Diet and body shape changes of pāroko Kelloggella disalvoi (Gobiidae) from intertidal pools of Easter Island
Abstract
This study assesses seasonal variation in the morphology and diet of juveniles and adults of the Easter Island endemic goby Kelloggella disalvoi from intertidal pools during September–October 2015 (spring) and June–July 2016 (winter), utilizing geometric morphometric and gut-content analyses. A set of 16 landmarks was digitized in 128 individuals. Shape changes related to size changes (i.e. allometry) were low (18·6%) and were seasonally similar. Body shape changes were mainly dorsoventral (44·2% of variance) and comprised posteroventral displacement of the premaxilla and bending of the body. The latter included vertical displacement of the anterior portion of the first and second dorsal fins and the entire base of the caudal fin. Diets mainly comprised developmental stages of harpacticoid copepods (from eggs to adults), ostracods, isopods, gastropods and bivalves. Also, trophic niche breadth remained constant throughout development and did not vary between seasons. Nonetheless, significant dietary differences were detected in specimens collected during spring (main prey items: harpacticoid copepods and copepod eggs) and winter (harpacticoid copepods and copepod nauplii). Finally, there was weak but significant covariation between diet and morphology: molluscivores were characterized by having an inferior mouth gape, whereas planktivores had an anteriorly directed premaxilla.
Introduction
Gobies (family Gobiidae) are one of the most diverse marine fish families globally, with at least 1120 species described in 170 genera (Thacker & Roje, 2011). They are commonly recognized by their small size, cryptobenthic nature and unique pelvic fins upon which they often perch (Tornabene et al., 2013). Populations of small, cryptic gobiids are able to reproduce at small sizes [c·10–11 mm standard length (LS) in Coryphopterus kuna Victor 2007 (Victor et al., 2010)] and ongoing recruitment occurs throughout the year, which compensates for their exceptionally short life span on the reef (59–140 days) (Depczynski & Bellwood, 2006; Victor et al., 2010; Lefevre et al., 2016). Additionally, rates of predation on most small gobies are high, resulting in their short life span (Munday & Jones, 1998; Hernaman & Munday, 2005).
Gobies occupy a diverse array of niches and microhabitat types, from coral reefs to sand, mud, rubble and algal flats. These different niches pose constraints on individuals, affecting their body morphology and even brain structure (Yamada et al., 2009; White & Brown, 2015). Nonetheless, gobiid speciation is correlated with diminished rates of morphological diversification, resulting in similar adult body shapes across species (Thacker, 2014).
Ecomorphology is defined as the relationship between morphological factors (i.e. form and function) and environmental factors (i.e. ecology) among individuals, populations, species and higher taxa, or communities (Motta & Kotrschal, 1992; Motta et al., 1995). Morphology influences ecology by limiting the ability of an individual to perform key tasks (Wainwright, 1991). Morphology also changes during the life of an organism; some fish species undergo gradual changes throughout their lives, while others experience ontogenetic threshold effects in their ecomorphological relationships (Russo et al., 2007; Bernal-Durán & Landaeta, 2017). In many marine fishes, body shapes change in a two-stage growth curve, where the first stage is characterized by dramatic morphological change over a small size interval (allometry) and the second by a change in size without a change in shape (isometry) (Loy et al., 1998; Russo et al., 2009). Additionally, closely related species that occupy different microhabitats may exhibit different morphologies (Herler, 2007).
One of the most remote inhabited islands in the Pacific Ocean is Easter Island (Rapa Nui). Easter Island has a relatively depauperate fish fauna, composed of only 139 shore-fish species representing 57 families, of which 21·7% are endemic (Randall & Cea, 2012). Six gobies are recognized from Easter Island (Randall, 2009; Randall & Cea, 2012), five in the family Gobiidae and one (Gnatholepis pascuensis Randall & Greenfield 2001) in the family Gobionellidae. One species, Disalvo's goby Kelloggella disalvoi Randall 2009, lives in high tide pools and sometimes occurs in the splash zone above high tide (Randall 2009). Tide pools in the volcanic rocks of Easter Island may be as shallow as 10 cm, with sharp borders and are exposed to large fluctuations in temperature, salinity and water level. Such an extreme environment may place constraints on the morphology and trophic interactions of K. disalvoi. Randall & Cea (2012) suggested that K. disalvoi must have ‘an extraordinary tolerance to the extremes of temperature and salinity’.
The main goal of this study is to describe seasonal variation in the morphology and diet of juveniles and adults of the Easter Island endemic goby K. disalvoi, utilizing geometric morphometric and gut contents analyses.
Materials and methods
Study area
Easter Island or Rapa Nui (Chile, south-east Pacific Ocean at 27° 09′ S; 109° 22′ W) is an intra-oceanic volcanic island within the Nazca Plate, c. 350 km east of the East Pacific Rise, with a total area of 166 km2 (Vezzoli & Acocella, 2014). The annual cycle of satellite-measured (resolution 4 km × 4 km) chlorophyll a shows highest concentrations in austral winter (July; mean 0·051 mg m−3 during 2002–2012) and reach their lowest values in late October–early November (0·017 mg m−3; Andrade et al., 2014). Sea-surface temperatures at Easter Island also follow an annual cycle, with a summer maximum in February (mean 25·6° C during 2002–2012) and a minimum in mid-August (20·2° C; Andrade et al., 2014). Subtidal substrates are dominated by scleractinian corals, mainly Porites lobata and Pocillopora sp., with 10% covered by crustose coralline algae (Friedlander et al., 2013; Wieters et al., 2014).
Field and laboratory work
Between September 27 and October 2, austral spring 2015 and June 29 and July 07, austral winter 2016, the intertidal zone of Easter Island was intensively sampled. From two locations, Hanga Roa and Vaihu (Fig. 1), 174 K. disalvoi individuals were collected with hand nets and immediately transferred to a plastic box filled with clear water. Later, individuals were treated with an overdose of benzocaine, following Ross & Ross (2008), photographed (Nikon AW130; www.nikon.com), anaesthetised and preserved in 70% ethanol. The institutional committee for Bioethics for the Research with Animals (CIBICA) at Universidad de Valparaíso approved this protocol under resolution BEA048-2015.
) and Vaihu (
), Easter Island, Chile, Southeastern Pacific Ocean.At the laboratory of the university at Viña del Mar, Chile, each individual was photographed twice with a Canon 500d T1 camera (www.canon.com) in order to measure it (mm; total length, LT; height; width; premaxilla length) and estimate measurement error. Individual measurements were not corrected for shrinkage. Stomachs were extracted under a Motic SMZ140 stereomicroscope (www.motic.com) connected to a digital video camera Motic 2500 (5.0 Mpixel) and each prey item was counted and measured (maximum length and width) under the same equipment or under a Motic BA310 light microscope, if the prey were too small to be identified under the stereomicroscope. The volume of each prey item was estimated using the three-dimensional shape that most closely resembled the item, following Cass-Calay (2003) and Sun & Liu (2003). Prey were identified according to Goddard (2003) for copepods, Whatley & Richard (1999) for ostracods, Raines & Markus (2012) for bivalves, Rehder (1980) for gasteropods and Foster & Newman (1987) for barnacles. Three adult K. disalvoi specimens were deposited at Museo Nacional de Historia Natural de Chile (MNHNCL ICT 7537) as reference material.
Morphometric relationships
Total length (LT, mm), body depth (DB, mm), width (WB, mm) and premaxilla length (LP, mm) of individuals collected during spring and winter were correlated and a Shapiro–Wilk test revealed that these measures were normally distributed. Least-squares simple linear regressions were performed in order to compare body dimensions between sampling periods (spring v. winter) using one-way ANCOVA (for a size range of 12–29 mm). All statistical analyses were performed using the software PAST 3.13 (Hammer et al., 2001).
Geometric Morphometrics
Sixteen landmarks (LM) were digitized from photographs of lateral views of the bodies of 128 individuals (Fig. 2) using the software TpsDig 2.17 (Rohlf, 2006; Stony Brook Morphometrics; http://life.bio.sunysb.edu/ee/rohlf/software.html), with a database having been created using the software TpsUtil 1.58 (Rohlf, 2006; Stony Brook Morphometrics; http://life.bio.sunysb.edu/morph/soft-dataacq.html). Poorly preserved individuals (n = 46) were excluded from analyses. Geometric morphometric analyses were performed using MorphoJ software 1.05f (Klingenberg, 2011; Apache Licence, Version 2.0; http://www.flywings.org.uk/MorphoJ_page.htm). A generalized Procrustes analysis was performed in which coordinates were scaled, translated and rotated to a common coordinate system to extract the shape variation from each individual and remove the influence of size and rotation (Rohlf & Slice, 1990). Residuals from the consensus configuration were modelled with the thin-plate spline (TPS) interpolating function (Bookstein, 1991). Principal component analysis (PCA) was performed to identify the main axes of shape change and the specific regions of shape change (reflected as movements of landmarks). Shape changes were visualized using warp transformation grids following Klingenberg (2013).
The digitizing procedure was repeated for 25 randomly selected specimens and two separate Procrustes ANOVAs were carried out for assessing the relative amounts of variation among individuals and of measurement error and to evaluate the effect of ethanol on shape. Ontogenetic allometry was calculated by multivariate regression of shape on size (i.e. centroid size) (Klingenberg, 2016), where allometry is the proportion of variation for which the regression accounts as a percentage of total variation. Regressions slopes (i.e. allometry) were compared between seasons using a one-way ANCOVA.
Diet
Feeding incidence (IF) was calculated as a percentage of the total number of individuals with gut contents, in relation to the total number of individuals examined (Sassa & Kawaguchi, 2004). Comparisons of IF among size ranges and between sampling dates were performed with contingency tables.
Diets were described using the percentage total number (%N) of items that were examined, the frequency of occurrence (%F) of a diet item in each individual with food in its gut and the percentage volume (%V) of each item out of the total volume of prey items. An index of relative importance (IRI) was calculated as follows: IRI = (%N + %V)%F. To readily allow comparisons among prey items, the IRI was standardized to %IRI for each prey item i (Cortés, 1997).
To quantify variability in the feeding success of K. disalvoi, two measures were calculated: the number of prey items per gut (NPPG, number) and total prey volume per gut (VPPG, mm3) (Reiss et al., 2002; Vera-Duarte & Landaeta, 2016). To determine whether these indicators of feeding success were related to LT, Spearman's correlations (rS) were performed between indicators (NPPG and VPPG) and LT. If the correlation was significant, feeding success was assumed to be size-dependent and a one-way ANCOVA was utilized for comparisons between stages and sampling dates; otherwise, Mann–Whitney U-tests were utilized.
Two-way PERMANOVAs were used to assess differences in the prey composition (in terms of number or volume of each prey item) between seasons (winter, spring) and among size groupings (<15, 15–20, 20–25, >25 mm LT). The Morisita index was chosen as a similarity index following Cortés (1997) and 9999 permutations were run using the software PAST 3.13.
Pearre's trophic niche breadth (Pearre, 1986) was adopted to analyse the relationship between prey size and predator size. This model uses the s.d. of the log10-transformed prey size as a measure of trophic niche breadth. In this analysis, individuals were classified according to total length at 0·1 mm intervals. Only individuals with >2 prey items in their gut were used for further analyses. The mean and s.d. of the log10-transformed prey width were calculated for each available K. disalvoi size class. The relationship between standard length (LS) and the corresponding mean and s.d. of the log10-transformed prey size was examined using linear regression analysis to test for shifts in niche breadth with LT.
Ecomorphological interactions
Covariance between body shape and prey composition (number and volume of prey) was calculated using a partial least-squares (PLS) ordination. PLS is a method for exploring patterns of covariation between at least two blocks of variables, which reduces the dimensionality of both data blocks, yielding axes that explain covariance between blocks, ordered from the pair that explains the maximal covariance to the pair that explains the least, all of which are mutually orthogonal (Zeldich et al., 2012). The analysis was done using the morphological matrix as block 1 and the diet composition matrix (number and volume of each prey item) as block 2 using the software MorphoJ (http://www.flywings.org.uk/morphoj_page.htm/).
Results
Morphometric relationships
Specimens of K. disalvoi collected in the intertidal tide pools of Easter Island varied from 11·5 to 32·4 mm LT (spring 2015) and between 11·0 and 28·6 mm LT (winter 2016). During spring, gobies were significantly larger (Mann–Whitney U = 1514, P < 0·001), had a longer premaxilla (one-way ANCOVA, F1,171 = 11·96, P < 0·001) and deeper bodies (one-way ANCOVA, F1,171 = 9·65, P < 0·001), but were thinner (one-way ANCOVA, F1,171 = 7·14, P < 0·001), compared with specimens collected in winter (Table I).
| Model | Intercept | S.E. | Slope | S.E. | r2 | P | |
|---|---|---|---|---|---|---|---|
| Spring | LP v. LT | 27·814 | 46·088 | 42·621 | 2·226 | 0·868 | <0·001 |
| DB v. LT | −0·138 | 0·112 | 0·102 | 0·005 | 0·849 | <0·001 | |
| WB v. LT | −0·318 | 0·126 | 0·144 | 0·006 | 0·898 | <0·001 | |
| Winter | LP v. LT | −51·375 | 24·127 | 44·765 | 1·315 | 0·936 | <0·001 |
| DB v. LT | 0·094 | 0·059 | 0·117 | 0·003 | 0·945 | <0·001 | |
| WB v. LT | −0·044 | 0·058 | 0·101 | 0·003 | 0·849 | <0·001 |
Geometrics Morphometrics
Procrustes ANOVA estimated a measurement error of 0·24% (F1,650 = 40·17, P < 0·001, Table II) and a moderate ethanol fixation effect, explaining 32·1% of all shape change (F1,598 = 2·12, P < 0·0001, Table II). Relationship between centroid size (CS) and LT was well explained by a linear model in spring (R2 = 0·86; P < 0·001) and winter (R2 = 0·81, P < 0·001) and therefore CS provided a good proxy for body size.
| Sum of squares | d.f. | Mean square | F | d.f. | P | |
|---|---|---|---|---|---|---|
| Individual | 0·085 | 624 | 1·36 × 10−3 | 40·17 | 1650 | <0·001 |
| Measurement error | 0·002 | 650 | 3·40 × 10−6 | |||
| Individual | 0·047 | 572 | 8·29 × 10−5 | 2·12 | 1598 | <0·001 |
| Fixation error | 0·023 | 598 | 3·91 × 10−5 |
Shape variation was explained by 28 principal components, of which 12 explained 95·7% of the variance. The first two principal components of shape explained 58·6% of the variance during the late ontogeny of K. disalvoi (Fig. 3 and Table III). PC1 reflected 45·3% of the total variance, characterized by displacement of the premaxilla from an anterior to inferior position (factor loading LM 1 = −0·365, LM 2 = −0·278) and a bending of the body, expressed by vertical displacement of the anterior portion of the first and second dorsal fins (factor loadings LM 8 = 0·278 and LM 9 = 0·328) and the centre and ventral base of the caudal fin (factor loading of LM 15 = −0·296, LM 16 = −0·235). Bending of the body may have been caused by ethanol preservation. PC2 explained 13·4% of the total variance, reflected in a stretching and lengthening of the body. The anterior portion of first and second dorsal fins (factor loadings of LM 8 = −0·307 and LM 9 = −0·246) moved to the anterior part of the body and posterior portion of the anal fin and second dorsal fins (LM 12 = 0·546 and LM 13 = 0·432) moved posteriorly (Fig. 3 and Table III).
) and winter (
).| Landmarks | X–Y | PC1 | PC2 |
|---|---|---|---|
| Anterior tip of the premaxilla | x1 | 0·059 | 0·146 |
| y1 | −0·365 | 0·117 | |
| Posterior tip of the premaxilla | x2 | 0·112 | 0·135 |
| y2 | −0·278 | 0·081 | |
| Centre of the eye | x3 | −0·017 | 0·156 |
| y3 | −0·216 | 0·085 | |
| Insertion of operculum | x4 | −0·158 | −0·166 |
| y4 | 0·048 | −0·009 | |
| Dorsal insertion of the pectoral fin | x5 | −0·067 | −0·085 |
| y5 | 0·121 | −0·019 | |
| Ventral insertion of the pectoral fin | x6 | 0·103 | 0·038 |
| y6 | 0·099 | 0·011 | |
| Anterior insertion of the pelvic fin | x7 | 0·189 | 0·122 |
| y7 | 0·091 | −0·014 | |
| Anterior insertion of the dorsal fin | x8 | −0·194 | −0·307 |
| y8 | 0·278 | −0·050 | |
| Anterior insertion of the second dorsal fin | x9 | −0·108 | −0·246 |
| y9 | 0·328 | −0·127 | |
| Anus | x10 | −0·073 | −0·162 |
| y10 | 0·286 | −0·121 | |
| Anterior insertion of anal fin | x11 | −0·102 | −0·198 |
| y11 | 0·269 | −0·140 | |
| Posterior insertion of anal fin | x12 | 0·042 | 0·546 |
| y12 | 0·044 | −0·070 | |
| Posterior insertion of a second dorsal fin | x13 | 0·143 | 0·432 |
| y13 | −0·012 | 0·032 | |
| Dorsal base of the caudal fin | x14 | 0·110 | −0·213 |
| y14 | −0·162 | 0·140 | |
| Ventral base of the caudal fin | x15 | −0·070 | −0·067 |
| y15 | −0·296 | 0·007 | |
| Centre of the caudal fin | x16 | 0·031 | −0·132 |
| y16 | −0·235 | 0·078 | |
| Eigenvalues | 0·0008 | 0·0002 | |
| Variance (%) | 45·248 | 13·384 |
Post-settlement ontogenetic allometry of K. disalvoi described 18·6% of the shape variation, which primarily comprised a change in the position of the premaxilla and a broadening of the base of the caudal fin (Fig. 4). Allometry did not vary between sampling seasons (one-way ANCOVA, F1,119 = 0·14, P = 0·7).
) and winter (
).Feeding incidence and diet composition
Feeding incidence of K. disalvoi was high, ranging between 97·1% in spring and 98·0% in winter. There was no significant difference in the IF between spring and winter (χ2 = 0·7, P > 0·05).
The diet of K. disalvoi comprised 31 prey items during spring and winter, with different life stage of harpacticoid copepods dominating (Table IV). In spring, the diet was dominated by harpacticoid copepodites (IRI = 23·9%), copepod eggs (IRI = 15·4%), harpacticoid adults of several species (e.g. Perissocope adiastaltus, IRI = 13·0%; Lourinia armata, IRI = 10·7%; Diossacus varicolor, IRI = 7·9%), ostracods (IRI = 5·3%) and other adult harpacticoid copepods (IRI = 4·9%). In winter, the diet was dominated by harpacticoid copepodites (IRI = 27·5%) and nauplii (IRI = 14·5%), isopods (IRI = 8·4%), L. armata (IRI = 8·01%), P. adiastaltus (IRI = 5·85%), ostracods (IRI = 5·78%) and marine halacarids (IRI = 5·70%).
| Prey item | Spring | Winter | ||||||
|---|---|---|---|---|---|---|---|---|
| %N | %F | %V | %IRI | %N | %F | %V | %IRI | |
| Diatom | 0·78 | 13·64 | 0·11 | 0·13 | — | — | — | |
| Invertebrate egg | 0·97 | 9·09 | 0·11 | 0·10 | 0·16 | 5·21 | 0·05 | 0·01 |
| Nematod | 0·13 | 6·06 | 0·02 | 0·01 | 0·42 | 8·33 | 0·29 | 0·07 |
| Trematod | — | — | — | 0·13 | 3·13 | 0·67 | 0·03 | |
| Polychaeta remains | — | — | — | 1·50 | 15·63 | 5·91 | 1·35 | |
| Bivalve | 0·10 | 6·06 | 5·43 | 0·34 | 0·26 | 7·29 | 6·39 | 0·57 |
| Gasteropod | 0·31 | 12·12 | 0·87 | 0·15 | 0·16 | 5·21 | 0·05 | 0·01 |
| Heliacus codoceoae | 0·05 | 3·03 | 0·39 | 0·01 | 0·42 | 7·29 | 0·60 | 0·09 |
| Nodilittorina pyramidalis | 0·13 | 4·55 | 3·67 | 0·18 | 0·20 | 5·21 | 2·16 | 0·14 |
| Nerita plicata | 0·91 | 16·67 | 18·61 | 3·32 | 0·16 | 4·17 | 0·67 | 0·04 |
| Barnacle nauplii | 0·21 | 7·58 | 0·10 | 0·02 | 0·46 | 6·25 | 0·12 | 0·04 |
| Cypris | 0·94 | 25·76 | 1·71 | 0·70 | 5·33 | 53·13 | 3·45 | 5·45 |
| Chthamalid ramus | 0·26 | 3·03 | 0·14 | 0·01 | 1·08 | 4·17 | 0·51 | 0·08 |
| Copepod egg | 25·65 | 57·58 | 0·32 | 15·41 | 8·98 | 30·21 | 0·30 | 3·28 |
| Harpacticoid nauplii | 4·85 | 39·39 | 1·02 | 2·41 | 19·41 | 57·29 | 2·23 | 14·48 |
| Harpacticoid copepodite | 20·38 | 83·33 | 9·86 | 23·92 | 21·33 | 82·29 | 7·23 | 27·45 |
| Harpacticoid copepod | 4·41 | 56·06 | 4·14 | 4·92 | 3·01 | 38·54 | 1·89 | 2·20 |
| Diossacus varicolor | 6·21 | 63·64 | 5·98 | 7·89 | 5·19 | 38·54 | 3·36 | 3·85 |
| Harpacticus sp. | 2·35 | 60·61 | 2·41 | 2·96 | 1·57 | 33·33 | 0·92 | 0·97 |
| Lourinia armata | 7·70 | 53·03 | 10·68 | 10·70 | 5·59 | 60·42 | 5·76 | 8·01 |
| Metis holothuriae | 0·76 | 18·18 | 0·13 | 0·17 | 0·59 | 15·63 | 0·20 | 0·14 |
| Perissocope adiastaltus | 9·94 | 77·27 | 6·45 | 13·00 | 6·08 | 59·38 | 2·35 | 5·85 |
| Tisbidae | 2·40 | 45·45 | 5·11 | 3·50 | 1·54 | 26·04 | 1·65 | 0·97 |
| Copepod remains | 0·65 | 37·88 | 0·51 | 0·45 | 0·36 | 11·46 | 0·22 | 0·08 |
| Crustacean remains | 1·02 | 15·15 | 2·67 | 0·57 | 0·78 | 25·00 | 2·41 | 0·93 |
| Gammaridae amphipod | 0·03 | 1·52 | 0·03 | 0·00 | 0·20 | 5·21 | 4·65 | 0·29 |
| Hyperidae amphipod | 0·16 | 7·58 | 2·64 | 0·22 | 0·33 | 8·33 | 3·22 | 0·35 |
| Isopod | 0·08 | 4·55 | 2·36 | 0·11 | 2·19 | 35·42 | 18·11 | 8·40 |
| Tanaidacea | 0·16 | 9·09 | 1·02 | 0·11 | 0·88 | 19·79 | 3·97 | 1·12 |
| Ostracod | 4·07 | 45·45 | 7·40 | 5·34 | 5·19 | 51·04 | 4·49 | 5·78 |
| Semicytherura sp. | 0·13 | 4·55 | 1·04 | 0·05 | 0·16 | 5·21 | 0·67 | 0·05 |
| Chironomid larvae | 0·08 | 4·55 | 0·51 | 0·03 | 1·05 | 16·67 | 8·34 | 1·83 |
| Apterygota | — | — | — | 0·03 | 1·04 | 0·01 | 0·00 | |
| Chironomid | 0·03 | 1·52 | 0·00 | <0·01 | — | — | — | |
| Trichoptera | 0·03 | 1·52 | 0·77 | 0·01 | — | — | — | |
| Halacarid | 3·52 | 43·94 | 3·17 | 3·02 | 4·57 | 63·54 | 3·11 | 5·70 |
| Pontarachnidae | — | — | — | 0·20 | 2·08 | 0·71 | 0·02 | |
| Unidentified prey | 0·63 | 19·70 | 0·60 | 0·25 | 0·49 | 8·33 | 3·29 | 0·37 |
During spring, the numerically most important prey items were copepod eggs (N = 25·7%), harpacticoid copepodites (N = 20·4%) and adult P. adiastaltus (N = 9·9%); during winter, main prey items were harpacticoid copepodites (N = 21·3%), harpacticoid nauplii (N = 19·4%) and copepod eggs (N = 9·0%). Based on frequency of occurrence, the most frequent spring prey items were harpacticoid copepods (F = 83·3%), adult P. adiastaltus (F = 77·3%) and adult D. varicolor (F = 63·6%) and most frequent winter prey items were harpacticoid copepodites (F = 82·3%), adult L. armata (F = 60·4%) and adult P. adiastaltus (F = 59·4%). Based on volume, Nerita plicata was the most important prey item in spring (V = 18·6%), followed by L. armata (V = 11·0%) and harpacticoid copepodites (V = 9·9%); during winter, the most important prey items were isopods (V = 18·1%), harpacticoid copepodites (V = 7·2%) and bivalves (V = 6·4%) (Table IV).
Feeding success and trophic niche breadth
Number of prey per gut (NPPG) varied between two and 290 prey items (mean ± s.d. = 56·4 ± 47·8) in spring and between one and 110 (30·8 ± 24·2) in winter. NPPG was related to LT during spring (R2 = 0·19, P < 0·01) and winter (R2 = 0·20, P < 0·001) [Fig. 5(a) and Table V); K. disalvoi ate more prey during winter than spring (U = 1870, P < 0·001). Total volume per gut (VPPG) varied between 0·004 and 0·877 mm3 (0·161 ± 0·169 mm3) in spring and between 0·007 and 0·817 mm3 (0·131 ± 0·134 mm3) in winter [Fig. 5(b)]. VPPG was not related to LT in spring (R2 = 0·04, P > 0·05), but it increased as LT did in winter (R2 = 0·17, P < 0·001). One-way ANCOVA indicated similarities in the VPPG consumed by K. disalvoi between seasons (F1,160 = 1·84, P = 0·171, Fig. 5).
) and winter (
).| Model | Intercept | S.E. | Slope | S.E. | r2 | P | |
|---|---|---|---|---|---|---|---|
| Spring | NPPG | 5·359 | 1·848 | 235·770 | 77·811 | 0·191 | <0·01 |
| VPPG (mm3) | −0·004 | 0·007 | 0·367 | 0·307 | 0·035 | >0·05 | |
| Winter | NPPG | −2·915 | 0·867 | 143·690 | 31·808 | 0·197 | <0·001 |
| VPPG (mm3) | 0·003 | 0·004 | 0·610 | 0·146 | 0·174 | <0·001 |
Two-way PERMANOVA detected significant differences in numerical prey composition between seasons (F1,9996 = 9·95, P < 0·001); similarly, diets differed among size groups (F1,9996 = 5·84, P < 0·001) (Table VI). The smallest individuals (<15 mm LT) significantly differed from other size groups in numeric prey composition (Bonferroni corrected P-value <0·001). Numerical prey composition differences between seasons were related to a decrease in copepod eggs (N = 25·0–9·0%) and an increase in harpacticoid nauplii (4·9–19·4%) and cypris (0·9–5·3%). Likewise, prey volume showed differences between seasons (F1,9996 = 5·94, P < 0·001) and among size groups (F3,9994 = 1·70, P < 0·001), while greater variability occurred between seasons (68·0%) than among size groups (19%) (Table V). Differences in prey volume included decreases in N. plicata from spring to winter (V = 18·6–0·7%), L. armata (10·7–5·8%), ostracods (7·4–4·5%) and P. adiastaltus (6·5–2·4%) and an increase in isopods (2·4–18·1%), chironomid larvae (0·5–8·3%), gammarid amphipods (0·03–4·7%) and tanaids (1·0–4·0%).
| Source | Sum of squares | d.f. | Mean square | Pseudo-F | P |
|---|---|---|---|---|---|
| Prey number per gut | |||||
| Season | 2·76 | 1 | 2·76 | 9·95 | <0·001 |
| Size group | 4·86 | 3 | 1·62 | 5·84 | <0·001 |
| Interaction | −12·27 | 3 | −4·09 | −14·74 | >0·05 |
| Residual | 42·73 | 154 | 0·28 | ||
| Total | 38·08 | 161 | |||
| Prey volume per gut (mm3) | |||||
| Season | 2·59 | 1 | 2·59 | 5·95 | <0·001 |
| Size group | 2·23 | 3 | 0·74 | 1·71 | <0·001 |
| Interaction | −17·34 | 3 | −5·78 | −13·28 | >0·05 |
| Residual | 67·01 | 154 | 0·44 | ||
| Total | 54·48 | 161 | |||
During winter, maximum prey width (WMP) ranged from copepod eggs (14·5 µm) to bivalves (771·9 μm). Similarly, in spring the WMP also ranged from copepod eggs (22·4 µm) to bivalves (739·9 µm). LT and trophic niche breadth were independent during spring (rS = −0·004, P > 0·05), but in winter both variables were positively correlated (rS = 0·52, P < 0·001, Fig. 6). Mean ± S.E. trophic niche breadth was 0·21 ± 0·05 in spring and 0·20 ± 0·05 in winter and did not vary between seasons (t-test, t = 0·54, P > 0·05). The widest prey items were bivalves, isopods, N. plicata and adult L. armata; most variability in prey size (=width) throughout ontogeny was due to small prey such as copepod eggs, invertebrate eggs and harpacticoid copepodites (Fig. 6).
) and winter (
).Covariation between shape and diet
The PLS indicated low but significant covariance between body shape changes and diet composition of K. disalvoi based on numerical abundance (RV = 0·11, P < 0·001). PLS 1 explained 88·7% of total covariance [Fig. 7(a)]. Relationships between shape changes (block 1) and diet (block 2) were weak but significant (rS = 0·47, P < 0·01). Using volume of each prey item, PLS analysis also showed low covariance between body shape and diet (RV = 0·07, P < 0·05). PLS 1 explained 64·5% of the total covariance [Fig. 7(b)]. There was a low but significant relationship between morphological changes and prey composition by volume (rS = 0·34, P < 0·001). Morphology of individuals that ate bivalves (i.e. large prey volume) was characterized by having an inferior mouth gape; whereas specimens that had a more anterior premaxilla ate mainly harpacticoid copepods (i.e. adult L. armata).
) and winter (
).Discussion
The goby Kelloggella disalvoi is a small endemic fish species that inhabits intertidal volcanic pools of Easter Island, south Pacific. After settlement, moderate morphological changes occurred that were mainly along the dorsoventral axis. These body shape changes showed a strong relationship with preservative (i.e. ethanol), but this does not account for all shape changes in this axis. No significant differences were detected in ontogenetic allometry between spring and winter. Juvenile and adult K. disalvoi are mainly zoobenthivorous and their diet is composed of the developmental stages of harpacticoid copepods such as eggs, nauplii, copepodites and adults and other small crustacean such as ostracods, cypris, isopods and tanaids, in addition to small gastropods and bivalves. During winter, specimens ate numerically more prey items, but the ingested volume was similar between spring and winter. No seasonal changes occurred in trophic niche breadth. Finally, weak but significant covariance was found between body shape and diet; specimens with an anterior mouth gape ingested more harpacticoid copepods, while specimens with an inferior mouth gape ate more bivalves.
The results found here indicate that K. disalvoi display low morphological variance during post-settlement and adult stages. This is expected because most ontogenetic allometry occurs during the early (pelagic) life stages (Loy et al., 1998; Russo et al., 2009; Zavala-Muñoz et al., 2016). Nonetheless, two morphological changes were observed: posteroventral displacement of the premaxilla and deepening of the caudal fin. The first seems to have implications in prey choice, whereas the second may help adult K. disalvoi move from pool to pool. A closely related species from Guam, K. cardinalis, is capable of moving over damp substratum from one pool to another, using a blenniid-like wriggling of its body. A similar behaviour was observed in situ in K. disalvoi occupying tide pools on Easter Island in both seasons.
The monkey goby Neogobius fluviatilis (Pallas 1844) exhibits larger ontogenetic allometry (26·3%; Čápová et al., 2008), with an increase in body depth in the trunk region and under the first dorsal fin, a decrease in caudal peduncle depth and consistent mouth gape. In the amphidromous goby Sicyopterus lagocephalus (Pallas 1770), a ventral migration of the mouth and cranial remodelling occurs during settlement (Keith et al., 2008). These morphological changes are similar to those described here for K. disalvoi.
Thacker (2014) described an inverse relationship between speciation and morphological diversity within Gobiiformes, with a decrease in the rate of shape change in Gobiidae, suggesting that diversification within this family is subject to morphological constraints and strong stabilizing selection.
Body shape and size are important adaptations to specific environments and changes in body shape can facilitate niche exploitation (Herler, 2007). Red Sea gobies of the genus Bryaninops Smith 1959, which inhabit encrusting or massive corals, have a more depressed body than species occupying branching corals (Gobiodon Bleeker 1856) (Herler, 2007). Kelloggella Jordan & Seale 1905 does not have an extreme body like those fishes, thus, body shape may be less important to adaptation–speciation in this genus (L. Tornabene, pers. comm.).
Crustaceans are a primary food source for K. disalvoi, as well as for most gobies in rocky intertidal pools (Hernaman et al., 2009; Compaire et al., 2016). The goby K. cardinalis feeds primarily on copepods, ostracods, amphipods, small gastropods, marine midges and their larvae, polychaetes and large amounts of algae (mostly Microcoleus, Enteromorpha, Ectocarpus and Sphacelaria; Larson, 1983). The K. disalvoi diet mainly comprised harpacticoid copepods, which are also targetted by several other goby species (Saeki et al., 2005; Hernaman et al., 2009). Interestingly, no algae were observed in K. disalvoi gut contents. Larson (1983) stated that K. cardinalis ingest algae by picking them from the water column, or biting directly at the substrate, while competing with Eviota Jenkins 1903 species. It is possible that the dual effects of lacking other tide-pool goby competitors and having only patchy algal cover around Easter Island (limited to short tufts of Cladophora, Giffordia, Ulva and Laurencia; Santelices & Abbott, 1987) combine to free K. disalvoi from having to rely on algae. Also, K. disalvoi have tricuspid teeth in both jaws (Kuramochi, 1980; Randall, 2005), which is extremely rare in gobies, recurved canines and straight pointed teeth, make it possible for them to feed on a broad range of food resources.
There were significant differences in the size and diet of specimens collected in July (austral winter) and September–October (austral spring). Size differences suggest that two cohorts were studied, indicating a short life-span, at least in the highest tide pools of Easter Island. Diet differences may arise from seasonal variation in the feeding rates, as has been observed in several coral-reef gobies (Hernaman et al., 2009) and because of the seasonal cycle of primary production around Easter Island (primary production peaks in July; Andrade et al., 2014).
We thank J. Bustos and Y. Figueroa for field work assistance during the sampling in Easter Island. M. Goddard (Universidad de Valparaíso) and B. Campos (Universidad de Valparaíso) helped in the identification of harpacticoid copepods and molluscs, respectively. An anonymous reviewer and L. Tornabene (University of Washington Fish Collection) gave important suggestions and comments to an early version of the manuscript. This research was funded by Comité Oceanográfico Nacional de Chile (CONA-Chile), grant CIMAR C21I 15-05 to M.F.L. and C.A.B.
