Developmental plasticity and variability in the formation of egg‐spots, a pigmentation ornament in the cichlid Astatotilapia calliptera

Vertebrate pigmentation patterns are highly diverse, yet we have a limited understanding of how evolutionary changes to genetic, cellular, and developmental mechanisms generate variation. To address this, we examine the formation of a sexually‐selected male ornament exhibiting inter‐ and intraspecific variation, the egg‐spot pattern, consisting of circular yellow‐orange markings on the male anal fins of haplochromine cichlid fishes. We focus on Astatotilapia calliptera, the ancestor‐type species of the Malawi cichlid adaptive radiation of over 850 species. We identify a key role for iridophores in initializing egg‐spot aggregations composed of iridophore‐xanthophore associations. Despite adult sexual dimorphism, aggregations initially form in both males and females, with development only diverging between the sexes at later stages. Unexpectedly, we found that the timing of egg‐spot initialization is plastic. The earlier individuals are socially isolated, the earlier the aggregations form, with iridophores being the cell type that responds to changes to the social environment. Furthermore, we observe apparent competitive interactions between adjacent egg‐spot aggregations, which strongly suggests that egg‐spot patterning results mostly from cell‐autonomous cellular interactions. Together, these results demonstrate that A. calliptera egg‐spot development is an exciting model for investigating pigment pattern formation at the cellular level in a system with developmental plasticity, sexual dimorphism, and intraspecific variation. As A. calliptera represents the ancestral bauplan for egg‐spots, these findings provide a baseline for informed comparisons across the incredibly diverse Malawi cichlid radiation.


| INTRODUCTION
Understanding how evolutionary changes in genetic, cellular, and developmental mechanisms generate variation in organismal adult form is a fundamental goal of evolutionary developmental biology (Carroll, 2008;Smith et al., 2018).Vertebrate pigment patterns are excellent traits in which to address such questions, being extremely diverse within and between species.Importantly, color patterns influence a wide range of ecological interactions and are vital for animal adaptation: they can affect courtship behavior, mate preference, predator avoidance and thermoregulation, among others (Elkin et al., 2023;Kratochwil & Mallarino, 2023).
Among vertebrates, teleost fish are especially suited for elucidating the cellular and developmental mechanisms underlying pattern formation, since they show a strikingly diverse array of pigmentation phenotypes and the contributing cells are readily visible in the skin throughout embryonic and juvenile development (Patterson & Parichy, 2019).Yet until now, we have little insight into the cellular and developmental processes underlying variation in teleost color pattern formation, with most of the data coming from a few model species, such as the zebrafish (Danio rerio) and medaka (Oryzia latipes) (Irion & Nüsslein-Volhard, 2019;Kelsh et al., 2009;J. Owen et al., 2021;Parichy & Liang, 2021;Parichy, 2021;Patterson & Parichy, 2019).
Data from these model species show that teleost pigmentation patterns result from different proportions and arrangements of specialized color-bearing cells, chromatophores, which synthesize and store pigments or reflective nanostructures.Common pigmented chromatophore classes are melanophores (black-brown melanin pigment) and xanthophores (yellow-orange carotenoid and pteridine pigments).The most common structural chromatophore is the reflective iridophore; in zebrafish there are at least two iridophore classes, that result in silver or blue coloration, depending on the size, shape and cytoplasmic organization of the reflecting intracellular crystals (Gur et al., 2020).Rarer chromatophore classes include erythrophores (pigmented, red) and two classes of leucophores (structural, white together with orange or black pigment) (Parichy, 2021;Salis et al., 2019;Schartl et al., 2016).
For periodic patterns such as zebrafish stripes, the adult color pattern is determined by chromatophore behaviors in response to interactions between chromatophores and their surrounding tissues.These are beststudied in zebrafish stripes, in which dark stripes consist of melanophores and blue iridophores, while the light interstripes consist of xanthophores and silver iridophores (Patterson & Parichy, 2019).In this system, iridophores respond to early positional information provided by the tissue environment: adult iridophores first appear near the horizontal myoseptum, creating a primary inter-stripe by extensive proliferation.Subsequent stripe formation is self-organizing, emerging from interactions between iridophores, xanthophores, and melanophores that influence the differentiation, migration, proliferation, and death of each chromatophore (Nakamasu et al., 2009;J. P. Owen et al., 2020;Volkening, 2020).
Despite the advantage of such detailed mechanistic studies, focus on the zebrafish stripe system has emphasized naturally selected traits and invariant periodic patterns in pigmentation development research.However, many fish color patterns are non-periodic, intra-specifically variable, and sexually dimorphic, such as distinctive male ornaments in guppies, cichlids and damselfish (Morris et al., 2020;Santos et al., 2014;Wacker et al., 2016).For such ornaments, positional information cues are hypothesized to outweigh the role of self-organizing interactions, unlike periodic patterns (Parichy, 2021).Yet the mechanisms underlying nonperiodic patterns and the emergence of sexuallydimorphic colouration remain unaddressed.Here, we study the development of a sexually-selected and sexually-dimorphic colouration ornament with interand intraspecific variation: the egg-spots of eastern African haplochromine cichlid fishes (Salzburger et al., 2007;Santos et al., 2014).
Eastern African cichlid fish represent one of the most extensive adaptive radiations among vertebrates (Kocher, 2004;Salzburger, 2018;Santos et al., 2023).We use "eastern Africa" instead of "East Africa" to exclude the use of colonial terminology (Pickering et al., 2023).Their remarkable pigment pattern diversity combined with low genomic sequence divergence are an ideal system to study diversification of pigmentation traits (Malinsky et al., 2018).Consequently, eastern African cichlids are increasingly used to study the genetic basis of variable color patterns, such as stripes, bars, blotches, nuptial colouration, and egg-spots (Albertson et al., 2014;Gerwin et al., 2021;Hendrick et al., 2019;Kratochwil et al., 2018;Liang et al., 2020;Salzburger et al., 2007;Santos et al., 2014).The underlying cellular and developmental basis of this diversity remains unknown.To address this, we focus on egg-spots: yellow-red circular markings on the anal fins of males (Figure 1a) of ~1500 haplochromine cichlid species, varying in number, sizes, combined area, color, and position both within and between species (Figure 1a and 1c) (Salzburger et al., 2007;Santos et al., 2014).Due to varying sizes of and distances between each egg-spot within a fin, egg-spots are best described as non-periodic patterns.
At the organismal level, egg-spots have a signaling function during the mating behavior of these mouthbrooding fishes (Wickler, 1962).After spawning, the female gathers the eggs into her mouth; the male then presents his egg-spots to which the female responds by snatching, bringing her mouth close to the genital opening that discharges sperm, fertilizing the eggs.Further, egg-spots function as an honest indicator of male condition, with larger males harboring on average a higher number of egg-spots (Lehtonen & Meyer, 2011).In some species, egg-spots are selected via female choice with females preferring males with many egg-spots (Hert, 1989(Hert, , 1991)).In others, egg-spots play a role in male-male competition, with males with a higher number of egg-spots having an intimidating effect on similarly-sized opponents (Theis et al., 2012(Theis et al., , 2015)).Taken together, egg-spots are sexually selected "badges of status" signaling male dominance and quality.As such, egg-spots offer a unique opportunity to study the cellular mechanisms underlying the development of intraspecific pattern variation, within the context of sexual selection.To study the developmental basis of such pigment pattern variation, we characterize egg-spot formation in the generalist species Astatotilapia calliptera, which is part of the lake Malawi radiation (Figure 1a and 1c).Phylogenetic analysis shows all Malawi cichlid species (~850 species) resulted from three cichlid radiations, each stemming from a generalist Astatotilapia -type ancestral lineage (Malinsky et al., 2018).Accordingly, this species represents the ancestral bauplan of egg-spots for the Malawi cichlid radiation (Turner et al., 2021).We focus on a population of A. calliptera from a crater lake north of Lake Malawi referred to in the literature as Lake Masoko (variant spelling Massoko, as used by the German colonial administration) and known locally as Lake Kisiba (Clark et al., 2022;Turner et al., 2019).
A. calliptera from Lake Kisiba are at an early stage of adaptive divergence into two ecomorphs-benthic (deepwater) and littoral (shallow-water) (Malinsky et al., 2015).Both the "deep-water" and "shallow-water" individuals display intraspecific variation in egg-spot number and color.Using these populations, we aimed to: (1) uncover cellular events contributing to egg-spot development and variation; (2) identify developmental divergence between sexes in anal fin pigmentation; and (3) determine if the external environment influences egg-spot development.Across multiple stages of fin development, we examined pigmentation in both ecomorphs, primarily for replication and secondarily to identify potential differences during egg-spot development.
We find that both iridophores and xanthophores are involved in egg-spot development, with a key role for iridophore aggregations in initializing egg-spot formation.We show that both males and females exhibit initial stages of egg-spot formation, with sexual dimorphism becoming apparent only at later juvenile stages.We find very similar progression of egg-spot formation events between shallowwater and deep-water replicate clutches.Furthermore, we find that the timing of the onset of egg-spot development is plastic and does not correlate with age or size.Instead, iridophore aggregations form sooner when the fish are isolated from conspecifics at early developmental stages.Finally, we characterize variability in egg-spot positions, and features of development which may contribute to adult variation in this pattern.
We propose cellular interactions explain much of eggspot development, suggesting that non-periodic pattern formation can be governed by similar mechanisms as periodic patterns.We also identify cellular behaviors not observed in other systems like zebrafish stripes.Additionally, we suggest that positional information and hormonal mediation regulate egg-spot position, timing plasticity and sex-limited maturation.Our results indicate that a combination of cellular interactions and these additional inputs could be developmental sources of variation in egg-spot patterns.

| Fish housing
Astatotilapia calliptera stocks were kept under constant conditions (25 ± 1°C, 12 h dark/light cycle, pH 8) in 220 L tanks.Fish were fed twice a day with cichlid flakes and pellets (Vitalis).Tank environment was enriched with plastic plants, hiding tubes, and coral sand substrate.Embryos were extracted from mouthbrooders and raised in tumblers until free-swimming juveniles.

| Experimental design
Three imaging series were completed, starting at three different developmental stages (Figure 2).For each developmental stage, two replicates were performed, one with shallow-water individuals A. calliptera 'masoko' and one with deep-water individuals.Individuals were kept in typical social groups of single or multiple clutches until the day imaging started.At the onset of each imaging series individuals were housed individually to keep track of each individual through time.Embryo cohort: Embryos were kept in separate egg tumblers and imaged daily, from when the anal fin condensation became visible (10 days post fertilization, dpf) until the transition to the free-swimming juvenile stage, marked by yolk closure where the skin closes over the yolk.Yolk closure cohort: Juveniles were housed individually upon reaching the free-swimming stage, at the point of yolk closure, and imaged every 3 days until 12 days after yolk closure and every 5 days thereafter.Juveniles were housed in cylinders (14 cm diameter) constructed of plastic mesh allowing for water flow.All cylinders were placed in the same aquarium (220 L) so that the water conditions and temperature were the same for all the individuals in each imaging series.For the first 15 days of imaging, a nylon net was placed within the cylinder, using material from tights and weighed down with sand, to prevent the fish escaping the cylinders due to their small size.Once larger, the net was removed and plastic plants were added to each cylinder as enrichment.Socialized cohort: Juveniles were kept in a social group until 6 weeks post yolk closure for deep-water individuals and until 10 weeks post yolk closure for shallow-water individuals.These individuals were then individually housed in the mesh cylinders described above and imaged every 5 days.Three to 5 days was the maximum imaging frequency for juveniles because daily imaging affected survival.The sample sizes for each cohort were as follows: (1) Embryo cohort: shallow-water n = 8, deepwater n = 8; (2) Yolk closure cohort: shallow-water n = 8 (5 male, 3 female), deep-water n = 6 (3 male, 3 female); (3) Socialized cohort: shallow-water n = 12 (5 males, 7 females), deep-water n = 9 (5 males, 4 females, individuals from two clutches).
The imaging series of the yolk closure cohort and the shallow-water socialized cohort spanned 50 days; the imaging series of the deep-water socialized cohort spanned 150 days.The shallow-water socialized cohort was conducted first as a pilot imaging series, with individuals selected for visibly initiated egg-spot development (existing aggregations) to maximize the likelihood of capturing the completion of egg-spot development.Images permitting measurement of standard lengths were not collected for this pilot cohort.Following assessment of the images, we planned the timing of other cohorts and deep/ shallow-water replicates with the aim of capturing all stages of egg-spot development from fin condensation to mature spots, as well as standard length data.Length of imaging series for each cohort was limited by survival rates and project licence conditions.It was not possible to continue imaging the embryo cohort after yolk closure, because repeated anaesthesia as juveniles after daily anaesthesia as embryos caused low survival rates.The project licence limited juvenile series to 50 days initially, but was amended to 150 days for the deep-water socialized cohort to capture as much of egg-spot development within the same individuals as possible.

| Sex genotyping
DNA was extracted from fin clips stored in 100% ethanol for <2 weeks using the PCR biosystems rapid extract lysis kit or Quick-DNA TM Miniprep Plus kit by Zymo research.Astatotilapia calliptera 'kisiba/masoko' has three XY sex-determining systems (Munby et al., 2021).A multiplex PCR assay was developed with three different primer sets to genotype the three different sex genotypes (Table S1).Two primer sets target a gsdf duplication and TE insertion on chromosome 7 and the other set a TE insertion on chromosome 19.PCR and subsequent gels for visualization were run with negative and positive controls for each of the sex determination systems present in the species (Munby et al., 2021)  presence is indicated by smaller (207 bp) and larger (614 bp) bands than the reference.

| Anaesthesia procedure
Fish were immersed in epinephrine (Sigma-Aldrich [Merck] Product number E4250) (1 mg/mL) to contract pigment cells alongside 0.01% tricaine (Ethyl 3aminobenzoate methanesulfonate, Sigma-Aldrich [Merck], product number E10521) for 5 min, then anaesthetized in 0.02% tricaine until immobilized (approximately 1 min).Fish were imaged for <30 min in a petri dish in 0.01% tricaine to maintain anaesthetization then allowed to recover in tank water.Exposure to epinephrine for 5 min is not sufficient to contract all pigment cells, but longer exposure greatly increases the chance of death, so fish would be unlikely to survive repeated exposure.

| Imaging and image processing
Images from whole mount specimens were taken with a Leica M205 FCA stereoscope with DFC7000T camera.Standard length images of the whole body were taken at ×1, with multiple tiles taken if the specimen was too large to be contained within a single frame.Anal fin images were taken at ×6.3 magnification at multiple focal planes and multiple tiles.All fins were imaged under both darkfield incident and brightfield transmitted light.Each tile of focal planes was merged using the batch process function in Helicon Focus 8, with Method B, radius 8, smoothing 4.Then, the tiles for a single individual were stitched together using the panorama function in Affinity Photo version 1.10.5.
To capture light reflection of guanine crystals contained in iridescent iridophores, anal fins were dissected from killed individuals and fixed overnight at 4°C in 4% paraformaldehyde in 1X phosphate buffered saline (PBS).Samples were rinsed in 1X PBS and mounted in 100% glycerol on a microscopy slide (Thermo Fisher) under a glass coverslip (Corning).Light reflection of iridophores was captured using an Olympus FV3000 point-scanning confocal microscope using a 638 nm laser and 628-648 nm collection window.Confocal imaging was performed with a ×30 (NA 0.95) silicon oil immersion objective to visualize reflection produced by guanine crystals within iridophores in the forming eggspot.Transmitted light was also collected to visualize the surrounding tissue.Raw confocal images were processed in Fiji (Schindelin et al., 2012) to crop and adjust brightness where necessary.

| Event scoring and standard length measurement
Events during fin development were defined as: Xanthophore appearance: any xanthophores visible in the fin, including both the contracted orange circle (responded to epinephrine treatment) and the non-contracted appearance of a dispersed orange hue (no epinephrine treatment or non-responsive) (Figure S1).Iridophore appearance: any iridophore visible in the fin, identified as gray lines in images taken under transmitted light and confirmed as reflective in images taken under incident light.Iridophore appearance at specific positions in the fin: any iridophore visible between the designated numbered fin rays.Xanthophore-iridophore association: when xanthophores directly overlap with densely-accumulated iridophores.Clear outer transparent ring: complete ring without chromatophores surrounding the egg spot.
For each individual, the recorded event timing reflects the earliest image showing the defined characteristic of each event.Standard length was measured from the tip of the snout to the caudal peduncle in whole-body stereomicroscope images or photographs, using the straight line tool in Fiji (Schindelin et al., 2012).Event timings relative to days and standard lengths were plotted in R using ggplot2 (Wickham, 2016).Survival (time-to-event) analyzes were conducted in R with ggsurvfit (Sjoberg et al., 2024) and survival (Therneau, 2023).Effect sizes of the differences in event timings between cohorts were tested in R with Hedges g using effectsize (Ben-Shachar et al., 2020).Significance of differences in event timings between cohorts were tested in R using the Kruskal-Wallis test when only two groups were being compared, and pairwise Wilcox with Bonferroni adjustment when multiple groups were compared (Table S3).Event scoring data is available in File S1 and R scripts on GitHub (https://github.com/Santos-cichlids/Developmental-plasticity-and-variability-inegg-spot-formation-in-Acalliptera).For egg-spot positions, positions are described by the fin rays that the aggregations are bounded by at either side.Positions are measured on the final imaging session for mature male juveniles.Only mature egg-spots with dense and defined cellular aggregations were considered.

| Image segmentation
Image segmentation for quantification of melanophores and xanthophores was performed using Affinity Photo, ilastik and Fiji.Melanophore coverage was quantified across the whole fin in stages before egg-spots began developing; for each image the background was removed using the selection brush tool in Affinity Photo.Cell segmentation was performed using the machine learning image segmentation tool ilastik (Berg et al., 2019).One image from each imaging day was used for the training dataset using the pixel classification tool.Feature selection parameters color/intensity, edge and texture were all given a sigma of 10.Iterative learning cycles were performed until the cell type of interest was appropriately classified.To compensate for background colouration changes between imaging sessions, images were batch processed by imaging session.Outputs were exported as simple segmentation tiff files and reimported to Fiji for quantification of chromatophore coverage and number using the Analyze Particle function.
Quantification of xanthophore distribution was performed in a region outside of the developing egg-spots to avoid noise from other chromatophore types.Only a defined region of the fin was measured as a proxy for xanthophore distribution across the anal fin: a 0.5 mm 2 square over the first segment of the 8 th fin ray was extracted from each image using Fiji.The outside of the image was cleared and the remaining 0.5 mm 2 image exported.To segment both contracted and uncontracted xanthophores, the extracted images were imported into ilastik.Two iterations of the pixel classification tool in ilastik on each xanthophore type were performed, with one individual from each imaging session selected for the training set as described above.Cell counts and area measurements were performed in Fiji as described above, and the sum of the area of both xanthophores was used as the total area coverage within the sampled region.To analyze the effect of sex and time since single housing on xanthophore coverage we used Pearson's correlation to measure the strength and direction of xanthophore area coverage over time for both the shallow and deep-water replicates.Further, we used a repeated measures linear mixed-effect model to account for the fixed effects of sex, time, their interaction and the random effect of individual variability.For further details see our GitHub page (https:// github.com/Santos-cichlids/Developmental-plasticity-andvariability-in-egg-spot-formation-in-Acalliptera).

| Male and female anal fins have different pigment cell composition
We first sought to characterize the cellular basis of sexual dimorphism in the adult anal fin, finding differences in the presence and arrangement of pigment cell types between males and females (Figure 1a-h).The conspicuous eggspots in sexually mature A. calliptera 'kisiba/masoko' males are composed of densely packed iridescent iridophores and yellow-to-orange xanthophores (Figure 1i,j; Figure S1), similar to Astatotilapia burtoni (Santos et al., 2014).The egg-spots' conspicuousness is enhanced by contrast from an outer transparent ring.This region has a reduced amount or is devoid of pigment cells altogether (Figure 1i,j).Few xanthophores are present outside the egg-spot in males, while iridophores are also sparsely distributed across the anal fin (Figure 1c and 1i,j).Iridophore identity was confirmed from iridescence under incident light (Figure 1i,j).Melanophores are distributed across the anal fin, but absent in egg-spots (Figure 1c, 1e,f, 1j).Additionally, males have red erythrophores on the fin's distal edge (Figure 1c and 1e,f).
In contrast, females display a homogenous distribution of xanthophores resulting in an overall yellow anal fin (Figure 1d and 1g,h).Upon close inspection, we found two morphologies of xanthophores in the female fins.One type contracted in response to epinephrine treatment, revealing a dark orange center surrounded by a yellow halo (Figure 1h orange arrow, Figure S1).The other type was lighter in color and unresponsive to epinephrine (Figure 1h, yellow arrow, Figure S1).This unresponsiveness persisted with higher concentrations and longer exposure times (Figure S1), but whether color and contractility differences represent a difference in cell type remains to be determined.The presence of two xanthophore morphologies in females may suggest that such cells are also present in male egg-spots, and perhaps the high density of egg-spot cells impedes identification.Adult female anal fins also have melanophores and iridophores sparsely distributed across the anal fin but lack erythrophores (Figure 1d and 1g,h).
To investigate the developmental origins of these patterns, we characterized the sequence of cellular events from the onset of overt embryonic fin development to adult stages.

| Iridophore aggregations initiate egg-spot formation in both sexes
To identify the chromatophores and cell behaviors that initialize egg-spot development, we characterized the earliest visible colouration events.For this purpose, we imaged the same individuals through time from the onset of overt anal fin development to juvenile stages (Figure 2).We imaged the fins of: (1) embryos, single housed at 10 dpf, up to yolk closure (embryo cohort, Figure 2a); (2) early juveniles, isolated at yolk closure for a 50-day period (yolk closure cohort, Figure 2b).For replication purposes, for each cohort we imaged a deep-water and a shallow-water clutch.We found a series of events during early anal fin pigmentation development that was consistent between individuals, replicate clutches, and cohorts.
First, background pigmentation develops in the fin, with melanophores appearing in embryos followed by xanthophores (Figure 3a,b).Melanophore and xanthophore coverage increases throughout embryonic stages (Figure 3c,d and 3m).At the transition between embryonic and juvenile development (yolk closure), the anal fin has melanophores homogeneously distributed throughout the fin (Figure 3c-f) and xanthophores spread across the anterior up to the 6 th fin ray, all contracting in response to epinephrine (Figure 3c-f).Accordingly, xanthophores were present in the fin at the start of imaging for all individuals in the yolk closure cohort.Melanophore numbers decline in early juveniles soon after yolk closure (Figure 3h and 3m,n), while contracting xanthophore coverage continues to increase (Figure 3e-j).
Iridophores first appear in the anterior hard-spined domain of the anal fin in the days surrounding yolk closure (19-23 dpf) (Figure 3d, 3f,g, 3n) coinciding with peak melanophore numbers (Figure 3n).The first appearance of iridophores occurs concomitantly with the yolk closure period, more specifically it occurred at the end of the imaging series for the embryo cohort and at the beginning for the yolk closure cohort.Iridophores are elongated cells, appearing as faint gray lines under transmitted light and reflective under incident light (Figure 3f,g and 3l,m).Iridophores first appear near the 3 rd fin ray (Figure 3c-g) in the first 0-3 days in the yolk closure cohort (Figure S3a,d) with a median of 0 (Kaplan-Meier survival median), and then appear between the 4 th and 5 th rays (Figure 3h,i) from 3 to 9 days after yolk closure (Figure S3a,e) with a median of 3 (Kaplan-Meier survival median).Notably, the first aggregation to develop always formed centered between the 4 th and 5 th fin rays (Figure 3j,k) making the arrival of iridophores to this location the earliest visible event of egg-spot formation.When iridophores are present between the 4 th and 5 th fin rays, non-contracting and contracting xanthophores are seen in the rest of the fin.
Next, an increasing number of iridophores accumulate between the 4 th and 5 th fin rays (Figure 3j,k), often spreading posteriorly up to the 6 th fin ray.Iridophores become so numerous that individual cells are hard to distinguish (Figure 3k,l).As their number increases, iridophores become closely associated with contracting xanthophores (Figure 3k-m) from 8 to 24 days after yolk closure (Figure S3a,f) with a median of 14 (Kaplan-Meier survival median).In epinephrine-treated fins, the xanthophores in these associations have a dark orange color, and a lighter halo is either much smaller or cannot be seen, which may be due to stronger contraction, obstruction by associated iridophores, or indicate a different cell type.This association between iridophores and xanthophores is observed with high frequency and consistency.However, it is unclear whether the xanthophore is simply overlaying the iridophore or if there is a stronger intercellular connection between the two cells in the association.
The described order of developmental events that contribute to the initiation of egg-spot development is consistent between individuals and replicate clutches (Figures S2 and S3).However, the timing of events was later in the shallow-water clutch than the deep-water clutch, with greater timing variation for later events between morphs and individuals (Figures S2 and S3; Table S2), indicating inter-clutch variation and less robust timing for later egg-spot initialization events.Whether this inter-clutch variation reflects differences between morphs would require further replicate clutches.
Taken together, these results suggest iridophores have a role in initializing egg-spot formation despite not being the first chromatophores in the fin.The initialization by iridophores is followed by the association with xanthophores, while melanophores and erythrophores do not seem to play a role in these initial stages, consistent with the cellular composition of adult egg-spots.Surprisingly, both males and females show the same developmental trajectory in these early stages.All individuals develop iridophore aggregations, with no difference in event order or growth rate (Figure S4) and no significant difference in event timing (p = .5by logrank survival test for all events), despite mature egg-spots being a sexually-dimorphic trait in A. calliptera.

| Male-limited transparent ring formation follows a shift in xanthophore density
At the end of the imaging of the yolk closure cohort (50 days post yolk closure) (Figure 2b), all individuals displayed large iridophore and xanthophore aggregations, but lacked distinctive features of adult male eggspots, such as the chromatophore-free transparent ring surrounding each egg-spot (Figure 1a, 1c, 1i,j).To determine the cellular events leading to the formation of mature egg-spots and to characterize developmental divergence between the sexes, we imaged anal fin pigmentation in late juveniles (socialized cohort, Figure 2c).These individuals were initially kept together in a typical social stock tank and transferred to single housing for imaging ~6-10 weeks after yolk closure.We followed two replicates: shallow-water individuals isolated after ~10 weeks showing early aggregations followed for 50 days, and deep-water individuals isolated F I G U R E 3 (See caption on next page).
after ~6 weeks followed for 150 days (Figure 2).The shallow-water socialized individuals were isolated at the latest stage as this was conducted first as a pilot imaging series, with individuals selected for visibly initiated eggspot development (see methods Section 2.3 for details).The deep-water socialized individuals were isolated earlier than the shallow-water socialized replicate to capture as much of egg-spot development as possible in a single series (see methods Section 2.3 for details).
The deep-water socialized cohort exhibited the same consistent order of events leading up to egg-spot initialization (Figures S5 and S6).Similarly to the yolk closure cohort, the initial aggregations formed in the socialized cohorts contain iridophore-xanthophore associations and are initially similar between males and females (Figure 4a and 4f; Figure S7).Subsequently in all males, xanthophore coverage gradually decreased outside the egg-spot region (Pearson's correlation R = −0.75,p < .0001)(Figure 4a-c; Figure S7) while female xanthophore distribution across the fin remained stable (Pearson's correlation R = −0.09,p = .53)(Figure 4f-i; Figure S7).A significant interaction between sex and days shows that there is significant decrease in xanthophores over time specific to males (LMM shallow-water: beta = −0.086683,t-value = −8.147708,p < .00001;LMM deepwater: beta = −0.019255,t-value = −5.829644,p < .00001)and occurred concurrently with an increase in size and density of the iridophore-xanthophore aggregations.These became increasingly dense in males (Figure 4a-c; Figure S7) but not in females (Figure 4f-h; Figure S7).As aggregations grow, dark contracting xanthophores similar to those in eggspots appear elsewhere in the fin, without association with iridophores, but only persist in females (data not shown).The appearance of the transparent ring was the final major event in egg-spot formation (Figure 4c) and is similarly limited to males, coinciding with reaching a near-total loss of xanthophores outside of the egg-spot (Figure 4c and 4i; Figure S7).As the transparent ring forms, melanophores and iridophores appear sparsely distributed elsewhere in the fin (Figure 4d,e), and red erythrophores appear at the distal edges (Figure 4d,e; Figure S7).
These results show divergence between sexes in late juvenile stages follows a common developmental program of cellular behavior from embryonic to early juvenile stages.In males, aggregations of iridophores and xanthophores mature into egg-spots, while in females aggregation growth plateaus and widespread xanthophore coverage persists.Therefore the earliest visible point of divergence between sexes is the concordant growth of aggregations and reduction in xanthophore distribution across the fin in males (Figure 4b  and 4g).Females at the end of our series continue to display small aggregations despite their absence in most lab-bred female adults (Figure 1b).Further imaging would be needed to characterize their complete loss.

| The onset of egg-spot development is plastic and dependent on the social environment
The social environment can impact cichlid pigmentation development with correlated hormonal and behavioral changes (Korzan & Fernald, 2007;Korzan et al., 2008;Maan et al., 2006), thus we sought to test if the social environment of A. calliptera individuals influences eggspot development.To test this, we took advantage of our three imaging series starting at three different developmental stages (10 dpf embryo, yolk closure juvenile and late juvenile; see Figure 2), with individuals being singly housed at different ages.For each event in egg-spot initialization, we contrasted the standard length and number of days since the first imaging session.The number of days reflects the elapsed time since the individual was single housed, and standard length is a descriptor of developmental stage, a preferred indicator over dpf (Parichy et al., 2009;Singleman & Holtzman, 2014) (Figure S8).We reasoned that if there is plasticity in egg-spot development induced by the timing at which individuals are single housed, then for such events, days since single housing would be a better predictor of event timing than developmental stage.
The timing of the two first events, appearance of xanthophores and iridophores in the fin, is better predicted by developmental stage than time in single housing (Figure S9).The first appearance of xanthophores is always during embryo stages (Figure S9; Table S3) and the first appearance of iridophores in the fin around yolk closure (Figure S9; Table S3) between 1 day prior and 3 days after yolk closure (Figures 3n  and 5c,d) causing iridophores to be visible in the fin typically from the first imaging day in the socialized cohort (Figure 5f; Figures S6 and S9).However, after the first appearance of each pigment cell, only xanthophorebased pigmentation progressed with developmental stage.Consequently, on the first day iridophores are observed in the fin for each cohort, the appearance of iridophores is similar between cohorts, with very few cells in only the anterior (Figure 5c,d and 5f).Meanwhile, the xanthophore-like coverage is lowest in the embryo cohort (Figure 5c), greater in the yolk closure cohort (Figure 5d) and greatest in the socialized cohort with both contracting and non-contracting morphologies visible (Figure 5e).Changes in iridophore-based pigmentation first occur only in the days after single-housing regardless of each cohort's developmental stage.Iridophores appeared between the 4 th and 5 th fin rays 0-9 days after single-housing in the yolk closure cohort (20-29 dpf) and 0-10 days after single-housing in the deepwater socialized cohort (~42-52 dpf) (Figure 5a, 5e, 5g).Associations appeared 8-24 days after single-housing in the yolk closure cohort (28-44 dpf) and 5-25 days after single-housing in the deep-water socialized cohort (~47-67 dpf) (Figure 5a, 5e, 5g).Only in the shallow-water socialized cohort were iridophores present between the 4th and 5th fin rays and associations present on the first day of imaging (n = 12 out of 12, n = 10 out of 12, respectively) (Figure 5a), due to these individuals being singly housed at the latest stage of all cohorts, with individuals selected for existing aggregations (see methods Section 2.3).Together, this demonstrates that when raised in a group, egg-spot development involving iridophores only progressed beyond appearance in the fin at later stages, but can occur sooner upon earlier isolation.Consequently, when socialized shallow-water individuals are excluded due to the bias towards existing aggregations, the standard lengths at the point of these later iridophore events are significantly different between yolk closure and socialized cohorts (Figure 5b; Table S3), with large effect sizes for the standardized differences between cohorts (Table S4) but the days since initial imaging sessions are not significantly different (Figure 5a; Table S3) and effect sizes are very small to moderate between cohorts (Table S4).Days in single housing are therefore a better predictor of the timing of iridophore-based egg-spot formation events than developmental stages.Furthermore, the duration of egg-spot initialization was unaffected by the developmental stage at which isolation commenced (Table S3).
We infer that individual isolation can induce aggregation development, regardless of developmental stage and fin background pigmentation.This indicates the initiation of iridophore aggregation is environmentdependent and the progression of xanthophore-based background fin colouration is independent from social environment, until it responds to iridophore events (see Figure 4).Therefore, we suggest that iridophores are the social environment-dependent cell type which cause a change in the fin colouration development thitherto.

| Competitive interactions between aggregations likely contribute to egg-spot pattern variation
In A. calliptera, there is variation in the number of eggspots on the anal fin.Egg-spots continue to be added as juveniles and adults grow, adult A. calliptera 'kisiba/ masoko' typically have 2-17 egg-spots (unpublished data).We asked whether aspects of development of later-forming egg-spots could facilitate number variation, and how early this variation arises.For this purpose, we examined the development of the first three egg-spots in males in our socialized cohort.
Unlike the first egg-spot, the formation of subsequent egg-spots is dynamic and variable between individuals.Multiple aggregations may initially form adjacent to each other, separated by single fin rays (Figure 6a, white arrows), or more distant to other aggregations, separated by multiple fin rays (Figure 6b, white arrows).In some fins, the first aggregation is initially large and spans multiple fin rays but in these cases, this first aggregation later reduces in spread (Figure 6c, white arrows).
When subsequent aggregations are initially adjacent, the middle aggregation later reduces in cell density and disappears, with flanking aggregations developing into separate egg-spots (Figure 6a, white arrows).When initially distant, if a growing aggregation becomes close to another, a similar process of density reduction and cell disappearance occurs at the aggregations interface (Figure 6b, white arrows).Curiously, in one fin an aggregation started forming in the anterior spiny ray region, but similarly disappeared when a later-appearing aggregation in the typical first egg-spot position began maturing (Figure 6d, white arrows).This shrinking of aggregations in close proximity to others indicates a form of competition between aggregations.Moreover, after an initial appearance, aggregations may disappear for multiple days before forming again and maturing (Figure 6c, gray arrows) or shift their position across a fin ray (Figure 6c, green arrows).
To understand the impact of the more variable formation of subsequent egg-spots, we compared eggspot positions among the socialized cohorts at the end of the series.At this stage most males show the adult phenotype: dense mature aggregations with an outer transparent ring (n = 10 out of 11) (Figure S10).Of these, most harbored three mature egg-spots (n = 9 out of 10) (Figure S10), one individual of which showed an additional, later-appearing aggregation between the first and second mature egg-spots (Figure S10k).The remaining male harbored two mature egg spots and a small aggregation (n = 1 out of 10) (Figure S10b).We found the position of the first egg-spot is the most consistent, always covering at least the region between the 4 th and 6 th fin rays (n = 10 out of 10), with most (n = 6 out of 10) extending more anteriorly, covering the region between the 3 rd and 6 th fin rays.One extended to the 7 th ray (Figure 6e; Figure S10; Table S5).The second and third egg-spots have more positional variability.The second egg-spot was the most variable, most frequently (n = 3 out of 10) occurring between fin rays 6b and 8b, followed by (n = 2 out of 10) 5b to 7b.All other positions for the second spot were unique (n = 6 out of 10) (Figure 6f; Figure S10; Table S5).The third spots' most frequently covered the regions between 8b and 9b (n = 2 out of 9), 9a and 10 (n = 2 out of 9), and 8b and 10 t (n = 2 out of 9) with the others unique (n = 3 out of 9) (Figure 6g; Figure S10; Table S5).All third eggspots were on the distal posterior edge of the anal fin (Figure S10).
This variability could result from the competitive interactions between growing aggregations; with earlier initiation and therefore greater size, the first aggregation may be less affected, explaining its greater positional consistency.Despite the variation in the precise position of the second and third egg-spots, the overall phenotype was similar (Figure S10), indicating that similar outcomes can be achieved by a noisy formation process.
In later adult stages, egg-spots continue to show apparently competitive interactions and can shift positions (Figure S11) indicating that egg-spot patterns are dynamically changing throughout adulthood even after each reaches a mature appearance.

| DISCUSSION
Elucidating the cellular and developmental processes underlying pigment pattern formation is essential to understand how variation can emerge.In this study, we (e-g) Proportion of males from the socialized cohort with an egg-spot at each position for the 1st (e), 2nd (f), and 3rd (g) egg-spots.Positions are described by the fin rays that mature egg-spots are bounded by on the final imaging session.Each of the three initial spots are diagrammed separately.Where lines bordering the bars would overlap, the exact height and width of the bars are jittered so that each bar can be distinguished-it is intended as a visual guide only, (see Table S5 for precise data).
were interested in characterizing the development of the haplochromine cichlid egg-spots-a non-periodic male ornament harboring high levels of intra and inter-specific variation.We focused on the egg-spots of A. calliptera, a cichlid fish part of the Malawi radiation that phenotypically resembles the radiation ancestor.We aimed to: (1) identify key cellular events contributing to egg-spot development and variation; (2) uncover when and how anal fin pigmentation development diverges between males and females; and finally (3) assess if external social conditions influence egg-spot development.
Overt egg-spot formation starts with the appearance of iridophores in the position of the future egg-spot (Figure 7a-c).Egg-spot development then proceeds via the association of iridophores with xanthophores (Figure 7d), a growing aggregation of associated cells (Figure 7e) while xanthophores are lost from elsewhere in the fin (Figure 7f), and finally the clearing of a transparent chromatophore-free ring (Figure 7f).Contrary to expectations, we found both males and females exhibit the earliest stages of egg-spot development (Figure 7c-e) with anal fin patterns diverging only later between the sexes (Figure 7f).Furthermore, the timing of egg-spot initialization is plastic, with iridophore aggregations responding to the social environment (Figure 7c).Finally, there is variation between individuals in the development (Figure 7g) and positions of the second and third spots.
Self-organization via cellular interactions plays a significant role in periodic color patterns in teleosts (Volkening, 2020).However, the role of positional information cues is hypothesized to be more important in the development of non-periodic patterns, to define their position in the tissue (Parichy, 2021).Such positional information cues may exist as morphogen gradients, physical tissue architecture, or growth patterns (Kratochwil & Mallarino, 2023).Though positional information and local self-organizing mechanisms can provide alternate explanations for patterning (Green & Sharpe, 2015), they are not mutually-exclusive (Kratochwil & Mallarino, 2023).To evaluate the role of these mechanisms in egg-spot positioning, we deduce two scenarios with primary roles for self-organizing cellular interactions or positional information, and consider cellular events during egg-spot formation for each scenario.
Iridophores first appear in the anterior hard-spined domain of the fin, likely migrating from the body as progenitors, but form aggregations in the soft-ray domain.For the first egg-spot, we observed a consistent accumulation of iridophores between the 4 th and 5 th fin rays, posteriorly adjacent to the spiny-soft ray boundary in the anal fin.This suggests that the transition from a spiny-rayed to a soft-rayed tissue environment (Figure 7a) plays a role in egg-spot patterning.This role is supported by work in the Tanganyika cichlid, Astatotilapia burtoni: disruption of the signaling network that establishes the soft-ray domain causes concordant posterior shifts in the spiny-soft ray boundary and eggspot positions (Höch et al., 2021).We reason the soft-ray domain acts as a permissive or inductive region for iridophore aggregation following migration from the hard ray domain.In this scenario, egg-spots could emerge mostly due to cellular interactions between chromatophores, with iridophores initially aggregating and forming associations with xanthophores between the 4 th and 5 th fin rays solely because this is the first point of entry to this permissive/inductive environment.In an alternative scenario, within the soft-ray environment there could be multiple signaling centers providing positional information and inducing iridophores to aggregate in each spot location, one always present between the 4 th and 5 th fin ray and others in variable posterior positions (Figure 7g-2, 3, 5).In either scenario, there may be tissue signals from fin rays as aggregations first form between rays, and mechanisms to regulate the typically medial proximal-distal positioning.
Both scenarios, a permissive/inductive environment and signaling centers, could explain the cellular events observed during egg-spot formation.Iridophorexanthophore associations could result from short-range attraction between the two cell types or differentiation of iridophores into a novel cell type; in either scenario, this could be first triggered by entering the permissive zone or within close range of the first signaling center.
Aggregation growth could be due to migration of xanthophores and iridophores into the aggregations or local differentiation from unpigmented progenitors.In the latter case, loss of xanthophores elsewhere in the fin may be due to cell death or depigmentation, however we find no debris typical of dead pigment cells in our images.These mechanisms are compatible with both scenarios: cellular interactions can act at short-and longrange to promote local differentiation, migration, or cell death (J.P. Owen et al., 2020;Patterson & Parichy, 2013;Volkening & Sandstede, 2018;Yamaguchi et al., 2007).Notably, in zebrafish, iridophores attract xanthophores at short range (Patterson & Parichy, 2013).The same cellular behaviors can occur in response to a morphogen gradient from a signaling center (Bier & De Robertis, 2015;Ninov et al., 2010).
Similarly, the clearing of a chromatophore-free transparent ring around the egg-spot in males could be due to repulsion, induced cell death or depigmentation in response to cellular interactions or signaling centers.In zebrafish, iridophores and xanthophores repel melanophores at long range while attracting each other at short range to separate light interstripes from dark stripes (Frohnhöfer et al., 2013).Similar interactions in A. calliptera fins could explain the transparent ring.Alternatively, a diffusible signal from the signaling center or the aggregation itself could be decoded by concentration threshold to induce cell death or repulsion in the ring but not beyond it.Both scenarios require an explanation for the change in aggregation behavior from growth to ring clearing.It may be that as the aggregation reaches a certain density or cell number, the signals/ interactions from the aggregation reach a critical threshold or change in nature.Consideration of these events shows that either scenario of self-organizing cellular interactions or tissue landmarks patterning mechanisms could underlie the development of eggspot patterns, and in fact demonstrates that each scenario would likely involve both mechanisms.Analyzes of pigment cell mutants in A. calliptera will be required to understand the relative contribution of each mechanism and the specific contribution of each cell type for eggspot formation.
That said, the variability in the formation and final position of the second and third egg-spots supports the scenario driven predominantly by cellular interactions in a permissive/inductive region.Most notably, aggregations shifting position across fin rays while forming is more parsimonious with dynamic self-organization than control by a predefined signaling center.The apparently competitive interactions between developing aggregations are further indication of interactions between pigment cells.Iridophore-xanthophore associations can initially form in varying anterior-posterior locations, and competitive inter-aggregation interactions determine which aggregations develop into egg-spots (Figure 7g-4).This may contribute to the adult intraspecific variation in A calliptera egg-spots, including spot number and precise position.
The question that arises is how this self-organizing process leads to variation in egg-spot position and number, unlike zebrafish colouration self-organization which gives rise to highly reproducible periodic stripes.Greater noise, or susceptibility to noise, in egg-spot formation could explain this (Maini et al., 2012), such as interindividual variation in external environment, hormonal levels, system responsiveness to these two factors, or pigment cell interactivity.There also remains the question of how this variability in egg-spot formation manifests after the third forming egg-spot.Growth may be a factor: as fin area increases this may create enough space between existing egg-spots for new aggregations to develop without competition (Figure 7g-6, 7).Longterm imaging of later egg-spot development is necessary to address this question.
Like many other aspects of male colouration in cichlids, egg-spots are most strongly expressed when a male is dominant.In our study, we found the timing of egg-spot initiation is plastic and dependent on the social environment, with iridophores aggregating in the position of the first egg-spot upon single housing.Thus social isolation can hasten the onset of egg-spot development.This suggests that the presence of conspecifics slows the onset of egg-spot development, and that this social constraint can be released by isolation.We speculate that the isolation of individuals in their own territory may trigger the same or similar developmental pathways as dominance and territoriality and concurrently trigger egg-spot development.This timing plasticity may be a mechanism by which smaller fish sometimes bear higher numbers of egg-spots than larger fish in A. calliptera: the earlier an individual is released from social constraints, the earlier they would start adding egg-spots at smaller standard lengths (Figure 7g-1).Orange pigmentation traits are often sexually-selected honest indicators of condition because xanthophore color depends on dietary carotenoids uptake (Leclercq et al., 2010;Olson & Owens, 1998).Therefore in this sense it is surprising that iridophores are the cells dependent on the social environment in egg spot developmental plasticity.
There is likely hormonal mediation between sensing social environment and iridophores aggregating, as hormones often mediate developmental plasticity (Brakefield et al., 1998;Ledón-Rettig & Ragsdale, 2021;Moczek & Nijhout, 2002).A likely candidate is thyroid hormone, a conserved regulator of chordate metamorphosis (Paris & Laudet, 2008) which in zebrafish promotes iridophore maturation as well as limiting melanophore population expansion and promoting accumulation of carotenoids in xanthophores (McMenamin et al., 2014;Saunders et al., 2019).Further, in clownfish thyroid hormone mediates between the environment and speed of formation of iridophore-based white bars (Salis et al., 2021).There may also be involvement of glucocorticoids, as these are linked to social status in many vertebrates (Godwin, 2010;Perry & Grober, 2003) and can act synergistically with thyroid hormone, enhancing its effect (Brown et al., 2014;Rousseau et al., 2022).
Contrary to expectations, we observed iridophore aggregations in both males and females.We therefore speculate that a sex-independent increase in thyroid hormone regulates iridophore aggregation behavior in response to changes in the social environment (Figure 7c).Differences between males and females emerge later in development when the iridophorexanthophore aggregations grow in size, xanthophore distributions change and the transparent rings form in males.A sex-independent capacity to initiate egg-spot development but a sex-limited capacity to develop mature egg-spots explains why immature egg-spots sometimes occur in dominant females (Heule & Salzburger, 2011).These late developmental differences between sexes are likely modulated by sex-specific hormones (Figure 7f) similar to regulation of femalespecific pigmentation in the brown anole lizard by the estrogen receptor-1 gene, coexpressed with ccdc170 (Feiner et al., 2022).Similarly, in cichlids, upregulation of androgens, gonadotropins and their respective receptors in males are associated with development of sexual dimorphism and dominance (Korzan et al., 2008;Maruska et al., 2011Maruska et al., , 2022;;Moore et al., 2022).

| CONCLUSION
Taken together, our data indicates that egg-spots are a promising system to investigate teleost color patterning development and evolution.It appears likely that eggspot development largely self-organizes by cellular interactions, and is initiated by iridophores, similar to the well-studied zebrafish stripes.With previously undescribed cellular morphologies and behaviors including variably contracting xanthophores, iridophorexanthophore associations, and transparent ring clearance, egg-spot formation lends further evidence to the suggestion that pigment cells and their interactions may be evolutionarily labile (Kelley et al., 2013;McCluskey et al., 2021;Singh & Nüsslein-Volhard, 2015).Moreover, that egg-spot formation can initiate regardless of the stage of background xanthophore-based pigmentation indicates that background pigmentation and egg spot patterning are not interdependent.This is in contrast to zebrafish stripe formation in which contributions from all chromatophores are required for patterning (Patterson & Parichy, 2019) indicating evolutionary lability in pigment cell interactions.In addition to selforganizing intercellular interactions, we propose multiple additional inputs to regulate the position, timing, and maturation of the egg-spot pattern: an inductive/permissive tissue domain to bound initial iridophore positioning, sensing and hormonal mediation of social environment influencing timing of pattern initiation, and a sex-limited hormone enabling egg-spot maturation only in dominant males.Therefore, egg-spots provide an exciting model for investigating how environmental and conspecific behavioral cues affect pigment pattern formation at the cellular and developmental level.Furthermore, A. calliptera represents the ancestral bauplan for egg-spots.As such our findings provide a baseline for informed comparisons across the Malawi radiation which harbors an incredible extent of inter-and intraspecific variation.Addressing the genetic and developmental underpinnings of such variation will yield insights into the genes, signals, and cellular interactions that underlie the mechanisms patterning non-periodic male pigmentation ornaments and their evolution.

F
I G U R E 2 Longitudinal imaging cohorts: embryo, yolk closure and socialized.Given that anal fin pigmentation takes approximately 200 days to fully develop and that originally the project licence only allowed repeated imaging of the same individual for 50 days, we imaged different cohorts spanning the entirety of egg-spot development, with individuals from each cohort single housed at different time points.Thus, to span all developmental events leading to egg-spot formation we performed three imaging series starting at different stages: 10 dpf embryos (a), juveniles at the point of yolk closure (b) and juveniles maintained in a social group for 6-10 weeks post yolk closure for deep-water and shallow-water individuals, respectively (c).Each cohort had two replicates, one with shallow-water and one with deep-water individuals.All individuals were kept in a group before being housed in individual compartments for imaging (b, c).During the imaging period of single housing, anal fins were imaged every few days (see methods Section 2.3 for details).Marks on the timelines indicate imaging days.Embryo and yolk closure cohorts span early stages of egg-spot development (initial chromatophore aggregations [Section 3.2]); the socialized cohort span the later events (transparent ring formation [Section 3.3]).dpf, days post fertilization.

F
I G U R E 3 Early anal fin pigmentation development and egg-spot initialization (a-j) Images of developing fins showing embryo and yolk closure cohorts (deep-water replicates).All images are with transmitted light except (g) and (m) under incident light.Events shown: melanophores and xanthophores present in the embryo fin (a, b), spread of chromatophores in the embryo fin and appearance of iridophores (c, d) xanthophores and iridophores first present in early juvenile fin under different lighting conditions (e-g), iridophores first present in between 4 th and 5 th fin ray (h, i) iridophores associate with xanthophores (j, k).(l, m) Iridophores associated with contracting xanthophores under different lighting conditions.(n) Number of melanophores in the fin against days post fertilization, with gray shading indicating timing of first iridophore appearance (19-23 dpf) for both morph replicates and both cohorts).Arrows indicate: melanophores (black), contracting xanthophores (orange), non-contracting xanthophores (yellow), iridophores (blue), location of aggregation (white).Gray scale bars 1 mm, white scale bars 250 µm.Early juvenile fins are treated with epinephrine.Equivalent series for shallow-water replicate in Figure S2.dpf, days post fertilization.

F
I G U R E 4 Divergence in egg-spot development between sexes.Shared development (a, f) and point of divergence (b, g) in juvenile males and females.Insets highlight the first egg-spot aggregation.Points of divergence are a reduction of xanthophore distribution across the fin in males (b), unchanging in females (g) and transparent ring formation (c) absent in females (h).Iridophores, erythrophores and melanophores (d, e) are found in mature males (c).(i) Area coverage of xanthophores (both contracting and non-contracting) measured in a 0.5 mm 2 square region centered on the first segment of the 8 th fin ray against Day with Pearson's correlation.Arrows indicate: melanophores (black), contracting xanthophores (orange), non-contracting xanthophores (yellow), erythrophores (red) and iridophores (blue).Equivalent series in Figure S7.

F
I G U R E 5 Iridophores aggregate soon after fish are singly housed at different developmental stages.(a, b) Violin plots of iridophore events in egg-spot initiation by cohort and morph.For the first imaging session in which the event was observed, each event is plotted against days since the first imaging session (Day) (a), and standard length (SL) (b).Each point is one individual.Standard lengths were not obtained for the shallow-water socialized cohort as images permitting this measurement were not collected (see methods Section 2.3 for details).(c, d, f) Transmitted light images of the fin when iridophores are first present, for a representative deep-water morph individual from the embryo (c), yolk closure (d), and socialized cohort (f).(e, g) Transmitted light images when iridophore-xanthophore associations are formed, for a representative deep-water morph individual from the yolk closure (e), and socialized cohort (g) Arrows indicate melanophores (black), xanthophores (orange) and iridophores (blue).Gray scale bars 1 mm, white scale bars 250 µm.

F
I G U R E 6 Variability and flexibility in subsequent egg-spot formation.(a-d) Examples of dynamic formation of second and third egg-spots.Arrows indicate emerging and fading aggregations.Each row is one male from the socialized cohort.Gray scale bars 1 mm.

F
I G U R E 7 Overview of egg-spot development in Astatotilapia calliptera.(a) Representation of the spiny and soft fin ray domains in the anal fin and a key illustrating chromatophore types present in the system (b-f) Trajectory of egg-spot development in yolk closure and socialized cohorts, showing stages of egg-spot formation with hypothesized factors contributing to variation at each stage represented as numbers in black circles.Circles 3, 6, and 7 represent factors hypothesized to contribute to variation at all stages of development.(g) Summary of hypothesized factors contributing to variation in adult phenotype.