Skip to main content
Dryad

Characterizing phenotypic diversity in marine populations of the threespine stickleback

Cite this dataset

Fraser, Ainsley L.; El-Sabaawi, Rana (2022). Characterizing phenotypic diversity in marine populations of the threespine stickleback [Dataset]. Dryad. https://doi.org/10.5061/dryad.bzkh189cx

Abstract

The threespine stickleback (Gasterosteus aculeatus) is an important model for studying the evolution of vertebrate morphology. Sticklebacks inhabit freshwater, brackish, and marine northern hemisphere waters. Anadromous and marine populations (hereafter marine) are assumed to have remained unchanged morphologically from ancestral marine sticklebacks, despite marine environments varying on regional and local scales. Recent studies suggest that genetic and phenotypic structure exists in marine populations, yet the scale of this variation, and its ecological causes remain unclear. Our goal was to assess morphological trait variation in marine stickleback populations around Southern British Columbia (BC) and determine if oceanographic and habitat characteristics were associated with this variation. Between May-July 2019, we sampled 534 sticklebacks from 15 sites around Vancouver Island, a region characterized by a large diversity of oceanographic and habitat features. We characterized trait variation using two-dimensional (2D) geometric morphometric analysis, comparing individuals between oceanographic regions and habitats. We focused on head and body shape. We found that marine sticklebacks varied morphologically among and between regions and habitats, but the variation did not appear to be related to environmental variation. Sexual dimorphism was the largest source of variation, but oceanographic and habitat variables influenced differences between sexes. We concluded that marine sticklebacks offer abundant opportunities for expanding our knowledge of drivers of morphology.

Methods

General Methodology of the Project

Between May – July 2019, a total of 534 fish were collected from 15 sites around Vancouver Island and Southern BC, Canada. At each site, sticklebacks were caught with beach seines at 1–2m depth and 2–3m offshore, or with un-baited minnow traps. We sampled in all four oceanographic regions of Vancouver Island: the Strait of Georgia, the Juan de Fuca Strait, and the western and northern coasts of Vancouver Island. Within each region, we sampled from three different coastal habitat types: tidal flats, salt marshes, and lagoons.

Stickleback Head and Body Photograph Files

The original image files were converted into tps format using the software tps.Util version 1.61, and then organized into files for superimposition. The tps images were uploaded into tps.Dig version 2.05 to digitize landmarks on the head and body photographs. The TPS files are included as well, labelled as “All_Body_Fraser_El-Sabaawi.TPS” and “All_Head_Fraser_El-Sabaawi.TPS”. For the head photographs 13 anatomical landmarks were placed around the left side of the head in tps.Dig. For body photographs, 15 landmarks were placed on the left side of the body.

After the landmarks were digitized, landmark coordinates (X and Y positions) were uploaded into R. We extracted linear distances from each image using the “linear.dist” function in the landvR package for R. Head length (mm) was extracted from headshots (landmarks 1 and 12) as a proxy for head size. Standard length (cm) was extracted from body shots (landmarks 1 and 8) as a proxy for body size.

These lengths, plus individual ID tag, sex, site names, habitat type, oceanographic region, plate morphology, salinity, temperature, mean head length/mean standard length and latitude were included in classifier Excel spreadsheets to upload into R. There is one for the head size/shape analysis and another classifier file for body size/shape analysis.

Generalized Procrustes Analysis

The head and body shape coordinates were analyzed separately. In the geomorph package, the “gpagen” function was used to perform a Generalized Procrustes Analysis (GPA), which is the most common approach for separating shape from size.  Following GPA, 26 vectors of shape were produced for the head dataset (X and Y coordinates for 13 landmarks), while 30 vectors of shape were produced for the body dataset (X and Y coordinates for 15 landmarks).  Additionally, an extra vector was produced for each dataset which described the geometric size of each specimen’s head or body (i.e. CS).

Usage notes

Programs: tps.Util version 1.61 and tps.Dig version 2.05. 

In R, we used the landvR package and the geomorph package.

Funding

Natural Sciences and Engineering Research Council