Temporal changes in hamlet communities (Hypoplectrus spp., Serranidae) over 17 years

Transect surveys

Hamlet population densities and relative abundances were estimated using scuba visual censuses. Belt transects were preferred over the roaming surveys adopted by Aguilar-Perera (2003) since they provide standardized population density estimates (i.e. number of individuals per unit of reef area) that can be compared across time and space and also used to estimate population sizes (Puebla et al., 2009, 2012b). Yet hamlet densities are quite low in La Parguera, resulting in a large proportion of empty 25 × 2 m transects (Aguilar-Perera, 2003). In order to address this limitation, 100 × 4 m transects were adopted instead, following the approach used in previous hamlet surveys (Puebla et al., 2007, 2008, 2009, 2012b), resulting in an eightfold increase of the area surveyed per transect (400 m² v. 50 m² for 25 × 2 m transects). Briefly, two divers swam in parallel a few feet above the reef with each diver counting all the hamlets observed within 2 m on each side of a 100 m transect tape. Fishes swimming across the transect tape were signalled to avoid counting the same fish twice. Coral-reef habitat was specifically targeted for the transects (i.e. avoiding sandy or seagrass areas) and an effort was made to cover a variety of depths and reef zones (e.g. reef slope v. reef flat), but the exact location of each transect was randomly chosen. An effort was made to broadly match the sampling effort of Aguilar-Perera (2003) and avoid comparatively under or over-sampling, with also a broadly similar proportion of surveys conducted in the inner v. outer reefs.

Data analyses

Densities (number of fish 1000 m⁻² of reef) and hamlet relative abundances (%) were estimated for each species, reef site, reef type (inner v. outer reefs) and overall. Similar data were recompiled from Aguilar-Perera (2003)'s raw species counts (it is noted that some minor numerical errors were found in the total counts in Aguilar-Perera, 2003; (Table I), hence the relative abundances reported here differ slightly from those in the original paper).

Differences in relative abundances were tested with permutational analysis of variance (PERMANOVA; Anderson 2001) with 1999 permutations per test and visualized using non-metric multidimensional scaling as implemented in the vegan package in R (Oksanen et al., 2017; www.r-project.org), using the Bray-Curtis measure of ecological distance in both cases (Bray & Curtis, 1957). This analysis was done at the reef level, i.e. considering relative abundances per reef (not per transect). Differences between the inner and outer reefs were first tested and given the significant outcome (see Results) changes in hamlet relative abundances between 2000 (Aguilar-Perera, 2003) and 2017 (this study) were tested for the inner and outer reefs separately. Hamlet diversity in the inner and outer reefs for the years 2000 (Aguilar-Perera, 2003) and 2017 were estimated using the effective number of species of the first order, corresponding to the exponential of the Shannon entropy and referred from here on as the effective number of species (Hill, 1973; Jost, 2006). Differences in diversity between 2000 (Aguilar-Perera, 2003) and 2017 were tested for the inner and outer reefs separately with a Mann–Whitney U-test and finally the depth distribution of H. chlorurus was tested for significant differences between shallow (<5·5 m) and deep (≥5·5 m) sections of the inner reefs using a χ²-test following Aguilar-Perera (2003). This test was not repeated for the other species due to the relatively low number of sightings.

Transect surveys

A total of 35 non-overlapping transects were conducted, covering an area of 14 000 m² across 16 reefs at depths ranging between 2 and 18 m (Table I). One hundred and seventeen hamlets from seven species were sighted within the transects, providing an overall hamlet density estimate of 8·4 ± 1·5 fish 1000 m⁻² of reef (mean ± S.E.). The most abundant species by far was H. chlorurus, representing 80·3% of all hamlets seen, followed by the butter hamlet Hypoplectrus unicolor (Walbaum 1792), 7·7% and the barred hamlet Hypoplectrus puella (Cuvier 1828), 6·0%. The shy Hypoplectrus guttavarius (Poey 1851), H. indigo, black Hypoplectrus nigricans (Poey 1851) and yellowbelly Hypoplectrus aberrans (Poey 1868) hamlets were rare, with only one or two individuals seen over all transects. No other hamlet species were sighted outside of the transects.

There was a significant difference in relative abundances between the inner and outer reefs [Fig. 2(a); PERMANOVA P < 0·01]. The inner reefs were dominated by H. chlorurus (91·1% of all hamlets seen) followed by H. puella (5·0%) and H. unicolor (3·0%) and H. indigo was rare (1·0%). No other species was sighted within the transects, but it is noted that H. nigricans was observed outside of the transects on the inner reefs, implying presence on these reefs at a density <0·1 fish 1000 m⁻². The outer reefs were dominated by H. unicolor (37·5% of all hamlets seen) followed by H. chlorurus, H. puella, H. nigricans and H. guttavarius (12·5% each) and finally H. indigo and H. aberrans (6·3% each). No other species was sighted outside of the transects on these reefs. The difference in community composition between the inner and outer reefs was accompanied by a significant difference in density, with an average of 10·5 ± 1·9 v. 3·6 ± 1·0 fish 1000 m⁻² of reef in the inner and outer reefs, respectively (mean ± S.E., Mann–Whitney U-test P < 0·05).

There was a significant change in relative abundances between the years 2000 (Aguilar-Perera, 2003) and 2017 on the inner reefs [Fig. 2(b); PERMANOVA P < 0·05]. The most notable difference was the relative increase of H. chlorurus, from 58·2% in 2000 to 91·1% in 2017 at the expense of H. nigricans (10·5% to <1%), H. aberrans (7·8 to 0%) and H. indigo (7·2 to 1·0%).

There was a marginally significant change in relative abundances between 2000 and 2017 on the outer reefs (Supporting Information Fig. S1; PERMANOVA P = 0·05), but it should be noted that this result is to be interpreted with caution given the low hamlet densities and counts on the outer reefs and the over-dispersion of the 2017 data compared with 2000 (due to the occurrence of both H. guttavarius and H. indigo in one outer reef in 2017, none of which were observed on the outer reefs in 2000). The most notable change on the outer reefs was a decrease of H. puella from 60·9% in 2000 to 12·5% in 2017 and an increase of H. chlorurus and H. nigricans from 0 to 12·5%. The presence of H. guttavarius on the outer reefs is also to be noted since this species was not reported by Aguilar-Perera (2003) in either the inner or outer reefs.

Effective number of species on the inner and outer reefs from Aguilar-Perera's data and this study are presented in Fig. 3. The increased dominance of H. chlorurus on the inner reefs in 2017 compared with 2000 is reflected by a significant decrease in effective number of species, from 3·1 ± 1·1 in 2000 to 1·4 ± 0·6 in 2017 (mean ± S.D., Mann–Whitney U-test P < 0·01). The opposite trend was observed on the outer reefs (1·62 ± 0·25 in 2000 v. 2·53 ± 0·66 in 2017), but this difference was not significant (Mann–Whitney U-test P > 0·05) and here again caution is warranted due to the low species densities and counts on the outer reefs.

Overall depth distributions of all hamlets sighted are presented in Fig. 4. Hypoplectrus chlorurus was significantly more abundant in the shallow (<5·5 m) sections of the inner reefs than in the deeper areas (≥5·5 m, χ²-square test P < 0·001).

Temporal changes in the hamlet community

Hamlet relative abundances have noticeably changed between the years 2000 and 2017 in La Parguera, with most notably a > 30% increase of H. chlorurus on the inner reefs, from 58·2% in 2000 (Aguilar-Perera, 2003) to 91·1% in 2017. Both surveys targeted the hamlets specifically, were extensive and highly replicated at the inner-reef level, conducted on the same reefs and in the same depth range. It is also noteworthy that the hamlets are quite conspicuous. Thus, differences in relative abundance between Aguilar-Perera (2003) and this study are unlikely to be due to methodological biases. The relative increase of H. chlorurus on the inner reefs occurred at the expense of the other, less abundant species, resulting in a significant decrease in effective number of species from 2000 to 2017.

While the relative abundance of H. chlorurus has increased since the year 2000, it also appears that its density has declined, implying that overall hamlet densities have declined. As mentioned above, Aguilar-Perera (2003) does not provide density estimates, only relative abundances, but densities of H. chlorurus in La Parguera have been estimated at 35 fish 1000 m⁻² of reef in 1988–1989 (McGehee, 1994), 18 ± 3·8 in 2005 (Bejarano & Appeldoorn, 2013) and 6·7 ± 0·6 in this study (mean ± S.E.). These estimates are to be interpreted with caution due to differences in methodologies and reefs surveyed between the three studies, but the magnitude and consistency of the decline suggests that they reflect a real trend.

Hypoplectrus chlorurus was significantly more abundant at shallow depths on the inner reefs in 2017, with a density of 4·5 ± 0·9 fish 1000 m⁻² of reef at <5·5 m v. 2·1 ± 0·6 at ≥ 5·5 m (mean ± S.E.). Interestingly, Aguilar-Perera (2003) did not find a significant difference in abundance of H. chlorurus between shallow (<5·5 m) and deep (>5·5 m) sections of the inner reefs using the same methodology. Further back in time, McGehee (1994) reports the opposite pattern in 1988–1989, with a significant increase in abundance of H. chlorurus with depth in La Parguera. Here again, the sequence from higher abundances of H. chlorurus in deeper waters in 1988–1989 (McGehee 1994), no difference with depth in 2000 (Aguilar-Perera 2003) and higher abundances at shallow depths in 2017 suggests that the relative increase of this species on the inner reefs has been accompanied by not only a decrease in density, but also a shift in depth distribution towards shallower depths.

Implications

One implication from these results is that it cannot be assumed that hamlet relative abundances and densities are stable over a temporal scale of <20 years. Retrospectively, this does not come as a surprise given the dynamic nature of the coral reefs with which these fishes are tightly associated and even more so in the face of the large-scale anthropogenic influence on these ecosystems. Thus, data on hamlet relative abundances and densities need to be updated when possible and relevant. Another implication is that given the rarity and uncertain taxonomic status of some species, global databases on hamlet distributions and abundance are to be interpreted with caution [not to mention the pervasive occurrence of errors in such databases (Robertson, 2008)]. Clearly, the distribution of e.g. the tan hamlet (Hypoplectrus randallorum Lobel 2011) depends on what is considered a tan hamlet in the first place (see below). Extensive surveys are required to capture rare species, which are commonplace in the hamlet as illustrated in the present study; for four species only one or two individuals were sighted over 35 transects covering a total of 14 000 m² across 16 reefs. The hamlets appear to be able to persist at low densities, which might be due to the fact that they are simultaneously hermaphroditic (implying that two individuals can reproduce regardless of their sex). The hamlets are also able to find conspecifics on the reef when present (Puebla et al., 2012a), which is confirmed here by the observation that three of the four rare species were sighted in pairs when found. Finally, given the data presented here, it is also possible that hamlet distributions might be dynamic at the regional scale over a few decades, which could have implications for the understanding of speciation in this group [population-centre hypothesis (Domeier, 1994)].

Ecology

What factors could have driven the observed temporal changes in the hamlet community? The simplest explanation that comes to mind is neutral, stochastic variation. Assuming ecological equivalence among species and individuals, relative species abundances are expected to fluctuate following a process analogous to genetic drift referred to as ecological drift, with the expectation that less abundant species have a higher probability of going extinct (Hubbell, 2001). Since H. chlorurus was already the most abundant species on the inner reefs in 2000, it is in principle plausible that stochastic fluctuations in the other, less abundant species led to their relative decline.

Nevertheless, specific patterns in the distribution and abundance of the hamlets from La Parguera suggest that the hamlet community is not behaving neutrally, but is at least in part shaped by ecological factors. The most notable of these patterns is the difference in densities and relative abundances between the inner and outer reefs, despite the fact that these are in close geographic proximity (<12 km). The inner reefs were clearly dominated by H. chlorurus, which represented 91·1% of all hamlets sighted. The outer reefs had in contrast a much more even species distribution (resulting in a higher effective number of species), with the most abundant species, H. unicolor, representing 37·5% of the hamlets sighted and H. chlorurus only 12·5%. Densities were also three times lower on the outer reefs than on the inner reefs, with an average of 3·6 ± 1·0 (mean ± S.E.) individual hamlets 1000 m⁻² of outer reef versus 10·5 ± 1·9 on the inner reefs all hamlet species confounded. This difference between the inner and outer reefs appears to be temporally stable since Aguilar-Perera (2003) also reports differences in relative abundance and densities between the inner and outer reefs (Supporting Information Fig. S2), with lower densities and fewer H. chlorurus on the outer reefs. In this context it is noted that the inner reefs are clearly structurally distinct from the outer reefs, justifying the decision to contrast these two types of reefs in both Aguilar-Perera (2003) and this study. The inner reefs are characterized by a shallow reef flat and a reef slope that goes down to c. 14 m while the outer reefs are exclusively deep (13–18 m) and characterized by a spur-and-groove formation. The outer reefs are also clear blue reefs while the inner reefs are more turbid. In addition to differences in hamlet communities between the inner and outer reefs, the non-random depth distribution of H. chlorurus also suggests a role for ecology in shaping its distribution.

If ecology drove the observed changes in the hamlet community from La Parguera, what ecological factors in particular might be involved? So far ecologists have failed to identify clear ecological differences among hamlet species. Sympatric species are commonly found in the same habitat, often within sight of each other. The hamlets also have broadly similar diets as revealed by stomach-content and stable-isotope analysis in a variety of locations across the wider Caribbean, including La Parguera, except for H. indigo that appears to feed mostly on fishes (Randall, 1967; Whiteman et al., 2007; Holt et al., 2008; G. M. Serviss, unpubl. data). This being said, broad differences in distribution and abundances between reef sections and types have been noted (Thresher, 1978; Fischer, 1980a; McGehee, 1994). Regarding H. chlorurus in particular, significant differences have been found in its distribution in Deep Water Cay (Grand Bahama) with higher abundances on the shallow Acropora cerviconis zone (G. M. Serviss, unpubl. data). This resonates with the observation that H. chlorurus was often (although not exclusively) found in association with A. cervicornis in La Parguera.

Turbidity

One intriguing point to be noted about H. chlorurus is that an extensive study on seawater turbidity and fish communities conducted between February and October 2005 on 21 reefs off La Parguera indicated that it is the only fish together with the sharknose goby Elacatinus evelynae (Böhlke & Robins 1968) that shows a positive association with turbidity, with higher abundances on more turbid reefs (Bejarano & Appeldoorn, 2013). In addition, sediment cores indicate that sedimentation of terrestrial origin has significantly increased over recent decades on the inner reefs in La Parguera (Ryan et al., 2008), providing a parallel between changes in turbidity regimes and the relative abundance of H. chlorurus. Ryan et al., (2008) also report that sedimentation rates are higher on the inner reefs (0·47 ± 0·02 cm year⁻¹) than on the outer reefs (0·19 ± 0·01 cm year⁻¹, mean ± S.D.), once more providing here a parallel between turbidity and relative abundances of H. chlorurus on the inner v. outer reefs. This pattern is consistent with the observation that the outer reefs were in clear, open blue waters while the inner reefs were noticeably murkier. It has been noted before that different hamlets tend to associate with different turbidity regimes (Thresher, 1978), but the hypothesis that water turbidity per se could constitute an important ecological factor in shaping hamlet communities has not been evaluated in depth.

If H. chlorurus responds positively to turbidity, an increase in turbidity on the inner reefs over recent decades would be expected to result in higher relative abundances of this species on these reefs, as observed here. Water turbidity is also negatively correlated with coral cover (r² = 0·50; Bejarano & Appeldoorn, 2013). Therefore, a decrease in coral cover associated with the increase in sedimentation rates on the inner reefs would be expected to result in a decrease in hamlet densities, as observed here. One point to consider is seasonality since this survey was conducted during the dry season while Aguilar-Perera's survey as well as Bejarano & Appeldoorn's study on turbidity were conducted over both the dry and rainy seasons. Relative abundances are not expected to vary seasonally at the reef level since the hamlets are reef-associated organisms and as such do not move between reefs after settlement (which would imply swimming over extensive non-reef areas). In addition, transect and tagging data over several seasons at other sites in Panama and Belize (Puebla et al., 2007, 2012a) indicate that hamlet communities do not change substantially at this time scale and that individuals are long-lived (several years) and quite sedentary. This being said, it is possible that the distribution of individuals within reefs might vary seasonally and the data on depth distribution are therefore to be considered with caution.

Aggressive mimicry

Several hamlets, including H. chlorurus, have been proposed as aggressive mimics, whereby the predatory hamlets (the putative mimics) gain an advantage in the approach and attack of prey by resembling and sometimes actively associating with other non-predatory fishes (the putative models; Randall & Randall, 1960; Thresher, 1978; Puebla et al., 2007), a hypothesis that is still debated (Robertson, 2013). One prediction generated by this hypothesis is that the distribution of H. chlorurus would be expected to match the distribution of its putative model, the yellowtail damselfish Microspathodon chrysurus (Cuvier 1830) (Thresher, 1978). This does not appear to be the case in La Parguera since M. chrysurus shows a negative association with water turbidity (i.e. more abundant on less turbid reefs), the exact opposite of the pattern found for H. chlorurus (Bejarano & Appeldoorn, 2013). This lack of association between putative-model and mimic distributions does not necessarily invalidate the aggressive mimicry hypothesis, but it is noted that specific aggressive mimicry behaviours were not observed in the hamlets from La Parguera during this survey.

Colour pattern variation

The hamlets from La Parguera form discrete phenotypic clusters that correspond to described species, but the taxonomic status of H. nigricans is still to be clarified. Hypoplectrus nigricans from La Parguera matches the description by Aguilar-Perera (2004) and resembles H. nigricans from Barbados (Puebla et al., 2008), but differs from H. nigricans from Panama, Belize and Mexico that is smaller, darker, with short and round pelvic fins (Aguilar-Perera, 2004; Puebla et al., 2008). This suggests that this nominal species might in fact constitute a species complex as proposed by Aguilar-Perera (2004); Puebla et al., (2008) and Lobel (2011). It is also noted that some individuals were tan coloured, but differed from the recently described tan hamlet H. randallorum in lacking the spots on the nose, at the base of the pectoral fin and on the upper part of the caudal peduncle that are diagnostic of this species (Lobel, 2011). Additional data from other locations are needed to clarify the taxonomic status of H. nigricans and establish whether or not there is another tan hamlet species.

In addition, a few individuals appeared intermediates between species, notably between H. chlorurus and H. nigricans and H. chlorurus and H. puella, respectively. Yet it is important to underscore that such individuals were rare, representing <2% of all hamlets sighted within and outside of transects. Considering that hybrid pairings and spawnings have been observed in natural populations (Fischer, 1980a; Barreto & McCartney, 2007; Puebla et al., 2007, 2012a), that there do not appear to be intrinsic post-fertilization barriers between species in the hamlets (Whiteman & Gage 2007) and that in the only case where hybrids were bred in aquaria, they appeared intermediate between parental species (Domeier, 1994), it is plausible that such intermediate individuals might be hybrids.

Perspectives

The hypothesis that hamlet species might respond differentially to turbidity is intriguing and deserves further evaluation since turbidity correlates with a variety of ecological factors whose effect on hamlet communities need to be disentangled. One possibility is that hamlets might differ in their visual sensitivities, which is currently being investigated using whole-genome analysis with a particular focus on opsin genes. Whole genomes will also allow testing whether individuals of intermediate appearance are actually hybrids and if so what type of hybrids (F₁, F₂, backcross, …) and also clarify the taxonomic status of H. nigricans.

The authors are grateful to V. and C. McMillan, R. Pappa, C. Milton, O. Espinosa and the Magueyes Island Marine Laboratories staff for their assistance. This project was funded by grants from the Global Genome Initiative and Smithsonian Institute for Biodiversity Genomics, DFG and The Future Ocean Cluster of Excellence.