Computational protocol: Coral architecture affects the habitat choice and form of associated gobiid fishes

Similar protocols

Protocol publication

[…] We assessed body shape in subadult or adult preserved specimens of G. histrio (n = 33, SL > 22 mm) and G. rivulatus (n = 21, SL > 21 mm) from the same reef in the northern Red Sea. All specimens were fixed in 5 % formalin and preserved in 70 % ethanol. The thick epidermal mucus layer, typical for Gobiodon, was scraped off with a scalpel to reveal the anatomical points (landmarks) required for morphometric analysis.A geometric morphometric (GM) approach was used to study shape differences between species of Gobiodon. All specimens were scanned with an EPSON Perfection 4990 flatbed scanner using a 3,200 dpi resolution, according to the protocol of Herler et al. (). A set of 15 anatomical landmarks and 6 sliding semi-landmarks per specimen (Fig. a; Table ) was digitized on the randomized images using the program tpsDig 2.14 and tpsUtil 1.44 (Bookstein ; Rohlf ; Rohlf ). The landmark configurations were standardized for location, orientation, and scale by General Procrustes Analysis (Rohlf and Slice ). The resulting Procrustes shape coordinates were used for statistical analysis and visualization of shape differences.In addition to the GM approach, we further measured body depth (Vd; dorso-ventral body height at ventral fin origin) and greatest head width (gHw; widest horizontal transversal head dimension) in a subsample of similar-sized (<28 mm SL) G. histrio (n = 16) and of all G. rivulatus (n = 21). The body volume was approximated from these two variables together with standard length (SL) by calculating the volume of an ellipsoid SL × Vd × gHw × π × 0.166; the lateral body display area was approximated as SL × Vd × π × 0.25.We assessed the relationship between coral morphology and fish shape using another sample of live fish (21 adult G. histrio, 9 adult G. rivulatus) taken from the occupied corals. Fishes were taken from colonies of A. digitifera and A. gemmifera with clove oil (Munday and Wilson ), allowed to recover, and stored in numbered plastic boxes. In the laboratory, fishes were kept in a 160-l aquarium and supplied with fresh seawater by a flow-through system. Fishes were narcotized again with clove oil, following the protocol of Munday and Wilson (), and laterally scanned on an EPSON Perfection V30 flatbed scanner using a 3,200 dpi resolution, following the protocol of Herler et al. (). SL and body depth were measured on the scans, whereas greatest head width (gHw; widest head dimension), head width (Hw; distance between the left and right upper opercular insertions), and greatest body width (gBw; widest body dimension) were directly measured on narcotized fishes with a digital calliper to the closest 0.01 mm. After scanning and taking measurements, fishes were allowed to recovered in aerated seawater and released back to the reef. [...] We compared the variables mean colony diameter (mean DIA), BL, BTD, and IBD at 7 mm height across occupied and unoccupied colonies of A. digitifera and A. gemmifera by three-way ANOVAs with occupation status, coral species, and reef zone as independent factors. Multivariate differences in these groups were explored by a principal component analysis (PCA) of the log-transformed measurements. To identify aspects of coral morphology that are most important for coral occupation by gobies, we performed a logistic regression of occupation status (0, 1) on the log-transformed variables BL, BTD, IBD, and mean DIA. These analyses were computed with Mathematica 8.0 and SPSS 17.0.In order to explore shape differences between the Gobiodon species, PCA was applied to the Procrustes shape coordinates of the 33 G. histrio and 21 G. rivulatus specimens using the software tpsRelw 1.49 (Rohlf ) and PAST 2.03 (Hammer et al. ). A MANOVA was performed on the first seven principal component scores with PAST 2.03 to test for group mean differences.An ANCOVA was used to test for species differences in allometric relationships between the variables body depth, greatest head width, body volume, and lateral body display area.To identify relationships between the shape of fishes and that of host corals, a two-block partial least squares (2B-PLS) analysis (Sampson et al. ; Streissguth et al. ; Rohlf and Corti ) was performed on the log-transformed coral measurements (BL, BTD, IBD, DIA) and fish measurements (gBw, gHw, Hw, SL, Vd; 21 G. histrio and 9 G. rivulatus) with the program PAST 2.03. Two-block PLS yields linear combinations (latent variables) for each block with maximum covariance between the two sets of variables (in this case, fish measurements and coral measurements). […]

Pipeline specifications

Software tools TpsDIG, TpsUtil, TpsRelw
Application Macroscope & basic digital camera imaging
Organisms Acropora digitifera, Danio rerio