Differences in Early Development between Parental Species and Asexual Hybrids forms of European Spined Loaches (Genus Cobitis)

Abstract

Introduction

Hybridization and polyploidy are increasingly recognized as major drivers of evolutionary innovation. By merging and multiplying genomes, they induce profound regulatory and structural perturbations that can alter phenotypes, yet the mechanisms by which they influence development remain debated. In particular, while theory predicts that polyploidy increases cause the increase in cell size, resulting in slower development, empirical findings are mixed. Moreover, the independent effects of polyploidization and hybrid genome admixture on early ontogeny are rarely disentangled. Here, we used the Cobitis taenia-C. elongatoides hybrid complex, a natural system comprising two sexual diploids and both diploid and triploid asexual hybrids to investigate how hybridization and polyploidy affect early development.

Methods

We examined two key developmental processes: (1) the timing of embryonic progression from the first cleavage to somitogenesis and (2) migration of primordial germ cells (PGCs), which are crucial for reproductive competence. By comparing both hybrid types to each of their sexual progenitors, we were able to isolate the effects of genome merging versus genome multiplication.

Results

We found distinct developmental delays linked to effects of the parental genome, hybridization, and ploidy. Initial cleavage dynamics was mostly attributable to differences between the two parental species, while later delays were more pronounced in hybrids, regardless of ploidy, highlighting hybridization-driven developmental shifts. Only at subsequent stages did ploidy-specific effects emerge, prolonging development in triploids against all diploid forms. In contrast, PGC number and migration were primarily influenced by ploidy, with triploids showing higher PGC counts than diploids.

Conclusion

Our findings reveal that hybridization and polyploidy exert both independent and interactive effects on early ontogeny. The Cobitis complex thus offers a powerful model for exploring developmental consequences of genome restructuring.

Introduction

The development of multicellular organisms, from the first zygotic divisions to the formation of a functional, reproducing individual, relies on a tightly regulated cascade of gene activation, signaling, and morphogenetic events [1, 2]. Disruptions to these processes typically result in deleterious outcomes. Among the most profound sources of developmental perturbation are large-scale changes to genome architecture, particularly hybridization (the merging of distinct genomes) and polyploidization (possessing multiple genomic copies). These events can trigger widespread regulatory mismatches, chromosomal instability, and dosage effects, phenomena sometimes described collectively as “genomic shock” [3–5]. Despite this, both processes are increasingly recognized as powerful evolutionary forces across plants and animals, contributing to major lineage radiations [6, 7] and the emergence of traits not present in the parental taxa [7]. This is evident in vertebrate diversification, putatively sparked by multiple rounds of ancient whole genome duplication [6, 8] and, in particular, among fishes, where polyploidy recurrently arose both through autopolyploidy and allopolyploidy, either naturally or via artificial induction [9–11].

Polyploidy has also been hypothesized to exert predictable effects due to increased DNA content. These include altered nucleocytoplasmic ratios and consequently modified metabolism, slower cell cycles, and altered DNA/RNA balance, all of which could impact developmental rate [12–15]. Indeed, delayed embryogenesis has been repeatedly observed in polyploid animals and plants, which has been interpreted as a general consequence of polyploidization.

However, the effects of increased genome content are not limited to cell physiology alone. Previous research has shown that hybrid and polyploid vertebrates often differ from their parental progenitor species in both phenotype and gene regulation. In particular, genome dosage effects and expression dominance can cause hybrids, especially triploids, to deviate from the simple intermediate patterns expected under additive inheritance [5, 16, 17]. Notable exceptions exist; for example, in fish, autotriploid specimens often do not differ morphologically from their diploid counterparts [9]. Independently arisen hybrid and/or polyploid lineages also often exhibit recurring features, including changes in cell and body size due to increased genome content [12, 13], altered expression balance between parental subgenomes [16], transgressive phenotypes arising from regulatory mismatches [18], and reproductive alterations such as sterility or the emergence of asexuality [17, 19].

Despite these insights and despite their central role in evolution (and substantial applications in agriculture and aquaculture), the developmental consequences of polyploidy and hybridization remain incompletely understood, especially regarding early embryonic development. This is the period when gene regulatory networks are highly sensitive to perturbation and when the initial consequences of genome merger or multiplication are most likely to manifest. For instance, asynchronous expression of duplicate genes has been proposed to affect polyploid organisms generally [19]. Such asynchrony can be detrimental, and increased rates of developmental deformities have been observed in polyploid fish embryos and larvae [9, 11], yet the underlying developmental mechanisms remain largely unexplored. Studying early ontogeny in natural hybrid and polyploid systems therefore offers a promising window into the proximate impacts of altered genome architecture on animal development.

In this study, we investigate how hybridization and polyploidy affect two key aspects of early ontogeny: (1) the timing of developmental progression from the first cleavage to somitogenesis and (2) the formation and migration of primordial germ cells (PGCs), a crucial process that determines reproductive potential in vertebrates. We focus on the loach fish of Cobitis hybrid complex (Teleostei, Cypriniformes) where natural co-occurrence of parental species with their diploid and polyploidy (triploid) hybrids allows a potential to at least partially disentangle the individual contributions of genome merging (hybridization) and genome multiplication (polyploidy) on fundamental developmental processes during embryogenesis.

In Central European hybrid zones, two parapatric species, C. taenia (hereafter we also denote its genome composition as TT, owing to the presence of two haploid sets of “T” genome) and C. elongatoides (hereafter labelled as EE), coexist and produce diverse hybrid forms, which occur in various genomotypes, depending on composition and dosage of their genomes [20, 21]. These genomotypes include sterile hybrid males [22, 23], as well as diploid and triploid hybrid females. Occasionally, hybrids may also possess the genome of a third species from Pontic-Caspian region, C. tanaitica (NN). Consequently, the diversity of hybrids in central Europe involves wide range of genomotypes including ET, EET, ETT, and ETN forms that reproduce clonally via gynogenesis. These lineages arise through interspecific crosses and produce unreduced eggs by premeiotic genome endoreplication (PMER), in which the genome is duplicated prior to meiosis, enabling pairing of identical sister chromosomes and clonal transmission of the maternal genotype. Recent work shows that PMER occurs in only ∼1–5% of oogonia, while most germ cells fail to duplicate their genome and undergo apoptosis, indicating a substantial developmental cost of asexuality in hybrids [24]. Embryonic development is triggered by sperm from a sexual species, but the paternal genome is typically excluded. Only rarely is it incorporated, producing additional polyploid offspring, including triploids and occasionally tetraploids, the fecundity of latter remaining uncertain [20, 25, 26]. Because clonal females depend on sperm from sexual males, asexual and sexual lineages coexist in a system resembling host-parasite dynamics [27]. However, clonal hybrids cannot fully outcompete their sexual hosts due to sperm limitation and the low efficiency of PMER. Phylogenetic analyses demonstrated that despite described above limitations, hybrids form a swarm of independently arisen clones, stable over relatively long time scales, with the oldest estimated to persist for up to ∼300,000 years [21, 28].

It remains unclear to what extent differences in early developmental timing reflect the effects of hybridization itself versus those of increased ploidy in this system. A pioneering embryological study by Juchno et al. [29] reported delayed cleavage and organogenesis in triploid clonal hybrids compared to one sexual species, C. taenia, suggesting that genome composition may affect ontogenetic rate from its earliest stages. However, interpretation of these results is limited because the study did not investigate the variability between parental species nor the effect of hybridization on diploid hybrids. Furthermore, the insemination with sperm from a parental species resulted in a sample containing both triploid and tetraploid embryos, complicating the comparison of the developmental data with parental lines.

To address these questions specifically, the present study explicitly includes both sexual species (C. elongatoides and C. taenia) alongside their diploid (ET) and triploid (ETT) hybrid derivatives and uses an insemination approach that generates progeny of uniform ploidy. Such an experimental design enables us to disentangle which developmental shifts are mostly attributable to parental divergence, which reflect general consequences of hybridization, and which arise more likely from genome polyploidization.

Methods

Material

Adult individuals of spined loaches (parental species and hybrids) were sampled in rivers and water reservoirs of the Czech Republic (Nová Říše reservoir) and Poland (Barycz river) during the spawning season from April to July (genomotypes EE, ET, and ETT) or in November. Materials used in the study consisted of the following number of females: five of parental species (EE, n = 3; TT, n = 2), three ET hybrids, and ten ETT hybrids, supplemented with four diploid males (EE, n = 3; TT, n = 1). The adult Cobitis males were identified by the presence of lamina circularis – the bony plate on the pectoral fin. Fish acclimatization in standardized breeding tanks was carried out to avoid temperature shock. All fish were fed the same diet consisting of Artemia sp. and Tubifex sp.

The ploidy level of all parental individuals was identified with flow cytometry and allozyme analysis [28, 30], which also allowed a rough differentiation of clonal identity [28]. The ploidy level of offspring was examined by flow cytometry only. Whole larvae (euthanized by tricaine overdosing) were processed using a kit for nuclei staining CyStain UV Precise T (Sysmex Partec GmbH, Germany) in accordance with the manufacturer's protocol. The relative DNA content was determined using a CyFlow Ploidy Analyzer (Sysmex Partec GmbH, Germany), against samples from diploid control groups.

Cross-Experiments Design

For the experiments, the ovulation and spermiation in males and females of loaches were induced by injecting Ovopel (a GnRH analog with dopamine receptor antagonist, Interfish Kft., Budapest, Hungary) [31, 32]. The treatment was administered in two doses. The first injection contained one pellet of the Ovopel dissolved in 20 mL 0.9% NaCl. Second injection, after 12 hours, included one pellet of the Ovopel dissolved in 4 mL 0.9% NaCl. The dose for both sexes was 0.05 mL/10 g of body weight. Before injection, the fishes were anaesthetized using water bath with tricaine solution. The eggs were collected 21–28 h after the first injection and placed in dechlorinated tap water at 18°C in Petri dishes with sperm. Parental species oocytes were fertilised with sperm of males from the same species, which was collected from 21 to 24 h after the first hormone injection and kept in Hank’s solution on ice until fertilisation. Fertilization (time zero) was defined as the mixing of eggs with sperm followed by immediate water activation.

In the case of diploid (ET) and triploid (ETT) hybrids, we needed to ensure the maintenance of constant ploidy level in the offspring. Hence, in order to prevent sperm genome incorporation and subsequent ploidy increase among the progeny, the eggs were inseminated with UV-irradiated goldfish sperm (Carassius sp.) – as proved in our previous experiment to be able to activate development of EET eggs without confounding effects of genome fusion [33]. The goldfish sperm was collected by stripping without hormonal stimulation, with the males continuously held in temperate aquaria. To inactivate the DNA in the goldfish sperm, it was diluted 1:9 in Kurokura 180 solution and placed on a glass Petri dish. It was then irradiated with UV using a Crosslinker UVP CL-1000 placed on a laboratory shaker at a dose of 300,000 μJ cm^-2 (Fig. 1a).

Fig. 1.

Visualization of the experimental procedure used in the present study. Fish silhouettes used in the figure were sourced from PhyloPic (https://phylopic.org) under open-use licenses. The syringe icon was obtained from the Microsoft PowerPoint built-in icon library. a Procedure for gamete collection in female and male Cobitis and Carassius males. b Design of the crossing experiments: eggs of parental species were fertilized with conspecific sperm; gynogenetic unreduced eggs of hybrid specimens were activated to develop using UV-irradiated Carassius sp. sperm in order to maintain the ploidy level of the offspring by preventing paternal genome incorporation into the embryos. EE, Cobitis elongatoides; TT, C. taenia; ET, diploid hybrid (Elongatoides-Taenia); ETT, triploid hybrid (Elongatoides-Taenia-Taenia). c Each clutch divided for manipulation and observation.

Observation and Definition of Developmental Stages

Developmental stages of loaches were determined according to morphological diagnostic features by comparison with the developmental stages of zebrafish (Danio rerio) and Misgurnus anguillicaudatus, both representing Cypriniformes [34, 35]. Development was divided into five major periods: cleavage, blastula, gastrula, segmentation and hatching, documented at regular intervals: cleavage period every 10 min; blastula period every 30 min; gastrula period every 1 h; segmentation period every 2.5 h; hatching period every 10 h. To compare the developmental rates among genomotypes, we subdivided each period into several stages (n = 14; see Table 1).

Table 1.

Mean, minimum, and final times from fertilization to reach developmental stages at 18°C in the progeny of each Cobitis hybrid biotype and parental species

Stages name	EE = 3 ♀			TT = 2 ♀			ET = 3 ♀			ETT*	ETT clone 1 (6 ♀)			ETT clone 2 (4 ♀♂)			Characteristic	Fig. 2
	mean	1st	last	mean	1st	last	mean	1st	last	mean	mean	1st	last	mean	1st	last
2 cells	00:55	00:45	01:10	1:13	1:05	1:20	1:13	1:00	1:30	1:12	1:11	0:50	1:30	1:13	1:00	1:30	1st cleavage furrow	a
4 cells	01:31	01:20	01:45	1:43	1:35	1:50	1:42	1:35	1:50	1:33	1:29	1:10	1:45	1:38	1:25	1:55	2 × 2 array of blastomeres	b
8 cells	1:59	1:40	2:20	2:06	1:55	2:10	2:12	2:10	2:25	2:01	2:05	1:45	2:30	1:57	1:40	2:15	2 × 4 array of blastomeres	c
32 cells	3:22	2:55	3:40	3:16	3:00	3:25	3:19	3:00	3:40	3:22	3:25	3:00	3:40	3:20	3:00	3:35	blastodisc with 32 cells assigned to the central part of the a. pole	d
Blastula
Dome – 30% epi	11:20	10:25	12:00	11:15	NA	NA	11:57	11:00	13:00	12:31	12:13	11:30	13:10	12:49	12:15	13:30	30% of the yolk covered by the blastoderm	e
Gastrula
50% epi	12:28	11:15	13:20	14:10	NA	NA	14:00	13:00	15:00	14:09	14:11	13:30	14:50	14:07	13:45	14:30	50% of the yolk covered by the blastoderm	f
100% epi	20:58	20:00	21:45	20:55	19:00	20:00	24:18	24:00	25:00	24:16	23:25	21:45	25:30	25:07	24:45	25:30	Tail bud prominent	g
2 somites	24:28	23:45	25:20	24:19	23:15	24:00	25:52	25:30	26:45	27:35	27:33	24:30	31:00	27:37	27:00	28:15	First somatic furrow	h
8 somites	29:12	28:15	30:00	28:56	27:15	28:30	28:12	27:40	29:00	31:39	31:26	28:45	34:30	31:52	30:30	33:00	Brain primordium, optic vesicle	i
18 somites	35:53	35:10	36:30	35:33	34:20	35:10	36:20	35:00	38:00	41:03	40:01	36:15	42:30	42:05	17:15	43:30	Arrow indicates the optic vesicle	j
34 somites	50:25	48:40	52:00	48:30	48:30	49:50	50:10	48:00	52:00	56:15	55:20	51:30	58:00	57:11	56:15	58:00	Median fin fold, tail extension, otolith [36]:	k
Hatching
Hatching	75:30	68:00	82:00	76:30	70:00	80:00	78:10	72:00	85:00	81:23	80:06	72:00	90:00	82:41	74:00	90:00	Pigmentation in retina and skin	l
1st fry	91:20	85:56	92:38	92:10	NA	NA	106:00	90:00	110:00	102:15	101:40	91:00	113:00	102:50	91:00	113:00	Pigments appear in whole body, formation of pect. fins and exter. gills	m
2nd fry	108:00	101:40	108:13	NA	NA	NA	121:30	115:00	125:00	119:47	117:15	112:00	124:00	122:20	114:00	134:00	Small pectoral fins, short exter. gills	n

Time format: h:mm.

EE, C. elongatoides; TT, C. taenia; ET, diploid hybrid elongatoides x taenia; ETT, triploid hybrid; ETT*, pooled mean for both triploid clones; epi., epibole; a. pole, animal pole; epi., epiboly; pect., pectoral; exter., external; NA, not analyzed; 1st, time of first embryo which achieved particular developmental stage; last, time of last embryo which achieved particular developmental stage.

All embryos were observed with chorion because loach embryos are more sensitive to enzymatic dechorionation with trypsin (based on our observations). Only several eggs were dechorionated mechanically with tweezers for better imaging. The temperature was regulated at 18°C in the incubator Q cell (Verkon, Czech Republic) throughout the experiment, and dechlorinated water was changed twice per day.

The embryos from the 18 resulting families were randomly subsampled into two Petri dishes, each containing 20–30 individuals, whose development was further evaluated. Embryonic and larval development was observed from 2 cells stage till 1st fry stage of each group at the same time by photographic records (Leica DFC425C). Altogether, around 14 developmental stages were documented in each family.

For each family at each observation interval, we recorded the time at which the first and the last embryo reached a given developmental stage. Because individual-level timing data were not achievable for all embryos owing to rapid developmental succession, the midpoint of this interval was used as an estimate of stage timing at the family level. Next, family-level estimates were considered independent biological replicates. For each genomotype, the overall developmental timing at a given stage was calculated as the mean of family-level values. These genomotype-level means were subsequently used for statistical comparisons among groups.

Visualization of PGCs

Randomly selected embryos from each family were used to observe PGC migration following Saito [37], using protocols optimized in our laboratory and described by Linhartova [38]. In brief, the method is based on visualization of PGC migration by examining Green Fluorescent Protein, GFP expression in embryos injected with synthesized mRNA. Injection can be performed either into the blastodisc or into the underlying yolk. In our case, Cobitis embryos were injected at the two-cell stage with GFP-nos1 3′UTR mRNA (300 ng/µL in 0.2 m KCl) beneath the blastodisc using a glass capillary mounted on a micromanipulator (Narishige, Japan), a microinjector (Eppendorf, Germany), and a fluorescence stereomicroscope (Leica M165FC). Micropipettes were pulled from glass capillaries (Drummond) using a needle puller (PG-10; Narishige, Tokyo, Japan). From each strain, a subset of embryos served as controls: some were left intact to assess potential effects of the microinjection procedure itself on survival and developmental rate, while others were injected with rhodamine dextran (MW 500,000) as an inert tracer to control for effects of the injection process and injected material, allowing us to distinguish these from specific effects of the GFP-nos1 mRNA (Fig. 1b).

PGC migration was documented from the onset of 90% epiboly until the 2nd fry stage using a fluorescence stereomicroscope (Leica M165FC) equipped with a digital camera (Leica DFC425C). All embryos were maintained in darkness at 18°C to minimize attenuation of GFP fluorescence. Once PGCs had settled in the germinal ridge, where they were clearly distinguishable and countable, their numbers were quantified in all individuals at three developmental stages: 38-, 42-, and 46-somite stages.

Statistical Analysis

Time measurements were analyzed using a generalized linear model (GLM). In this model, a single clutch (i.e., embryos from a single mother within one Petri dish) was treated as the unit of comparison. To account for clustered and repeated measurements, a generalized estimating equations (GEEs) approach was applied using the geepack package [39]. We tested whether observed time values were significantly predicted by the following categorical variables: developmental stage (i.e., achievement of individual stages), ploidy level (diploid vs. triploid), and genomic constitution (pure species vs. hybrids in general). The model formula was as follows: time ∼ stage + ploidy + genomic_constitution.

Significance of categorical predictors was assessed using a multivariate Wald χ² test implemented in the car package [40]. Due to incomplete coverage for some stage × genomotypes combinations, we fitted the GLM-GEE only to stages and groups with sufficient data; underpowered combinations were excluded a priori. Finally, Tukey-adjusted post hoc comparisons under a nested design (genomotypes | stage) were computed from estimated marginal means using the emmeans package [41].

The effect on the number of PGCs was examined in relation to hybridization and polyploidization. To test this, a generalized linear mixed model was fitted using the lme4 package [42], with genomotype as a fixed effect and single clutch as a random intercept to account for non-independence within clutches. Genomotypes were not compared directly; instead, orthogonal contrasts were applied to the estimated marginal means obtained with the emmeans package [41]. As mentioned above, two contrasts were used: (I) hybridization – comparing non-hybrids against hybrids, and (II) polyploidization – comparing diploids against triploids. Model diagnostics, including checks for residual dispersion, were performed using the DHARMa package [43] to assess the adequacy of model fit. All analyses were conducted in R software v. 3.4.2 [44].

Results

In total, we obtained complete developmental sequences for three families of C. elongatoides, two families of C. taenia, three diploid ET hybrids, and ten triploid ETT hybrids. Allozyme analysis of previously characterized diagnostic loci indicated that all diploid ET individuals share the same multilocus genotype, suggesting they may belong to a single clonal lineage. Among triploids, the ten maternal individuals could have been assigned to two distinct allozyme-defined multilocus genotypes: 6 individuals belonged to one clone and 4 to the other, suggesting they are composed of at least two separate clonal lineages.

General Characteristics of Developmental Stages and PGC in Cobitis hybrid complex

Embryonic Development

Cleavage Period. After insemination, the chorion of the egg expanded from the embryo. During cleavage, embryos divided synchronously, starting at approximately 60 min post-fertilization in all observed groups. The timing of the stages is provided in Table 1, and the stages are shown in Figure 2.

Fig. 2.

Embryonic development during the cleavage, blastula and segmentation period in the European loaches at 18°C. a 2 cells. b 4 cells. c 8 cells. d 32 cells. e Dome stage, 30% epiboly. f 50% epiboly. g 100% epiboly, arrow indicates the tail bud. h 2 somites. i 8 somites. j 18 somites, arrow indicates the otic vesicle. k 34 somites. l Hatching. m 1st fry. n 2nd fry, arrow indicates the small external gills. Scale bar = 500 µm (a–k) and 1 mm (l–n).

Two-cell stage (50 min–1 h 15 min): we observed typical meroblastic cleavage with the first cleavage starting approximately at the same time in all genomotypes except significantly earlier initiation in EE. The cleavage furrow began to form around 50 min post-fertilization (mpf) in EE and around 60 mpf in remaining groups. Two-cell embryos were observed at about 60 mpf in diploids EE and around 70 mpf in hybrids and TT (Fig. 2a).

Four-cell stage occurred between 1 h 25 min and 1 h 50 min (Fig. 2b) with second furrow forming at 85 mpf in EE and 95 mpf in hybrids and TT. Subsequent cleavage furrows formed vertically to the second cleavage furrow around 2 hpf. With ongoing divisions, the differences between EE and other groups diminished as the other groups began to “accelerate”.

Thirty-two-cell stage was achieved by all genomotypes at approximately the same time (Table 1; Fig. 3a, b) with embryos possessing layers of blastomeres in the central part of their blastodisc (Fig. 2d).

Fig. 3.

Difference in developmental rates between Cobitis parental species and their hybrids. a Visualization of developmental trends (y-axis: log-transformed minutes) for parental species and their hybrids. Points represent mean values, and vertical lines indicate the range (minimum–maximum) from the first to the last embryo of each genomotype reaching a given developmental stage. EE, C. elongatoides; TT, C. taenia; ET, diploid hybrid; ETT, triploid hybrid. b Comparison of developmental rates between genomotypes in hours. Letters indicate the results of Tukey’s post hoc test across genomotypes; bars represent standard errors. Detailed results are provided in the text.

Blastula Period. During this period, the blastodisc became spherical, and we could clearly distinguish the early blastula stage, characterized by synchronous cleavage of the cells, and the late blastula stage, with asynchronous cleavage and the appearance of some morphological characteristics. For the most accurate comparison between genomotypes, we considered only the time at which the most easily recognizable stage (i.e., the dome stage at 50% epiboly [Fig. 2f]) was reached. This stage was first achieved by TT, followed by EE, and only later by diploid and triploid hybrids, which represented the slowest groups (Fig. 3).

Gastrula Period. This period was characterized by involution, convergence, and extension movements, which began just after 50% epiboly (Fig. 2f), when the blastoderm covered half of the yolk cell, and continued until 100% epiboly (Fig. 2g), when the blastoderm completely covered the yolk surface. At this stage, the tail bud became apparent in the posterior region of the embryonic body. Overall, these stages were reached earlier by both parental species, followed by ET and ETT hybrids, which reached them at nearly the same time (Fig. 3a, Table 1).

Segmentation Period. During somitogenesis, the embryos elongated their body axis, the tail bud became more prominent, and primary organs began to differentiate from the germ layers. Somite formation progressed from the trunk toward the tail. By the end of the segmentation period (38 somites and beyond), we observed a considerable reduction of the yolk mass. Compared to the parental species, which both reached the formation of somites around 22 hpf, hybrids developed more asynchronously: diploid ET embryos reached somitogenesis at around 24 hpf, and triploids at around 25 hpf (Fig. 3a).

The 2-somite stage was reached earliest by both parental species and latest by triploid hybrids (Table 1; Fig. 2h). The optic primordium began to develop at the 6-somite stage and became clearly visible in the cephalic region by the 8-somite stage (Fig. 2i), accompanied by the appearance of a horizontal crease. On average, diploid hybrids reached the 8-somite stage slightly faster than both parental species, whereas triploid hybrids were the slowest.

At the 18-somite stage (Fig. 2j), we observed a constriction forming in the posterior region of the yolk and the emergence of otic placodes in the head region. These placodes subsequently hollowed, giving rise to otic vesicles, whose cavities expanded as development proceeded. This stage was reached by all diploids at approximately the same time and significantly later by triploids.

By the 34-somite stage (Fig. 2k), the yolk region ceased elongating, otoliths within the otic vesicles were distinguishable, and the tail region was segmented and separated from the yolk. Triploids again lagged behind, whereas among diploids, TT developed faster than both EE and ET. Later stages of segmentation were marked by the appearance of the anus and the initiation of heartbeat (around 38 somites). By the 44-somite stage, melanophores became faintly visible in several regions of the body, and the yolk mass had retreated toward the anus. The segmentation period concluded with a marked reduction of the yolk sac and the freeing and movement of the tail.

Hatching and Post-Hatching Period. At hatching, many organs were already well developed, but larvae of all genomotypes still contained a considerable amount of yolk. Eyes were pigmented, and blood circulation was visible in all genomotypes. Melanophores were also present at this stage. Parental species and their di- and triploid hybrids differed in the general abundance of melanophores, with hybrid larvae typically exhibiting more than EE and TT.

All sibling embryos from a single mother hatched almost simultaneously, regardless of whether they were raised in one or multiple batches. Some asynchronicity was observed among families within the same genomotype, and considerable variability existed among genomotypes. The average onset of hatching in polyploid hybrids was significantly delayed compared to both parental species and diploid hybrids (Table 1; Fig. 3a, b).

Migration of PGCs

To gain insight into early germline development in Cobitis, we utilized in vivo labelling of PGCs. We were able to detect and trace PGCs from the beginning of 75% epiboly (the earliest stage) until the early post-hatching stage (2nd fry; Fig. 4). The onset of PGC migration was noted at 90% epiboly, and migration continued until the early post-hatching stage. PGCs were observed on both sides of the larvae.

Fig. 4.

Primordial germ cells (PGCs) development of European loaches from late gastrula to the 2nd fry stage, at 18°C. Arrows indicate the position of PGCs during each stage. a1–a2 75% epiboly. b1–b2 Early somatogenesis stages. c1–c2 4 somites. d1–d2 8 somites. e1–e2 About 16 somites. f1–f2 About 21 somites. g1–g2 About 30 somites. h1–h2 Hatching. i1–i2 1st fry. j1–j2 2nd fry.

During these favourable stages, constant changes in the positions of PGCs were observed, and all genomotypes exhibited a very similar process of PGC migration. At the onset of somitogenesis, PGCs were observed to be aggregated on both lateral sides of the trunk. Subsequently, during the 10–20 somite stage, they began alignment into a linear configuration along the anterior and posterior aspects of the body. After the separation of the tail from the yolk, the distribution of PGCs occurred at the junction of the yolk sac and their position remained largely unchanged near the site of hatching. During the hatching period, the PGCs achieved the central part of the yolk extension and proceeded dorsally (Fig. 4). No significant developmental differences were observed between GFP-injected and control embryos.

Differences in PGC Counts

At all examined stages, we observed differences in both the number and position of PGCs among diploid EE and ET embryos (mean PGC count = 13) and polyploid ETT embryos (mean PGC count = 23; Table 2, Fig. 5). Differences in number of PGCs were statistically supported by a Wald χ² test from a generalized linear mixed model, with genomotype as a fixed effect and maternal identity as a random intercept (χ² = 21.64, df = 2, p < 0.001). The model explained 56% of the variance (R² marginal = 0.417; R² conditional = 0.56), and model diagnostics showed moderate overdispersion (dispersion parameter = 1.2), indicating that the model was well fitted. Post hoc contrasts revealed a significant effect of ploidy (z = −2.606, p = 0.009), while the effect of hybrid status was marginally nonsignificant (z = 1.858, p = 0.063), suggesting that differences in PGC numbers were primarily driven by ploidy rather than hybrid origin.

Table 2.

Comparison of mean number of PGCs between EE and ET and ETT hybrid embryos

Biotype/stages	N	PGC	38					42					46
			SD	min	max	N	PGC	SD	min	max	N	PGC	SD	min	max
EE	10	11	3.25	6	16	5	11	4.11	6	16	6	14	4.89	8	20
ET	4	12.5	1.65	10	14	3	12.6	4.1	8	18	8	12	0.5	8	16
Pooled ETT mean	8	21	3.6	18	28	8	23.5	4.09	16	28	5	21	4.11	14	26
ETT clone 1	6	20.3	3.54	18	28	5	24	3.57	18	28	4	23	2.23	20	26
ETT clone 2	2	23	3	20	26	3	22.6	4.71	16	26	1	14	–	–	–

Stages are defined based on number of somites. Pooled ETT mean: calculated for all ETT clones.

PGCs, primordial germ cells. EE, Cobitis elongatoides; ET, diploid; ETT, triploid.

Fig. 5.

Difference in mean number of primordial germ cells (PGCs) between Cobitis elongatoides (EE), diploid (ET), and triploid (ETT) hybrids (calculated for both ETT clones). Vertical lines indicate the minimal and maximal number of PGCs. Different letters denote significant differences.

Differences in the Rate of Embryonic Development

Developmental timing among different genomotypes raised at 18°C was visually compared using a line chart (Fig. 3a). The y-axis represents time in minutes on a log10 scale, while the x-axis corresponds to sequential developmental stages.

The GLM model fitted by GEE showed that developmental stage itself was indeed the most influential factor explaining variation in reported times (χ² = 121.234, df = 11, p < 0.001). After accounting for this primary effect, additional effects of ploidy (χ² = 12, df = 1, p < 0.001) and genomic constitution (χ² = 16, df = 1, p < 0.001) also significantly contributed to variation in developmental timing among clutches.

These findings indicate that differences in developmental rate are also influenced by genomic background. Interestingly, post hoc comparisons revealed that these differences do not follow a uniform or stage-independent pattern (Fig. 3b, online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000552676). Some genomotypes reached certain stages more quickly, while others caught up or surpassed them at later points, making the dynamics of developmental pacing nonlinear and context-dependent.

Specifically, the earliest developmental events (the onset of cleavage) occurred most rapidly in EE, whereas TT and both hybrid genomotypes (ET and ETT) showed a modest delay. By the 32-cell stage, these early differences largely diminished, and all genomotypes converged in timing. At subsequent stages, divergence among genomotypes became apparent again. Since gastrulation, both hybrid types (ET and ETT), regardless of ploidy, progressed more slowly than the two parental species.

From the segmentation period onward, this trend reversed, and triploid hybrids (ETT) consistently lagged behind all other groups across multiple somite stages. In contrast, diploid hybrids (ET), despite earlier delays, appeared to progress more rapidly during somitogenesis, eventually aligning in timing with both parental species during late segmentation and hatching. As a result, final hatching times were broadly similar among EE, TT, and ET, while ETT embryos required the longest developmental intervals to reach this endpoint.

Discussion

General Patterns of Embryonic Development and PGC Migration in Cobitis

Embryonic development in all examined Cobitis genomotypes (both parental species [C. elongatoides and C. taenia] and their ET and ETT hybrids) followed the canonical teleost sequence, progressing through cleavage, blastula, gastrula, segmentation, and hatching. Each stage exhibited characteristic features such as cleavage furrows, epiboly, somitogenesis and organogenesis, and the acquisition of larval morphology. These patterns are consistent with observations in model teleosts such as D. rerio [35] and other Cobitis loaches, including C. taenia and its polyploid hybrids [29], as well as M. anguillicaudatus [34, 37, 36].

PGC migration in Cobitis paralleled that in other teleosts, though with species-specific differences in timing. In our study, PGCs appeared around 75% epiboly. This is earlier than in Tinca tinca (just before 100% epiboly [38]) and Leucopsarion petersii [37], but later than in D. rerio and M. anguillicaudatus, where PGCs are detectable by 50% epiboly [37]. The migration trajectory was highly conserved: PGCs localized bilaterally near the dorsal region by 90% epiboly, aligned along the anterior-posterior axis, and settled near the yolk extension during hatching. This pattern was consistent across species and hybrids, suggesting strong phylogenetic conservation of germline development, largely unaffected by hybridization or polyploidy.

Variation in Developmental Rates among Cobitis Species and Hybrid Genomotypes

Although all Cobitis genomotypes underwent a conserved sequence of embryonic stages, we detected significant variation in the rate at which these stages were achieved. Notably, the tempo of development varied among ploidy types and genomic constitution of investigated specimens.

In the earliest stages, C. elongatoides (EE) embryos stood out by initiating cleavage significantly earlier than all other groups (Fig. 3a, b). However, beginning at the gastrulation stage, hybrids (both diploid [ET] and triploid [ETT]) consistently lagged behind both parental species. Ultimately, from the segmentation period onward, triploid hybrids exhibited a more pronounced and sustained developmental delay compared to both diploid hybrids and parental species. In contrast to triploids, the diploid hybrids, while initially slower, appeared to partially catch up during somitogenesis and reached hatching at similar times to diploid species.

While triploid hybrids came from at least two independent clonal lineages, we were able to investigate diploid ET hybrids from a single clonal lineage only. This suggests that observed differences between diploid and triploid hybrids may not be unambiguously attributed to effects of ploidy, but also to maternal effects stemming from inherent similarity of clonal siblings within the diploid group of hybrids. Nevertheless, we observed the overall pattern of generally slower development in investigated triploids encompassing at least two clonal lineages determined by allozymes. Hence, our findings broadly support the widely reported trend of polyploidy slowing down embryonic development observed in another polyploid clonal Cobitis [29] and across various fish and amphibian systems [9, 45, 46]. Such a trend of prolonged embryonic stages and slower post-embryonic growth in polyploids has been extensively investigated in plants (reviewed in [8, 13]), but also in animal triploids and tetraploids [15, 47], and is generally attributed to increased cell and genome size, which lead to slower cell division rates and extended tissue differentiation periods [12, 13]. In addition, regulatory disruptions due to dosage imbalance or hybrid genomic architecture may further contribute to developmental instability, particularly in non-model or evolutionarily young polyploid systems [8].

However, at a finer scale our results reveal a more nuanced picture, since triploid embryos also exhibited unexpectedly faster progression through certain early stages, suggesting that developmental dynamics may shift over time and that genome size or composition alone cannot fully account for the entirety of observed patterns. Similar stage-specific shifts in developmental pace have been reported in some artificially induced triploid salmonids, like Oncorhynchus mykiss, with overall faster embryonic development in triploids compared to diploids, though hatching timing remained similar in one study [48], while in the other, triploids initially lagged in post-hatching growth but later matched diploids by 11 months of age [49]. Likewise, in Clarias gariepinus, triploid larvae lagged behind diploids until the blastula stage, after which they accelerated and hatched earlier [50]. Studies on sturgeons further highlight how developmental timing may result from a complex interplay of ploidy, egg size, and maternal origin. Detlaff et al. [51] reported that from early gastrulation stages, Acipenser ruthenus, a tetraploid with small eggs, developed faster than both other species, a tetraploid Huso huso and an octoploid Acipenser gueldenstaedtii, which have considerably larger eggs. However, at the hatching stage, the octoploid A. gueldenstaedtii exhibited a marked slowdown in development, while both tetraploid species continued developing at a comparatively faster pace, irrespective of differences in egg size.

Another factor potentially interfering with observed partial acceleration in hybrids could involve heterosis (hybrid vigor). This phenomenon, widely documented in both animals and plants [52], may imply enhanced performance in hybrids relative to their parental species, potentially also enhancing developmental rates. However, heterosis is a multifaceted phenomenon with no universal mechanism [52]. For instance, intergenus and intragenus crosses of Danio and Devario have shown that hybrid traits and offspring viability depend on phylogenetic distance, with greater divergence often increasing lethality [53]. In other study, conversely, intergeneric hybrids between xenocyprinids – Megalobrama amblycephala and Chanodichthys dabryi showed faster development and hatching times than one of the parental species and outperformed both parents in growth [54]. In our case, diploid hybrids (ET) did not exhibit any obvious developmental acceleration over the parental species, yet they also did not lag significantly behind them, unlike triploid hybrids (ETT) that initially performed similarly to diploids, but exhibited pronounced delays at later stages, consistent with previous findings [29]. Interestingly, in this hybrid complex, recent comparisons among triploid forms of various combinations of parental genomes suggest that developmental timing and the occurrence of malformations during early life stages indeed depend on the interaction between particular genome composition and temperature [55].

Overall, our data indicate that, unlike some cases of heterosis reported in other fish systems, Cobitis hybrids do not exhibit consistent early developmental acceleration, and their developmental dynamics appear primarily shaped by ploidy and maternal or lineage-specific effects rather than hybrid vigor.

PGCs Migration and Numbers

While the influence of ploidy on the overall tempo of embryonic development proved to be stage-specific and more nuanced than expected, it had a more pronounced effect on PGCs. Triploid hybrids (ETT) exhibited significantly higher numbers of PGCs compared to both diploid hybrids (ET) and diploid parental species (EE), whose PGC counts during somitogenesis closely matched those previously reported for M. anguillicaudatus [36].

One possible explanation is a maternal, clone-specific effect. Indeed, variation in PGC number has been documented even within single species [39], and since Cobitis diploid and triploid hybrid lineages are clonal and genetically identical to their maternal progenitors, we may have sampled triploid clones predisposed to higher PGC production, as such traits could be maternally inherited in agreement with the concept of maternal effects [56]. However, our dataset included embryos from at least two clonal triploid lineages, making it unlikely that this effect alone explains the observed pattern.

It is also possible that clonal hybrids in general may be under selection for an elevated number of PGCs. This selection may stem from the highly inefficient process of PMER, whereby only a minority of hybrid germ cells successfully duplicate their genomes and proceed through meiosis, while most oogonia (∼95–99%) undergo defective synapsis and subsequent apoptosis [24]. Therefore, a potential to expand the initial pool of PGCs may help asexual hybrids increase the absolute number of oocytes that successfully bypass this cytogenetic bottleneck. However, it is not straightforward to evaluate how such a process may apply to our case. First, unlike in mammals, where the initial PGC count likely sets the upper limit for the ovarian reserve [57, 58], a proportion of PGCs in teleost fish acts as a reserve and contributes later [59]. Second, this hypothesis does not fully explain why diploid hybrids despite also undergoing the same PMER as triploids, do not show similarly elevated PGC numbers. This discrepancy could have several reasons. For instance, it may still reflect a lineage-specific maternal effect in a single sampled diploid clone, or point to differences in reproductive strategy: triploid hybrids, which produce fewer but larger eggs [60], may benefit from maximizing PGC input to ensure sufficient oocyte output, whereas diploids may offset germline loss through higher fecundity and the production of numerous, smaller eggs. In any case, the difference between diploid and triploid hybrid counts PGCs suggests that hybridization and/or clonality alone may not universally explain the observed phenomenon.

Another possible explanation is that differences in PGC abundance between all-female clonal hybrids and bisexual parental individuals (C. elongatoides) are linked to sex differentiation, as in many teleosts higher PGC numbers tend to bias development toward females, while their reduction, or even depletion promotes all-male progeny [61–64]. However, the relevance of this mechanism in Cobitidae remains uncertain. Experimental depletion of PGCs in related species M. anguillicaudatus did not affect sex ratios [65], and our recent work indicates that C. elongatoides possesses a differentiated XY sex chromosome system that is non-homologous to that of C. taenia [66], suggesting a primarily genetic rather than germline-driven mode of sex determination. Importantly, both diploid and triploid hybrids are expected to produce all-female progeny, yet only triploids exhibit elevated PGC numbers, whereas diploid hybrids show PGC counts similar to bisexual C. elongatoides. If low PGC numbers promoted maleness, C. elongatoides should display the lowest values, which argues against a central role of PGC abundance in sex determination in Cobitis.

Taken together, our comparison of parental species with diploid as well as triploid hybrids indicates distribution of PGCs is most obviously dictated by the differences in ploidy. If so, this opens a potential for some mechanistically grounded hypothesis, such as role of germplasm. In teleosts, PGCs arise from maternally deposited germplasm, and recent work has demonstrated a dose-dependent effect of germplasm components on PGC induction [67, 68]. Since triploid eggs tend to be larger and contain more cytoplasm, they may also harbor greater quantities of germplasm, thereby supporting the formation of more PGCs. Similar relationships have been observed in sturgeons and zebrafish, where germplasm availability directly influences PGC number [69, 70].

In this context, it is noteworthy that recent findings have also elucidated a role for mitochondria in the regulation and rejuvenation of PGCs in sturgeons. PGC number was shown to increase in proportion to mitochondrial abundance within the germplasm [71]. The elevated cytoplasmic volume observed in triploid hybrids may therefore facilitate an increased mitochondrial content in oocytes, potentially stimulating higher PGC proliferation. Although this remains untested in Cobitis, the hypothesis offers a compelling mechanistic explanation for elevated PGC numbers in triploids, and possibly in polyploids more generally.

However, we note that our dataset – while including embryos from at least two independent triploid clonal lineages – remains limited, and the observed pattern should be interpreted with caution. Across earlier developmental stages (38- and 42-somite), both triploid lineages consistently exhibited significantly elevated PGC numbers relative to diploid forms. At the 46-somite stage, however, this pattern was less uniform: while one lineage (N = 4) retained higher PGC counts, the second lineage was represented by a single individual whose PGC number fell within the diploid range. Given the extremely limited sample size for this lineage at this stage, it is not possible to distinguish whether this observation reflects stochastic variation, methodological constraints (e.g., partial overlap of PGCs leading to undercounting), or biological processes such as differential proliferation or apoptosis. Experimental work in this system is inherently demanding, as it requires genetically verified individuals from morphologically indistinguishable lineages, multiple controlled crosses, and technically challenging embryological manipulations. Consequently, even datasets based on multiple lineages may remain constrained in sample size at particular stages.

Despite these limitations, the overall trend of increased PGC numbers in triploids is consistently supported across earlier stages and independent lineages. At the same time, the observed variability suggests that the relationship between ploidy and PGC number is not strictly deterministic, but likely modulated by additional lineage-specific or developmental factors. Future studies based on larger sample sizes and complementary approaches – such as histological analyses of earlier stages – will be necessary to clarify the mechanisms underlying this pattern.

Conclusions

In conclusion, our results suggest that early development in Cobitis is largely conserved; however, developmental tempo and progression through particular developmental stages depends on ploidy and genomic composition and may even be species-specific.

Namely, while hybrid forms (diploid and triploid) develop slightly slower at early ontogenetic stages as compared to their parentals, diploid ET hybrids caught up during somitogenesis and reach hatching at comparable times with their parental species. In contrast, triploid hybrids (ETT) exhibit pronounced delays at later stages, reaching the hatching period as the last among all investigated forms. This indicates that polyploidy may predominantly affect later developmental dynamics but is not deterministic at early stages, as also indicated from some literature reports on sturgeons [51]. Overall, we found that both types of hybrids represent developmentally distinct entities, likely shaped by interactions among genomic architecture, maternal effects, cytological traits, and potentially heterosis.

Second, we observed the elevated number of PGCs in both investigated clones of triploid hybrids compared to their diploid counterparts. This effect is novel to the best of our knowledge and may reflect synergistic operation of several key mechanisms, such as a direct mechanistic consequence of cell architecture implied by ploidy (e.g., scaling of germplasm or cytoplasmic volume) and/or an adaptive compensation for inefficiencies associated with clonal reproduction.

These findings call for an increased attention to early developmental processes, which are usually difficult to investigate but crucial for the establishment of hybrid, polyploid and asexual lineages. Thus, they open new avenues for future research on the effects of polyploidy on cellular metabolism and its potential role in resolving aspects of the so-called “The sex paradox” in relation to hybrid origin of asexuality.

Acknowledgments

Authors are profoundly obliged to our greatest technicians without whom all laboratories would be closed and no science would have been done (Š. Pelikánová, J. Machová, and P. Šejnohová). As none of the authors are native English speakers, the manuscript was reviewed for grammar and language clarity using ChatGPT-4 [72].

Statement of Ethics

All procedures were performed in accordance with the Animal Research Committee of the Faculty of Fisheries and Protection of Waters (Vodnany, Czech Republic) and were approved by the Ministry of Agriculture of the Czech Republic (reference No. MZE-3972339723/2025–13143). Fish were maintained according to the principles based on the EU-harmonized Animal Welfare Act of the Czech Republic and principles of laboratory animal care in compliance with the national law (Act No. 246/1992 on the Protection of Animals Against Cruelty). Fish from the Polish population were collected in accordance with the permission of the General Director for Environmental Protection in Poland (permit No. DZP-WG.6401.117.2022.TŁ.2). Fish from Czech Republic were collected and maintained in accordance with the permission of Ministry of Agriculture of the Czech Republic (permit No. MZE-24154/2021-18134).

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

The study was supported by the Czech Science Foundation Project No. 24-12217S. Institute of Animal Physiology and Genetics receives support from Institutional Research Concept, Grant/Award No. RVO67985904. The study was also financially supported by the Ministry of Education, Youth and Sports of the Czech Republic, projects CENAKVA (LM2018099), Biodiversity (CZ.02.1.01/0.0/0.0/16_025/0007370), VVI CENAKVA Research Infrastructure (ID 90238, MEYS CR, 2023–2026).

Author Contributions

A.Z.: investigation, methodology, formal analysis, conceptualization, funding acquisition. G.S.: investigation, formal analysis, visualization, validation, and writing – original draft. R.F.: methodology, resources, supervision, visualization, and funding acquisition. T.T.: investigation and formal analysis. O.B. and D.K.: formal analysis. M.P.: methodology, resources, and supervision, J.K.: resources and supervision. K.J.: conceptualization, resources, supervision, project administration, funding acquisition, and writing – original draft. All authors contributed during the writing – review and editing.

Funding Statement

The study was supported by the Czech Science Foundation Project No. 24-12217S. Institute of Animal Physiology and Genetics receives support from Institutional Research Concept, Grant/Award No. RVO67985904. The study was also financially supported by the Ministry of Education, Youth and Sports of the Czech Republic, projects CENAKVA (LM2018099), Biodiversity (CZ.02.1.01/0.0/0.0/16_025/0007370), VVI CENAKVA Research Infrastructure (ID 90238, MEYS CR, 2023–2026).

Data Availability Statement

Data on developmental stages, PGC development, and migration have been deposited in Zenodo (https://doi.org/10.5281/zenodo.19661553).