Quantifying Threespine Stickleback Gasterosteus aculeatus L. (Perciformes: Gasterosteidae) Coloration for Population Analysis: Method Development and Validation

Abstract

Fish coloration plays an important role in reproduction and camouflage, yet capturing color variation under field conditions remains challenging. We present a standardized, semi-automated protocol for measuring body coloration in the popular model fish threespine stickleback (Gasterosteus aculeatus). Individuals are photographed in a controlled light box within minutes of capture, and color is sampled from eight anatomically defined standard sites in human-perception-based CIELAB space. Analyses combine univariate color metrics, multivariate statistics, and the ΔE* perceptual difference index to detect subtle shifts in hue and brightness. Validation on pre-spawning fish shows the method reliably distinguishes males and females well before full breeding colors develop. Although it currently omits ultraviolet signals and fine-scale patterning, the approach scales efficiently to large sample sizes and varying lighting conditions, making it well suited for population-level surveys of camouflage dynamics, sexual dimorphism, and environmental influences on coloration in sticklebacks.

1. Introduction

Fish coloration patterns are the focus of numerous studies of marine and freshwater taxa, and they utilize a broad range of methods [1,2,3,4,5]. Experimental studies are often carried out, yet direct observational field datasets are also common. Studies of coloration in model fish species, including zebrafish, guppies, cichlids, and threespine stickleback, have enabled detailed investigations of the physiology and genetics of coloration [6,7,8,9,10,11,12,13,14,15,16,17].

Fish body coloration serves as a versatile communication channel, closely tied to both behavior and environmental contexts [18,19]. Whether through disruptive or countershading camouflage that helps avoid predators [20], or vivid nuptial displays that attract mates during spawning [21], coloration underpins key ecological and evolutionary processes in fish. Even in pelagic species inhabiting relatively homogeneous environments, variable body coloration, controlled by specialized cell structures, provides adaptive camouflage [8,22].

The threespine stickleback, Gasterosteus aculeatus L., is a popular model species in behavioral and evolutionary biology due to its ecological plasticity, high abundance, and ability to thrive in aquarium experiments. G. aculeatus exhibits color dimorphism during the spawning season, when males become reddish with orange hues and blue eyes, and females mostly remain cryptically colored [23,24]. Studies of nesting behavior in male stickleback demonstrated the adaptive role of nuptial coloration in sexual selection and speciation [25], and those ultraviolet cues are important signals in communication among individuals [15,26,27]. Moreover, variation in spatial distribution and grouping of melanophores, such as the formation of vertical stripes in the posterior body, is likely linked to camouflage coloration important for the adaptation to contrasting niches [28,29].

In the marine environment, sticklebacks spend the non-breeding season in offshore habitats, displaying only cryptic pelagic coloration for both males and females [30]. For spawning, sticklebacks migrate to coastal habitats and streams, where a high proportion of nests are built among the vegetation. Males form a combination of nuptial and camouflage coloration patterns at spawning coastal grounds [24,31]. This allows them to avoid predation and prepare to display when courtship begins. The female stickleback coloration also shifts with the maturity stage, yet has received less attention despite its potential contribution to mating and nest defense [32].

In the subarctic White Sea, stickleback densities on spawning grounds often exceed 100 fish per m², with roughly two females for each male [30,33,34]. Yet nest densities rarely exceed 3–5 nests per m² [35], so only a minority of fish can spawn concurrently. This spatial limitation likely creates variation in hormonal states that translate into differences in nuptial coloration intensity. Moreover, the very high rates of egg cannibalism observed in this population [36,37] can accelerate turnover among nesting fish, further contributing to individual variation in coloration patterns.

Quantifying nuptial coloration has evolved from simple visual scales assigning discrete visual red colored gradients [26,38] to calculating photographic “redness indices” by measuring relative contribution of red, green, and blue in RGB color space [39], segmenting colored pixels [40,41], or using spectrophotometry to derive hue, chroma, and brightness from reflectance spectra [25,42]. Further methods have included focusing on specific landmarks across the body [43].

Despite the range of techniques available for studying stickleback coloration, none readily capture the full complexity of population-level dynamics. Simple visual scores or single-spot measurements may record obvious nuptial flashes or broad camouflage patterns, but cannot accommodate the mixtures of faint pre-spawning hues, cryptic coloration patterns, and bright breeding colors that often coexist within a single fish and vary with habitat. Likewise, laboratory-based spectrometry and landmark-sampling methods lack the flexibility required for large-scale surveys of hundreds of individuals under changing light, background, and behavioral conditions.

In this study, we aim to develop a universal technique for quantifying coloration in population studies of G. aculeatus. Our goal was a method that (1) can track color shifts in individuals and sexes during the spawning season and (2) scales to large field datasets without reducing precision.

We validated this technique on two datasets in which sex-specific coloration is not immediately apparent: fish maintained under standard laboratory conditions, and fish immediately upon capture in the wild. Specifically, we asked the following questions: (i) Can our method reliably distinguish males from females by coloration during the pre-spawning period? (ii) Which body regions provide the strongest discrimination between sexes? (iii) Which coloration components contribute most to these differences?

2. Materials and Methods

The study was conducted at the Educational and Research Station “Belomorskaia” of Saint Petersburg State University, located in the Kandalaksha Bay, the White Sea (66°17.606′ N, 33°38.747′ E). A beach seine measuring 7 × 1.5 m was used to catch approximately 1100 sticklebacks in the Seldianaya Inlet on 5 June 2021. Fish were caught three days after the first sticklebacks were observed in the inshore zone, following migration from open waters where they wintered. At the time of capture, spawning coloration was not observed.

Then, these individuals (hereafter referred to as “experimental” fish) were placed in a pool with dimensions of 2.5 × 1.5 × 0.6 m, with no spawning substrate, and kept for 11 days at ambient outdoor conditions, i.e., running sea water temperature around 12 °C, and under polar day photoperiod conditions. The fish were daily fed with TetraMin^® granules (Spectrum Brands, Sulzbach, Germany). No breeding behavior was recorded during this experiment.

After eleven days, 34 females and 17 males from the pool were randomly selected to comprise the experimental dataset. In this dataset, the standard length (SL) of males was 60.1 ± 1.2 mm, and that of females was 68.6 ± 0.9 mm, i.e., slightly larger than published White Sea values (57.5 mm and 62.8 mm in males and females, respectively) [30]. According to our previous studies, roughly half of the spawners in the White Sea were two years old, with one- and three-year-old fish also represented [33,44]. Sex was first determined by morphological traits and then verified by dissections of all specimens after the experiment was completed. Generally, females of marine White Sea stickleback are larger than males, and they have smaller heads and more fusiform body shapes (see [45] for detailed characteristics of sexual dimorphism). Males and females can also be distinguished by their inner mouth color: males have bright orange lips, while females have pale lips [46].

We also sampled fish (hereafter referred to as “wild” fish) from the Seldianaya Inlet (66°20.284′ N, 33°37.354′ E) on 4 June 2023, four days after migration to the nearshore areas but still before males started the nest building. The same fishing gear was used during the sampling. The total field dataset sample consisted of 30 females and 30 males, all of which lacked typical nuptial coloration, such as a red throat, an opercular region, and bright blue eyes. In this sample, we balanced the sex ratio. Sex was determined in live fish using morphological criteria [45], and, when necessary, confirmed by gonadal inspection via dissection. Mean SL in females was 65.6 ± 0.6 mm, and that in males was 60.1 ± 0.4 mm.

Each individual was photographed alive using a specially designed photo light box measuring 60 × 60 × 60 cm within 1–2 min after capture. The box was painted with white matte paint both inside and out. Two AA-cell-powered bracket lights were used for lighting. The photos were taken with a Canon EOS 650D DSLR camera (f = 18 mm, ISO = 100) and a RICOH WG-6 (f = 4.2, ISO variable). The camera was positioned perpendicular to the shooting plane with the lens facing downwards, always 17 cm from the photographed surface. Fish were placed in the center of the box just after capture (5 specimens per series); this helped reduce distortion at the edges of the image and ensured accurate size representation of the photographed individuals (Figure 1). Pictures were taken in. RAW format with an image size of 3456 × 5184 pixels using EOS D650D and in .jpeg using RICOH WG-6 (image size: 5184 × 3888 pixels). In the resulting images, the pixel-to-mm ratios were 18:1 for D650D and 24:1 for WG-6. For color corrections, we used Datacolor SpyderCheckr24 color card, followed by processing in Adobe Lightroom [47] and SpyderCheckr 1.3 software [48]. Corrected images were saved in .JPEG format and analyzed using ImageJ 1.54g software [49,50].

Measurements of fish body coloration were conducted using CIELAB color space, where L* represents lightness, ranging from 0 (black) to 100 (white); a* represents the green to magenta parameter, ranging from −100 (fully green) to +100 (fully magenta); and b* represents the blue to yellow parameter, ranging from −100 (fully blue) to +100 (fully yellow). A combination of L = 50, a* = 0, and b* = 0 means a point of neutral gray [51,52,53]. General color description in this color space is presented in Figure S1. This model was chosen because it aligns with human perception of color, rather than the fish’s visual apparatus. Prior studies of fish color variability, including mating coloration in male sticklebacks, have also relied on human perception [54,55,56].

Color parameters L*, a*, and b* were measured in ImageJ [49,50] by converting images to Lab channels and sampling 20 × 20-pixel standard sites (SS) (≈1.14 mm² area in the fish body) via the Region of Interest Manager tool. Channel values were extracted with the micaToolbox plugin [57], and SS positions could shift within a 60 × 60-pixel window to avoid glare.

To estimate the per-site pairwise differences in total color change between males and females, and between experimental and wild individuals, ΔE* pigmentation index was calculated using the following formula:

$$E^{*}={(L_2^{*}-L_1^{*})}^2+{(a_2^{*}-a_1^{*})}^2+{(b_2^{*}-b_1^{*})}^2,$$

(1)

where L*, a*, and b* are values of CIELAB chromatic parameters in pairwise comparisons.

For each SS, mean color values were used for ΔE* calculation. Distances between mean color values were categorized based on ΔE* values as follows: irrelevantly perceptually different (ΔE* < 1), slightly perceptually different (1 < ΔE* < 2.3), or clearly perceptually different (ΔE* > 2.3) [58]. For interpreting results, the smallest perceivable difference between the colored patches is considered as 0.5–1.0 ΔE units. For the experiment data calculations, a 34 × 17 comparison matrix was used, and for field data, a 30 × 30 comparison matrix was used.

Statistical analyses were conducted using PAST 5.2.2 [59], JASP 0.19.3.0 [60], Python 3.13 [61] in Anaconda Navigator 2.6.3 software [62], and R 4.5.0 [63] in RStudio 2025.05.0+496 [64]. Because the distribution of parameters L*, a*, and b* differed from normal in some cases (Shapiro-Wilk test, p < 0.05), we used Mann–Whitney tests to compare means from samples with non-normal distribution and Student’s t-test when comparing means from samples with normal distribution. Differences for which the p-value is p < 0.05 are mentioned as significant.

Principal component analysis (PCA) was performed to check the hypothesis that there is sexual dimorphism in body coloration. Prior to analysis, a* and b* metrics were added +100 units to avoid negative values, and each coloration metric values (L*, a*, b*) were standardized across individuals for each standard site (SS). This involved calculating the mean and standard deviation for L*, a*, and b* for each SS, and adjusting individual coloration values accordingly. Once standardized, means or medians (if distribution is not normal) were calculated for each SS in each individual, and used for analysis. Student’s t-test was used to test scores significance between males and females in the experimental data and also within each sex group when comparing experimental and field data. Loadings exceeding 0.3 were included and interpreted.

3. Results

3.1. Standard Sites Selection and Coloration Description

To describe stickleback coloration patterns, we identified eight standard sites (SS) on different parts of the body (Figure 2), taking into account five criteria: (i) variability in coloration intensity across the body, (ii) variability in male coloration during the breeding season, (iii) differences in coloration between males and females during the breeding season, (iv) suitability for measurements on two-dimensional computer images, and (v) practicality to ensure that the analysis would not be excessively labor-intensive. In selecting the SS, we considered both gradient areas (as in [65]) and threespine stickleback-specific coloration patterns ([27,66,67] and references therein).

SS1–SS3 and SS8 (yellow squares in pink rectangles on Figure 2) were previously found to be important for examining changes in cryptic coloration [68], and the coloration of SS4–SS7 is sex-related (red squares in pink rectangles on Figure 2) according to the literature [27,66,67,68].

3.2. Lightness (L*)

In the experimental individuals, color parameter L* ranged in females from 11.3 to 84.3; in males, L* ranged from 3.7 to 69.5. The darkest SS was SS1 (located close to the dorsum) in both sexes: in females, L* was 22.1 ± 4.7 (mean ± SD), and in males, it was 20.4 ± 5.0 (Figure 3). Low L* values were also found at the eye iris (SS6 and SS7). In females, the mean was 34.1 ± 11.2 and 35.5 ± 7.4 for SS6 and SS7, respectively, and in males, these mean values are 18.1 ± 8.3 and 22.3 ± 10.3, respectively. Conversely, the lightest areas were found on the midbody (SS2, SS4) and abdomen (SS3, SS5), with L* mean values ranging from 34.6 ± 7.4 to 55.8 ± 6.6 in females and from 35.9 ± 7.1 to 51.7 ± 7.2 in males. Females generally demonstrated higher lightness (L*) compared to males. This trend was consistent across the particularly cryptic SS, such as SS1 (dorsal side) and SS3 (ventral side).

In the field dataset, the overall L* in females varied from 12.80 to 66.79, and in males, it varied from 8.1 to 62.9. Both females and males had brighter sex-related SS, such as SS4 (mean in females was 54.9 ± 5.5, and that in males was 51.2 ± 7.8), SS5 (mean in females was 60.4 ± 2.5, and that in males was 55.2 ± 9.6), and SS6 (mean in females was 44.3 ± 8.9, and that in males was 25.0 ± 8.7). The darker SSs slightly differed between sexes. The lowest L* in females was in the dorsum (SS1, 17.6 ± 2.4), while males had both dark dorsum (SS1, 20.4 ± 7.3) and iris (SS6 and SS7, 25.0 ± 8.7 and 24.4 ± 10.9, respectively) (Figure 3). No obvious patterns were registered for sex-related and cryptic coloration SS.

Detailed information on each SS for L* and other CIELAB parameters and pair comparisons in males and females, for both experimental and field data, can be found in Table S1.

3.3. Green to Magenta Axis (a*)

In the experimental dataset, parameter a* varied from −6.8 to 10.3 in females and from −9.1 to 4.6 in males. Females demonstrated a pronounced bias toward magenta at several SSs, notably the iris (SS6, SS7), operculum (SS4), and preoperculum (SS5), and these SSs varied from 3.6 ± 1.1 to 5.3 ± 1.8. Experimental females also had greener coloration on the dorsum (SS1)—the mean was −0.6 ± 2.2. Interestingly, individual SS3 values in females varied greatly across the axis. Experimental males predominantly manifested greenish tints, particularly on the frontal body parts (SS4, SS6, SS8), ranging from −2.4 ± 1.8 to −1.9 ± 2.9 (see Table S1 and Figure 4). Coloration significantly differs between females and males for all, except SS1.

The range of the a* parameter in females from the field conditions ranged between −5.0 and 6.2. In males, the range was biased to greenish colors, from −6.9 to 2.4. All mean values from cryptic coloration SS (SS1, SS2, SS3, and SS8) were below zero, ranging between −2.8 ± 1.3 and −0.8 ± 0.9 in females and between −4.2 ± 1.3 and −1.4 ± 1.5 in males. Sex-related coloration SS for both sexes tended to be less greenish, with means in females ranging from −1.6 ± 1.1 to 1.4 ± 2.1 and that in males ranging from −2.4 ± 1.9 to −0.4 ± 1.4. There were significant differences in a* between males and females for all SSs except SS1. When comparing sex-related and cryptic coloration SSs, no clear patterns emerged.

3.4. Blue to Yellow Axis (b*)

Within experimental data, values for the b* parameter (Figure 5) in females ranged from 1.0 to 27.8, and in males, they ranged from −6.9 to 31.4. In females, a distinct yellow bias (positive b* values) was noted at several SS, including the dorsum (SS1, SS8), lateral side (SS2), operculum (SS4), and iris (SS6). Mean values on these SS varied from 14.7 ± 2.9 to 19.2 ± 4.4. In experimental males, a general yellow bias was observed across the six SSs, with the means ranging from 15.9 ± 3.6 to 20.7 ± 7.0 (SS1, SS2, SS3, SS4, SS5, SS8). However, the iris in males had a bias to neutral gray (b* = 0) with means 7.4 ± 8.9 (SS6) and 10.8 ± 5.2 (SS7). Notably, some males had a bias to blue colors in iris in SS6. Some significant differences between males and females were found for SS2, SS3, SS4, SS5, and SS6 (Figure 6).

Across the field data, the b* values in females ranged between −3.7 and 23.0, but all the means were positive and biased to yellow color. The most yellow-biased SS were SS8 (mean: 14.0 ± 3.2), SS2 (mean: 12.6 ± 3.9), and SS3 (mean: 11.1 ± 3.2). In males, the b* parameter ranged from −2.2 to 27.8, and all the means were also positive (Figure 5). The most yellow-biased SSs were SS8 (mean: 15.9 ± 5.8), SS2 (mean: 14.9 ± 6.3), SS3 (mean: 13.6 ± 4.8), SS4 (mean: 13.6 ± 6.4), and SS5 (mean: 11.2 ± 4.8). Statistically significant differences between males and females were found for SS3, SS4, SS5, and SS1 (Figure 6).

3.5. Principal Component Analysis

Ordination of individual fish color profiles L*, a*, and b* broadly separated males and females within experimental and field datasets. There were slight overlaps between the two groups of males, and between males and females from the field conditions (Figure 7). The first component (PC1) explained 26.1% of the total variation, the second component (PC2) explained 22.0%, and the third component (PC3) explained 14.7%. There were high loadings along PC1 for the parameter a* (0.8), along PC2 for the parameter b* (0.7), and along PC3 for the parameter L* (0.5).

Scores in males and females significantly differed along PC1 in the experimental (t = 10.82, p < 0.001) and field datasets (t = 7.12, p < 0.001), along PC2 (t = 5.32, p < 0.001 in the experimental and 4.70, p < 0.001 in the field dataset) and along PC3 (t = 2.35, p = 0.02 only in the experimental dataset). Significant differences between females from the two groups were found only along PC1 (t = 11.96, p < 0.001) and PC2 (t = 11.57, p < 0.001). However, males demonstrated significant difference for PC1 (t = 2.37, p = 0.02), PC2 (t = 7.398, p < 0.001) and PC3 (t = 2.21, p = 0.03). In addition, we observed positive correlations between the a* parameters and PC1, positive correlations between parameter b* and PC2, and positive correlations between the L* parameter and some SSs. There was a group of positive a*-loadings along PC1 exceeding 0.8 for all SSs except SS1. In addition, we found positive b*-loadings along PC2 in all SSs except for SS6. Finally, at SS6 and SS7, loadings for both L* and b* parameters varied in a range from 0.5 to 0.6, and loadings for the L* parameter at SS2, SS4, and SS8 exceeded 0.6.

3.6. Pigmentation Index ($\Delta$E*)

Comparisons of coloration between sticklebacks using ΔE* indicated pronounced variation across experimental and field datasets. In both cases, ΔE* suggested differences in coloration between males and females (

Table 1. Pigmentation index (ΔE*) comparisons between sexes within and between experimental and field datasets. Mean ± SD are in the numerator; minimum and maximum values are in the denominator.

SS	Comparisons
	Females vs. Males, Experiment	Females vs. Males, Field	Experiment vs. Field, Females	Experiment vs. Field, Males
1	6.17 ± 2.77	6.54 ± 5.39	10.30 ± 3.05	9.54 ± 3.28
	(0.81–17.75)	(0.24–23.51)	(3.35–20.40)	(1.64–21.43)
2	13.72 ± 6.54	8.99 ± 3.81	11.33 ± 4.49	13.12 ± 6.55
	(0.87–38.58)	(0.43–23.64)	(1.11–33.19)	(1.89–41.06)
3	11.34 ± 5.64	6.15 ± 2.96	9.71 ± 3.00	11.15 ± 4.27
	(1.69–32.29)	(0.43–16.09)	(2.74–21.73)	(3.34–25.90)
4	14.66 ± 4.49	9.00 ± 4.13	13.19 ± 5.93	17.94 ± 8.31
	(2.15–26.12)	(0.56–23.43)	(2.69–35.39)	(2.31–44.77)
5	9.70 ± 4.33	6.81 ± 4.80	10.66 ± 3.10	12.64 ± 5.59
	(1.20–30.59)	(0.55–23.74)	(4.45–26.49)	(1.55–49.06)
6	17.00 ± 5.96	14.81 ± 6.19	13.85 ± 6.51	11.91 ± 5.58
	(1.60–31.21)	(11.68–37.35)	(1.46–41.84)	(1.01–33.43)
7	12.12 ± 6.55	12.47 ± 6.73	14.64 ± 7.28	12.15 ± 5.30
	(1.39–37.82)	(0.22–35.37)	(3.04–44.43)	(1.61–31.25)
8	10.67 ± 3.99	7.89 ± 3.86	8.94 ± 3.43	10.23 ± 4.85
	(1.45–23.03)	(0.69–21.73)	(1.03–19.83)	(0.64–27.50)

). Most of the SS showed values exceeding 10 units. In experimental fish, we observed high differences in the posterior body SS (SS2 and SS4) and the iris (SS6 and SS7), exceeding 12 units. SS3, SS5, and SS8 had a range between 9 and 12 units. The lowest mean was noted for SS1—lower than 7 units. Fish from the field conditions also show obvious variation across the SS, despite their mean values mostly being lower than 10 units. For instance, iris (SS6, SS7) had the highest mean units, then a group of SS2, SS4, and SS8 had a range of means between 7 and 9 units. All the other SS values were approximately at the same level and were below 7 units.

Then, we compared females and males separately for the experimental and field conditions (Table 1, columns 3 and 4). Females’ means ranged from 8.94 to 14.64 units, but standard deviations were high (from 3.00 to 7.28), indicating substantial variability. All SS series showed were broad (e.g., 1.11–33.19 for SS2 and 3.04–44.43 for SS7). Females exhibited the highest means in SS4 (13.19 ± 5.93), SS6 (13.85 ± 6.51), and SS7 (14.64 ± 7.28), whereas mean values in the remaining SSs were below 12 units.

Mean values of ΔE* in males in the experimental and field datasets varied from 9.54 to 17.94 units, with the highest variability in SS4 (SD = 8.31), as well as the widest range (2.31–44.77). SS5 also showed a broad range (1.05–49.06), mirroring the high overlap observed in females. SS4 had the highest mean value (17.9 ± 8.3), while SS2, SS5, and SS7 formed an intermediate cluster, each with means between 12 and 14 units. In other SSs, mean ΔE* values remained below 12 units.

4. Discussion

Here, we present a novel method for describing coloration in threespine stickleback, designed for use in population analyses. We consider this technique to be effective for (i) describing camouflage coloration, (ii) capturing nuptial coloration dynamics during the spawning season in both sexes, and (iii) tracking the seasonal development of sexual dimorphism. Below, we discuss the validation of this method and examine its advantages and limitations.

4.1. Image Capture Timing: Minimizing Color Shifts

Fish possess highly dynamic chromatophore systems that enable rapid shifts in skin pigmentation, allowing them to match a new substrate for improved camouflage and reduced predation risk. Therefore, it is essential to minimize the interval between capture and image recording. Benthic threespine sticklebacks modify their body coloration very quickly, within 2–2.5 min after changing substrate [27]. In contrast, pelagic stickleback, which typically experience slower habitat shifts, show slower changes, maintaining largely stable coloration for at least 20 min after a change of background [27]. Accordingly, researchers tried to minimize imaging delays: Candolin (1999) completed all stickleback photographs within one minute [69], and French et al. (2018) within 15 min of retrieval [32]. In our study, although we worked with fish recently migrated inshore from the offshore pelagic habitats, and their coloration is, therefore, expected to be closer to that of pelagic fish, we aimed to keep the interval between capture and taking images to just 1–2 min. When designing our protocol, we sought an optimal trade-off between minimizing imaging delays and processing sufficiently large sample sizes for robust population analyses. Nevertheless, more research is required to accurately quantify the rate and extent of stickleback coloration shifts across different habitats and life stages, taking into account their large ecotype and geographical variation, which will allow researchers to better balance rapid imaging with adequate sample sizes.

4.2. Detection Thresholds: Human vs. Analytical Sensitivity

Our method reliably detected mean overall ΔE* differences of ~9–10 units between males and females in samples of 20–30 individuals. By comparison, the classic “just-noticeable difference” for human vision is ΔE* ≃ 2.3 units [70], indicating that the eye is inherently more sensitive than our current approach. However, direct translation of laboratory-derived thresholds obtained under optimal conditions to field settings remains challenging: ambient lighting variability and natural among-individual color variation among individual fish introduce additional noise that can reduce the sensitivity of the human eye and thus obscure small ΔE* shifts.

Despite these limitations, our ΔE* benchmarks suggest that, under favorable conditions, experienced observers familiar with threespine stickleback sexual dimorphism patterns should reliably distinguish male and female groups in modestly sized samples. We believe, however, that in the case of the White Sea stickleback with its well-pronounced sexual dimorphism, sexing based solely on external body shape (e.g., anterior body markings) may be preferable [45]. The effectiveness of such morphological cues, however, depends on the magnitude of sexual dimorphism, which varies among populations [26,71]: populations with weaker dimorphism will be harder to sex using morphology alone.

In our case studies, lightness (L*) accounted for most of the overall ΔE* contrast, overshadowing chromatic parameters a* and b*. Examining chromatic components separately (a* and b*), our method achieved a detection threshold of just 3–6 units, comparable to human sensitivity under ideal conditions. This underscores the need for caution when interpreting earlier coloration studies: while their broad findings definitely remain robust, subtle hue shifts may have gone undetected.

4.3. Nuptial Coloration as a Baseline

As in many other fish species, nuptial coloration is the brightest and most colorful during the annual cycle of threespine stickleback [24,26,31,67,72,73], making it a logical baseline for assessing the extent of coloration change in our study. Because detailed descriptions of sea-spawning White Sea stickleback nuptial coloration are unavailable, we rely on generalized data from other populations. Numerous studies have documented substantial variation in nuptial coloration among threespine stickleback populations [24,26,67,72,73] while also identifying common features that serve as reference points.

Male nuptial coloration has been characterized most thoroughly: males develop a bright red throat and anterior ventral region, along with vivid blue eyes [74,75,76]. Female nuptial coloration is generally much less intense, although some females can also display a red throat [77,78]. To capture breeding patterns, we selected SS4–SS7 on the anterior part of the body. Both sexes exhibit overall body darkening during the breeding period [79,80]. The posterior body region undergoes less dramatic change than the anterior, but still contributes substantially to camouflage coloration, and this region is addressed by SS1–SS3 and SS8.

Although most research involving threespine stickleback has focused on male coloration, recent studies increasingly address female nuptial coloration as well [32,81]. Our data corroborate the study by French et al. (2018) [32] in showing that integrating both sexes within a single study is essential for comprehensive population analyses of sexual dimorphism. Accordingly, our study demonstrates a method that can be used for estimating coloration in male and female stickleback using the same reference points without sex-specific adjustments.

4.4. Method Validation in Experiment and in the Field

Our case study was designed primarily to validate the method’s capabilities rather than to perform an in-depth biological analysis. We examined two datasets of White Sea threespine stickleback captured during the pre-spawning period, when nuptial coloration was not evident. The first experimental dataset consisted of fish maintained, after capture, under homogeneous conditions for 11 days at ~12 °C to minimize among-individual and among-sex variation. Inshore migration in this species is neither synchronous among individuals nor between sexes [33,82], and, indeed, our experimental fish showed no mating behavior, nor did they exhibit the blue iris or red body patches characteristic of full nuptial coloration [83]. The second dataset comprised wild fish sampled immediately upon capture at the onset of inshore migration.

When comparing the datasets, experimental fish of both sexes were overall darker, and males displayed a slightly greener iris, although differences in iris blueness of iris were minimal. These shifts toward breeding coloration likely reflect the extended time that experimental fish spent inshore under cooler conditions, allowing spawning-related physiological processes to advance, albeit at a slower rate due to reduced temperature. Notably, the brighter lighting of the pool compared to the sea (although not formally quantified) did not impede this progression.

Interestingly, among-individual coloration variability—as measured by the SD of initial coloration parameters, PCA, and ΔE*—was higher in experimental fish than in wild individuals, despite the uniformity of laboratory conditions. However, because experimental fish were more advanced in their spawning preparation (as evidenced also by their darker overall coloration; see above), this pattern is not unexpected and likely reflects asynchronous progression toward spawning among individuals. Also, despite the pool’s blue bottom, we observed no increase in the blue component at cryptic SSs. Such effects were reported by Clark & Schluter (2011) [27]. This suggests that background hue alone, in our case, does not drive detectable color changes.

4.5. Site-Specific Variation: Sex-Related vs. Cryptic Standard Sites

According to expectations from generic patterns of stickleback coloration (detailed description for the White Sea stickleback is absent), sex-related SS4, SS6, and SS7 demonstrate larger differences between sexes even in pre-spawning fish, with the notable exception of SS5, which typically represents the well-known red throat coloration of spawning male sticklebacks (Figure 6). Males at SS5 exhibit a slight shift toward greenish rather than reddish hues compared to females, suggesting that red coloration in males may develop later and relatively rapidly over the subsequent weeks.

Males also show red coloration in SS4, although usually less pronounced compared to SS5. Again, we observe no signs of breeding coloration here in pre-spawning fish. Rather, females display more intense yellow and magenta hues, shifting in a direction opposite to what is typically observed during stickleback spawning. These shifts at SS4 are even more pronounced than those at SS5.

A different pattern is observed in iris coloration (SS6 and SS7). In males, the iris is notably darker than in females, accompanied by distinct chromatic differences: along the a* axis, male iris coloration shifts toward greener shades, while along the b* axis, it moves toward bluer tones. These shifts are highly significant in both datasets, suggesting relatively early development of sexual dimorphism in iris coloration prior to spawning. Changes in iris coloration may be attributed to physiological restructuring of iris pigmentation in males, typically transitioning toward blue hues during spawning [84].

Differences at the cryptic SS are generally less pronounced and more consistent across SSs: males are darker and greener. However, since previous stickleback coloration research has primarily focused on the anterior body regions, the extent to which our findings align with the literature remains unclear. The observed female coloration appears cryptic in the vegetated littoral zone dominated by macrophytes and benthic substrate rather than matching the saturated open-sea blues and greens of pelagic waters [15,27]. This coloration pattern aligns with observations of freshwater sticklebacks from Paxton Lake, Canada [27]. In that environment, benthic sticklebacks exhibit green–yellow dorsal coloration, adapting to the spectral heterogeneity of the littoral zone. However, in dense macrophyte thickets, males with bright red coloration often outcompete others, as demonstrated experimentally [85].

4.6. Method Limitations and Future Developments

One limitation of our method is the use of the CIELAB color space, which is based on human color perception [50]. While this approach is widely used in ichthyology [54,55,56], it does not fully reflect the characteristics of the fish visual system, which can perceive a broader spectral range, particularly in the ultraviolet spectrum [15,27,86]. This discrepancy may lead to inaccuracies in the interpretation of biologically relevant signals and communication strategies used by sticklebacks. Accounting for the ultraviolet portion of the spectrum could improve the biological relevance and precision of the analysis. The spectral sensitivity of the threespine stickleback ranges from wavelengths of 350 to 700 nm with a decrease [87], whereas in humans it ranges approximately from wavelengths of 380 to 750 nm [88,89]. Because the spectral sensitivities of humans and sticklebacks still overlap considerably, especially with regard to the main visual signals used in stickleback communication and camouflage—red, green, blue, and yellow. Our method remains valid and informative for analyzing color patterns associated with most behavioral and ecological aspects of this species.

Another limitation of our approach is its inability to capture fine-scale striping and other complex body patterns, which are common in both juvenile [90,91] and adult [27,32,92] stickleback. In these fish, patterned areas (for example, rows of melanophore stripes or nuptial bands) contrast with background areas (the uniform interstripe skin between markings), and pattern-specific variation refers to differences in hue, brightness, or contrast between those two zones. Because our SS are fixed in position and fixed in position, each SS often spans both a patterned stripe and its adjacent background, averaging out any stripe vs. interstripe differences. Consequently, although our method reliably quantifies overall coloration, it may overlook biologically meaningful pattern elements. Addressing this limitation will require special methodological developments, such as increasing the number of SS while reducing their size, or implementing automated image analysis with color- or shape-based segmentation, which is quite common now in the literature [56,93].

Finally, our technique is not yet fully automated and relies on manual placement of anatomical landmarks on fish images to define each standard sampling site. Although this manual step increases processing time, it guarantees precise site localization and consistent color extraction across our heterogeneous image set. This is especially important given shadows and glares on three-dimensional wet objects. Automated landmark-detection methods have been explored in geometric morphometrics [94,95] and fish-specific applications [96,97], but varied fish posture, lighting, and background complexity often still require human correction, and most studies on stickleback rely on expert methods. However, as computer-vision techniques advance—particularly in automated segmentation and landmarking frameworks [57,93], we plan to integrate them into future research.

5. Conclusions

We developed and validated a semi-automated approach for quantifying threespine stickleback coloration. By combining rapid image capture (<2 min post-capture) with standardized CIELAB color measurements at eight anatomically defined standard sites (SS), our method reliably detects both overall (including L*, a*, and b*) and chromatic (including only a* and b*) differences between sexes. In our tests, a sample size as small as 20–30 individuals per sex was sufficient to detect sex differences in coloration, despite some overlap between sexes. Although larger samples may include more overlapping individuals, this is outweighed by the resulting increase in statistical power. Validation on both experimental and wild fish demonstrated that the method allowed us to track subtle shifts associated with preparation for spawning, such as increased body darkness and iris greening in males. Nevertheless, it relies on human-centered CIELAB space, which omits ultraviolet signals crucial to stickleback vision; fixed, relatively large SSs preclude fine-scale pattern analysis; and manual landmarking can introduce observer biases. Despite these limitations, it represents a step toward standardized, quantitative analyses of fish coloration under field conditions. It offers a framework for investigating camouflage dynamics, nuptial color development, and sexual dimorphism. Although optimized for threespine sticklebacks, it can probably be adapted to other fish species to explore the ecological and evolutionary drivers of color variation.