Reproductive cycles in tropical intertidal gastropods are timed around tidal amplitude cycles

Abstract Reproduction in iteroparous marine organisms is often timed with abiotic cycles and may follow lunar, tidal amplitude, or daily cycles. Among intertidal marine invertebrates, decapods are well known to time larval release to coincide with large amplitude nighttime tides, which minimizes the risk of predation. Such bimonthly cycles have been reported for few other intertidal invertebrates. We documented the reproduction of 6 gastropod species from Panama to determine whether they demonstrate reproductive cycles, whether these cycles follow a 2‐week cycle, and whether cycles are timed so that larval release occurs during large amplitude tides. Two of the species (Crepidula cf. marginalis and Nerita scabricosta) showed nonuniform reproduction, but without clear peaks in timing relative to tidal or lunar cycles. The other 4 species show clear peaks in reproduction occurring every 2 weeks. In 3 of these species (Cerithideopsis carlifornica var. valida, Littoraria variegata, and Natica chemnitzi), hatching occurred within 4 days of the maximum amplitude tides. Siphonaria palmata exhibit strong cycles, but reproduction occurred during the neap tides. Strong differences in the intensity of reproduction of Cerithideopsis carlifornica, and in particular, Littoraria variegata, between the larger and smaller spring tides of a lunar month indicate that these species time reproduction with the tidal amplitude cycle rather than the lunar cycle. For those species that reproduce during both the wet and dry seasons, we found that reproductive timing did not differ between seasons despite strong differences in temperature and precipitation. Overall, we found that most (4/6) species have strong reproductive cycles synchronized with the tidal amplitude cycle and that seasonal differences in abiotic factors do not alter these cycles.


| INTRODUCTION
Evolutionary ecological theory predicts that reproductive investment responds to diverse ecological drivers so that it is timed and partitioned to optimize offspring survival. In many marine and terrestrial organisms, reproduction is timed to coincide with seasonal increases in resource availability and thus to maximize offspring growth and survival. For example, many birds time reproduction to coincide with maximum seasonal abundances of insects (Thomas, Blondel, Perret, Lambrechts, & Speakman, 2001), or zooplankton and fishes (Durant, Anker-Nilssen, & Stenseth, 2003;Regehr & Montevecchi, 1997;Suryan, Irons, Brown, Jodice, & Roby, 2006). Reproduction may also be timed to coincide with the availability of ephemeral habitats necessary for embryonic or larval development. Insects and amphibians that rely on seasonally available pools of water for larval development time reproduction with the wettest times of the year (Shine & Brown, 2008), and parasitoids time reproduction to coincide with the availability of hosts (Hood et al., 2015). Reproduction may also be timed to minimize mortality due to harsh environmental conditions, or predation on mothers or offspring (Christy, 2011). For example, many marine organisms release larvae or spawn during the night to avoid visual predators (Christy, 2011). It has been suggested that predation may drive the evolution of synchrony of emergence of cicadas, annual mass spawnings of corals, and synchronized reproduction of sea turtles as predator-swamping strategies (Ims, 1990;Williams, Smith, & Stephen, 1993).
Among marine organisms, intertidal crabs provide the bestunderstood example of the adaptive optimization of reproductive timing (Christy, 2003(Christy, , 2011Christy & Stancyk, 1982;Morgan & Christy, 1994, 1995Skov et al., 2005). More than 80 species of crabs have been shown to time larval release to coincide with times when predation by visual predators is lowest (i.e., nighttime) and when tidal currents are most likely to quickly move larvae away from shallow water (Christy, 2011). In general, tidal currents are greatest when the tides have the largest amplitude. As tidal amplitude follows a lunar or semilunar cycle (depending on the location), this produces a lunar or semilunar cycle in the timing of larval release (Christy, 2003(Christy, , 2011Christy & Stancyk, 1982). Describing biological cycles as "lunar" or "semilunar" simply reflects the duration of the cycle and should not be read to imply a causal relationship with the lunar phases, as reproductive cycles for a number of crabs show precise matching to details of the tidal amplitude cycle, adjusting reproduction so that larvae are released during the tides with the largest amplitude in the month, regardless of whether these occur during the full or new moon (Skov et al., 2005). In addition, some species can adjust the timing of mating to compensate for the temperature dependence of development and ensure that hatching occurs during the optimal time regardless of temperature (Kerr, 2015;Kerr, Christy, Collin, & Guichard, 2012;Kerr et al., 2014). Overall, intertidal crabs follow the same cycles regardless of habitat or taxonomic group, while the exceptions are explained by particular features of the larvae thought to reduce predation in other ways, or constraints on the ability to release larvae at the optimal time imposed by the habitat (Morgan, 1996a,b;Morgan & Christy, 1994, 1995.
Semilunar or lunar cycles of various kinds have been reported only occasionally for other groups of iteroparous marine invertebrates.
Those species for which cycles have been commonly reported are primarily subtidal and include lunar cycles in sea urchins, where synchronized reproductive timing is thought to ensure fertilization success (e.g., Lessios, 1991;Mercier, Ycaza, & Hamel, 2007;Pearse, 1975) and lunar release of larvae from brooding corals (e.g., Richmond & Jokiel, 1984;Tanner, 1996). It is surprising that semilunar or lunar reproductive cycles have been reported for so few intertidal invertebrates, because avoidance of visual predators by offshore advection of larvae could benefit species other than crabs. In addition, the lives of intertidal animals revolve around the progression of the tides. The tidal amplitude and the time of the tide relative to the diel cycle affect all environmental variables in this habitat, thus influencing activities including foraging, access to mates, and molting of a wide range of organisms (Weisberg, Whalen, & Lotrich, 1981;Iwasaki, 1995;Tran et al., 2011;reviewed in Naylor, 1999). Therefore, it seems intuitively likely that the reproduction of intertidal invertebrates could be shaped by factors other than offshore advection of larvae that could also result in reproductive cycles synchronized with the tides.
We surveyed the reproductive timing of 6 species of tropical intertidal gastropods with diverse natural histories to determine whether reproduction is synchronized among individuals in the population and whether this synchrony follows a regular cycle relative to the lunar or tidal amplitude (semilunar) cycle. We aimed to answer the following questions for each species: (1) Does reproduction follow a synchronized monthly or bimonthly cycle? (2) If reproduction is periodic, is the cycle more closely related to the lunar or tidal amplitude cycle? (3) If present, are reproductive cycles timed to release larvae into the water column during spring tides, as they are in crabs? and (4) Is the timing of reproductive cycles altered by seasonal changes in environmental conditions. Cycles with timing similar to those reported for crabs would be consistent with the idea that the same risk factors drive reproductive timing in gastropods. Cycles with different timing would point to different selective factors shaping reproductive synchrony. A lack of synchrony would suggest that the timing of egg deposition and/or hatching relative to lunar or tidal amplitude cycles are not selectively important.

| METHODS
The 6 snail species were studied at sites where they are abundant in the intertidal of the Bay of Panama. Two species, the nerite Nerita scabricosta and pulmonate limpet Siphonaria palmata, were studied in the rocky intertidal of Culebra Island, near Panama City at the entrance to the Panama Canal (Figure 1a,c). Four other species (Cerithideopsis californica, Natica chemnitzi, Crepidula cf. marginalis, and Littoraria variegata) were studied at Playa Venado, near the town of Veracruz, 8 km from Culebra Island (Figure 1a,b; Collin & Ochoa, 2016). Tides in the Bay of Panama are semidiurnal with an average amplitude of 3.9 m and a maximum amplitude of 6.9 m. The largest amplitude, or spring, tides of the 14.8-day amplitude cycle occur consistently 2-3 days after the full and new moons, but the largest spring tides of the 29.5day lunar cycle may occur after either a full or a new moon. We used the nighttime ebb amplitude, which was calculated as the difference between the height of the nighttime high tide and the height of the following low tide. In general, high tides on large amplitude (spring) tides occur between 5:00 and 7:00 in the morning and evening, while the high tides on the small amplitude (neap) tides occur around noon and midnight. The tight relationship between the tidal amplitude and the time of day of the tides means that a number of abiotic factors vary in concert with the amplitude cycle, making it difficult to attribute causality of semilunar cycles to any specific abiotic factor without employing experimental manipulations.
The Bay of Panama experiences distinct seasonal variation in precipitation, wind patterns, and upwelling of cold waters causing seasonal changes in temperature and salinity. The dry season, when upwelling occurs, generally extends from mid-December to early May.
Late May to early December is characterized by frequent rain and relatively warm and constant sea surface temperature (Collin & Ochoa, 2016;Robertson & Collin, 2015). These seasonal changes in abiotic factors affect the timing and intensity of reproduction of several organisms. Hence, we compared the timing of reproduction during the wet and dry season to explore the potential for seasonal changes in abiotic factors to alter timing.

| Study species and methods
To determine whether reproductive events of these snail species follow a cycle, we surveyed the abundance of egg masses, or the timing of egg deposition or release of larvae, depending on the species. Each species was studied for several months, and for four of these species, this was done in parallel with a study of seasonal differences in egg mass abundance and offspring size (Collin & Ochoa, 2016). We determined the timing of reproductive events relative to the tidal amplitude and lunar cycles in different ways based on the idiosyncrasies of the natural history of each species.

| Littoraria variegata
Littoraria variegata live on the trunks and lower branches of mangrove trees in the mixed mangrove forest (Avicenia germinans, Avicenia bicolor, Laguncularia racemosa and Rhizophora sp.). They release pelagic egg capsules each of which surrounds a single egg. Nine to 36 (mean 19.5) female Littoraria were collected 1-2 times a week. They were maintained in the laboratory, in the dark, in a temperature-controlled incubator at 28°C in 50-ml Falcon tubes half-filled with seawater.
Every day the water was emptied from the tubes, and the presence or absence of pelagic egg capsules was recorded. The % of females releasing capsules each day was used in the analyses. The data from this species are similar to those generated by studies designed to determine whether cycles can be maintained in the absence of environmental cues (e.g., Forward, 1987). Data were collected between October 2013 and August 2014.

| Crepidula cf. marginalis
Broods of the slipper limpet Crepidula cf. marginalis were studied on the beach at Playa Venado, which is a mix of muddy sand and rock rubble. Crepidula cf. marginalis are abundant on small rocks, at the same tidal height as the moonsnail egg masses (see below). Slipper limpets brood egg masses between their neck, propodium, and the substrate. Every Monday, Wednesday, and Friday 10-20 large (>9 mm) Crepidula cf. marginalis that were paired with a male were removed from the rocks, and the presence or absence of a brood was recorded. Mature broods turn brown when they are within 1-2 days of hatching, and the presence of brown broods was scored as a proxy for hatching and used in the statistical analyses. These snails were surveyed between November 2013 and August 2014.

| Cerithideopsis californica
Cerithideopsis californica var. valida (referred to subsequently as C. californica) were studied in the high intertidal in the mixed mangrove forest (Avicenia germinans, Avicenia bicolor, Laguncularia racemosa, and Rhizophora sp.). Cerithideopsis californica inhabit the mudflat and the base of the mangrove trees and deposit gelatinous egg strings on the surface of the mud (Miura, Frankel, & Torchin, 2011). Three 1 m × 1 m permanent quadrats were installed on the mud at the base of mangrove trees. Every Monday, Wednesday, and Friday, we marked each new egg mass with labeled sticks and removed the sticks when the masses had hatched. The number of sticks deployed and retrieved each day was recorded. Deployment and removal of sticks were used as proxies for egg deposition and hatching, respectively, and used in the statistical analyses. These actual events occurred <2 days prior to the reported event. Data were collected between November 2013 and August 2014.

| Nerita scabricosta
Nerita scabricosta deposit small blister-shaped capsules in small tide pools in the high rocky intertidal of Isla Culebra (Collin, Roof, & Spangler, 2016), slightly above those in which S. palmata deposit egg masses. Photographs of two quadrats in 15 pools were taken every Monday, Wednesday, and Friday. The number of bright white, newly deposited egg capsules was counted from each photograph to record deposition. Newly hatched capsules were counted as those where a circular scar occupied a location that had a capsule the previous day.
The actual events occurred <2 days prior to the reported proxy for the event. Data were collected between August and October 2014.

| Natica chemnitzi
Egg masses of the moonsnail Natica chemnitzi were studied on the beach at Playa Venado, which is a mix of muddy sand and rock rubble.
Natica chemnitzi deposit sand-covered egg collars typical of moonsnails on top of the sand or muddy sand in the high midintertidal (Collin & Ochoa, 2016). Every Monday, Wednesday, and Friday, we placed three 100 m × 2 m band transects perpendicular to the coast in the zone where Natica egg masses occur. As masses could not be followed individually, but clear cycles were present in this species, we predicted peak hatching times for each tidal amplitude cycle and analyzed them as reproductive events. The peak in hatching should occur approximately n/2 days after the peak number of masses and n days after the peak in egg deposition, where n is the time to hatching.
All the masses in each transect were counted between January 2013 and August 2014.

| Siphonaria palmata
Siphonaria palmata deposit gelatinous masses in small tidal pools in rocks in the high midintertidal of Isla Culebra. Every Monday, Wednesday, and Friday, we counted the number of egg masses in twenty 7 cm × 7 cm quadrats. As the loose ribbons could not be tracked individually, but clear cycles were present in this species, we predicted peak hatching times for each tidal amplitude cycle and analyzed them as reproductive events. The peak in hatching should occur approximately n/2 days after the peak number of masses and n days after the peak in egg deposition, where n is the time to hatching. Data were collected between September 2015 and June 2016.

| Statistics
Most previous studies of invertebrate reproductive cycles plot events relative to time and draw conclusions by visual inspection of the tidal and lunar phases relative to the reported events (see section 4). These graphs are intuitively easy to interpret, but they make detection of small differences difficult, and conclusions drawn this way are limited to indicating which general part of the cycle seems to fit with peaks in the data. We used circular statistics to quantify how reproduction was clustered relative to the semilunar tidal amplitude cycle and the lunar cycle. Reproductive events were plotted on circular plots as an angle in degrees relative to the tidal amplitude or lunar cycle in R using the package GGplot 2 (Wickham, 2009). Each day in the cycle of interest was converted to degrees by dividing 360° by the number of days in the cycle. If reproductive peaks occur at the same point in the cycle across multiple cycles, the points will appear in a cluster covering only a few degrees of the circle. We visually inspected the data plotted on circular graphs representing the ~15-day tidal amplitude cycle, the ~29-day lunar cycle, and the ~monthly cycle of larger and smaller spring tides (two tidal amplitude cycles) with the day of the maximum amplitude nocturnal tide, the day of the full moon, and the day of the maximum amplitude tide of a month (largest spring tide) at 0°, respectively.
Because a 2-week cycle was evident for most species, we used this tidal amplitude cycle to calculate descriptive circular statistics and tested for clustering of events using the CircStats2010d package in MATLAB (Berens, 2009). Mean angle, resultant vector length (R, a measure of dispersion or synchrony that ranges from 0, randomly scattered, to 1, identical values), and circular standard deviation were used to quantify the timing and synchrony of events relative to the tidal amplitude cycle. We used the Omnibus test for circular uniformity to determine whether events were unevenly distributed around the circle for each of the 6 species (Berens, 2009;Zar, 2010). A p-value of <.05 indicates a significant difference from a uniform distribution of events.
For those species with moderate to high synchrony (R > 0.4), we used a v-test to determine whether reproductive events occurred near the timing of the maximum tides. The v-test for circular uniformity tests between the null hypothesis of a uniform distribution and an alternate hypothesis of a nonuniform distribution with a known mean direction.
To test whether events occurred near the large amplitude tides, we tested an alternate hypothesis with an angle of 0°.

| Lunar versus tidal amplitude cycles
The largest spring tides of a lunar month are sometimes associated with the new moon and sometimes with the full moon. To determine whether snails with semilunar cycles time reproduction with the phases of the moon or with the size of the spring tides, we plotted the timing of reproduction during the month relative to the largest spring tides (0°) and relative to the lunar phase (full moon = 0°). Plotting the data this way will produce 1 cluster if they follow a 1-month cycle and 2 clusters of points ~180° apart if the events follow the 2-week cycle.
If more reproductive events occur during one phase of the moon compared to the other, we would expect more points in one cluster on the circular graph or for the points in one cluster to extend further from the center than the points in the other cluster. The same logic applies when the data are mapped relative to the 1-month amplitude cycle, which includes both the large and small spring tides of a month. If the snails are targeting the largest amplitude tides of the month, more points will occur around 0° on the tidal amplitude graph, rather than two equal clusters at 0° and 180°. If the snails are targeting a specific phase of the moon, points will also occur in one large cluster and one smaller cluster. For the species with bimodal patterns in reproduction relative to the monthly or lunar cycle, we tested for a difference in intensity of reproduction between the two hemispheres of the circle using binomial tests. We defined the two hemispheres by dividing them with a line perpendicular to the axis running through the clusters. A significant difference indicates that the species preferentially reproduces during one half of the lunar or monthly tidal amplitude cycle.

| Impact of environmental conditions on timing
To determine whether seasonal variation in abiotic conditions alters the timing of reproduction relative to the tidal amplitude cycle, we compared the average timing of events relative to the tidal amplitude cycle in the dry season (December-May) to that of the wet season (late May-early December).

| Overview of results
The duration of development varies from 4 to 28 days among the species studied (Table 1). For all six species, the omnibus test rejected a uniform distribution of reproductive events across the tidal amplitude cycle (omnibus test: p < .05, Figures 2, 3, Table 2). However, only 4 species (L. variegata, C. californica, N. chemnitzi, and S. palmata) showed clear cycles with periods of reproductive activity separated by periods in which reproduction did not occur (Figures 2, 3). When data from these 4 species were plotted relative to the lunar cycle, they exhibited a bimodal pattern of reproduction, indicating that they reproduce twice per lunar month, rather than targeting one particular phase of the lunar cycle  and C. californica showed differences in reproduction between the two events that occurred each month. For L. variegata, virtually all events occurred during the larger of the two spring tides ( Figure 6).  Nearly all females released egg capsules during the larger spring tides of a lunar month regardless of lunar phase (Figures 2,4,6, binomial test: p ≪ .001). This indicates that tidal amplitude is more important for this species than is the lunar phase.

| Cerithideopsis californica
Field observations of marked egg masses of C. californica showed development times of 4.0 days (SD = 1.7; N = 90). Egg masses were laid with very high synchrony (R = 0.79), on average 1 day before the maximum amplitude tide (347°, Omnibus test p ≪ .05, Figure 3, Table 2, 3). Hatching occurred approximately 4 days later (90°) with slightly lower synchrony (R = 0.60, Figure 3, Table 2, 3). This matches the 4-day period of development observed in the field (Table 1). Intensity of reproduction was lower during the dry season, synchrony occurred during both seasons, and average timing of reproduction was <1 day later in the dry season than in the wet season ( Figure 5). This species showed no difference in the number of egg masses laid or hatched during full or new moons (binomial test (laid): p = .55, hatched: p = .18, Figure 4). However, a significantly larger number of reproductive events occurred near the time of the smaller spring tides than the larger spring tides (binomial tests: p ≪ .001, Figure 6).

| Nerita scabricosta
Nerita scabricosta hatch after an average of 28 days in the field (

| Siphonaria palmata
Observations of S. palmata masses collected immediately after laying showed development times of 3.0 days (SD = 0; N = 10). The maximum number of egg masses per tidal amplitude cycle occurred with very high synchrony (R = 0.85), 1-2 days after the lowest amplitude tides (average timing = 214°, Omnibus test p < .05, v-test relative to 180° p < .05, Figures 2, 3). Based on a 3-day incubation period, we estimated the peak in hatching to be ~3 days after the minimum amplitude tide (average timing = 255°, v-test relative to 0° p = 1.0, v-test relative to 180° p < .05, Figure 3). Peaks in egg masses occurred ap-

Omnibus test for uniformity (p)
Peak release/hatching-high synchrony The mean angle is the average timing of events in degrees relative to the tidal amplitude cycle (1 day = ~24° and 0° is the point of maximum tidal amplitude). R, the resultant vector length, is a measure of synchrony of timing of events and ranges from 0 (no clustering) to 1 (all events happening at the same relative angle

| DISCUSSION
Four of the six species examined here show well-synchronized cycles in reproduction that are clearly separated by nonreproductive periods.
All of these show two reproductive periods per month, supporting the view that these are synchronized with the tidal amplitude cycle rather than the lunar cycle. Larval release occurs during the largest amplitude tides of the tidal amplitude cycle in two of these species, L. variegata and N. chemnitzi. One other species, Cerithideopsis californica, laid eggs during the maximum amplitude tides and hatched near the end of the half of the 2-week cycle with larger amplitude tides. Only S. palmata hatched during the half of the cycle with the smaller amplitude tides, on average, 3 days after the minimum amplitude tide. In L. variegata, reproduction was concentrated during the largest spring tides of the month regardless of lunar phase and minimal reproduction was observed during the smaller of the spring tides in the month.
The two species that do not show well-synchronized cycles in reproduction are those that have the longest development periods. Time to hatching in Crepidula cf. marginalis ranges from 8.5 days at 28°C to 10.2 days at 23°C in the laboratory (Collin, 2012), and time to hatching in Nerita scabricosta averages 24-33 days in the field . Clear biweekly cycles in crab reproduction have been observed to decrease with increasing latitude and concomitant increases in F I G U R E 5 Circular plots of timing of reproductive events relative to the tidal amplitude cycle separated by season. The dry season occurs from mid-December to early May, the wet season occurs from mid-May to early December. Angles around the circumference of the circle represent the tidal amplitude cycle in days, with 0° representing the day of the maximum amplitude tide. Black dots and gray triangles represent individual occurrences of reproduction (hatching, release of larvae, presence of egg masses). For Natica chemnitzi and Siphonaria palmata, the gray triangles represent the dates of peak mass occurrence for each cycle and the black dots represent the predicted peak hatching for each cycle. The arrow represents the average timing of reproductive events. The length of the arrow is scaled to the descriptive statistic R (resultant vector length), a measure of synchrony of events (the more synchronous the events, the longer the arrow). p-values for the omnibus test for circular uniformity (p(o)) and the v-test (p(v)), which tests whether clustering of events are timed with a specified angle (0°) are shown on the figure (p(v) < .05 indicates significant clustering around the maximum amplitude tide. If the p-value for the omnibus test is >.05, the timing relative to a specific angle cannot be tested (p(v) = N/A). *Nonsignificant p-values for the omnibus tests for S. palmata are due to small sample sizes

| Are biweekly cycles common in intertidal gastropods?
Monthly or biweekly cycles have been previously reported for only a few marine gastropods, and most of these are littorinids or siphonarids (Tables 4, 5). Among littorinids, and Littoraria in particular, there T A B L E 3 Circular statistics and tests for associations between peak release and spring tides The mean angle is the average timing of events in degrees relative to the tidal amplitude (1 day = ~24° and 0° is the day of the maximum tidal amplitude) cycle. The 95% confidence interval (CI) is provided as the lower and upper confidence interval limits. The v-test tests for circular uniformity relative to a predefined angle, in this case 0° (representing the maximum amplitude tide). Rejection of the null hypothesis of circular uniformity indicates that the events are significantly clustered near 0°. A significant omnibus test (obtained for all species; Table 1) and a nonsignificant v-test indicate that events are significantly clustered, but that the average timing of events differs significantly from 0°. Significant p-values are represented in bold, and species are categorized as in Table 2. a For N. chemnitzi and S. palmata, hatching timing was estimated by adding one half of the development time to the timing of the maximum count of egg masses for each tidal amplitude cycle. Circular statistics were then calculated based on projected peaks of hatching per cycle.
F I G U R E 6 Circular plots of reproductive events for the 4 species which show clear synchrony relative to ~monthly cycle of larger and smaller spring tides (two tidal amplitude cycles). The circle represents the time in days between the largest spring tide of a month (0°) and the largest spring tide of the following month, 180° represents the day of the smaller spring tide, and 90° and 270° are the approximate timing of the neap tides.

Monthly tidal amplitude cycle (30 days)
is ample evidence that release of pelagic egg capsules or brooded larvae usually occurs around the spring tides (Table 4) Nodilittorina lineolata collected along a Brazilian coast showed no relationship between capsule abundance and tidal height, but did show a close match to a time series of wave heights, with release occurring during rough seas (Bueno, Moser, Tocci, & Flores, 2010). This pattern was interpreted to support a pattern of propagule release that promotes onshore retention of capsules (Bueno et al., 2010). Egg capsules of Melarapha cincta and M. oliveri collected from tide pools also showed associations with rough weather, as well as high salinity, rather than any obvious lunar cycle (Pilkington, 1971). The only field study to support a relationship between tidal amplitude and propagule release showed that the presence of littorinid capsules in plankton samples coincided with times when actual water levels (as opposed to the predicted levels) exceeded the mean high water spring tide level (Borkowski, 1971). This study was conducted in Florida, which experi-  (Berry, 1986a), the magnitude of the tidal amplitudes, and/or  Ohgaki (1981) T A B L E 4 Reproductive cycles documented as peak larval or capsule release in littorinid species. Bold highlights results from this study the typical wave heights experienced by each species (Bueno et al., 2010).
Among pulmonate limpets, Siphonaria gigas, S. diemenensis, and S. capensis, all produce egg masses around the neap tides (Levings & Garrity, 1986;Pal, Erlandsson, & Sköld, 2006;Parry, 1977), while S. denticulata and S. pectinata lay around the spring tides (Creese, 1980;Zischke, 1974; Table 5). There is some evidence that the timing of deposition may vary among populations of the same species. For example, at one site in Japan, Siphonaria japonica lays during the second and fourth quarters of the moon (Hirano, 1980), but may deposit eggs only during the full moon at another site (Abe, 1940). Hirano (1980) pointed out that spawning only occurs at night and that periods of low spawning correlate with periods of low nighttime movement induced by the tidal cycle, suggesting that the complex interaction between diurnal cycle, tidal cycle, and patterns of movement may control the pattern of egg deposition.
The timing of hatching can be inferred for three of the species listed in Table 5. Siphonaria denticulata in Australia (Creese, 1980) lay their egg masses 2-3 days following the new or full moon and hatch approximately 6 days later, which suggests that hatching occurs around the neap tides. In contrast, Parry (1977) observed that S. diemenensis produce egg masses synchronously at the onset of neap tides and that these masses take 10 days to hatch, suggesting that hatching occurs near the spring tides. Likewise, Siphonaria sirius lays eggs around the neap tides and they hatch 5-6 days later around the spring tides (Iwasaki, 1995). This variation in projected hatching time further supports the conclusion that reproductive cycles in Siphonaria are not all driven by factors selecting for hatching at a particular point in the tidal amplitude cycle. Iwasaki (1995) suggested that egg deposition during neap tides limits emersion of S. sirius egg masses during early development, which may increase embryo survival. Siphonaria gigas lays egg masses during the neap tides, primarily during the smaller amplitude neaps of a lunar month, and hatching occurs primarily during the smaller amplitude spring tides of a month (Christy, 2013;John Christy, unpublished data). This species is found on exposed rocky shores in the upper midintertidal, where they are submerged by all spring tides. S. sihpo Japan Half moon Abe (1935, 1941in Iwasaki, 1995 S. atra Japan Half moon Abe (1935Abe ( , 1941in Iwasaki, 1995) S.

15
Laboratory Zischke (1974) were designed to test for seasonal patterns in reproduction and sampled monthly rather than at the higher frequencies necessary to detect biweekly or monthly cycles. It is therefore unclear whether cycles are more common in Siphonaria species than can currently be appreciated from the literature, or whether the species listed in Table 5  Umbonium vestiarium spawns around neap tides (Berry, 1986b(Berry, , 1987, while Trochus niloticus spawns at the new moon (Hahn, 1993) and the tropical abalone Haliotis asinina spawns just after the full and new moons (Counihan et al., 2001). The presence of larvae of Melagraphia aethiops, Zediloma atrovirens, and Lunella smaragda in the plankton correlates with storms, and not with tides, lunar cycles, or temperature (Grange, 1976). As most vetigastropods are free-spawners that likely rely on synchronized reproductive timing to ensure fertilization success, timing in this group may be driven by different factors than in the species studied here, all of which have internal fertilization.

| Why do the timing of cycles vary?
Littorinids, snails that release their eggs and larvae directly into the water column, like crabs, most often time their release around the large amplitude tides. This suggests that mother snails are targeting propagule release for the same periods that maximize offshore transport and potentially minimize predation on the larvae or planktonic capsules, as do intertidal crabs. This timing may also be optimal for larval or capsule release as it is the moment when the seawater is closest to the normal habitat of the mother snails, reducing travel costs and predation risk for mothers as well.
Snails that deposit benthic egg masses may face different selection pressures. Timing and location of deposition that ensure survival of embryos to hatching may supersede timing to optimize offshore transport of larvae. Embryos in benthic egg masses may face significant mortality from thermal stress, desiccation, osmotic stress, and predation (Podolsky, 2003;Przeslawski, 2004;Rawlings, 1994). Detailed research on the intertidal bubble snail Melanochlamys diomedea shows that timing of egg mass deposition relative to the time of day of the low tides significantly impacts thermal stress experienced by the mass and can therefore be used to predict egg mass survival (Podolsky, 2003). Masses laid during the spring tides are predicted to have significantly higher loss than those that were laid during neap tides. Such strong selection for successful hatching could easily shape synchrony of reproductive timing. It is pertinent to note that both species that do not show clear synchrony in our study are those whose embryos are buffered from external stressors; C. cf. marginalis brood their embryos, protecting them from predation, desiccation, and osmotic stress, and N. scabricosta capsules are well protected from predation with calcium carbonate spherules and are buffered from desiccation by being deposited in pools. Siphonaria palmata, which show very clear synchrony, deposit masses in similar pools very slightly lower in the intertidal.
They are not so well defended as the capsules of N. scabricosta, which suggests that predation during larger amplitude tides may play a role in shaping reproduction around the neap tides. C. californica, the species here with the highest exposed egg masses, occurs at a height (~4.25 m) that is always inundated during spring tides, but is at the extreme edge of the high tides during the neap tides; high tides during the smallest high neap tides reach between 3.6 and 4.3 m depending on the month. This suggests that masses that mature much later than average risk being ready to hatch during a period when they are not inundated, or when hatchlings risk being stranded in only a few centimeters of water. Although this does not explain why they reproduce preferentially around the small amplitude spring tides, detailed experimental work and additional comparative studies of related species living at different tidal heights or in different environments are necessary to assess these scenarios before a clear understanding of the factors that influence reproductive synchrony can be reached.