Sharing morphospaces: early ontogenetic shape changes in two clingfish larvae (Pisces: Gobiesocidae) from the south-east Pacific Ocean
Abstract
Larval body shape changes and developmental timing were examined in two clingfish species from the south-east Pacific Ocean, Gobiesox marmoratus and Sicyases sanguineus. Ontogenetic allometry showed no interspecific variation and <7 mm standard length (LS) larvae of both species occupied similar morphospace, but larger G. marmoratus showed increased body depth while larvae of S. sanguineus developed a flattened head and maintained a hydrodynamic body. Estimated developmental timing suggests that larval body shape changes were faster in G. marmoratus than in S. sanguineus prior to settlement.
Clingfishes (family Gobiesocidae) are globally distributed in many different habitats, in both tropical and temperate waters (Briggs, 1955, 1986). This family includes nine subfamilies, 43 genera and 160 species (Briggs, 1993; Conway et al., 2015). These fishes have complex life cycles, with a switch from pelagic larvae to benthic juveniles and adults. Along the coast of continental Chile, three species of clingfishes have been described: Gobiesox marmoratus Jenyns 1842, Sicyases sanguineus Müller and Troschel 1843 and Tomicodon chilensis Brisout de Barneville 1846 (Pérez, 1981). Gobiesox marmoratus inhabits boulder fields, is carnivorous (Muñoz & Ojeda, 1997; Pardo-Gandarillas et al., 2004), has a hatching size of 2·6 mm and a growth rate of 0·24 mm day−1 (Contreras et al., 2013) and reaches a maximum length of 16 cm. Sicyases sanguineus inhabits the upper intertidal zone; is omnivorous (Paine & Palmer, 1978; Muñoz & Ojeda, 1997; Bernal-Durán & Landaeta, 2017); has a hatching size larger than 3 mm and a growth rate of 0·14 mm day−1 (Contreras et al., 2013); and reaches a maximum length of 30 cm, which is exceptionally large for members of this family (Briggs, 1955). The third species, T. chilensis, is distributed in northern Chile and has an herbivorous diet (Berrios & Vargas, 2004); no information about larvae of this species is available in the literature.
Because of their high abundance in plankton samples off central Chile, G. marmoratus and S. sanguineus were used to study how body shape trajectories and allometry interspecifically vary during larval (i.e. pelagic) stages using geometric morphometrics, which provides a powerful means of visualizing shape differences (Russo et al., 2009; Klingenber, 2013; Mitteroecker et al., 2013). Any association between body shape and size is defined as allometry (Gould, 1966; Mosimann, 1970). The goal of this study is to compare body shape changes from early larvae to presettlement in two clingfish species.
A bongo net was used to collect larvae from nearshore waters off Montemar (32° 57′ S; 71° 33′ W) and El Quisco (33° 24′ S; 71° 43′ W), central Chile, during the austral spring of 2011–2013. All samples (n = 314) were initially fixed in 5% formalin buffered with sodium borate; after 12 h, they were preserved in 96% ethanol. Identification was performed following Pérez (1981). Individuals were photographed and measured under an Olympus SZ-61 stereomicroscope (http://www.olympusamerica.com/index.asp) with a Moticam 2500 (5·0 megapixel) video camera connected to a PC with Moticam Image Plus 2.0 software (www.motic.com).
Nine landmarks (LM) were digitized in lateral views of 96 larval G. marmoratus and 126 larval S. sanguineus (Fig. 1) using the software TpsDig 2.17 (http://life.bio.sunysb.edu/morph/) (Rohlf, 2006). A Procrustes ANOVA was carried out to estimate digitization error, for which the digitizing procedure was repeated for 20 randomly selected specimens (each specimen was slightly rotated and photographed twice). Shape information was extracted from the landmark coordinates with a generalized Procrustes analysis (Dryden & Mardia, 1998). These new coordinates were used for further statistical comparisons. Allometry was estimated by a multivariate regression of shape, represented by Procrustes coordinates, on size, represented by centroid size (CS) (Loy et al., 1998; Klingenberg & Marugán-Lobón, 2013; Mitteroecker et al., 2013). Morphometric data were classified by size group and species for a canonical variate analysis (Table I) that evaluated shape variance among these. Configurations of wireframe graphs were visualized in a two-dimensional coordinate reference system. All multivariate analyses were performed using MorphoJ software 1.05f (http://www.flywings.org.uk/morphoj_page.htm) (Klingenberg, 2011).
| Species | LS (mm) | Size group | n |
|---|---|---|---|
| S. sanguineus | <7 | 1 | 58 |
| S. sanguineus | 7 to 10 | 2 | 47 |
| S. sanguineus | >10 | 3 | 21 |
| G. marmoratus | <7 | 1 | 43 |
| G. marmoratus | 7 to 10 | 2 | 29 |
| G. marmoratus | >10 | 3 | 24 |
- n, Sample size.
To obtain a standardized measurement of time (Kováč, 2002), larval age was estimated utilizing growth models described by Contreras et al. (2013) for larval G. marmoratus and S. sanguineus. Therefore, a length-at-age key was used to estimate individual ages of each individual examined via geometric morphometrics. One-way ANCOVA was run to compare the estimated developmental timing (Procrustes coordinates v. larval age) of G. marmoratus and S. sanguineus, using the software PAST 3.13 (Hammer et al., 2001).
Individuals used for this study varied from 4·6 to 13·2 mm standard length (LS) for S. sanguineus and from 4·6 to 12·4 mm for G. marmoratus. Digitization error was 3·4% (Table II). The relationship between centroid size and LS was well described by a linear model (R2 = 0·96, P < 0·01); hence, CS is a good proxy of body size. Larval body shape of S. sanguineus changed with body size (R2 = 0·81, P < 0·01); the most noticeable shape changes consisted of head elongation (distance between LM1 and LM8; Fig. 1) and elongation of the visceral cavity (horizontal displacement of LM3) and an increase in body height at the level of the head (vertical displacement of LM8 and LM9). Ontogenetic allometry for S. sanguineus accounted for 39·3% of body shape variation. Larval body shape of G. marmoratus also changed with body size (R2 = 0·82, P < 0·01), with principal shape changes being head elongation (increased distance between LM1 and LM8; Fig. 1) and increased body height at the head (expressed by vertical displacement of LM8 and LM9). Body size explained 34·5% of shape variation in G. marmoratus. Ontogenetic trajectories for both species showed the same pattern (F1,219 = 0·004, P > 0·05), with body shape exhibiting mostly allometric variation during early stages [< 13 mm; Fig. 2(a)].
| Effect | Sums of squares | d.f. | Mean square | F | d.f. | P |
|---|---|---|---|---|---|---|
| Individual | 0·2620 | 266 | 9·85 × 10−4 | 29·08 | 1280 | < 0·01 |
| Error | 0·0095 | 280 | 3·39 × 10−5 |
) and Sicyases sanguineus (
) and (b) the relationship between their Procrustes co-ordinates and larval age, showing developmental timing during pelagic larval stages.The first two canonical variates (CV) explained 89% of larval body shape variance, with significant differences among size groups of the two species (P < 0·01 for all combinations). Variation in CV1 (69·7%) was mostly driven by ontogenetic shape changes, such as lengthening of the head, consisting of increased distance between the tip of the premaxilla and the upper margin of the operculum and increased body depth at the level of the operculum (Fig. 3). CV2 (19·3%) described separation of both species in the morphospace of specimens >7 mm LS. This axis was mostly influenced by elongation of the visceral cavity and a flattening of the body to approximate the adult S. sanguineus morphotype (Fig. 3).
, 
; 2, 
; 3, 
). Wireframe diagrams show shape changes (
) from (b) negative to (c) positive values along CV1 and from (d) negative to (e) positive values along CV2. 
, Wireframes corresponding to the consensus shape.Shape changes related to larval age varied significantly between the two clingfish species; developmental timing was faster in G. marmoratus than in S. sanguineus [one-way ANCOVA, F1,219 = 51·40, P < 0·01; Fig. 2(b)]. Additionally, body shape of S. sanguineus larvae did not change after a certain age (approximately 60 days).
This study did not examine subadult or adult specimens and these could have potentially influenced the results. Bernal-Durán & Landaeta (2017) found that allometric growth occurred early in the ontogeny of S. sanguineus (between 4 and 12 mm LS). Above this size, there was little change in body shape. Growth trajectory in that study showed a two-stage pattern: the first stage was characterized by faster shape changes in a small size interval and the second by growth with little change in body shape (Loy et al., 1998). The present study did not observe a two-stage pattern of body shape trajectories in either examined species, but this may have been because juveniles were not examined.
Interspecific differences in larval clingfish body shape change are related to early life traits: S. sanguineus larvae have a hydrodynamic body during most of their planktonic stage and display slower shape changes during larval development. These results are consistent with the slow growth rates (Contreras et al., 2013) and long pelagic larval duration (PLD; Mansur et al., 2014) previously estimated for this species. Additionally, extended developmental timing may be related to the large adult body size attained by this species (up to 30 cm LS), which is an outlier for the Gobiesocidae. Furthermore, G. marmoratus larvae changed rapidly from a slender larva adapted to pelagic life to a deep-bodied individual ready to settle in the benthos. Again, these results are consistent with estimated faster growth rates and shorter PLD (Contreras et al., 2013; Mansur et al., 2014). Therefore, these two common fishes of the rocky intertidal zone along the central Chilean coast display different strategies during their pelagic, dispersive phase prior to settlement.
We appreciate the comments and suggestions of V. Garmendia (Pontificia Universidad Católica de Chile, Chile), C.P. Klingenberg (The University of Manchester, U.K.) and two anonymous reviewers to an early version of the manuscript. Projects Fondecyt 1120868 and Fondecyt 1150296 (Comisión Nacional de Investigación Científica y Tecnológica, Chile) partially funded the writing of this manuscript.
