After a catastrophe, a little bit of sex is better than nothing: Genetic consequences of a major earthquake on asexual and sexual populations

Abstract Catastrophic events can have profound effects on the demography of a population and consequently on genetic diversity. The dynamics of postcatastrophic recovery and the role of sexual versus asexual reproduction in buffering the effects of massive perturbations remain poorly understood, in part because the opportunity to document genetic diversity before and after such events is rare. Six natural (purely sexual) and seven cultivated (mainly clonal due to farming practices) populations of the red alga Agarophyton chilense were surveyed along the Chilean coast before, in the days after and 2 years after the 8.8 magnitude earthquake in 2010. The genetic diversity of sexual populations appeared sensitive to this massive perturbation, notably through the loss of rare alleles immediately after the earthquake. By 2012, the levels of diversity returned to those observed before the catastrophe, probably due to migration. In contrast, enhanced rates of clonality in cultivated populations conferred a surprising ability to buffer the instantaneous loss of diversity. After the earthquake, farmers increased the already high rate of clonality to maintain the few surviving beds, but most of them collapsed rapidly. Contrasting fates between sexual and clonal populations suggest that betting on strict clonality to sustain production is risky, probably because this extreme strategy hampered adaptation to the brutal environmental perturbation induced by the catastrophe.


| INTRODUC TI ON
Catastrophic events, such as wildfires, storms, tsunamis, earthquakes, or volcanic eruptions, dramatically affect both population demography and genetic diversity (Lande, 1988(Lande, , 1993, especially in sessile and sedentary organisms. The sharp reduction of the census size following these events generally leads to severe demographic bottlenecks and to subsequent impoverishment of genetic variation (Carson, 1990;Nei, Maruyama, & Chakraborty, 1975). Signatures of such bottlenecks result in the loss of allelic variants, especially the rarer ones, but the signal only persists a few generations after the catastrophic event (Luikart & Cornuet, 1998). In addition, the distribution of the remaining genetic diversity among individuals may also be modified, as reduced effective size increases the odds of mating among relatives (Frankham, 1998;Spielman, Brook, & Frankham, 2004). Such high stochasticity at both the genetic and demographic levels constitutes a fertile ground for shifts in reproductive mode.
For example, in plants, a reduction of self-incompatibility was observed in bottlenecked populations (Reinartz & Les, 1994) and increased rates of self-fertilization are often detected at the retracting edges of a species' range (Levin, 2012;Pujol, Zhou, Sanchez Vilas, & Pannell, 2009).
Documenting the consequences of catastrophic events on genetic variation and reproductive system in the wild is difficult and relies on chance since they are, by nature, unpredictable. Indeed, most studies have focused on the consequences of a catastrophic event on genetic structure by analyzing the impacted populations a posteriori (Jacquemyn, Roldán-Ruiz, & Honnay, 2010;Russello, Gladyshev, Miquelle, & Caccone, 2004) or sometimes benefitting from banks of ancient data (e.g., data from museum specimens: Nyström, Angerbjörn, & Dalén, 2006;archeological data: Hadly et al., 2004;Weber, Stewart, & Lehman, 2004). Population-based studies in which information was gathered before and after the extreme event in order to quantify the changes in genetic diversity caused by the catastrophe are rare (but see Gallardo, Köhler, & Araneda, 1995;Hsu et al., 2017;Pujolar et al., 2011;Wilmer et al., 2011). These studies have reported a loss of genetic diversity by directly comparing the standing variation before and after the event.
If loss in genetic variation is expected regardless the rate of sex, the severity and the duration of the recovery phase should vary (Hörandl, 2006). As stated in Hörandl (2006), "sexuals can probably re-establish genetic variation more effectively after a bottleneck (…), apomicts [i.e., asexuals or clonals], in contrast, need longer time periods and/or multiple colonizations for the creation of clonal diversity." Indeed, if clonality could favor demographic resilience by maximizing reproductive insurance, the propagation of new alleles, originating from mutation or migration, is slowed down by the rarity of sex. On the other hand, large clonal lineages are more prone to survive severe demographic bottleneck. We thus hypothesized that clonality may buffer, at least to some extent, the loss of variation caused by extreme demographic reduction. Testing the hypotheses of Hörandl implies working along a gradient in the rate of clonality (corresponding to the relative frequency of the descendants resulting from clonal reproduction within a population; Marshall & Weir, 1979).
The red alga Agarophyton chilense (formerly Gracilaria chilensis) represents a good candidate to tackle these predictions about the evolutionary shifts caused by catastrophic events in partially clonal species. This species, as other species of Gracilariaceae (Krueger-Hadfield et al., 2016), is able to shift between purely sexual reproduction (involving the alternation of haploid gametophytes and of morphologically similar diploid tetrasporophytes, both fixed to rocky substrate via holdfasts, forming fixed populations) and asexual propagation (via vegetative fragments of fronds capable of generating new free-floating individuals growing on sandy/muddy bottoms; Guillemin et al., 2008). While asexual reproduction has never been reported in fixed populations of A. chilense (Guillemin et al., 2008;Guillemin, Valero, Faugeron, Nelson, & Destombe, 2014), humans have made use of this clonal ability for aquaculture development (Buschmann, Hernandez-Gonzalez, & Varela, 2008). Indeed, A. chilense is one of the very few domesticated algae (Valero et al., 2017). Its farming began in the 1980s and relies on planting vegetative cuttings of both gametophytes and mostly tetrasporophytes (Guillemin et al., 2008).
This recent anthropogenically induced high rate of clonality has already affected genetic and genotypic diversities of A. chilense.
Notably, diploid heterozygous lineages may have been indirectly selected by farming practices (Guillemin et al., 2008). All information available about genetic diversity and reproductive system in A. chilense was gathered during sampling campaigns performed between 2004 and 2009 (Guillemin et al., 2008(Guillemin et al., , 2014  1 hereafter). Coastal communities, including fishermen which were economically dependent on the farming and harvesting of A. chilense, were impacted by the huge waves of the resulting tsunami and the changes of coastal configuration (i.e., coastal uplift, Castilla, Manríquez, & Camaño, 2010). Ecosystems, housing, vessels, and infrastructure were heavily damaged in this region and led to permanent shifts in economic activities, including the desertion of algal farming in some of the previously most intensively harvested areas (Marín, Gelcich, & Castilla, 2014). Southward, the region of Puerto Montt was totally spared. These two regions benefited from a genetic survey starting before 2010 (see Guillemin et al., 2008) and were resampled immediately after the earthquake and tsunami of February 2010 and again 2 years later, in 2012.
These questions about the respective advantage of sex versus clonality in buffering the catastrophe and promoting population resilience, and the farmer's evolutionary role in this socio-ecological system is central to develop conservation and management policies in A. chilense. They are also interesting for fundamental research on partially clonal organisms, fueling the current debate about genetic rescue (Whiteley, Fitzpatrick, ChrisFunk, & Tallmon, 2015;Willi, Kleunen, Dietrich, & Fischer, 2007).
The aim of this paper was to empirically address the predictions of Hörandl (2006) that sexuals should be able to restore genetic diversity more rapidly after a bottleneck than asexuals.
Such studies are, to our knowledge, rare in primary producers and mostly nonexistent for partially clonal species. More specifically, we will test the following predictions: (a) The genetic variability is less impacted by the catastrophic event in farms, due to the enhanced buffering capacity of clonality; (b) the restoration of genetic variability is more rapid in sexual populations; and (c) putative shifts of reproductive strategies occurred in impacted populations. Finally, the current health status of natural and cultivated populations will be discussed in light of the impact of farmers have on their evolutionary trajectories.

| Model species and sampling strategy
Agarophyton chilense is characterized by a complex isomorphic life cycle with an alternation of independent haploid and diploid individuals (i.e., male and female gametophytes and tetrasporophytes, respectively) with a similar morphology. In the field, only fertilized female gametophytes are easily detected by eye. Distinction among nonmature individuals, reproductive but nonfertilized female gametophytes, male gametophytes, and tetrasporophytes is not possible.
The sampling was therefore blind, and it was not possible to control, a priori, the number of diploid individuals collected in each site. The sex and stage (ploidy level) of the individuals was determined a posteriori by observations of the reproductive organs under a binocular microscope (Guillemin et al., 2008). For nonmature individuals, phase was determined using sex markers developed by Guillemin, Huanel, and Martínez (2012). Only diploid individuals were kept for analyses, for two main reasons. Diploids indeed represented the main portion of the samples, notably in farms (Guillemin et al., 2008).
In addition, most of the methods of inference in population genetics were developed for diploids.
Temporal sampling of natural fixed populations (N = 7) and farms (N = 6) began in 2002 and was undertaken during the Austral summer months (i.e., January to March). Before the earthquake, sites were sampled at one or two time steps between 2002 and 2009 (Table 1). Sampling in 2010 was completed in March, <1 month after the earthquake. All sites were sampled in 2012.
Sampling was carried out in two Chilean regions where both A. chilense natural and farmed opulations are encountered: Concepción, heavily impacted by the earthquake of 2010 through both tsunami waves and coastal uplift (Castilla et al., 2010;Vargas et al., 2011), and Puerto Montt where this event went almost unnoticed ( Figure 1).

| DNA extractions, PCR amplifications, and locus scoring
DNA extractions followed the protocol recommended by Cohen et al. (2004). Due to the eventful evolutionary history of A. chilense (an initial founder event followed by recurrent demographic bottlenecks explained by overharvesting and domestication; Guillemin et al., 2014), only a few microsatellite markers have been shown to be variable in Chile (Guillemin, Destombe, Faugeron, Correa, & Valero, 2005).
Indeed, previous studies showed that genetic diversity of A. chilense is highly reduced along the Chilean coast compared to the species region of origin (New Zealand) probably because of the combined effects of the founding event (that likely took place at the end of the Quaternary) and the recent human impacts of harvesting and cultivation practices (Guillemin et al., 2014). PCR amplification of five microsatellite loci and allele size scoring was performed according to Guillemin et al. (2005).
For loci 2B2, 6C7, 7F12, and 8B2 (locus names follow Guillemin et al., 2005), PCR products were visualized on an ABI 3100 Sequencer fragment analyzer (Applied Biosystem). For the locus 7D3 (Guillemin et al., Note: In the column "type of populations," we further indicate whether populations were natural (P) or cultivated (F) and whether they were impacted by the earthquake (in February 2010, gray column)-if so noted Impacted. A semiquantitative assessment of the population size (S) before the earthquake is provided, with 1 indicating a size a few meters square, 10 for few tens of m 2 , 100 for few hundreds of m 2 , and 10e6 for big farms covering many hectares. For each sampling date, the number of diploid samples used in this work is provided. The symbol "X" indicates that concerned populations disappeared after 2010.
2005), PCR products were run on 6.5% polyacrylamide denaturing gels in a LI-COR DNA sequencer model 4200™ (LI-COR). For all five loci, PCR products of five to ten individuals already genotyped for the study of Guillemin et al. (2008) were scored jointly with the new samples to ensure correspondence in allele size between temporal sampling. A second and third round of PCR and electrophoresis were performed for individuals with missing data.
Genotypes with missing data were not retained for the following analyses. Most genotypes, from sampling done before 2009, were already published in Guillemin et al. (2008). Temporal datasets for farms and wild A. chilense used in this study are available in DRYAD: https://doi.org/10.5061/dryad.9kd51 c5d8.

| Clonal and genetic diversities
Due to cultivation practices of A. chilense, the occurrence of clonal replicates is expected to be high in farms (Guillemin et al., 2008). The dataset was then first screened for identical multilo-  (Nei, 1978) using GEnEtiX v4.05 (Belkhir, Borsa, Chikhi, Raufaste, & Bonhomme, 2004). For each sampling site, the departure from panmixia was estimated through the F iS of Weir and Cockerham (1984) and its significance was assessed with a procedure of 1,000 permutations.

| Estimation of rates of clonality in farms
The rate of clonality c was assessed using ClonEstiMate, an efficient and robust Bayesian method to infer rates of clonality from populations genotyped at two time steps (Becheler et al., 2017). We hypothesized that our populations of A. chilense evolve according to an extended Wright-Fisher model (hereafter noted eWFM) explicitly taking into account partial clonality (assuming stable population size, mutation rates, and reproductive mode; no migration and selection; equations available in Stoeckel & Masson, 2014). Based on the transition of genotypic frequencies between two time steps, ideally corresponding to two successive generations, ClonEstiMate allows inferring the most likely rate of clonality explaining such a transition. According to the recommendations provided by the authors, c was not calculated when the sampling dates were too distant (i.e., more than 2 years) as this time lapse probably involved more than two generations and could lead to underestimation of c (Becheler et al., 2017). Inferences were thus calculated between 2009 and 2010, and between 2010 and 2012, for each farm where data were available.

| Data analysis
Two kinds of analysis were performed. First, the temporal behavior of population descriptors of clonal and genetic diversities was ana-  Table 1).
• Recovery: This group encompassed the populations from the region of Concepción sampled in 2012, 2 years after the catastrophic event.
We first performed Mann-Whitney Wilcoxon's rank tests to detect any differences among groups in the distribution of each descriptor of diversity. We compared nonimpacted versus im- The concept of effective size is hardly extendable to clonal populations, as it was developed for purely sexual organisms; therefore, it is not recommended to infer demography from genotypes in partially clonal populations. We thus propose an alternative methodology to evaluate the consequences of the catastrophic event on the genetic evolution of populations.
Transitions of genotypic frequencies measure the changes in frequencies of genotypes at one locus, in a population, from time t to t + 1. Assuming that populations fit with a mathematical population genetics model, with known and stable evolutionary forces (here, the eWFM described above), the probability of these transitions can be computed using CloNestiMate (Becheler et al., 2017).
This approach aimed at testing whether the earthquake altered the evolutionary forces (e.g., an increased strength of drift due to strong demographic reduction).
were computed assuming a fixed value of c = 0. Mann-Whitney Wilcoxon's rank tests were then performed to compare the probabilities of transitions of the nonimpacted versus impacted groups and the impacted versus recovery groups.

| RE SULTS
3.1 | Genetic diversity of natural sexual populations, before and after the earthquake Among the seven natural populations sampled, genetic diversity was low, independent of sampling year (0.28 < H E < 0.59 and 1.6 < Â < 3.4; Table 2). Most populations (75%) did not depart from Hardy-Weinberg equilibrium (HWE, nonsignificant F iS in 15 out of 20). Five F iS values were significantly different from 0: two of them were positive and three were negative (Table 2).
In the impacted region of Concepción, the two small populations of Tubul (P-TUB) and Lenga (P-LEN) Figure 1) did not survive the tsunami and coastal uplift caused by the earthquake, while the two slightly larger natural populations (P-DIC and P-COL) persisted (Table 1). Recovery dynamics were therefore only studied in these two surviving populations.
Changes in genetic diversity (H E ), between sampling years, were generally small for these two sexual populations (Table 2).
However, substantial changes just after the catastrophic event  (Table S1). In these impacted populations, the F iS increased between 2010 and 2012, from 0.01 to 0.22 in P-DIC and from 0.08 to 0.14 in P-COL (Table 2)

| Clonal and genetic diversities of farmed populations, before and after impact
As expected, the farmed populations were highly clonal (Table 3), often dominated by one or few asexual lineages, as attested by low values of R and β (Table 3, Figures 2 and 3). The same MLGs were detected over the whole sampling period (

| D ISCUSS I ON
This study documents the genetic consequences of a major cata- However, our results also showed that human practices in these farms played a role (while questionable from an evolutionary perspective) on population recovery and resilience.

| Strong impact of the mega earthquake on natural populations and quick recovery due to migration
In the days following the earthquake, two natural populations from  (Table 1, Figure 2), as expected in populations which sizes recently decreased (Carson, 1990;Luikart & Cornuet, 1998;Nei et al., 1975).
Surprisingly, genetic diversity in these two natural populations (estimated using allelic richness or expected heterozygosity) recovered to pre-earthquake levels after only 2 years. Extreme mortality events have been shown to lead to a decrease in genetic diversity due to drift (Chan, Lacey, Pearson, & Hadly, 2005;Gallardo et al., 1995;Hsu et al., 2017;Pujolar et al., 2011;Reynolds, Waycott, & McGlathery, 2013). However, impacts of catastrophic events on population size and genetic diversity as well as the speed of rebound are highly idiosyncratic and depend on the magnitude of the disturbance, local geography (e.g., existence of barriers to gene flow), and species dispersal capacity (Brante et al., 2019;Hadly et al., 2004;Spiller, Losos, & Schoener, 1998). Less prolific species, or with lower dispersal capacity, may show more lasting effects of extreme mortality events on genetic diversity, with slow genetic recovery (Brante et al., 2019;Hadly et al., 2004;Wilmer et al., 2011).
The Gracilariales, as most macroalgae, have generally been considered as poor dispersers since their sexual free-living stages are restricted to short-lived gametes and spores that generally recruit in close proximity (few meters) to parents (Engel, Destombe, & Valero, 2004;Engel, Wattier, Destombe, & Valero, 1999 The status of each population is described in the column "effect EQ" (effect earthquake). For impacted populations, we distinguished three substatus, before earthquake (status before), directly after it (status EQ), and 2 years after (status after). We estimated rates of clonality (c) from transitions of genotype frequencies between two sampling dates using ClonEstiMate. The number of multilocus genotypes (N MLG ) over the number of sampling units (N SU ), clonal diversity (R), and the parameter β of the Pareto's distribution was computed using GEnclonE v2.0. Expected heterozygosity (H E ) associated with its standard deviation, allelic richness (Â), and inbreeding coefficient F IS was estimated with GEnEtiX v4.05 (ns: nonsignificantly different from 0, *: p < .05, **: p < .01 and ***: p < .001).
depend not only on the ability to emigrate from source populations, but also to become established into sink populations. This is often compromised in sessile organisms, such as algae, when the density of already established individuals is high (monopolization hypothesis, De Meester, Gómez, Okamura, & Schwenk, 2002). The quick genetic recovery observed in the two natural populations studied here may be explained by the fact that mass mortality due to the earthquake opened up space for migrant settlement. Our results argue for moderate-to-high connectivity of A. chilense in the Concepción region, in agreement with a previous study (Guillemin et al., 2014) that proposed that the species is capable of colonizing areas with few or no established conspecifics. before the earthquake disappeared in 2010 (Table S1) Table 1. Numbers of sampled individuals are given in brackets tuco-tuco following a volcanic eruption (Hsu et al., 2017). The resilience of the polymorphism is explained by the arrival of new rare alleles during the recovery, likely immigrating from neighboring populations where they occurred at higher frequencies (Table S1). This suggests that resilient populations functioned as open ones, violating a second hypothesis of the eWFM (i.e., absence of migration) as disturbing local gene frequencies (Wright, 1931). This is in line with former studies revealing an increased weight of the migration after disturbance that provide opportunities for recruitment (Becheler, Benkara, Moalic, Hily, & Arnaud-Haond, 2014;Becheler, Diekmann, Hily, Moalic, & Arnaud-Haond, 2010;Eriksson, 1993;Reusch, 2006).
Increase in available open space lead, in some extreme cases, to mosaic patterns where impacted areas colonized by external sources are genetically distinct from the surviving local patches (Parvizi, Craw, & Waters, 2019) or to a reset of the genetic composition (Hsu et al., 2017;Wilmer et al., 2011) after the near-total crash of populations.
The surprising consequences of such profound genetic reset rely in a resulting diversity comparable to the original one, as the rapid recolonization is achieved through intense long-distance migration (i.e.,  (Frankham, 1998). Agarophyton chilense being a species presenting biphasic life cycle where independent haploid and diploid generations are encountered, the putative inbreeding depression could be mitigated by an effective purge of deleterious mutations during the haploid phase (Valero, Richerd, Perrot, & Destombe, 1992; but see Tortajada, Carmona, & Serra, 2009or Szövényi et al., 2013. In this species, we could expect the deleterious effects of the transition phase (i.e., at most a few generations after the perturbation) would come to an end once the frequency of new migrant alleles rises in these affected populations.  (Guillemin et al., 2008;Krueger-Hadfield et al., 2016).

| High level of clonality in farms buffer
Indeed, in highly clonal populations, each possible genotypeheterozygous or homozygous-can be fixed by drift (Reichel, Masson, Malrieu, Arnaud-Haond, & Stoeckel, 2016 (Marín et al., 2014;Rojas et al., 2017), it is also possible that farmers from Tubul have generated a mismatch between historically selected genotypes growing in the salt marsh and the requirements of the newly planted habitats. A study realized in the Puerto Montt region showed that A. chilense monoclonal stands performed better than more genotypically diverse beds (Usandizaga et al., 2020), suggesting the existence of multi-purpose genotypes (Baker, 1974;Lynch, 1984) in this species farms. Such super clones are expected to display a large plasticity, enabling them to cope with environmental fluctuations, and in some cases to be maintained over extensive time periods (Arnaud-Haond et al., 2012;Reusch, Bostrom, & Stam, 1999).
It seems that even if super clones populated F-TUB, they were not able to cope with the drastic habitat change from a salt marsh to a sandy beach with direct oceanic influence. This evolutionary failure may be a first indication of the importance of the genetic background of A. chilense populations in farm resilience. This issue does not only concern the focal species of this study, but most of the cultivated red Coupled with the weeding of plants germinated from seed presenting low growth rate, this strategy maintains high levels of genotypic diversity and high heterozygosity in these crops (Delêtre, Hodkinson, & McKey, 2017;Duputié, David, Debain, & McKey, 2007;Pujol, David, & McKey, 2005).

| CON CLUS IONS
This case study reported distinct fates of sexual versus clonal populations after a natural disaster. This is attributable to differences in reproductive strategy and is in line with several theoretical expectations about the respective advantages of sex and clonality. While strict clonality (i.e., rate of clonality, c, close to 1) conferred a noticeable ability to buffer genetic impoverishment after a catastrophic event, it lowered the odds of survival at larger environmental and time scales.
In contrast, populations reproducing sexually were more sensitive to the immediate genetic impacts of catastrophic events. Yet, genetic recoveries of these populations were fast, probably promoted by longdistance migration and rapid integration, by sexual recombination, of new alleles into the surviving local genetic pool (Keller et al., 2001).
Partial clonality could, however, allow combining the advantages of the two reproductive modes. Indeed, it was recently shown that trajectories of partially clonal species (i.e., for 0 < c < 0.9) were very similar to purely sexuals (Reichel et al., 2016;Stoeckel et al., 2019). This suggests that only a small amount of sex is sufficient to benefit from its evolutionary advantages after some generations.
Partial clonality is widely distributed within the tree of life, including ecologically central species (such as seagrasses, corals), many species of commercial or agricultural interest (tree-species), or human parasites. The implications of the presented results and the recent theoretical developments in population genetics of partially clonal species go beyond the specific case of algal farming.
Understanding the influence of this mixed reproductive mode on evolution is absolutely necessary for a world, where the frequency of dramatic fluctuations in population size is increasing. Notably, caricatural representations of clonality as a binary trait (i.e., purely sexual versus purely clonal) should no longer persist. On the contrary, future studies devoted to partially clonal species should consider the rate of clonality as a continuous variable and more fully appreciate its variation among populations and species, as well as its evolutionary influences. It is a prerequisite for relevant management plans of harvested species, conservation actions of threatened species, or efficient measures in public health. Finally, while even a small amount of sex is evolutionarily advantageous, clonality also constitutes an opportunity for successful genotypes to spread and persist, notably the general purpose genotypes. On this point, several questions remain unresolved. How do such genotypes arise? What are the evolutionary consequences of the presence, or introduction, of superclones into a population? It would be wise to resolve these questions prior to promoting general purpose genotypes.

ACK N OWLED G EM ENTS
We wish to thank Jonathan Cahn and Oscar Huanel for their help in the field. We are grateful to Flavia Nunès and the first referee for useful comments and suggestions on the earlier version of this manuscript. We thank the Biogenouest genomics core facility and Clonix2D (ANR-18-CE32-0001); and the international research network "Diversity, Evolution and Biotechnology of Marine Algae" (GDRI CNRS No. 0803).

CO N FLI C T O F I NTE R E S T
None declared.