Spawning aggregations of checkered snapper (Lutjanus decussatus) and blackspot snapper (L. fulviflamma): seasonality, lunar-phase periodicity and spatial distribution within spawning ground

Study site

This study was conducted at Sekisei lagoon in the Yaeyama Islands, Okinawa, located in the southernmost part of Japan (Fig. 1). One fish aggregation site was located in the Sekisei lagoon, although the precise location is not shown. This is because the study site is not protected during the spawning periods of the two snapper species yet and there is concern that showing the precise location might lead to overfishing the fish aggregations of the two snapper species.

Temporal variations in the fish abundance within the aggregation site

To clarify the monthly and weekly variation in the fish abundance at the aggregation site, daytime underwater observations were conducted. After preliminary surveys between September and October 2020, one line transect (600 m ×5 m) was set to extend over the whole aggregation site between November 2020 and December 2021. Underwater observations on the transect were conducted using SCUBA. During the main spawning season of the two snapper species (between May and October), weekly observations were conducted (Table 1). The observation days were adjusted to be carried out during each of the four lunar phases whenever possible (new moon, first-quarter moon, full moon and last-quarter moon). When rough sea conditions prevented such adjustments, the observation days were set as several days earlier or later of the four lunar phases (Table 1). During non-spawning seasons (between November 2020 and April 2021, and between November and December 2021), underwater observations were conducted around two lunar phases (last-quarter moon and new moon). This was because: (1) preliminary surveys between September and October 2020 revealed the fish aggregations of the two snapper species were found around the last-quarter moon and new moon and (2) Nanami et al. (2010) showed that Lutjanus decussatus showed a greater peak of ovary development during the last-quarter moon.

The number of individuals of the two snapper species on the above-mentioned 600 m ×5 m line transect was counted per every 1 min. During the observations, a portable GPS receiver, sealed in a waterproof case and attached to a buoy, was towed and the course and distance of the tracks were obtained in accordance with the protocol detailed in Nanami et al. (2017). The water depth range at which the underwater observations were conducted was c.a. 10 m to 15 m.

Analysis of periodicity in the increase of fish abundance in relation to lunar phase

To determine the statistical significance in periodicity of fish increasing at the aggregation site, a time-series analysis (correlogram) was applied. In this procedure, data obtained between May 3 and October 20 was used for the analysis (Table S1), since greater fish abundance at the aggregation site was found in the study period (see RESULTS). In the analysis, 24 time periods were established, in which each time period includes one lunar phase (last-quarter moon, new moon, first-quarter moon and full moon) (Table S1). Correlogram was applied (Fig. S1, Table S2) by using R statistical computing language (function “acf”: R Core Team, 2022). Statistical significance of auto-correlation coefficient was determined by using 95% confidence intervals (CI) as follow: ${\displaystyle 95\text{\%}\text{CI}=1.96/\sqrt{}\left(T\right)}$ where T is the number of observations.

It can be regarded that the auto-correlation coefficient is significant when the following equation was found: ${\displaystyle \left|pk\right|>95\text{\%}\text{CI}=1.96\sqrt{}\left(T\right)}$ where p_k is the value of auto-correlation at k th time lag, and — p_k— is the absolute value of p_k. If —p_k— is greater than the 95% CI, auto-correlation at k th time lag was significant. Since T was 24 in the analysis (Table S1), 95% CI was calculated as 1.96/√ (24) = 0.40. For example, auto-correlation coefficient at 4th time lag (p₄) was over 0.40, the p₄ is a significantly positive value. Namely, the periodic increase in fish abundance was found at the same particular lunar phase.

Since data for July 22 (full moon phase) could not be collected due to a typhoon, the missing value was imputed. In this procedure, other fish abundance data obtained around full moon phase (May 26, June 25, August 24, September 21 and October 20) was averaged and the average value was applied to impute the missing value.

Fine-scale spatial variations in fish density at the aggregation site

The above-mentioned underwater observations revealed that fish abundance showed positive peaks around last-quarter moon or new moon (see RESULTS). At the two lunar phases, fine-scale spatial variations in fish abundance (variations in fish abundance at intervals of several-tens meters) within the aggregation site were examined. The above-mentioned 600 m × 5 m line transect was divided into 1-minute sub-transects (average distance ± standard deviation = 20.5 m ± 4.1). Then, the number of individuals of the two snapper species and the distance for the 1-minute sub-transect were obtained. From the data, the number of individuals was converted to density (per 20 m × 5 m) for each 1-minute transect. Fish density on the 1-minute sub-transect was individually plotted by bubble plot along the 600 m × 5 m line transect.

The fine-scale spatial variation in fish density per 20 m × 5 m was shown as frequency data (histogram). Then, Kolmogorov–Smirnov test was applied to test the significant difference in fine-scale spatial variation in fish density between the two fish species.

Verification of spawning by ovarian development

Domeier (2012) and de Mitcheson & Erisman (2012) have indicated that presence of females with matured eggs in the hydrated stage (hydrated eggs) in aggregation sites is one of the direct evidence that the fish aggregation can be regarded as a spawning aggregation. To clarify whether females have hydrated eggs in the aggregation site, individuals of the two species were caught by spear-gun just after the above-mentioned underwater observations around the last-quarter and new moon (Table 1). In total, 73 and 92 individuals were caught for L. decussatus and L. fulviflamma, respectively (Table 1). In the laboratory, fork length (FL), whole body weight and gonad weight were measured. The gonadsomatic index (GSI) was calculated by using the formula: ${\displaystyle \text{GSI}=\text{Gonad weight (g)}/\left[\text{whole body weight (g)}-\text{gonad weight (g)}\right]\times 100.}$ To obtain histological observations, the gonads were preserved in 20% buffered formalin over 48 h and then kept in 70% ethanol baths. Embedded pieces of gonads were sectioned and stained with Mayer’s hematoxylin–eosin. Under microscopic observations, developmental stages of ovaries were examined whether the gonads were sufficiently developed for spawning (hydrated eggs). The categorization of ovarian developmental stages followed Ohta & Ebisawa (2015) and Ohta et al. (2017). According to the categorization of Ohta & Ebisawa (2015), oocytes with migration nuclear stage, pre-maturation stage and maturation stage were defined as hydrated stage.

Size and age frequency distributions

To determine the size frequency distribution of fishes at the aggregation site, histograms of fork length for the above-mentioned fish samples was plotted. Male and female were separately plotted, and probability density of size frequency was analyzed by R statistical computing language (function “density”: R Core Team, 2022).

To determine the age frequency distribution of fishes, sagittal otoliths of the above-mentioned fish samples were extracted from each fish, and cleaned in water and dried. Then, one otolith was embedded in epoxy resin and transversely sectioned into 0.5 mm-thick sections using ISOMET low speed saw (Buehler) and attached on a glass slide. The sectioned otoliths were observed under a microscope with reflected light at ×4 magnification, and the number of opaque rings was counted. The procedure of opaque rings count followed Ohta et al. (2017). The number of opaque rings on each otolith was counted twice. If the two counts coincided, the number of rings was used. However, if the two counts did not coincide, the number of opaque rings was counted once more and any two coinciding counts were used. Since Nanami (2021) and Shimose & Nanami (2015) revealed that each opaque ring was formed annually for the two species, number of opaque rings can be considered as age (year). Age frequency distribution for male and female was separately plotted, and probability density was analyzed by R statistical computing language (function “density”: R Core Team, 2022).

Mann–Whitney U-test test was applied to determine the significant differences in average fork length and average age between males and females.

Comparison of fish density between inside and outside the aggregation site

Domeier (2012) has proposed the definition of spawning aggregations, indicating that fish abundance in aggregation sites is at least 4-fold greater than that outside the aggregation site. In order to verify the definitions, 65 study sites outside the aggregation site were established in Sekisei Lagoon (Fig. S2) and the number of the two snapper species was counted. Above-mentioned 20-minutes underwater survey with a portable GPS receiver was conducted at each site (details about method was shown in Nanami (2020). Fish abundance per 600 m ×5 m was estimated by using the fish count data and the measured distance. The estimated fish abundance per 600 m ×5 m among the 65 sites were averaged and regarded as average fish abundance outside the aggregation site. Then, the fish abundance per 600 m ×5 m inside the aggregation site was compared with that outside aggregation site.

Temporal variations in the fish abundance and reproductive activity in relation to lunar phase

Greater fish abundance (number of individuals per 600 m ×5 m) of Lutjanus decussatus was found between May and October (Fig. 2A), when the water temperature exceeded 25 °C. During the six months, clear peaks in the fish abundance were only found around the last-quarter moon (Fig. 2B). The peak fish abundance ranged from 529 (October) to 1565 (June). Average GSI values ± SD (standard deviation) of females that were caught at the aggregation site was 10.34 ± 4.95 (ranged from 0.42 to 24.89) (Fig. 2C). About 69% of individuals (34 out of 49) had hydrated stage oocytes (Figs. 2C–2E, Table S3).

Greater fish abundance of L. fulviflamma was clearly found between April and October (Fig. 3A), when water the temperature exceeded 25 °C. During the seven months, clear peaks in the fish abundance were found around the last-quarter moon in April, May, June and October (Fig. 3B). In contrast, clear peaks were found around the new moon in July, August and September (Fig. 3B). The peak fish abundance ranged from 660 (September) to 6337 (June). Average GSI values ± SD of females were 6.33 ± 2.21 (ranged from 3.44 to 14.51) (Fig. 3C). About 90% of individuals (37 out of 41) had hydrated stage oocytes (Figs. 3C–3E, Table S4).

Periodicity of the increase of fish density

Correlogram of L. decussatus revealed that significant positive auto-correlation coefficients were found when time lags were 4 and 8 (Fig. 4A). Correlogram of L. fulviflamma revealed that significant positive auto-correlation coefficient was found when time lag was 4 (Fig. 4B).

Fine-scale spatial variations in fish density at the aggregation site

Fine-scale spatial distributions revealed that L. decussatus showed relatively uniform distribution at the aggregation site, especially around the last-quarter moon between May and October (Fig. 5). Fish densities per 20 m ×5 m were generally less than 140 individuals in most cases (Fig. 6).

In contrast, greater fish density of L. fulviflamma per 20 m ×5 m was found at particular areas (south-eastern side) in the aggregation site (Fig. 7). This tendency was clearly found around last quarter moon in April, May and June, and around new moon in July and August. In some cases, fish density within 20 m ×5 m was over 300 (Fig. 8).

Kolmogorov–Smirnov test revealed that a significant difference in fish density frequency (i.e., patterns in fine-scale spatial distribution) was found between the two fish species from May to October (p < 0.05, Table 2).

Size and age frequency distributions

For L. decussatus, fork length of most individuals was between 200.0–279.5 mm for both males and females (Figs. 9A, 9B). Average fork length of females (243.9 mm FL ± 21.4 SD: standard deviation) was significantly greater than males (229.2 mm FL ± 19.6 SD: Mann–Whitney U-test, p < 0.01). Highest probability of occurrence in fish length was about 210 mm FL for males whereas about 260 mm FL for females. Although average age was not significantly different between males (9.4 years ± 5.0 SD) and females (8.1 years ± 4.4 SD: Mann–Whitney U- test, p = 0.26), highest probability of occurrence in age was about 6 years for both males and females (Figs. 9C, 9D).

For L. fulviflamma, fork length of most individuals was less than 250 mm for males whereas 280 mm for females (Figs. 10A, 10B). Average fork length of females (246.9 mm FL ± 23.0 SD) was significantly greater than males (233.9 mm FL ± 17.8 SD: Mann–Whitney U-test, p < 0.01). Probability density revealed that highest probability of occurrence in fish length was about 240 mm FL for males whereas about 260 mm FL for females. Although average age was not significantly different between males (6.8 years ± 3.1 SD) and females (8.1 years ± 4.5 SD: Mann–Whitney U-test, p = 0.29), highest probability of occurrence in age was about 7 years and 5 years for males and females, respectively (Figs. 10C, 10D).

Comparison of fish density between inside and outside the aggregation site

Fish abundance per 600 m ×5 m of L. decussatus inside the aggregation site was 148.2 to 438.4-fold (average = 266.8-fold) greater than that outside the aggregation site (Table 3).

Fish abundance of L. fulviflamma inside the aggregation site was 33000.0 to 316850.0-fold (average = 141557.1-fold) greater than that outside the aggregation site (Table 3).

Verification of spawning aggregation of two snapper species

For the two snapper species Lutjanus decussatus and L. fulviflamma, this study was the first attempt to examine whether fish aggregations at a particular site can be regarded as spawning aggregations. The present study showed: (1) repeated greater fish abundance of the two species and this is predictable in time (particular month and lunar-phase) and space, (2) the fish abundance inside the aggregation site is over 4-fold greater than that outside the aggregation site, and (3) most females inside the aggregation site had hydrated eggs. Thus, the fish aggregations of the two snapper species can be regarded as spawning aggregations.

Spawning season and spawning day

The present study revealed that fish aggregations of the species were found between May and October. Nanami et al. (2010) showed similar results from fish samples by commercial catch, i.e., estimated main spawning season of L. decussatus was between June and October. Since developed- or matured-oocytes (tertiary yolk stage, migration nuclear stage, pre-maturation stage and maturation stage) were obtained for most female individuals between May and October, it is suggested that the main spawning season of L. decussatus can be regarded as between May and October in the study region. Since peaks of fish abundance were found only around the last-quarter moon and the periodicity in the increase of fish abundance during each last-quarter moon phase was significant, it is suggested that spawning occurs around the last-quarter moon.

The present study revealed that spawning aggregations of the species were found and the most females had hydrated eggs between April and October. Shimose & Nanami (2015) showed similar results from fish samples by commercial catch, i.e., estimated main spawning season of L. fulviflamma is between April and August. Therefore, it is suggested that the spawning season of L. fulviflamma can be regarded as between April and October in the study region. The peak fish abundance at the spawning ground was found during the last-quarter moon (in April, May, June and October) and new moon (in July, August and September), suggesting that spawning occurs around the last-quarter moon and new moon. This lunar-related periodicity in the increase of fish abundance at during a particular lunar phase (last-quarter moon or new moon) was also supported by the correlogram in the present study. This evidence for lunar-related spawning is the first finding for L. fulviflamma.

Since western Atlantic snapper species (Lutjanus cyanopterus (Cuvier, 1828) and L. jocu (Bloch & Schneider, 1801)) spawned for a period of four to seven consecutive days (Heyman & Kjerfve, 2008), the two snapper species in this study might spawn during several consecutive days around last-quarter moon and new moon. Intensive daily observations around the two lunar phases would clarify more precise ecological aspects about aggregation formation of the two snapper species.

Water temperature is one of the main factors for gonad development and reproduction of marine fishes (Wang et al., 2010). The present study revealed that reproductive activity of the two snapper species is likely to occur when water temperature is over 25 °C. In addition, decline of temperature (beginning of October) might be one of the factors leading to reduced ovarian development. This trend is consistent with cubera snapper (Lutjanus cyanopterus) in the Caribbean (Heyman et al., 2005; Motta et al., 2022), showing that spawning aggregation of L. cyanopterus related to increasing water temperature in the summer.

Fine-scale spatial variation in fish density within spawning ground

Some previous studies have shown fine-scale spatial variations of fish density at intervals of several-tens meters within spawning grounds for groupers (e.g., Colin, 2012; Nanami et al., 2017; Sadovy de Mitcheson et al., 2020). These studies have shown species-specific spatial variations among multiple grouper species and each species showed species-specific core sites, in which a very high density was found in a limited area within the spawning ground (Nanami et al., 2017; Sadovy de Mitcheson et al., 2020). Spatial variations of fish density at intervals of 250 m was also observed for two snapper species (Biggs & Nemeth, 2016), showing fine scale movement of two snapper species (L. cyanopterus and Lutjanus jocu) within the aggregation site in relation to spawning time.

The present study revealed a significant difference in the fine-scale spatial distributions in fish density between the two snapper species. Although the reasons why the two species showed species-specific spatial variations within the spawning ground remain unknown, this might be related with preference to a particular environmental condition or mating behavior.

Size and age frequency of fish aggregation

The present study showed that the maximum fork length of fish individuals at the spawning ground was 304 mm and 295.5 mm for L. decussatus and L. fulviflamma, respectively. In contrast, previous studies revealed that maximum fork length was 316.5 mm and 347.0 mm for Lutjanus decussatus and L. fulviflamma, respectively (Shimose & Nanami, 2015; Nanami, 2021). Since average fork length of fish individuals at the spawning ground was less than 250 mm for both species, most of the fish individuals forming the spawning aggregations were relatively small sized individuals.

A similar trend was also found for age frequency. Maximum age was respectively 24 and 23 for Lutjanus decussatus and L. fulviflamma (Shimose & Nanami, 2015; Nanami, 2021), whereas maximum age of fish individuals at the spawning ground was respectively 21 and 17 for L. decussatus and L. fulviflamma. Since average age of fish individuals at the spawning ground was less than 10 for both species, most of the fish individuals that form the spawning aggregations were relatively young individuals.