The effects of allospecific mitochondrial genome on the fitness of northern redbelly dace (Chrosomus eos)

Abstract Instantaneous mitochondrial introgression events allow the disentangling of the effects of hybridization from those of allospecific mtDNA. Such process frequently occurred in the fish Chrosomus eos, resulting in cybrid individuals composed of a C. eos nuclear genome but with a C. neogaeus mtDNA. This provides a valuable model to address the fundamental question: How well do introgressed individuals perform in their native environment? We infer where de novo production of cybrids occurred to discriminate native environments from those colonized by cybrids in 25 sites from two regions (West‐Qc and East‐Qc) in Quebec (Canada). We then compared the relative abundance of wild types and cybrids as a measure integrating both fitness and de novo production of cybrids. According to mtDNA variation, 12 introgression events are required to explain the diversity of cybrids. Five cybrid lineages could not be associated with in situ introgression events. This includes one haplotype carried by 93% of the cybrids expected to have colonized West‐Qc. These cybrids also displayed a nearly complete allopatric distribution with wild types. We still inferred de novo production of cybrids at seven sites, that accounted for 70% of the cybrids in East‐Qc. Wild‐type and cybrid individuals coexist in all East‐Qc sites while cybrids were less abundant. Allopatry of cybrids restricted to the postglacial expansion suggests the existence of higher fitness for cybrids in specific conditions, allowing for the colonization of different environments and expanding the species’ range. However, allospecific mtDNA does not provide a higher fitness to cybrids in their native environment compared to wild types, making the success of an introgressed lineage uncertain.

pathways (Le Bras, Clément, Pervaiz, & Brenner, 2005). Protein complexes of the respiratory chain that allow the production of energy by mitochondria are encoded by both the mitochondrial (mtDNA) and the nuclear genomes (nucDNA). A strong co-evolution between these genomes is expected to maintain the highly specific interactions between protein subunits required to be fully operational (Burton, Pereira, & Barreto, 2013;Osada & Akashi, 2012).
Mitochondrial introgression generally results from asymmetric hybridizations and repeated backcrossing and is therefore a slow process lasting several generations (Rieseberg & Wendel 1993). This phenomenon may also occur in a single generation in some hybrid complexes (Goddard & Schultz, 1993;Yamada et al., 2015). Hybrids of these complexes are perpetuating lineages with their own evolutionary fate.
They can occasionally produce individuals in which a diploid nuclear genome and mtDNA from different species are brought together in a single generation. Such instantaneous mitochondrial introgression events allow the disentangling of the effects of hybridization from those of allospecific mtDNA.
In the northern redbelly dace (Chrosomus eos), individuals can harbor either C. eos or C. neogaeus mtDNA, referred to hereafter as wild types and cybrids, respectively ( Figure 1a). Instantaneous mitochondrial introgression resulting in de novo production of cybrids ( Figure 1b) is possible due to the presence of all-female hybrids C. eosneogaeus (Goddard, Dawley, & Dowling, 1989). These hybrids reproduce clonally by gynogenesis; sperm of either C. eos or C. neogaeus is required, but only to trigger the development of the unreduced eggs (Goddard, Megwinoff, Wessner, & Giaimo, 1998). However, a high proportion of triploid hybrids may occur when the nuclear genome of C. eos sperm is incorporated in unreduced hybrid eggs . Triploid hybrids are expected to occasionally produce eggs with a haploid C. eos genome and C. neogaeus mtDNA (Goddard & Schultz, 1993). Fertilization of such an egg by a C. eos haploid sperm reconstitutes the diploid nuclear genome of C. eos but with a C. neogaeus mitochondrial genome and results in an instantaneous mitochondrial introgression (Angers & Schlosser, 2007;Binet & Angers, 2005;Goddard & Schultz, 1993).
Cybrids produced de novo inherit of a combination of nuclear and mitochondrial genomes that have not co-evolved since speciation time (ca. 5 Myears, Deremiens, Schwartz, Angers, Glémet, & Angers, 2015). Moreover, they must compete with C. eos wild type that are expected F I G U R E 1 Instantaneous mitochondrial introgression in the fish Chrosomus eos. (a) Individuals of the complex Chrosomus eos-neogaeus including C. eos and hybrids. (b) Hybridization between a male C. eos and a female C. neogaeus results in diploid hybrids. These all-female hybrids reproduce clonally by gynogenesis; sperm is only required to trigger egg development. Occasional incorporation of the genome of C. eos sperm results in triploid hybrids. Triploid hybrids can produce eggs with a haploid C. eos genome but a C. neogaeus mtDNA. Instantaneous mitochondrial introgression occurs when such an egg is fertilized by the C. eos sperm, leading to de novo production of cybrids. E and N refer to nuclear genome of C. eos and C. neogaeus, respectively, superscript to mitochondrial genome to be well adapted to environmental conditions. These Chrosomus eos cybrids can therefore provide a valuable model for addressing a fundamental question about mitochondrial introgression: How well do introgressed individuals perform through time in their native environment compared to wild types?
This study aimed to assess the effects of allospecific mtDNA C. neogaeus on the fitness of C. eos in their native environment. More specifically, we compared the long-term demography of C. eos cybrids to that of the wild types from which they occurred. Once produced, cybrids can reproduce sexually as wild types do. However, de novo production of cybrids represents an additional input of individuals that can demographically favor cybrids. On the one hand, cybrids are expected to exclude wild types if they display fitness similar or higher to that of wild types, as theoretically demonstrated by Barron, Lawson, and Jensen (2016). On the other hand, a lower fitness of cybrids can be demographically compensated by de novo production of individuals so that both cybrids and wild types can coexist in sympatry. To test these predictions, we determined the relative abundance of wild types and cybrids in a survey of 25 C. eos populations from southern Quebec (Canada). The relative abundance integrates the long-term demography by taking into account both fitness and the additional input of cybrids.
Determining where de novo production of cybrids occurred is of primary importance when assessing the fitness of cybrids in their native environment. This is particularly relevant in geographic contexts strongly modeled by postglacial expansion such as northeastern North America (April, Hanner, Dion-Côté, & Bernatchez, 2013;Gagnon & Angers, 2006), because cybrids of a given site may also originate from postglacial colonization. We thus inferred where de novo production of cybrids occurred to discriminate between the native environments and those colonized by cybrids during the postglacial expansion. At a given site, we inferred de novo production of cybrids when cybrids and one sympatric hybrid lineage (the C. neogaeus mtDNA donor) shared the same mtDNA sequence. In the absence of sympatric hybrids or if the mitochondrial haplotype of cybrids did not match with that of sympatric hybrids, we assumed that an introgression event occurred prior to postglacial colonization or that hybrid lineage that gave rise to the cybrid was extinct at this site.
We performed the survey in two regions (West-Qc and East-Qc) known to display contrasting patterns of C. eos-neogaeus hybrid diversity. In West-Qc, one hybrid lineage is widespread throughout this region resulting from postglacial colonization (Angers & Schlosser, 2007).
In East-Qc, multiple hybridization events occurred in situ and lineages displayed a narrow geographic distribution (Vergilino, Leung, & Angers, 2016) and may therefore represent different sources of cybrids.

| Prevalence of wild types and cybrids
A total of 664 individuals visually identified as Chrosomus eos (New, 1962) were collected from 18 sites in West-Qc and seven sites in East-Qc (Table 1, Figure 2) in southern Quebec (Canada). The sampling of wild type and cybrid individuals was random considering as they could not be visually discriminated. We first genetically confirmed the identity of individuals using markers specific to the C. eos nucDNA according to Binet and Angers (2005). We then used a PCR-RFLP-based method to identify individuals as wild type or cybrid according to mtDNA. In a single PCR, we used two pairs of primers (CR-eos and CR-neogaeus; Table 2)  We partitioned the diversity of mtDNA detected in C. eos (wild type or cybrid) by calculating the diversity overall sites (H T ), the diversity within population (H S ) using the average of Nei's gene diver-

| Inference of in situ production of cybrids
We inferred the locations of de novo production of cybrids to discriminate between native environments and those founded during the postglacial expansion. A correspondence between the cybrid and sympatric hybrid lineage mtDNA was considered as a de novo production of cybrids and the site as a native environment of cybrids.
The procedure was achieved in three steps: (1) a wide-scale survey using the single strand conformation polymorphism method (SSCP; Orita, Suzuki, Sekiya, & Hayashi, 1989) to analyze a large number of individuals; (2) the identification of mtDNA haplotypes according to previous studies using a reference gene; and (3) the confirmation of the correspondence between cybrid and hybrid mtDNA detected at step 1 by sequencing a large and variable region of the mtDNA.
We first performed the survey of mitochondrial DNA diversity using three gene segments (D-loop, ND3, and COI; T A B L E 1 Characteristics of the sampled sites. Geographic coordinates, relative abundance of wild-type and cybrid individuals per site, sample size (n), and Nei's gene diversity We then assigned each of the different haplotypes recovered by SSCP to the haplotypes A, B, and F previously found by Angers and Schlosser (2007) and Mee and Taylor (2012). We sequenced a segment of the cytochrome oxidase I (COI) gene (Angers & Schlosser, 2007) and searched for sequence similarity in GenBank.
Finally, we confirmed the correspondence of C. neogaeus mitochondrial haplotypes between sympatric cybrids and hybrids by sequencing a mitochondrial segment (hereafter designed as ND3-4L) encompassing a portion of the ND3 and ND4L genes and the tRNA-Arg gene using the ND3-4L primers (Table 2). We retrieved 522-bp quality sequences that were aligned using the MUSCLE algorithm available with the MEGA7 software (Kumar, Stecher, & Tamura, 2016).

| Relative abundance of wild types and cybrids
The relative abundance of wild type and cybrid individuals in a given environment was used as a measure integrating both fitness and demographic inputs of cybrids. More specifically, we assessed the cooccurrence of wild types and cybrids according to the origin of cybrids (de novo or migrants) using the C-score index (Stone & Roberts, 1990).
The C-score statistic calculates the number of sites for which wild types and cybrids never appear together. High C-score values indicate an increasing degree of mutual exclusivity between biotypes. The observed degree of allopatry was tested against 999 random communities generated according to a null model as described by Jonsson (2001).
We also constructed Mantel correlograms for establishing the spatial distribution of wild types and cybrids. We used straight line distance and waterway distance between sites. We used F ST values as dependent variables and a geographic distance matrix as an independent variable. The number of distance classes was determined according to Sturges' rule, and Mantel statistics was tested with 999 permutations using the correction for multiple tests proposed by Holm (1979). C-score and Mantel statistics were performed in R (R Development Core Team 2004) using the vegan package (Oksanen et al., 2015). Cybrids were detected in 12 of 18 sites in West-Qc and in all the sites in East-Qc (Figure 2). For a comparable regional diversity be-

| Large-scale survey of cybrid diversity
A total of six distinct haplotypes for C. neogaeus mitochondrial DNA were detected by combining variations in D-loop, ND3, and COI segments detected by SSCP (Figure 3; The haplotypes are designated using the letters of each cluster and a roman numeral (Figure 3; Table 3).
The distribution of mtDNA in cybrids revealed a strong geographic break as none of their haplotypes is shared between regions. Most of the cybrids from West-Qc were characterized by the haplotype A IV, while those from the East-Qc region displayed the haplotype A I (Figure 3). This led to a high Nei's gene diversity for overall sites (H E = 0.58). However, most populations displayed no haplotypic diversity. Only four populations revealed more than one haplotype, including site AS-1 from West-Qc that harbored three divergent haplotypes (A IV, B III, and F I).
The survey of the 13 hybrid lineages revealed a total of six haplotypes, with three closely related haplotypes for each of the groups A and B (Figure 3; Table 3). The distribution of hybrids diversity was similar to that observed in cybrids, as both regions displayed a different genetic composition: Haplotype B III was dominant in West-Qc and haplotypes A, and to a lesser extent haplotype B, characterized hybrids from East-Qc.
Cybrids displayed three haplotypes shared with sympatric hybrids (A I, A II, and B III; Figure 3). In West-Qc, no correspondence between cybrid and hybrid haplotypes was detected, except for haplotype B III, which was present in very low abundance in cybrids at sites AS-1 and AS-3. In East-Qc, in all sites but one (SF-4,5), both cybrids and hybrids that are sympatric shared the same haplotype (A I at four sites and A II at one site). This suggests that de novo production of cybrids likely occurred multiple times from sympatric hybrids.

| Inference of in situ production of cybrids
Correspondence between mtDNA haplotypes shared between sympatric cybrids and hybrids as putative donor was confirmed at the sequence level using the ND3-4L segment (GenBank accession numbers [MG793359 to MG793378]). The sequences are designated by a lowercase letter when distinct sequences of a given SSCP haplotype were recovered. Cybrids without matching mtDNA in sympatric hybrid displayed distinct sequences at the ND3-4L segment: haplotype A I-b from SF-4,5, A IV and F from AS-1 and NO-10, and A V from SF-14 (Table 4).
However, identical sequences were recovered from cybrids and one of their sympatric hybrid lineages at seven sites: two in West-Qc and five in East-Qc (

| Relative abundance of wild types and cybrids
In West-Qc, 96% of the cybrids were not produced in situ and most of them (93%) belong to the A IV haplotype (Figure 3; the West-Qc region is not explained by either geographic or hydrologic proximity, as no significant spatial autocorrelations (p > .108) were observed for the straight line or waterway distances (data not shown).
At the opposite, 69.9% of the cybrids from East-Qc were produced de novo (Figure 3; Table 5). Distribution of wild types and cybrids in East-Qc strongly contrasted with that of West-Qc as both forms were found in sympatry in all sites ( Figure 2). Moreover, wild types are more abundant than cybrids at all sites but one (SF-10; Table 1).

| Diversity of cybrids
The diversity of C. neogaeus mitochondrial DNA detected in C. eos cybrids individuals revealed multiple mitochondrial introgression events. A comparison of these different haplotypes with sympatric hybrid lineages revealed two distinct histories of mitochondrial introgression.
We first confirmed that the cybrids had been independently produced at several sites during the Holocene. The correspondence of haplotypes between cybrids and sympatric hybrids suggests seven de novo transfers of C. neogaeus mtDNA to C. eos. In West-Qc, cybrids carrying B III haplotype were detected in two geographically close sites. As we cannot rule out the migration of cybrids from one site to another, this result suggests at least one occurrence. In East-Qc, C. neogaeus mtDNA transferred to C. eos differed among sites and originated from distinct hybrid lineages, supporting five distinct introgression events. While cybrids from sites YA-1,2 and RI-3,4 displayed the same sequence, the distinct composition of hybrids did not supported migration between those sites. As hybrids of this region occurred in situ (Vergilino et al., 2016), we can conclude these transfers to C. eos also occurred during the Holocene.
However, five haplotypes detected in cybrids had no counterpart in sympatric hybrids and four of them displayed correspondence with none of the hybrids analyzed in this study. This could be the result of two nonmutually exclusive events. Firstly, cybrids may have been produced before the extinction of the hybrid lineage that gave birth to them or the hybrid is rare and has not been previously sampled. These hypotheses are likely for four of five haplotypes localized to a single or a few numbers of sites. A nonmutually exclusive hypothesis is that introgression events predated the end of the Pleistocene and cybrids colonized these sites during the postglacial expansion. This appears the most parsimonious scenario for cybrids with haplotype A IV as they were widespread T A B L E 4 Sequence identity of C. neogaeus mtDNA based on ND3-4L locus between cybrids and hybrids in sympatry. Haplotypes indicative of in situ formation of cybrids are shaded across West-Qc and account for 93% of the cybrids from this region, even in the absence of hybrids harboring this haplotype.
A comparison of the distribution of cybrids from the two surveyed regions revealed two contrasting pictures that strikingly paralleled those observed in C. eos-neogaeus hybrids. Most of the cybrids from West-Qc are expected to have originated from postglacial expansion. Postglacial colonization also appeared as the main source of hybrids, with one lineage colonizing the entire region (Angers & Schlosser, 2007;Vergilino et al., 2016). Colonization by a unique founder group for either hybrids or cybrids resulted in the low haplotypic diversity observed throughout the West-Qc region. However, 69.86% of the cybrids in East-Qc were produced in situ, as confirmed by the presence of hybrids with an identical haplotype. Most of the 36 hybrid lineages detected in East-Qc also occurred in situ during the Holocene and displayed restricted distribution areas (Vergilino et al., 2016). This high diversity of hybrids geographically organized is also reflected in the locally produced cybrids.

| Fitness of cybrids in native environment
As introgression events are expected to continuously occur via hybrids, this additional input of individuals can demographically favor cybrids. They have then the potential to exclude wild types even if both display similar fitness (Barron et al., 2016). One prediction is that most of the sites only harbored cybrids and hybrids. However, a striking result is that hybrids and cybrids did not coexist together without wild types. This leads us to propose two nonmutually exclusive hypotheses about the demography of cybrids.
A first hypothesis is that de novo production of cybrids did not frequently occur and does not represent a substantial contribution to the population growth of cybrids. The low abundance of cybrids in their native environment suggests that effective production of cybrids from hybrids is likely limited and only occasional. These empirical results strongly contrast with the expectations of Barron et al.'s theoretical model (2016). Under their model, an abundance of hybrids allowing the formation of cybrids as low as 5% is expected to generate high amount of cybrids, leading the sympatric wild type to extinction (Barron et al., 2016). In the current study, wild-type C. eos and hybrids are found together in 13 different sites but only seven of them harbored locally produced cybrids. Moreover, we failed to detect genetic signature of de novo production of cybrids for eight hybrid lineages ( Figure 3). Therefore, de novo production of cybrids is not only dependent on the presence of hybrids, as proposed by Barron et al. (2016), but also other factors, such as genetic, ecological, and environmental conditions, may hinder the efficiency of cybrid production.
The second hypothesis is that cybrids did not have higher fitness than wild types in their native environment. A mtDNA haplotype under positive selection is expected to increase in abundance and reach fixation (Smith & Haigh, 1974). In sites where in situ production of cybrids has been inferred, C. eos wild types were always present, indicating that in their native environment, cybrids could not exclude wild types.
Furthermore, wild types were more abundant in East-QC, where most of the de novo introgression events were observed. We can therefore conclude that allospecific mtDNA do not provide an immediate fitness advantage to locally produced cybrid. This prevents the competitive exclusion of wild types well adapted to these local conditions, even if cybrids can be occasionally produced by hybrids.

| Fitness of colonizing cybrids
The postglacial origin of most of the cybrids from the West-Qc sites provides a valuable system to assess their fitness across different environmental conditions. Cybrids and wild types were usually detected in allopatry in West-Qc, and their distribution could not be explained by either geographic distance or hydrologic network. Such a distribution could be the result of random processes associated with postglacial dispersal or the genetic drift of mtDNA. However, we can rule out these hypotheses when taking the presence of hybrids into account. Because hybrids require the sperm of one closely related sexual species to reproduce (Goddard et al., 1998), they are expected to co-occur as frequently with wild types as do cybrids in a null model of distribution. However, when co-occurrences were computed and tested by permutation following the method described in Borcard, Gillet, and Legendre (2011), the results revealed that hybrids T A B L E 5 Mitochondrial diversity in the cybrid populations. Relative abundance of the different C. neogaeus haplotypes detected in 202 introgressed C. eos individuals. Asterisk refers to cybrids produced in situ co-occurred more frequently with wild types (p = .003) than with cybrids (p = .340). For instance, considering sites where wild types and cybrids where allopatric, hybrids coexisted with wild types in three of the six sites but occurred in none of the seven sites where only the cybrids were found. The strong co-occurrence of wild type and hybrids coupled with the allopatry of cybrids make random processes unlikely to explain their respective distribution.
An alternative hypothesis is that wild types/hybrids and cybrids have colonized (or survive in) sites with different ranges of environmental conditions. In the absence of hybrids with haplotype A IV, we can rule out contemporaneous production of that cybrid in West-Qc region. This indicated that cybrids may have higher fitness than wild types in a given range of environmental conditions. This does not contradict the hypothesis that cybrids did not have higher fitness than wild types in their native environment. For instance, the higher abundance of cybrids reported in the northern part of C. eos distribution led Mee & Taylor (2012) to suggest that C. neogaeus mitochondria provide a fitness advantage over the C. eos mitochondria in colder habitats. Chrosomus neogaeus mtDNA has a significant influence on the phenotype of individuals as different epigenomes and proteomes were detected between the wild types and cybrids (Angers, Dallaire, Vervaet, Vallieres, & Angers, 2012). Cybrids also present a higher mitochondrial respiratory chain complex IV activity and higher swimming capacity than the wild-type C. eos (Deremiens et al., 2015). The hypothesis that cybrids may have a better fitness in specific conditions is therefore consistent with the literature reporting that mitochondrial introgression may allow one species to increase its range of distribution or its tolerance to environmental changes (Blier, Breton, Desrosiers, & Lemieux, 2006;Boratyński et al., 2011;Toews et al., 2014).
In conclusion, this study revealed the importance of analyzing multiple instantaneous introgression events as the fitness advantages associated with allospecific mtDNA may strongly differ between native environments and those colonized during the postglacial expansion. The allospecific mtDNA does not provide better fitness in the native environment of cybrids when compared to that of wild types expected to be well adapted to local conditions, making the success of an introgressed lineage uncertain. However, cybrids may have higher fitness than wild types in specific conditions and can allow for the colonization of environments different from those of the wild type, thus expanding the range of a species.