Skip to main content

Phenotypic correlates of pelvic spine coloration in the Threespine Stickleback (Gasterosteus aculeatus): Implications for function and evolution

Cite this dataset

Anderson, Christopher; McKinnon, Jeffrey (2022). Phenotypic correlates of pelvic spine coloration in the Threespine Stickleback (Gasterosteus aculeatus): Implications for function and evolution [Dataset]. Dryad.


Animal color patches may be static or plastic in expression and concealable or continuously visible, yet these aspects of coloration, and their consequences, have been little studied. We address them here using the threespine stickleback (Gasterosteus aculeatus). Despite a rich history of study of stickleback nuptial color pattern evolution, disagreement persists regarding selection pressures and function. However, little research has addressed the role of pelvic spine coloration, a potentially important, and substantially concealable, color pattern element. We investigated (i) whether male pelvic spine (along with throat and body) coloration is relatively static or plastic across the reproductive cycle, (ii) when pelvic spines are raised versus concealed across behavioral contexts, and (iii) associations between color patches and behavior in males. We found no significant variation in spine color across reproductive stages whereas body color was more plastic and intensely red during courtship and egg/fry care. Conspicuousness of pelvic spine coloration instead varied behaviorally, through increased erection frequency during social interactions and in response to a model predator. Spine erection frequency was positively associated with behaviors that enhance spine color visibility, i.e. flees and leads to nest. These findings suggest that stickleback use pelvic spines to display an intensely red color patch facultatively, either as a complement to similar body coloration or possibly as a substitute. In addition, elevated spine raising in the presence of a model predator, together with the presence of red spine coloration in females, raises the possibility that red spine coloration may also have an anti-predator function.


All data collected for this experiment was collected on male Threespine stickleback (Gasterosteus aculeatus) collected from Little Campbell River- Anadromous Site at the start of the reproductive season (May 2017 and May 2018) Please see publication for exact information on fish collection and husbandry.

All data was uploaded into JMP versions 14.1.0 and 15.0.0 for statistical analyses.

Comprehensive behavioral data as a complete data set can be found in the “BehavioralDATA_BORIS_WithPerMinCalculations_DRYAD” data sheet. Behavioral data was collected and analyzed in four steps:

Step 1: fish were videotaped using an HD Sony camcorder positioned in front of the tanks during each of the unique behavioral interactions (including experimental contexts of a nesting male under various presentations, i.e. (i) presentation of a stimulus female within a UV transparent plexiglass container, (ii) presentation of a gravid female for mating, (iii) presentation of a rival male, and (iv) presentation of a model predator. Further, videos were recorded of control males within control tanks being presented with an empty UV transparent plexiglass container).

Step 2: All videos were given random video identifiers by behavioral interaction type so as to prevent observer bias. No videos were analyzed until all videos were completed. Video analysis occurred during 2020 without direct knowledge of the color scores from each fish to again prevent observer bias. All videos were analyzed using BORIS behavioral software, which allowed for ethograms to be keyed in the system and behaviors quantified withing given videos. Not all behaviors could occur for each interaction type, so only videos that could be quantified were (see Supplemental Tables S1 and S2). After all videos were analyzed, the data from each fish across all interaction types was transferred to a comprehensive data sheet.

Step 3: A master spreadsheet was created from all fish and videos, with data of behaviors scored imported from BORIS output and n/a denoting behaviors that could not occur (see Supplemental Tables S1 and S2). These behaviors were keyed in the data sheet as “_total”, which includes the total number of times that a behavior occurred within the video timeline scored. From there, “behavior per minute” calculations were employed to standardize values used in analysis. Each “total” value was divided by the total minutes in which the video was scored and recorded in the data sheet as “_perminute” to indicate the standardized values.

Step 4: All statistical analysis of behaviors occurred within JMP versions 14.1.0 and 15.0.0. To analyze behavioral data by specific interaction type, the master spreadsheet was further broken down into the specific behavioral contexts. For example, if analyzing specific behavioral associations within the “female stimulus context”, only the behavioral values from the Stimulus Female interaction were included in analysis through “include/exclude and hide/unhide” row functions in JMP. Further, all Log1P values were generated within JMP from the “_permin” behavioral values, as well as all Principal Component Analysis (PCA) values for behavioral contexts noted within the publication.

Color data was collected for three different areas of the body, including the ventral throat, lateral body, and pelvic spine. The comprehensive data with complete datasets can be found in the “ColorData_Stickleback_DRYAD” data sheet. It was collected as follows:

1. Ventral Throat Chroma: Following methods outlined within the publication; spectrophotometry was employed to collect reflectance values for two spots at 1nm intervals from 350-700nm along the ventral throat. The stimulation of the cones in the stickleback eye was then calculated for the 350-700nm range in order to provide quantam photon catches from each cone type used in matrix calculations of hue (h.theta, h.phi), and saturation/chroma (r.vec, i.e. the distance from the achromatic center). R.vec values were of particular interest here so were recorded for each fish at each stage, with the r.vec values from each fish averaged to create the final r.vec value used in analysis here (i.e. “MeanThroatColor_r.vec_ByStage”)

2. Lateral Body Coloration: Following methods outlined within the publication, photographs of the lateral body of each fish at each stage were taken. From there, photographs were uploaded into Adobe Photoshop, and RGB values were recorded for each of 9 spots across the body within 5 areas commonly known to express nuptial coloration, including the Lower jaw, operculum, pectoral plate, ventral area between the lower jaw and pelvic spines (i.e. “VPS”), and lateral plates. Color measurements were corrected relative to a gay card in order to standardize the RGB values by dividing the R, G, and B values from each point by the RGB values generated from four points along the grey card, creating Rstand, Gstand, and Bstand. From here, Red Intensity (IR) was calculated by dividing the Rstand by the sum of Rstand , Gstand, and Bstand. The final Red intensity (IR) values were recorded for each spot along the male at each stage and uploaded into the color spreadsheet Please see Fig. 2 of the main text for geographic location along the body for each spot analyzed here.

Principal Component Analysis (PCAs) was used to create an overall value for lateral body coloration. All 9 spots were included and PCAs were generated for experimental males and control males separately, as outlined in the publication.

3. Pelvic Spine Coloration: Following methods outlined within the publication, photos of the erected pelvic spines were taken and uploaded into Adobe Photoshop. Using photoshop, R, G, and B values from each of 14 spots across 8 sections of the pelvic spine were recorded following the same steps outlined above. Final Red Intensity (IR) values were recorded for each spot and averaged to create the value analyzed here denoted as “SpineColor_Iaverage”.

Usage notes

Please refer to the README file.