during the day and 26% of the population at night).
Juvenile abundance showed the strongest statistical association with season, with 82% of all juvenile seahorses observed in the study seen at night during the dry season (Contingency Table   Analysis, X 2 = 20.194, df = 1, p < .0001; Figure 5). Overall, males differed seasonally and by periods, with 62.5% of all males observed by this study surveyed at night in the dry season (Contingency Table   Analysis, X 2 = 4.226, df = 1, p = .040). Nongravid males were equally common day and night in the wet season but were much more common at night during the dry season (Contingency Table Analysis, For each time of day, there was no effect of season on the frequency of females in the population; they were generally uncommon during the day and more common at night (Contingency Table Analysis, X 2 = 1.732, df = 1, p = .188). Gravid males comprised 18%-25% of the population during the day, and 75%-82% of the population at night, with no differences by season (Contingency Table Analysis, X 2 = 0.754, df = 1, p = .3852). Although this appears contrary, what shifts is the abundance of nongravid males, with many more during the wet season and many fewer overall during the dry season; thus there is a higher ratio of gravid to nongravid males in the dry season than in the wet.

| Body size
Size frequency distributions for the various time of day and season combinations differed in range and shape, with mean body length smaller in the dry season than during the wet season ( Figure 6). The Table 1), differences were driven by the interaction between the time of day (X 2 = 5.051, df = 1, p = .025) and season (X 2 = 4.440, df = 1, p = .035), with no difference observed between males and females (X 2 = 0.01, df = 1, p = .99).

| Seahorses
We investigated seahorse body posture, perch height, and holdfast use during night and day transects. Animals of both sexes were more likely to be upright during the night and prone during the day. Body posture, ranging from 0 (lying prone relative to the substrate) to 1 (fully upright relative to the substrate) differed significantly by time of day (GLM, X 2 = 131.90, df = 5, p < .0001; X 2 = 21.887, df = 1, p < .0001) but not by distance from shore (GLM, X 2 = 2.797, df = 2, p = .247). This pattern did not vary by sex (Wilcoxon test, X 2 = 2.181, df = 2, p = .336).
When compared between day and night transects, mean perch height was significantly higher during the night than during the day with fish located 286% vertically higher on holdfasts than during the night. When we investigated perch height relative to the time of day (GLM, X 2 = 580.015, df = 34, p < .0001), fish were significantly higher in the vegetation at night (X 2 = 166.942, df = 30, p < .0001), but there was no difference in perch height by the fish's sex or distance from shore (both p < .05).
Holdfast preference differed significantly between the com- Algae was strongly preferred both daytime and nighttime, whereas sponges were solely preferred during the day but were avoided at night.
Predator densities on transects were significantly higher during the night (Wilcoxon Signed Rank test, S = 55.00, n = 15, p = .0007; Masonjones & Rose, 2019). In addition to the higher densities, we also observed behaviors that increased the visibility of seahorses, including a nocturnal shift of animals moving shallower in the water column paired with upright body postures. This change in behavior increased the detectability for all members of the population, revealing a robust population of juveniles during the dry season, which indicates that the recruitment of the population has yet to be measured by the daytime surveys. Our findings are important for the broader understanding of seahorse ecology, which we address below for each major result. Additionally, our system highlights potential sampling bias created by only surveying populations during the day, which has critical implications across a diverse set of species and ecosystems. The conclusions from this study indicate the need for sampling populations at times that best reflect the biology of the species studied to assess and manage them for conservation targets. looking for them, demonstrates consistency in the findings between the two studies. While the daytime densities recorded in Sweetings

| Population density
Pond are higher than the worldwide average seahorse density , the nighttime density from this study is one of the highest ever reported for seahorses, especially from randomly established transects that are not on artificial habitats (Correia et al., 2013;Simpson et al., 2019). The drastic difference between the day and night densities did not change between the dry and wet seasons suggesting this phenomenon is not solely due to seasonal recruitment.
This study is one of the few to document nocturnality in seahorses and therefore, the dramatic increase in nighttime den-

| Sex ratios
The ratio of adult males to adult females (the adult sex ratio, ASR) is often used as an assessment of the mating system of a species.

| Females
We observed far more females at night, which suggests that females are either more cryptic than males during the day or are migrating from the depths to the shallows at night. Because a more even sex ratio with more females was observed during exhaustive searches of this region in the dry season, there is support for the hypothesis that females spend their day burrowing in the algae, resulting in more cryptic behaviors than males. However, an equally plausible explanation for the lack of females during daytime sampling is that they move to deeper habitats during the day. Although a small sample size, the two deeper transects not included in our analysis had only large females (n = 6, mean body size 95.83 (8.88) mm), so there is the potential that females have different daytime habitat preferences for deeper water or are more mobile than males during the day ( Figure S1a,b). Evidence from other studies suggests that seahorse females often possess larger home ranges and thus move more than males, which has been seen in the tiger-tail seahorse, H. comes (Perante, 2002), White's seahorse, H. whitei (Foster & Vincent, 2004;Harasti et al., 2014aHarasti et al., , 2014bVincent et al., 2005;Vincent & Sadler, 1995), short-snouted seahorse, H. breviceps (Moreau & Vincent, 2004), and other species (Foster & Vincent, 2004;Harasti et al., 2014aHarasti et al., , 2014bVincent et al., 2005;Vincent & Sadler, 1995).
In addition, H. reidi females were found to be the most active group whereas gravid males were the least active (Freret-Meurer et al., 2012), mirroring our observation that females and nongravid males were not observed as often during the day, whereas gravid males were the most commonly seen group.

| Body size shifts reflect juvenile recruitment
Across multiple lines of evidence, one of our most important discoveries was the abundance of newborns and juveniles at night during the dry season. This was confirmed by our 11% decrease in average body size in nighttime samples when localized to the shallowest parts of transects close to shore ( Figure S1c). indicates that younger adults, other than newborns are also present more often at night than during the day and that the population is composed of generally smaller animals during the dry season.

The greater representation of pregnant males and juveniles in
Sweetings Pond H. erectus both occurring during the dry months is also seen in H. guttulatus when both juveniles and pregnant males peaked during the same month of June (Gristina et al., 2017). Little information exists on juvenile recruitment patterns in seahorses and juveniles are rarely observed before their settlement as midsized juveniles into the habitats where adults are observed (Curtis et al., 2017;Foster & Vincent, 2004). It is also possible that juveniles in other systems are rarely observed because they are often dispersed by water currents from adult areas via rafting, as seen in a range of seahorse species (Bertola et al., 2020;Boehm et al., 2013;Luzzatto et al., 2013;Teske et al., 2007). Because it is an enclosed body of water with minimal hydrodynamical movement, Sweetings Pond seahorse juveniles may not engage in rafting in this system, but this has not yet been assessed. In the only other study of seahorses at night (H. comes on reefs in the Philippines), animals were observed up at the top of the reef crest at night and were not visible during the day across the 24-month study (n = 32 seahorses; Perante, 2002). In that study, nocturnality was hypothesized to be driven by human fishing practices, but it is possible based on the present study that this nocturnal vertical migration might be more common in seahorses than previously thought. However, crypsis during the day could suggest a predator avoidance behavior, and so, understanding the potential predator pool of the system is important. It is also possible that seahorses move up in the water column to better exploit ambient light and continue feeding, given observations of full guts both during the day and at night (Masonjones et al., unpublished). In European minnows, nighttime feeding is the primary foraging mode except in fish with malnutrition, so nocturnal foraging may be critical for fish (Metcalfe & Steele, 2001).

| Behavioral changes in habitat use
This work is the first to report holdfast preferences in this unique high seahorse-density Bahamian habitat. Of the potential holdfasts in the pond's northern region, we found that macroalgae were both the most abundant and most strongly preferred holdfast independent of time of day, whereas sponges were the second most preferred holdfast during the day but were rarely used at night. Seahorses using the sponges during the day were typically darkened in coloration and exhibited thanatosis behaviors, which could be due to increased predation. In H. whitei, adults prefer sponges and soft coral as holdfasts during the day, some of the taller holdfasts in the areas where they are found, while juveniles prefer gorgonians (Harasti et al., 2014a(Harasti et al., , 2014b. In Sweetings Pond, juveniles were found in the shallows where the first 15 m of transects closest to shore had the greatest amount of macroalgae. The preference for macroalgae parallels those found in H. comes where juveniles were most abundant on macroalgae, while adults were found equally on macroalgae and corals (Morgan & Vincent, 2007). Gristina et al. (2017)

| Potential predators
Due to increases in their population densities at night, the potential exists for Caribbean reef octopus (Octopus briareus) and West Indian spider crabs (Maguimithrax spinosissimus) to be nocturnal seahorse predators in Sweetings Pond, and thus impact seahorse behavior. In addition, large introduced Nassau grouper (Epinephelus striatus) are also found in the system and have been observed during the day. Given their gape size, the lack of many food species fish, and the abundance of seahorses, they are also candidate predators (Aronson, 1985). Octopus range in density across space and seasons (Aronson, 1985;O'Brien et al., 2020) and have been demonstrated in Australian systems to be active seahorse predators (Harasti et al., 2014a(Harasti et al., , 2014b. Seahorses and pipefishes are included in the diets of a wide range of species, including grouper, octopus, and crabs (Kleiber et al., 2011). Although the specific species found in the pond have never been described to actively hunt H. erectus, it is conceivable they could. Spider crabs (Maguimithrax spinosissimus) have been observed feeding on seahorses at the site (personal observation); however, no evidence of predatory behavior of crabs on seahorses was observed in this study, with crabs and seahorses regularly in close physical contact. These crabs feed predominantly at night, which corresponds to seahorse peak conspicuousness, where they move to the top of the benthic canopy.
A second potential explanation for the Sweetings Pond seahorses laying flat during the day could be in response to avian predators.
Surveys of predators indicate that seahorse and pipefish populations found in higher abundances have greater opportunistic bird predation (Kleiber et al., 2011). The high-density population in Sweetings Pond and the isolation of the population from oceanic habitats could potentially serve as an optimal feeding ground for avian predators, leading to intense selective pressures on the seahorses' perch height and the increased daytime crypsis in the pond. Given that seahorses are more accessible along the shoreline in shallow water, we would expect that if this were a factor the nearshore halves of the transects would exhibit fish lower in the macroalgal canopy during the day, with more prone body postures. However, there was no difference with distance to shore either at night or during the day for these two behavioral variables. There could be a selective advantage for the seahorses adapting thanatosis behaviors, regardless of the depth, because of the limited turbidity and pristine water clarity in the isolated pond. of this subpopulation. Although it is highly unlikely that males receive eggs from more than one female in a breeding event based on the morphological restrictions posed by the seahorse ovary (Sogabe et al., 2008), given the large number of males in the population, it is possible that fish may switch mates between breeding events.

| Future directions for Sweetings Pond
However, recent computational approaches suggest, that there is an interplay between female dispersion in the landscape (as we observed during the day) and male-biased populations that support the evolution and maintenance of monogamy (Gomes et al., 2018). As a result, Sweetings Pond is an excellent model in which to investigate the relationship between a population's social system and the potential for sexual selection to shape reproductive behaviors and mating system. This is a particularly important question given the Vulnerable Red List status of this species and the implications that reproduction has on population estimates and conservation management strategies (Pollom, 2017).

| CON CLUS IONS
The implications from our findings and the lessons learned from our nocturnal surveys provided valuable insight into the need for a more critical assessment of survey methods used in other systems. Reflecting on our study, the results indicated when and how you census your population matters a lot. An exploratory night dive made it abundantly clear why this population, with an unusual sex ratio, required more investigation and led to the discovery of these unique nocturnal behaviors and extreme diel density fluctuations.
Nocturnal "blindness" could be one of the largest overlooked components of sampling designs and it is supported by that the fact that we know very little about the nocturnal biology for the majority of species, particularly those requiring conservation measures. In some cases, this discrepancy is because the species itself is understudied, but in many cases, it is often due to our diurnal bias or because the ease of working during the day or sampling restrictions drove the science and limited the experimental design.
Evidence from across systems supports the assertion that population surveys need to include nighttime sampling. Studies of other fish species using acoustic techniques and trawling to assess fish populations have had mixed results sampling during the day compared with at night, with some species or water bodies with higher densities during the day and others higher at night (Draštík et al., 2009;Yule et al., 2007). Additionally, some studies indicate the importance of surveying fish populations at night for accurate counts, often because of differences in fish habitat use between day and night and the challenges of surveying fish in more complex habitats like coral reefs (Fox & Bellwood, 2011). An advancement in technology will also likely provide more opportunities to survey nocturnal populations as low-light cameras and underwater drones.
For example, the recent advent of techniques like BRUV's (baited remote underwater videos) underwater and camera traps for terrestrial communities that can capture images in low-light settings at night have helped to illustrate just how important nocturnal dynamics are within species and across habitats (Harvey et al., 2012(Harvey et al., , 2021Swann & Perkins, 2014). Further magnifying the need to survey at night, global increases in daytime human activities are shifting some animal populations to a more nocturnal lifestyle (Gaynor et al., 2018). This shift has led to key ecosystem effects like prey changes to nocturnal species, alterations in foraging patterns, and changing competition regimes, all factors that dramatically affect both the functioning of these ecosystems and the management of populations within them.
Duncan A. O'Brien: Investigation (supporting); writing -review and editing (supporting). In-kind support provided by the Leon Levy Native Plant Preserve and the Bahamas National Trust (staff, resources). Special thanks to Dr. Ethan Freid, who was the driving force in the initial discovery of nocturnal seahorses in the pond.

CO N FLI C T O F I NTE R E S T
The authors declare that there are no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data will be archived and made available on Dryad upon acceptance of the manuscript at the following location: DOI https://doi. org/10.5061/dryad.z612j m6g7. The temporary link to the data set is at https://datad ryad.org/stash/ share/ 0EJDa Wjl_182k7 xc_SHEEj lCzXf HqKtS 2A7nK az4BtA.