bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1 Influence of historical and human factors on genetic structure and diversity 2 patterns in peripheral populations: implications for the conservation of Moroccan 3 trout
4 Perea S1*, Al Amouri M2, Gonzalez EG1, Alcaraz L1, Yahyaoui A2, Doadrio I1
5 1Biodiversity and Evolutionary Biology Department, Museo Nacional de Ciencias 6 Naturales, CSIC. Madrid, Spain.
7 2Laboratory of Biodiversity, Ecology and Genome, Faculty of Sciences, Mohammed V 8 University. Rabat. Rabat. Morocco.
9 *Correspondence: Silvia Perea. Biodiversity and Evolutionary Biology Department, 10 Museo Nacional de Ciencias Naturales, CSIC, José Gutiérrez Abascal, 2. 28006 Madrid. 11 Spain. E-mail: [email protected]
1 1 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
12 Abstract
13 1.The brown trout s.l. has been the focus of numerous phylogeographic and conservation 14 studies due to its socioeconomic importance, its marked genetic and phenotypic 15 differentiation and its broad distribution range. Especially interesting evolutionary patterns are 16 observed for populations occupying peripheral areas of a distribution range, such as in the 17 case of the highly isolated trout populations in Morocco.
18 2.Continuous stocking programs may conceal natural genetic patterns, making it challenging 19 to discern evolutionary patterns. In Morocco, trout stocking programs have been implemented 20 to increase the genetic diversity of native populations by pooling fish of different origins in 21 the Ras el Ma hatchery (Azrou region) and then stocking them in the different basins. In this 22 study, phylogenetic and phylogeographic patterns, as well as genetic structure and diversity, 23 of Moroccan trout populations were analyzed to evaluate the impact of continuous fish 24 stocking on evolutionary processes in order to better distinguish between natural and human- 25 mediated patterns.
26 3.Two mitochondrial and nine microsatellite markers were analyzed for all populations along 27 the entire distribution range of brown trout in Morocco. Phylogenetic and phylogeographic 28 analyses rendered two highly divergent evolutionary lineages, one comprising populations in 29 the Drâa Basin and a second grouping the remaining Moroccan populations. Divergence of 30 the Drâa lineage occurred during the Upper Pliocene, whilst differentiation within the second 31 lineage coincided with the onset of the Pleistocene.
32 4.Genetic structuring among populations was evident. Nevertheless, populations exhibiting 33 higher levels of genetic diversity were those affected by human-mediated processes, making it 34 difficult to associate this diversity with natural processes. In fact, highly geographically 35 isolated, not stocked populations showed the lowest values of genetic diversity. Although 36 stocking management may increase the genetic diversity of these populations, it could also 37 lead to the loss of local adaptive genotypes. Hence, current trout conservation programs 38 should be revised.
39 Keywords: biodiversity, conservation evaluation, fish, genetics, rivers, streams
40
2 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
41 1. Introduction
42 Biodiversity loss in freshwater systems is occurring at a faster rate than in terrestrial 43 environments (Dudgeon et al., 2006; Strayer & Dudgeon, 2010). Exploitation of water 44 resources, habitat alterations, mainly due to water extraction for agriculture, hydraulic 45 infrastructure construction and urban water use, along with the introduction of exotic species, 46 is exacerbating this loss of freshwater biodiversity (Dudgeon et al., 2006). In the 47 Mediterranean region, climate models predict an increase in temperature and a decrease in 48 precipitation that will decrease the availability of water resources and intensify the effects of 49 increasing human pressure upon freshwater ecosystems (García-Ruiz, López-Moreno, Vicente 50 Serraron, Lasanta-Martínez & Beguería, 2011; Guiot & Cramer, 2016). Populations located in 51 the periphery of the species distribution ranges would be especially vulnerable to these global 52 warming effects, especially in the southern periphery, that could increase the extinction risk of 53 these populations (Gibson, van der Marel & Starzomski, 2009). Within this context, 54 freshwater fish faunas are considered useful indicators of trends in aquatic ecosystems due to 55 their intrinsic characteristics such as being at the top of food webs or their long longevity and 56 high mobility (Li, Zheng & Liu, 2010; Estevez et al., 2017). Some groups are especially good 57 indicators of water quality due to their specific ecological and habitat requirements. For 58 example, salmonids (e.g. trout, salmon and char) have very restricted ecological requirements 59 such as cold, clean and well-oxygenated waters and, therefore, are highly sensitive to habitat 60 change (Almodovar, Nicola, Ayllon & Elvira, 2012; Merrian, Fernandez, Petty & Zegre, 61 2017; Young et al., 2018).
62 Within salmonids, the brown trout is distributed widely across Europe, North Africa 63 and western Asia. Morocco constitutes the southwestern limit of the brown trout distribution 64 range and suitable habitats for these organisms in this region are found at higher altitudes than 65 in the northern latitudes. Consequently, these southern peripheral Moroccan trout populations 66 have a fragmented distribution and are mainly located in headwaters of high-altitude rivers, in 67 most cases only in the sources of these rivers, and in oligotrophic lakes in the Atlas 68 Mountains. Some populations are also found at lower altitudes in the Mediterranean slope of 69 the Rifian region but only in areas characterized by steep slopes with fast currents and 70 oxygenated waters (Pellegrin, 1924). Peripheral populations of widely distributed species, as 71 Moroccan trout, are expected not only to be more geographically isolated, showing a strong 72 population structure but also to have smaller effective population sizes and lower genetic
3 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
73 diversity (see Eckert, Samis & Lougheed, 2008 for a review of genetic diversity in peripheral 74 populations). For these reasons, knowing the genetic structure and diversity of Moroccan trout 75 populations is vital in order to evaluate the conservation status of these populations and to 76 design appropriate conservation strategies according to the needs and characteristics of 77 organisms from the peripheral areas of distribution ranges; Siler, Oaks, Cobb, Ota & Brown, 78 2014; Thorton et al., 2017).
79 For these conservation strategies, parameters such as genetic structure and diversity of 80 populations, or demographic trends, are critical to design accurate plans. The patterns of 81 genetic structure and diversity of populations are influenced by both historical and human- 82 mediated contemporary factors that determine changes in population size and gene flow 83 (Vucetich & Waite, 2003; Muhfeld et al., 2017), and that, ultimately, have driven the current 84 genetic patterns found in Moroccan trout. Therefore, to understand how the genetic structure, 85 diversity and demography of Moroccan trout populations have been shaped along its 86 evolutionary history, it is essential to consider on the one side the complex geomorphological 87 landscape of the High Atlas Mountains with its rivers flowing between steep canyons and the 88 climatic events that cause drastic floods in Morocco, that is a result of high tectonic activity 89 occurred during the Neogene and Quaternary eras (Michard, Frizon de Lamotte, Saddiqi & 90 Chalouan, 2008; Babault, Van den Driessche & Texeill, 2012; El Fels, Alaa, Bachnou & 91 Rachidi, 2018), and on the other side contemporary factors related to human activity in this 92 country. This complex geological history of Morocco has generated its highly diverse 93 mountains and river basins, and their highly isolated streams. Therefore, a high level of 94 genetic structure is expected in Moroccan trout since the formation of these freshwater 95 ecosystems.
96 Contemporary factors related to human activity are increasingly crucial to understand 97 current genetic structure and diversity of Moroccan trout populations. Aquatic resources are 98 being damaged worldwide due to increased infrastructure construction and overexploitation of 99 these resources for agricultural purposes and tourism (Tekken & Kropp, 2015; Molle & 100 Tanouti, 2017). The effects of these activities have been especially detrimental in Morocco, 101 where the annual total renewable water resources per capita is six-fold lower than the global 102 average, and water shortages are frequent (Doukkali, 2005; Tekken & Kropp, 2015). 103 Overfishing of trout populations, a direct consequence of Morocco’s increased infrastructure 104 and tourism, is another contemporary factor affecting the genetic structure and diversity of
4 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
105 populations (Almodovar & Nicola, 2004; Sánchez-Hernández, Shaw, Cobo & Allen, 2016). 106 The genetic erosion of populations due to the introduction of non-native lineages, a common 107 worldwide practice for sport fishing, is another threat for trout populations in general 108 (Almodovar, Nicola, Elvira & García-Marín, 2006; Vera, García-Marín, Martínez, Araguas & 109 Bouza, 2013; Arthington, Dulvy, Gadstone & Winfield, 2016). Introgression of different 110 genetic lineages due to stocking may conceal natural genetic patterns as a consequence of 111 population structure erosion and the introduction of non-native populations (Petereit et al., 112 2018; Vera, Martínez & Bouza, 2018). Indeed, the brown trout is one of the most important 113 riverine fishes, particularly for its role in local and sport fishing, which have favored massive 114 artificial translocations of individuals among basins. Consequently, over the last several 115 decades, populations have been mixing over entire brown trout distribution range, increasing 116 the complexity of the evolutionary processes affecting this group (Horreo, Abad, Dopico, 117 Oberlin & García-Vázquez, 2015; Sanz, 2017; Závorka et al., 2017).
118 Given their fragmented and peripheral nature, some Moroccan trout populations have 119 drastically decreased in the last decades (Clavero et al., 2017). Since 1957, the Moroccan 120 government has maintained a trout pool comprised of different native populations in the Ras 121 el Ma hatchery (Azrou region). Nevertheless, little information exists about the origin and 122 contribution of each population to the Ras el Ma hatchery stock; therefore, the impact of 123 stocking processes on the genetic diversity and structure of wild trout populations in Morocco 124 is unknown. The majority of studies evaluating the impact of trout stocking in southern 125 Europe from central European populations are based on the analysis of a variant of the lactate 126 dehydrogenase C (LDH-C) locus that was not initially present in the southern European 127 populations (McMeel, Hoey & Ferguson, 2001; Kohout, Papousek, Sediva & Slechta, 2012). 128 However, the management policy in effect in Morocco differs from Europe in that native 129 Moroccan trout populations are used for stocking. Therefore, an analysis of LDH-C is not a 130 viable tool to assess hybridization in these populations.
131 The main aims of this study are to (1) determine the phylogenetic and phylogeographic 132 structure of Moroccan trout populations (Salmo spp), (2) estimate their genetic diversity and 133 demographic parameters and (3) assess the impact of fish stocking from the Ras el Ma 134 hatchery on wild trout populations through the analysis of their genetic variation. For this 135 purpose, trout populations from different locations across the highlands of Morocco were 136 studied genetically through the sequence analyses of the mitochondrial MT-CYB gene and D-
5 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
137 loop region and the genotyping of nine microsatellite loci. The identification of the historical 138 and contemporary factors driving the genetic and demographic patterns found in Moroccan 139 trout populations will provide insight on the evolutionary processes that have occurred, or are 140 occurring, at the southwestern limit of the distribution range of brown trout. The knowledge 141 gained from this study may help inform relevant conservation strategies for this freshwater 142 fish group of great socioeconomic importance.
143 2. Material and methods
144 2.1. Sampling
145 A total of 475 individuals of the genus Salmo were collected between 2009 and 2015 from 19 146 localities across the Mediterranean, the Atlantic and desert river basins located at the southern 147 end of the Atlas Mountains (Figure 1, Table 1). Brown trout populations from the different 148 basins, and samples representing all recognized lineages except for the Tigris (Bernatchez, 149 2001; Suárez, Bautista, Almodovar & Machordom, 2001; Bardakci et al., 2006), were 150 included in the phylogenetic and phylogeographic analyses (Table 1). Moreover, sequences 151 from a population in Sicily (Italy), which has been shown to be closely related to Moroccan 152 trout populations (Tougard et al., 2018), and from three specimens belonging to the species 153 Salmo macrostigma obtained from GenBank (LT617630–LT617632, including the holotype 154 housed at the Natural History Museum of Paris; Tougard et al., 2018) were included in the 155 phylogenetic analyses. Specimens were captured by electrofishing or mesh nets with the 156 permission of local authorities, fin clipped and then returned to the stream, except for two or 157 three individuals per locality, which were kept as reference specimens. Fin clips were 158 preserved in 95% ethanol. All vouchers samples are stored at the Museo Nacional de Ciencias 159 Naturales, CSIC, Madrid, Spain. Locations that are supposed to remain not stocked were 160 selected based on information provided by the fishing committee of Haut Commissariat aux 161 Eaux et Forêts et à la Lutte Contre la Désertification (HCEFLCD).
162 2.2. Mitochondrial DNA amplification and sequencing
163 Total genomic DNA was extracted from fin tissue using the BioSprint15 DNA Blood Kit 164 (Qiagen). The complete mitochondrial cytochrome b gene (MT-CYB; 1140 bp) and a fragment 165 of D-loop, a region of the mtDNA control region (999 bp final alignment including gaps), 166 were amplified by polymerase chain reaction (PCR). For the D-loop region, the primers LN20 167 (5´-ACCACTAGCACCCAAAGCTA-3´) and HN20 (5´-GTGTTAGTCTTTAGTTAAGC-3´)
6 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
168 were used (Bernatchez & Danzman, 1993). The D-loop region could not be amplified in some 169 samples with the above primers. Therefore, a new primer (cytb-dloop-st: 5´- 170 ATCGGTCAAGTTGCCTCTG-3´) was designed using OLIGO 7 (Rychlik, 2007). For MT- 171 CYB, the primers Glu-F and Thr-R (Zardoya & Doadrio, 1998) were used. The PCR reaction 172 consisted of a final volume of 25 l and included 0.2 M of each primer, 0.25M of each 173 DNTP, 10X PCR buffer, 1.5 U Taq polymerase (5 PRIME) and 40 ng of genomic DNA. The 174 following thermocycling conditions were used: initial denaturation at 95 ºC for 5 min 175 followed by 40 cycles of denaturation at 94 ºC for 45s, annealing at 52 ºC for D-loop and 48 176 ºC for MT-CYB for 45s and extension at 72 ºC for 90s, and a final extension at 72 ºC for 10 177 min. Amplified DNA fragments were checked on 1.5% agarose gels, purified using ExoSAP- 178 IT (USB) and directly sequenced on an ABI 3730XL DNA Analyzer by Macrogen Europe 179 Inc. (http://www.Macrogen.com). 180 181 2.3. Microsatellite amplification and genotyping 182 A total of 445 samples were genotyped for nine previously described dinucleotide 183 microsatellite markers: STR60 and STR15 (Estoup et al., 1993), STR85 and STR543 (Presa 184 & Guyomard, 1996), STR131 (Estoup et al., 1998), STR591 and STR541 (Estoup et al., 185 2002), SSA103 (Thorsen et al., 2005) and SSA100 (Unpublished, see Giger et al., 2006 186 supplementary material). Different multiplex reactions were amplified: STR591+STR541, 187 STR543+SSA100 and SSA103+STR60+STR131+STR15. Locus STR85 was amplified 188 separately because it required a different annealing temperature. DNA amplifications were 189 performed in 20 l reactions that contained 2X QIAGEN multiplex PCR Master Mix*, 190 following manufacturer conditions. Forward primers were labeled with fluorescent dyes, and 191 reactions were run on a Veriti Thermal Cycler (Applied Biosystems). PCR products were 192 checked on 2% agarose gels. The amplified fragments were sequenced with an ABI 3730 193 DNA sequencer (Applied Biosystems) by Secugen (Madrid, Spain), and allele sizes were 194 assigned using the program GeneMapper v3.7 (Applied Biosystems). 195 196 2.4. Phylogenetic, phylogeographic and divergence time estimation analyses
197 MT-CYB and D-loop sequences were aligned in Geneious (Kearse et al., 2012). The best-fit 198 model of evolution for each marker (MT-CYB and D-loop) and for each codon position of the 199 MT-CYB gene was estimated via Akaike Information Criterion in PartitionFinder v.1.1.1 200 (Lanfear, Calcott, Ho & Guindon, 2012). The best scheme partition is represented in Table S1
7 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
201 (Supporting Information). The species Salmo salar and Salmo ohridanus were used as 202 outgroups. Bayesian inference (BI) was performed in MrBayes 3.2 (Ronquist et al., 2012). 203 For the analysis, two simultaneous runs were performed for 107 generations, each one with 204 four MCMC chains, sampling every 100 generations. Convergence was checked with Tracer 205 v.1.7. (Rambaut, Drummond, Xie, Baele & Suchard, 2018). After discarding the first 10% of 206 generations as burn-in, the 50% majority rule consensus tree and the posterior probabilities 207 were obtained. Maximum Likelihood (ML) reconstruction was conducted with RAxML in the 208 Trex-online server using the substitution model GTRGAMMAI and the rapid bootstrapping 209 algorithm (Stamatakis, 2006; Stamatakis, Blagojevic, Nikopoulos & Antonopoulos, 2007). 210 We assessed node confidence using 10,000 non-parametric bootstrap replicates. Uncorrected- 211 p genetic distances were also estimated in MEGA v.7 (Kumar et al., 2016) in order to check 212 for differences between and within populations. To assess the phylogeographic history of the 213 genus Salmo in Morocco, we reconstructed a haplotype network with MT-CYB and D-loop 214 sequences independently to evaluate the shallow relationships among closely related 215 haplotypes. The haplotype network was constructed using POPART (Leigh & Bryant, 2015), 216 and the median-joining algorithm was used following default parameters, as recommended for 217 multiple state data (Bandelt, Forster & Röhl, 1999).
218 Divergence time estimation was carried out using the concatenated mitochondrial 219 dataset in BEAST v.1.8.3 (Drummond, Suchard, Xie & Rambaut, 2012). Only one individual 220 per brown trout population was included in the analysis. Salmo ohridanus and Salmo salar 221 were also included in the analyses. The phylogenetic and phylogeographic analyses performed 222 in this study showed the artificial origin of the Tamda population; therefore, this population 223 was excluded from this analysis. Some of the other studied populations could be affected by 224 restocking from the Ras el Ma hatchery; however, on the basis of a native population 225 hypothesis and on the lack of knowledge about its impact in native populations, all remaining 226 populations were taken into account in the divergence time estimation analysis. A random 227 local clock, which has rate heterogeneity into account, and a Birth-Death speciation model 228 were used in the analysis. Due to the lack of old and reliable fossils, the molecular clock was 229 calibrated using two different biogeographical points. Given that the S. ohridanus lineage was 230 included in the phylogenetic analysis, the first point was the age of the formation of Lake 231 Ohrid in the Pliocene, estimated to have occurred between 5-2 Mya (Albrecht & Wilke, 232 2008). The second point was the formation of Isli Lake in the Early-Middle Pleistocene 233 (Ibouh, Michard, Charrière, Benkkadour & Rhoujjati, 2014). These calibration points were
8 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
234 incorporated into the analysis as normal priors with a distribution encompassing the range of 235 the age estimated for the formation of both lakes. A secondary calibration point based on the 236 divergence of Salmo salar and S. trutta around 14-17 Mya (Horreo, 2017) was also 237 considered in the analysis. MCMC analyses were run for 400 million generations, with 238 parameters logged every 10,000 generations. Default settings were used for the remaining 239 parameters. Convergence of the analysis was evaluated in Tracer v1.7 (Rambaut et al., 2018), 240 and results were summarized in TreeAnotator v.1.8.3 (Drummond et al., 2012). 241 242 2.5. Genetic structure analyses of the microsatellite and mitochondrial data
243 Two Bayesian clustering methods were used to determine the population structure of Salmo 244 trutta based on microsatellites. The number of populations (K) with the highest posterior 245 probability (mean lnProb(D)) under an admixture model was calculated with the program 246 STRUCTURE 2.3 (Pritchard, Stephens & Donelly, 2000). MCMC simulations consisted of 247 104 burn-in iterations followed by 106 sample iterations. Each simulation was run 10 times, 248 exploring K-values from 1 to 17 (total number of analyzed populations). The most likely 249 number of homogeneous clusters (best value of K) was assessed through the modal value of 250 delta, ΔK, using the online application STRUCTURE HARVESTER (Evanno, Regnaut & 251 Goudet, 2005; Earl & VonHoldt, 2012). CLUMPAK software (Kopelman, Mayzel, 252 Jakobsson, Rosenberg & Mayrose, 2015) was used to summarize results from the 10 253 independent runs, and the results were represented graphically using DISTRUCT (Rosenberg, 254 2004). Complementary Bayesian clustering analysis was performed in TESS v.3 (Caye, Deist, 255 Martins, Michel & François, 2016), considering both CAR and BYM admixture models, 256 which address spatial autocorrelation and complex spatiotemporal processes (Durand, Jay, 257 Gaggioti & François, 2009). Exploratory data analyses for different numbers of K were run to 258 evaluate the maximum number of clusters (Kmax), setting 50,000 sweeps, a burn-in of 10,000 259 and starting with a neighbour joining tree. The model with the lowest deviance information 260 criterion (DIC) value and that stabilized at the lowest number of clusters was chosen as the 261 one that best explained the genetic variation in the data for both admixture models.
262 To assess the relative contribution of genetic variation to structure within and among 263 Moroccan trout populations, we performed an analysis of molecular variance (AMOVA) 264 based on microsatellites and mitochondrial markers, implemented in Arlequin v.3.1.5.2 265 (Excoffier & Licher, 2010). Genetic differentiation among populations was addressed through
266 ΦST pairwise comparisons (mitochondrial; Hudson, Slatkin & Maddison, 1992) and through
9 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
267 the fixation index (FST) (microsatellites; Weir and Cockerham, 1984) using Arlequin v.3.1.5.2
268 (Excoffier & Licher, 2010). Statistical significance of ΦST and FST estimates was determined 269 by 1,000 permutations of individuals among populations. Significant deviations from the null 270 hypothesis of no differentiation were assessed with 10,000 permutation tests. As multiple 271 paired tests were performed, p-values were adjusted by Bonferroni’s correction (Rice, 1989). 272 For microsatellites, genetic relationships among samples were also examined using a factorial 273 correspondence analysis (FCA) implemented in GENETIX v.4.05 (Belkhir, Borsa, Chikhi, 274 Raufaste & Bonhomme, 2004), which places the individuals in a two-dimensional space 275 according to the degree of similarity in their allelic state. 276 277 2.6. Mitochondrial and Microsatellite genetic diversity analysis 278 The following genetic parameters of diversity for MT-CYB and D-loop were estimated using 279 the software DNAsp v5.0 (Librado & Rozas, 2009): number of haplotypes (h), nucleotide
280 diversity (), haplotype diversity (HD), number of polymorphic sites (S) and number of 281 pairwise differences (K). For microsatellites, the presence of null alleles was estimated using 282 the Oosterhout’s estimator as implemented in MICRO-CHECKER 2.2.3 (Oosterhout, 283 Weetman & Hutchinson, 2006). Allelic frequencies at the different loci were estimated with 284 the program FreeNA (Chapuis & Estoup, 2007). Microsatellite genetic diversity was 285 quantified for each locus and population based on the average number of alleles per locus
286 (NA), number of alleles standardized to those of the population sample with the smallest size
287 (NAR) (Nei & Chesser, 1983) and the observed (HO) and expected (HE) heterozygosities (Nei, 288 1978) using GENETIX 4.05 (Belkhir et al., 2004) and FSTAT (Goudet, 2001). Deviations 289 from Hardy-Weinberg (HW) proportions were tested using the exact probability test for 290 multiple alleles (Guo & Thompson, 1992), available on the web-based version of GENEPOP 291 4.2 (Rousset, 2008), at each locus for each population and over all loci for each population. 292 Genotypic linkage disequilibrium between loci pairs was estimated by Fisher’s exact test with 293 the web-based version of GENEPOP 4.2 (Rousset, 2008). 294 295 2.7. Population size changes and gene flow based on mitochondrial and microsatellites 296 markers 297 To evaluate population size changes based on mitochondrial data, deviations from a model of
298 mutation-drift equilibrium were tested for both mitochondrial markers using the Fu’s FS (Fu,
299 1997), R2 (Ramos-Onsins & Rozas, 2002) and Tajima’s D (Tajima, 1989) neutrality tests as 300 implemented in DNAsp v5.0 (Librado & Rozas, 2009). To assess demographic trends,
10 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
301 mismatch analyses were carried out in DNAsp v5.0 (Librado & Rozas, 2009). Initial values
302 were set at θ0 = 0 and θ1 = 99,999. The fit of the data to the sudden demographic expansion 303 model was tested by the probability of obtaining smaller raggedness values (r) than those 304 observed under coalescent algorithm simulations over 1000 pseudo-replications and with no 305 recombination (Rogers & Harpending, 1992). Gene flow among Moroccan trout populations 306 based on mitochondrial markers was estimated through the virtual number of migrants (Nm) 307 exchanged among populations per generation (Slatkin & Barton, 1989) using Arlequin 308 v3.5.2.1 (Excoffier & Lischer, 2010). 309 310 Possible decreases in effective population size on the basis of microsatellites were assessed 311 using the software BOTTLENECK (Piry, Luikart & Cornuet, 1999). Analyses were 312 performed considering three mutation models: 1) infinitive allele (IAM), 2) stepwise mutation 313 (SMM) and 3) two-phase (TPM) considering 70% stepwise and 30% variable mutations. A 314 one-tailed Wilcoxon signed-rank test for statistical detection of Hardy-Weinberg 315 heterozigosity excess was performed, as this test is more accurate in cases with a low number 316 of polymorphic loci (Luikart, Allendorf, Cornuet & Sherwin, 1998). Estimation was based on 317 10,000 replicates. The mode-shift test (Luikart et al., 1998) in BOTTLENECK was also 318 conducted to determine whether the observed distribution of allele frequencies among 319 Moroccan trout populations differed from expectation under drift-mutation equilibrium (L- 320 shaped distribution). As with the mitochondrial markers, gene flow among Moroccan trout 321 populations based on microsatellites was estimated through the virtual number of migrants 322 (Nm) exchanged among populations per generation (Slatkin & Barton, 1989) using Arlequin 323 v3.5.2.1 (Excoffier & Lischer, 2010). 324 325 3. Results 326 3.1. Phylogenetic and phylogeographic analyses 327 We obtained a total of 462 individuals for the complete MT-CYB gene (1140bp) and 472 328 individuals for a fragment of the mitochondrial D-loop region (999 bp final alignment 329 including gaps). The concatenated mtDNA dataset used for phylogenetic analyses comprised 330 a total of 387 individuals and 2139 bp (Table 1). Bayesian and Maximum Likelihood 331 topologies for this dataset were concordant (Figure 2). Phylogenetic analyses revealed that 332 Moroccan trout populations are not monophyletic and the presence of three main highly 333 supported clades. Clade I included samples from the Moroccan Drâa Basin (Dades and 334 M’Goun tributaries, corresponding to S. multipunctatus), the Adriatic, Danubian and
11 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
335 Mediterranean lineages of S. trutta, S. marmoratus and Iberian populations from the 336 Guadalquivir Basin. Clade II comprised the other Moroccan populations analyzed along with 337 the Algerian S. macrostigma (holotype: GenBank mtDNA sequence LT617631) and 338 populations from Sicily (Figure 2). Clade II was in turn subdivided into several well- 339 supported subclades (Figure 2). Thus, Isli Lake (corresponding to S. viridis) and Rifian (Farda 340 and Kannar) populations resolved as monophyletic groups. Populations within the Oum er 341 Rbia Basin were not monophyletic as individuals from one of its tributaries (Lakhdar R.) were 342 included in two different subclades: one comprised of populations from other Oum er Rbia 343 basin tributaries and the other consisting of the endorheic Sidi Hamza population (Ziz Basin) 344 and the Mediterranean Moulouya basin population. This latter population was also recovered 345 as polyphyletic as some Moulouya individuals were nested within the Oum er Rbia, Ourika 346 and Ifni–Tifnoute subclades. Within the subclade comprising populations of the Oum er Rbia 347 basin tributaries, the population Tessaout constituted an independent and well-supported clade 348 (data not shown; pp = 1.0, bootstrap = 95). Sidi Rachid population (Sebou Basin) was 349 clustered within the Sidi Hamza subclade. Populations from the Tensift Basin (adscribed to S. 350 pellegrini) were not monophyletic, as populations from its two tributaries, Ourika and 351 Rheraya, did not clustered together. Although Rheraya constituted a well differentiated clade, 352 some of its individuals clustered within the Oum er Rbia basin subclade. Ourika individuals 353 grouped with some from the Moulouya River (River is herein abbreviated as R.). The Ifni lake 354 population (corresponding to S. akairos) was closely related to the population found in the 355 Tifnoute R., a tributary of the Souss Basin that is geographically close to Ifni Lake, and some 356 representatives of the Moulouya and Tamda populations. Brown trout from the Iberian Duero 357 Basin constituted and independent lineage (Clade III). Phylogenetic relationships among the 358 three clades were not fully solved.
359 Overall percentages of uncorrected-p distances ranged from 0.0 to 1.2% for MT-CYB 360 and from 0.0 to 1.8% for the D-loop region (Table S2 in Supporting Information). Genetic 361 distances between trout populations from Drâa Basin (Clade I) and the other Moroccan 362 populations (Clade II) ranged from 0.9 to 1.2% for MT-CYB and from 1.1 to 1.8% for D-loop 363 (Table S2 in Supporting Information). The genetic distances among the Moroccan populations 364 within Clade II ranged from 0.0 to 1.0% and 0.1 to 0.8% for MT-CYB and D-loop, 365 respectively.
366 Haplotype network analysis of MT-CYB supported the phylogenetic analyses (Figure 367 3). The analysis of all sequences obtained for MT-CYB showed the existence of five clearly
12 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
368 differentiated Moroccan populations that did not share any haplotypes with the other 369 populations: Drâa Basin, Rheraya R., Tessaout R., Sidi Rachid R. and Isli Lake. In the other 370 Moroccan populations, most of the haplotypes were shared. Rifian populations (Farda and 371 Kannar rivers) showed only one haplotype shared by all individuals from these basins. This 372 haplotype, surprisingly, was also the most common one found in the Ifni and Tifnoute 373 populations. It was also observed in the Ourika river population. On the other hand, most 374 tributaries of the Oum er Rbia Basin were clustered in a differentiated haplogroup. 375 Nevertheless, haplotypes of some of the tributaries of the Oum er Rbia Basin showed a closer 376 relationship with those of other trout populations. For instance, a haplotype from Melloul 377 were more closely related to one found in Ifni, Tifnoute and Rifian populations, or Lakhdar 378 and Ait Nacer haplotypes were related to but not shared with those found in Sidi Hamza. The 379 Sidi Rachid population also constituted a haplogroup related to Sidi Hamza. Tamda and 380 Moulouya shared their two most common haplotypes. One of the museum specimens from the 381 Beth R. (Sebou Basin) showed the same MT-CYB haplotype as the most frequent one found in 382 the Sidi Hamza and Tamda populations, whereas the two museum specimens from Algeria 383 did not shared haplotypes with any of the analyzed populations. The D-loop haplotype 384 network analysis (Figure 4) differed slightly from the MT-CYB analysis for some of the 385 relationships for the Moroccan populations excluding the Drâa Basin. The most common 386 haplotype of Isli Lake was shared with Sidi Hamza, Lakhdar, Ait Nacer and Moulouya 387 populations; however, the remaining haplotypes of Isli Lake were unique. In contrast with the 388 MT-CYB network, Rifian populations did not share its unique haplotype with the Ifni and 389 Tifnoute populations, and some of the Sidi Rachid haplotypes were shared with Oum er Rbia 390 tributary populations. Ourika was closely related to some Moulouya basin individuals. 391 Finally, Rheraya and Tessaout populations appeared as differentiated haplogroups, as was 392 observed in the MT-CYB network. The relationships of the three museum specimens were the 393 same as in the MT-CYB network.
394 According to divergence time estimates, Clades I, II and III diverged near the 395 boundary of the Pliocene and the Pleistocene, ca. 5.3 Mya (3.6–6.6 95% HPD) (Figure 5). 396 Within Clade I, the Dades and M’Goun populations (Drâa Basin) diverged from their sister 397 lineages during the Upper Pliocene ca. 4.4 Mya (2.8–5.8 95% HPD), and from each other ca. 398 0.7 Mya (0.1–1.4 95% HPD) during the Pleistocene. Diversification of Moroccan popuations 399 in Clade II started in the Upper Pliocene ca. 2.9 Mya (1.6–4.3 95% HPD) (Figure 5).
400 3.2. Microsatellites and Mitochondrial structure analyses
13 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
401 The STRUCTURE analysis based on microsatellites rendered the best clustering at 402 K=17 according to the ΔK parameter (Figure 6A). This grouping was highly congruent with 403 the mitochondrial data, supporting Drâa Basin, Isli Lake and both tributaries of the Tensift 404 Basin (Ourika and Rheraya) as differentiated genetic groups. The population structure within 405 the Oum er Rbia Basin was higher than that inferred by the mitochondrial data. Tessaout 406 constituted a well-differentiated group, while the other Oum er Rbia populations showed 407 different alleles, although with a high level of admixture among populations (Figure 6A). This 408 Bayesian clustering analysis also supported the mitochondrial relationship of the Rifian 409 populations (Farda and Kannar) with those from the Ifni and Tifnoute; although Farda had 410 low probability of assignment (Q) values due to admixture. In addition to Farda, admixture 411 was evident in the Sidi Hamza, Moulouya, Tamda and Sidi Rachid populations and in those of 412 the Oum er Rbia Basin. The ΔK also supported the clustering of K=7 (Figure 6A) and, hence, 413 genetic differentiation among some of the basins. The Bayesian clustering analysis using 414 TESS (Figure 6B) also indicated K=17 as the optimal value for both admixture models (CAR 415 and BYM), as supported by the lowest DIC (9158.23 and 9482.29, respectively). This result 416 was highly congruent with the STRUCTURE results (Figure 6A). TESS estimation based on 417 the CAR admixture model recovered Ifni and Tifnoute as different populations.
418 In the factorial correspondence analyses (FCA), the first fourth components explained 419 38.4% of the genetic variance. According to the FCA, four well supported genetic groups 420 could be identified: populations from the Ifni and Tifnoute basins, the Drâa Basin (Dades and 421 M’Goun), the Mediterranean (Farda and Kannar) and the rest of the populations (subgroup I) 422 (Figure 7A). When only subgroup I populations were considered, populations from the 423 Tensift Basin (Ourika and Rheraya) were differentiated from each other and from the other 424 populations (subgroup II) (Figure 7B). Finally, when Tensift basin populations (and previous 425 groups) were removed from the analysis, populations from Isli Lake and Tessout, a tributary 426 of the Oum er Rbia Basin, were recovered as differentiated genetic groups, whereas Sidi 427 Hamza, Sidi Rachid, Tamda Lake and the Mediterranean Moulouya Basin grouped with the 428 remaining tributaries of the Oum er Rbia Basin (Figure 7C).
429 All FST comparisons based on microsatellite data were significant (Table S3 in Supporting
430 Information). For mitochondrial data, the ΦST pairwise comparisons among Moroccan trout 431 populations ranged from 0.000 (between the two Rifian populations) to 1.000 (between Sidi 432 Hamza and Farda) for MT-CYB and from 0.027 (Dades and M’Goun) to 1.000 (Tifnoute and 433 Kannar) for D-loop (Table S4 in Supporting Information). The majority of comparisons were
14 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
434 significant after Bonferroni correction. Significant levels of population differentiation were 435 also found when the entire dataset was analyzed as one gene pool or when basins were 436 considered as independent in the AMOVA analyses based on microsatellites and 437 mitochondrial data (Table 2 and 3). The highest partition of genetic variance was found 438 among basins: 58.8% for MT-CYB and 71.2% for D-loop. Similar to the mitochondrial 439 analysis, the highest percentage of genetic variance (61.9%) was explained among basins 440 when one gene pool was considered. When all basins were considered as independent, 441 variance in all partitions was significant, with the highest percentage of genetic variance 442 explained, again, among basins in all performed analyses.
443 3.3. Mitochondrial and microsatellite genetic diversity analyses
444 Within the Moroccan populations, a total of 46 haplotypes were recognized in MT-CYB and 445 74 in D-loop. However, global haplotype diversity was very high and similar for both 446 mitochondrial markers: 0.929 and 0.937 for MT-CYB and D-loop, respectively (Table S3 in 447 Supporting Information). Global nucleotide diversity was also high (>0.5%); however, when 448 populations were considered independently, none exceeded 0.3%, the highest value found in 449 Tamda population (Table S5 in Supporting Information). Tamda Lake had the most 450 haplotypes for both markers. This population also showed the highest genetic diversity values, 451 followed by the Mediterranean Moulouya Basin. The populations from Lakhdar and Rheraya 452 rivers also exhibited high values of mitochondrial genetic diversity. Some of the lowest 453 genetic diversity values were found for Isli, Tifnoute, Sidi Hamza, Sidi Rachid and both 454 Rifian populations. In fact, only one haplotype for each mitochondrial marker was found in 455 each of the Rifian populations, as was the case for the Sidi Hamza and Sidi Rachid 456 populations for the MT-CYB gene.
457 For microsatellites, deviation from Hardy-Weinberg Equilibrium (HWE) was not consistently 458 observed across loci and populations except for the STR15 locus and Farda population after 459 Bonferroni correction (adjusted α = 0.05/9 = 0.005). According to MICRO-CHECKER, null 460 alleles were only found for STR15, accounting for its deviations from HWE. Nevertheless, 461 the FreeNA analysis, after applying the ENA correction, indicated that potential bias on
462 global FST calculations caused by null alleles was insignificant (Global FST = 0.627; Global FST 463 with correction for null alleles = 0.617). Therefore, locus STR15 was included in analyses of 464 population structure. The test for linkage disequilibrium showed a very low number of
15 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
465 significant pairwise comparisons, indicating that all examined loci can be considered 466 independent.
467 The total number of nuclear alleles detected was 91. The number of alleles per locus 468 ranged from 3 (STR541 and STR60) to 28 (STR103), with an allelic richness of 2.85 and 469 8.40, respectively. Seven of the nine loci analyzed showed a total of 41 private alleles at 470 different frequencies in the populations, from between 0.0167 and 0.0357 in Melloul, Ifni, 471 Dades and M’Goun to 1.000 in Isli, Dades-M’Goun and Farda-Kannar. Indeed, Farda and 472 Kannar (Rifian populations) shared private alleles at high frequencies in six of the loci. Drâa 473 Basin tributaries also shared six private alleles in four loci. Oum er Rbia Basin tributary 474 populations also showed private alleles, including, remarkably, four in Tessaout and three in 475 Melloul. The Tamda lake population did not have any private alleles, and Mediterranean 476 Moulouya Basin only had one private allele but at a low frequency (0.05). In contrast to the 477 mtDNA data, global genetic variability was low (Ho = 0.249) across loci and populations, 478 with low allele numbers and heterozygosity values. Farda-Kannar, Dades-M’Goun and 479 Tessaout populations had the lowest values of genetic diversity while the highest were found 480 in Tamda, Moulouya and Miaami (Table S6 in Supporting Information). Allelic richness per 481 loci ranged from 1.4 to 3.77 or from 1.19 to 2.97 when this parameter was weighted by
482 sample size (Table S7 in Supporting Information). FIS values ranged from -0.135 in Miaami to 483 0.236 in M’Goun (Table S6 in Supporting Information). The significant negative value for the 484 Miaami population indicates an excess of heterozygotes in contrast to the significant positive 485 value, representing heterozygote deficiency, in both Farda and M’Goun.
486 3.4. Population size changes and gene flow based on mitochondrial and microsatellites 487 markers
488 Based on the null hypothesis of population expansion, the neutrality tests supported 489 deviations from the mutation-drift model in both populations of the Drâa Basin and in 490 Tessaout (Oum er Rbia Basin) for both mitochondrial markers, and in Ourika (Tensift Basin)
491 for D-loop (Table S7 in Supporting Information). A significant negative value of Fu’s FS was 492 also estimated for the Mediterranean Moulouya Basin for D-loop. Estimates of the virtual 493 number of migrants exchanged among populations and per generation were greater than one 494 for some of the pairwise comparisons, particularly between the two tributaries of the Drâa 495 Basin (Nm = 18) (Table S8 in Supporting Information). Positive gene flow (Nm > 1) was also 496 estimated for populations from some of the tributaries of the Oum er Rbia Basin (Miaami,
16 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
497 Melloul and Lakhdar), as well as for the Rifian populations, Ifni and Tifnoute basins. Gene 498 flow between Tamda Lake and the other basins differed depending on the marker analyzed: a 499 Nm > 1 was observed between Tamda and two tributaries of the Oum er Rbia Basin (Melloul 500 and Lakhdar) for MT-CYB but not for D-loop. In contrast, for D-loop, a Nm >1 was found in 501 pairwise comparisons of this lake population with those from Ifni Lake and the Rifian region. 502 A Nm > 1 between Tamda and Moulouya considering both markers was also estimated (Table 503 S8 in Supporting Information).
504 The Wilcoxon test detected a recent bottleneck (p<0.05) in the Tifnoute population under all 505 three mutation models analyzed for microsatellites (IAM, SSM and TPM) (Table S9 in 506 Supporting Information). Significant values were also observed for Moulouya, Rheraya and 507 Sidi Rachid under the IAM model. The mode-shift indicator test revealed a distortion of allele 508 frequency distributions characteristic of a recent bottleneck for Miaami, Rheraya and Sidi 509 Rachid. Additionally, analysis of the virtual number of migrants with the microsatellite 510 dataset (Table S3 in Supporting Information) showed Nm values greater than one in pairwise 511 comparisons of Tamda Lake with some of the Oum er Rbia tributaries (Melloul and Lakhdar), 512 Sidi Hamza, Moulouya and Sidi Rachid. The Nm was also greater than one in comparisons of 513 Sidi Rachid with Sidi Hamza, Molouya and Miaami populations, as well as between Sidi 514 Rachid and Moulouya. The Drâa populations, Dades and M’Goun, also showed a Nm value 515 greater than one.
516 4. Discussion
517 Understanding the evolutionary processes of species of socioeconomic interest like trout is a 518 challenge due to the difficulty of isolating natural versus human-mediated patterns. 519 Additionally, the Moroccan trout populations analyzed in this study are located at the southern 520 periphery of their distribution range; thus, the evolutionary forces operating on these 521 populations may lead to different patterns relative to those in more northern and central 522 populations. These differences may, for instance, increase fragmentation and reduce genetic 523 diversity, which would have important implications for the conservation of these populations 524 and would become these peripheral populations more vulnerable to human impacts.
525 4.1. Phylogenetic and phylogeographic relationships of Moroccan trout populations
526 In general, five lineages have been traditionally recognized in the brown trout (Adriatic, 527 Atlantic, Danubian, Marmoratus and Mediterranean), which, in turn, have been classified into
17 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
528 three larger groups (Bernatchez et al., 1992; Bernatchez, 2001). Brown trout populations in 529 Morocco, including the Mediterranean ones, have been considered as part of the large Atlantic 530 lineage (Bernatchez, 2001; Snoj et al., 2011; Ninua et al., 2018). However, other authors have 531 recognized a Siculo-North African lineage, as sister to the Atlantic one, and have included the 532 Moroccan trout within it (Tougard et al., 2018). Considering only the analyzed Moroccan 533 trout populations, two divergent evolutionary lineages have been identified in the present 534 study, one comprising the Drâa basin populations in the southernmost river drainage within 535 the distribution range of trout populations in Morocco. According to our results, this lineage is 536 sister to the Danubian, Adriatic and Mediterranean lineages, Salmo marmoratus and the trout 537 population from the Iberian Guadalquivir Basin. The distinctiveness of the Drâa population is 538 in agreement with previous studies (Snoj et al., 2011; Doadrio et al., 2015; Sanz, 2017; Ninua 539 et al., 2018; Tougard et al., 2018). The second lineage is composed of the remaining 540 Moroccan trout populations, as well as the Algerian and Sicilian populations. These last 541 relationships had been previously proposed in other phylogeographic studies (Berrebi et al., 542 2018).
543 4.2. Genetic structure of Moroccan trout has been driven by the geological and climatic 544 history and by contemporary human-mediated processes
545 Analyses of concatenated mtDNA sequences and microsatellite data have revealed a genetic 546 structure of brown trout populations in Morocco that can be mainly explained by historical 547 factors associated with the complex palaeohydrology of North Africa. These historical factors 548 have proven to have greatly influenced the genetic structure of other North African freshwater 549 fishes (Doadrio, 1994, Machordom & Doadrio, 2001). However, in the case of brown trout, 550 the degree and the temporal scale of divergence among populations are different to those of 551 primary freshwater fishes, such as cyprinids or cobitids (Machordom & Doadrio, 2001; Casal- 552 López & Doadrio, 2018). In addition, the patterns revealed in this study demonstrate the great 553 impact of contemporary human-mediated processes on the genetic structure of the Moroccan 554 populations of brown trout, and along with historical factors have been responsible for the 555 current genetic structure found in these populations.
556 The Atlantic Drâa Basin is estimated to have diverged from other brown trout lineages at the 557 Upper Pliocene, which is slightly older than previous estimations for the isolation of the Drâa 558 lineage (Snoj et al., 2011). The High Atlas Mountains are thought to have experienced an 559 uplifting pulse event during the Plio-Pleistocene period less than 5 Mya that has been
18 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
560 associated with drainage reorganization (El Harfi, Guiraud & Lang, 2006; Babault et al., 561 2012; Boulton, Stokes & Mather, 2014). This tectonic pulse, along with the geomorphological 562 configuration of the western Mediterranean region since the Upper Miocene–Pliocene periods 563 (Krijgsman et al., 2018), was likely responsible for the segregation of the Drâa lineage. The 564 complex geomorphological pattern and isolation of this region has prevented the artificial 565 introduction of trout in the Drâa populations.
566 The remaining Moroccan populations indicated two different conditions. Some of the 567 populations showed a high and recent level of geographic isolation, since the Pliocene– 568 Pleistocene period, and no human influence due to stocking, a hypothesis supported by all of 569 the molecular analyses. Such is the case of the Tessaout R. located on the left margin of the 570 Oum er Rbia Basin, with a steep geomorphology, geographically distant from the remaining 571 tributaries of the basin (Melloul, Lakhdar or Miaami), which showed a smoother landscape 572 (Babault et al., 2012), sharing mitochondrial haplotypes and nuclear alelles among them. The 573 two tributary populations analyzed in the Tensift Basin (Rheraya and Ourika) also 574 demonstrated a high level of genetic differentiation and Ourika resolved as phylogenetically 575 close to the Ifni Lake and Tifnoute R. (Souss Basin) than to Rerhaya. Ourika headwaters are 576 very close to some tributaries of the Souss Basin, including the Tifnoute, and to the Ifni Lake. 577 Consequently, the close phylogenetic relationship and shared haplotypes between Ourika and 578 these populations may be explained by fluvial captures, which are known to occur all over the 579 western High Atlas Mountains during Pleistocene (Babault et al., 2012; Boulton et al., 2014). 580 The population from Isli Lake is genetically differentiated from the other populations. The 581 geological origin of this lake is still under debate due to its geomorphology (rounded contour 582 and great depth). Some authors have proposed that it originated as a consequence of a 583 meteorite impact approximately 40,000 years ago (Ibhi, Nachit, Abia, Ait Touchnt & 584 Vaccaro, 2013; Nachit, Ibhi & Vaccaro, 2013), while others have hypothesized a similar 585 tectonic origin as other Moroccan lakes that is associated with karstic phenomena during the 586 Lower–Middle Pleistocene (Chaabout, Chennaoui-Aoudjehane, Reimold, Aboulahris & 587 Aoudjehane, 2013; Ibouh et al., 2014). The degree of genetic differentiation observed in the 588 Isli lake population and its estimated time of divergence support its ancient origin and tectonic 589 origin (Ibouh et al., 2014); nevertherless, hydrographical past connections (Babault et al., 590 2012; Ibouh et al, 2014) would explain the presence of shared haplotypes between Isli and 591 those from some of the tributaries of the Oum er Rbia and Ziz basins.
19 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
592 In contrast to these highly isolated and non stocked populations, trout stocking influence is 593 evident in other populations. Thus, the Ziz basin population (Sidi Hamza) shares haplotypes 594 and shows some level of gene flow with populations in the Oum er Rbia, Tamda and 595 Moulouya basins. The river source of Ziz Basin is close to those of some Oum er Rbia 596 tributaries (Babault et al., 2012), suggesting the possibility of secondary contact between 597 these two basins. However, given the degree of admixture human-mediated contact is a more 598 plausible explanation for the relationship among these basins. The Ziz population has been 599 one of the main populations used in stocking programs at the Ras el Ma hatchery over the last 600 several decades. Therefore, the presence of its mitochondrial haplotypes in the Oum er Rbia 601 and Mouloya basins can be attributed to an artificial process. Some Rheraya and Tamda 602 individuals were also related to some Oum er Rbia populations, as well as the most common 603 haplotype of Ourika R. was also found in the Moulouya basin, probably as a consequence of 604 stocking as well (Fekhaoui, Yahyaoui, Perea & Doadrio, 2016). Another population 605 influenced by stocking is the Sidi Rachid R. (Sebou Basin); this river flows nearest to the Ras 606 el Ma hatchery, therefore, it has been stocked with various trout populations originating from 607 the hatchery (El Hassen et al., 2011).
608 The relationship found between the Mediterranean (Farda, Kannar and Algeria) and the Ifni 609 and Tifnoute basin populations is more challenging to explain without taking into account 610 human-mediated processes. The presence of only one MT-CYB haplotype in Farda and 611 Kannar that is shared with the distant populations of Ifni and Tifnoute calls into question the 612 natural condition of the Rifian populations, even though several mitochondrial analyses 613 supported the Rifian populations as distinctive groups. There is evidence that Farda R. has 614 been stocked with trout from the Ras el Ma hatchery (Table S10 in Supporting Information; 615 Fehkaoui et al., 2016). Hence, the possibility of Tifnoute trout serving as a source in 616 Mediterranean Rifian populations due to stocking cannot be discarded. Nonetheless, results of 617 the Bayesian clustering analyses based on TESS supported the Rifian populations as distinct 618 groups. Genetic differentiation of the Algerian population further supports the native status of 619 the three Mediterranean populations. These populations probably constitute the remnants of 620 an ancestral trout lineage that was widely distributed in the past.
621 4.3. Genetic diversity of Moroccan trout populations is associated with historical factors 622 and trout stocking
20 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
623 Overall genetic diversity values based on mitochondrial data were high, taking into account 624 the concatenated dataset. Nevertheless, genetic diversity levels differed among individual 625 populations and while some populations exhibit high values in others genetic diversity was 626 extremely low. Tamda and Moulouya showed the highest levels of haplotype diversity as well 627 as heterozygosity according to the results of the microsatellite analyses. Human-mediated 628 origins for Tamda and Moulouya could explain the genetic sub-structure as a consequence of 629 haplotype sharing with other basins (Sidi Hamza, Oum er Rbia and Ourika) and the high 630 levels of diversity observed in these populations, although the Moulouya haplotypes related to 631 those in Ourika probably constitute a natural lineage of this population. However, Moulouya 632 was restocked two years prior to being sampled for the present study, and the high level of 633 genetic diversity found in this population is in agreement with the use of different native trout 634 populations to increase genetic diversity, as employed by the management stocking programs 635 of Ras el Ma hatchery. In the case of Tamda Lake, the high genetic diversity but lack of 636 distinctive genetic features supports the artificial origin of this population, as has been 637 maintained by local inhabitants of the region. Hence, the high virtual number of migrants 638 found in the Moulouya and Tamda populations may be a consequence of restocking events 639 rather than reflecting a real gene flow process.
640 Lakhdar and Rheraya also displayed high levels of genetic diversity. Lakhdar showed the 641 highest microsatellite allelic richness after weighting per sample size. Lakhdar haplotypes 642 related to those in Sidi Hamza, a probable consequence of restocking as indicated by the 643 clustering analyses, may explain the high mitochondrial genetic diversity uncovered in this 644 population relative to other Oum er Rbia populations. Population expansion was not inferred 645 for Lakhdar (non-significant neutrality tests and significant raggedness index), and stable 646 population sizes may favor the maintenance of high genetic diversity in this population. 647 Rheraya also does not appear to have undergone population expansion. In fact, a significant 648 recent bottleneck was estimated to have occurred in this population on the basis of molecular 649 analyses. Stochastic and unpredictable events associated with drastic climatic episodes during 650 the Quaternary (Benito et al., 2015) may have influenced the genetic and historical 651 demography of Rheraya. Given that recurrent bottlenecks can lead to the loss of genetic 652 diversity in freshwater organisms (Allendorf, 2017; Carim, Eby, Barfoot & Boyer, 2017), the 653 high diversity values obtained for Rheraya from the analysis of MT-CYB is difficult to explain 654 without considering it a possible effect of stocking.
21 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
655 In contrast, highly geographically isolated and not stocked populations including Drâa Basin 656 tributaries, Tessaout, and Isli Lake showed low levels of genetic diversity and low allelic 657 richness. Neutrality tests supported deviations from the mutation-drift model in both 658 populations of the Drâa Basin and in Tessaout for both mitochondrial markers. These 659 deviations could be associated with recent population expansion, a hypothesis supported by 660 the star-shape of the Dades-M’Goun and Tessaout haplogroups in both mitochondrial 661 haplotype networks in which several low frequency haplotypes are connected to the most 662 frequent one. In addition, no evidence of bottlenecks was recorded for these populations. 663 However, the mismatch distribution (Figure S1 in Supporting Information) did not fit a 664 population expansion model (multimodal shape), except for M’Goun (unimodal shape and 665 significant raggedness index). Genetic drift is an evolutionary force driving loss of genetic 666 diversity due to the random fixation along time of specific alleles, and its effect is related to 667 effective population sizes (Kliman, Sheehy & Schulz, 2008; Frankham, Bradshaw & Brook, 668 2014). Given this context, genetic drift may explain the low genetic diversity found in the 669 Drâa Basin and Tessaout populations. The geographic isolation of these three populations in 670 rivers flowing through deep canyons in the High Atlas Mountains, along with severe climatic 671 conditions during the Pleistocene (Hughes et al., 2011; Babault et al., 2012), could have 672 favored small but constant effective population sizes, thereby increasing the effect of genetic 673 drift, which would lead to decreased genetic diversity (e.g. Hare et al., 2011). Indeed, species 674 occupying narrow altitudinal ranges, such as the trout in the Drâa Basin and Tessaout R., are 675 particularly vulnerable to extreme climatic events (La Sorte & Jetz, 2010; Clavero et al., 676 2017).
677 Isli Lake constitutes a population that has been isolated since the Lower–Middle Pleistocene 678 (Ibouh et al., 2014; this study). Genetic drift may have also led to the low genetic diversity 679 observed in this population. Although Isli Lake does not appear to have experienced any 680 bottlenecks, five of the nine microsatellites analyzed were monomorphic for this population, 681 probably as a direct effect of inbreeding. Therefore, the bottleneck analyses should be 682 considered with caution. In contrast, Ifni Lake showed moderately higher levels of genetic 683 diversity relative to Isli Lake, probably due to an inflow of genotypes from ancient 684 connections with geographically close rivers/basins such as the Tifnoute R. in the Souss Basin 685 (Babault et al., 2012). Genetic diversity values were extremely low in Sidi Hamza 686 (mitochondrial), Tifnoute (mitochondrial and nuclear) and both Rifian (mitochondrial and 687 nuclear) populations. A significant bottleneck was estimated for Tifnoute based on the three
22 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
688 microsatellite mutation models tested; however, as in Isli Lake, bottleneck events could not be 689 inferred for Sidi Hamza and Rifian populations as most of the loci were monomorphic for 690 these populations. However, due to the narrow geographic range of these populations, small 691 effective population sizes would be expected.
692 4.4. Implications for conservation of Moroccan trout populations.
693 The conservation status of Moroccan trout is, in general, poor, and some populations have 694 been proposed to be Endangered or Critically Endangered, according to the IUCN 695 categorization (e.g. Doadrio et al., 2015; Clavero et al., 2017). Management schemes for 696 conservation often require an understanding of population dynamics in order to achieve 697 effective long-term results. A central concept in biodiversity conservation is that genetic 698 diversity is crucial to ensure the survival of species (Frankham et al. 2014). Therefore, its 699 conservation has become an explicit goal of strategic plans such as the one implemented at the 700 Convention on Biological Diversity (http://www.cbd.int). In fact, in conservation biology, 701 genetic structure and genetic diversity are recognized as important criteria to consider when 702 prioritizing populations for protection. They are also biodiversity components to be preserved 703 according to the IUCN (Frankham et al., 2014; McGowan, Traylor-Holzer & Leus, 2017). 704 Genetic diversity can determine factors such as species viability, resilience to environmental 705 stressors and adaptation to changing environmental factors (Frankham et al., 2014; 706 Rominguer et al., 2014). The preservation of genetic diversity is based on the relationship of 707 these parameters with an organism’s potential to evolve, thus generating a background from 708 which new variants can arise. These new variants can then potentially colonize new 709 environments and respond adaptively to environmental changes (Duglosh, Anderson, 710 Braasch, Cang & Gillette, 2015; Szucs, Melbourne, Tuff & Hufbauer, 2017).
711 The low level of genetic diversity shown here, particularly for some of the native and highly 712 isolated populations, such as those in the Drâa Basin, Tessaout R. and Isli or Ifni lakes, is one 713 of the main threats to Moroccan trout populations. The geographical isolation of these 714 Moroccan trout populations, which are located in the southern periphery of the distribution 715 range of the brown trout, also contributes to reduce genetic diversity. The isolated trout 716 populations of the High Atlas Mountains inhabit fragmented and reduced and unstable 717 habitats (Doadrio et al., 2015; Clavero et al., 2017). For this reason, they are more vulnerable 718 to the effects of catastrophic events, such as flooding following an unpredictable torrential 719 rainfall (Zkhiri, Tramblay, Hanich & Berjamy, 2017), which, in turn, can lead to changes in
23 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
720 demographic and genetic patterns. Such changes likely have caused the low level of genetic 721 diversity found in these isolated Moroccan trout populations along the evolutionary time. Low 722 levels of genetic diversity increase the vulnerability and the potential extinction risk of 723 freshwater fish populations (Faulks, Kerezsy, Unmack, Johnson & Hughes, 2017; Pavlova et 724 al., 2017), particularly as they relate to inbreeding and reduced reproductive fitness, and may 725 ultimately lead to a loss of evolutionary potential (Allendorf, England, Luikart, Rithcie, &
726 Ryman, 2008). Although FIS values were not significant for homozygote excess in the 727 majority of the studied Moroccan trout populations, some were monomorphic for some of the 728 microsatellite loci analyzed, including Isli, Kannar, Ourika, Tessaout, Dades and M’Goun. 729 Inbreeding may be especially high in Isli Lake, where fin and snout malformations in adults 730 are common, and the presence of juveniles is scarce (authors’ personal observation).
731 Due to their isolated and southern peripherial nature, the instability of their habitats and the 732 low genetic diversity of Moroccan trout populations, the management policy of the Ras el Ma 733 hatchery was to maintain trout stocks mixing different native populations with the aim of 734 increasing the genetic variability of restocked populations. The main source populations kept 735 at the hatchery are from Sidi Hamza, Sidi Rachid and some tributaries from the Oum er Rbia 736 Basin. From this mixed pool, Moroccan native populations have been frequently reinforced in 737 several basins (Table S10 in Supporting Information). The conservation policy executed by 738 Moroccan environmental authorities in relation to native trout populations involves habitat 739 improvement, introduction of new populations in regions with suitable habitats and 740 population reinforcement (Fehkaoui et al., 2016). Overexploitation and catastrophic events 741 that have occurred in some Moroccan regions inhabited by trout over the last 20 years have 742 justified their reinforcement (Roman & Ait Hssane, 2014; Gaume et al., 2016; El Fels et al., 743 2018). These events are thought to have also reduced population sizes in trout populations in 744 other regions of the world (George, Baldigo, Smith & Robinson, 2015; Pujolar, Vicenzi, Zane 745 & Crivelli, 2016). Nevertheless, the current management policy of stocking trout using native 746 Moroccan populations (i.e. translocations) to increase population size and genetic diversity in 747 the stocked populations, which has been also implemented in other regions for the brown trout 748 (e.g. Prodöhl, et al., 2019), may be misguided mainly as a consequence of two main reasons: 749 the negative impact of fish stocking and the loss of local adaptations.
750 Trout stocking for fishing is a common practice worldwide (Petereit et al., 2018; Vera et al., 751 2018). In Mediterranean countries, the impact of releasing trout from hatcheries as a means to 752 reinforce natural populations is one of the main threats for the conservation of these
24 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
753 populations, especially in rivers in which demand for sport fishing outweighs productivity 754 (Škraba et al., 2017; Saint-Pé et al., 2018). In Morocco, stocking of High Atlas trout 755 populations is infrequent, supported also by the performed molecular analyses, as most 756 populations are located in difficult to access areas, such as the tributaries of the Drâa basin or 757 Tessaout R. Nevertheless, stocking is relatively frequent in more accessible rivers, such as 758 some tributaries of the Oum er Rbia basin, Moulouya basin or Rheraya R. in the Tensift basin 759 and the Tamda Lake (Fekhaoui et al., 2016). The stocking activities in these basins are 760 responsible for the high levels of genetic diversity observed in these populations, which 761 cannot be associated with natural evolutionary processes according to our analyses. Within 762 the Rifian populations, a signal of introgression was found in Farda but not in Kannar, 763 supported also by literature (Fekhaoui et al., 2016). The high proportion of shared 764 mitochondrial haplotypes and nuclear microsatellite alleles between Sidi Hamza and Sidi 765 Rachid with other trout populations is also difficult to explain unless one considers these two 766 populations as the main sources of a trout stock that is repopulating the other introgressed 767 populations. Research efforts are also hampered by stocking practices as they make it even 768 more difficult to accurately infer the evolutionary history of salmonids (Valiquette, Perrier, 769 Thibault & Bernatchez, 2014; but see White, Miller, Dowell, Bartron & Wagner, 2018). 770 Morocco, a country important for trout diversity (Delling & Doadrio, 2005; Doadrio et al., 771 2015; Tougard et al., 2018), is not an exception, and the evolutionary history of Moroccan 772 brown trout could be misinterpreted due to the difficulty of discerning the impacts of 773 continuous stocking since 1957, mainly from the Ras el Ma hatchery.
774 The negative impacts of trout stocking may be summarized in aspects such as genetic erosion, 775 introduction of pathogens and parasites, predation, ecological competition or even alteration 776 of stream ecosystems (Alexides, Flecker & Kraft, 2017). Some management policies focused 777 on stocking in alpine lakes have revealed limited genetic impact on the wild stock, even after 778 several years of stocking (Heggenes, Roed, Hoyheim & Rosef, 2002). However, this is not the 779 case of Moroccan trout, as admixture level in some stocked populations is high, as is 780 suggested in this study. Moreover, several studies have demonstrated that interbreeding 781 between farmed and wild populations may lead to loss of local adaptation to specific 782 environmental conditions due to the introduction of “maladaptive” genotypes (Skaala et al., 783 2006; Bourret, O’Reilly, Carr, Berg & Bernatchez, 2011). The highly isolated Moroccan trout 784 populations have been probably subject to local selective pressures that have led to local 785 adaptation processes, as has been suggested to occur quite frequently in other salmonids
25 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
786 (García de Leaniz et al., 2007; Fraser, Weir, Bernatchez, Hansen & Taylor, 2011). 787 Differentiation of the brown trout populations in Morocco dates from the Upper Pliocene 788 (Drâa lineage) or Pleistocene (remaining populations). In the time since their divergence, 789 evolutionary adaptation to habitats found along the distribution range could have given rise to 790 evolutionary trajectories linked to different environments. Therefore, artificial introductions, 791 even if could mean an increase of genetic diversity, may negatively impact the evolutionary 792 potential of populations and their genetic integrity due to introgressive hybridization, as has 793 been frequently described in salmonids (Bourret et al., 2011; Sušnik, Pustovrh, Jesenšen & 794 Snoj, 2015; Muhlfed et al., 2017). For this reason, maintaining the genetic integrity of 795 populations such as Isli, Ifni, Drâa or Tessaout, despite their low genetic diversity, is essential. 796 Thus, conservation measures different than those currently in effect in Morocco, and aimed at 797 keeping local adapted populations or at creating genetic refuges, as has been proposed for the 798 Iberian Peninsula (Araguas et al., 2017; Vera et al., 2018), should be planned, especially for 799 those geographically isolated populations from High Atlas Mountains.
800 5. Conclusion
801 The native populations of brown trout in Morocco constitute a very singular entity due to their 802 intrinsic characteristics and restricted ecological requirements of cold and well-oxygenated 803 waters, only found in some freshwater systems of the High Atlas and Rifian mountains. These 804 restrictive conditions limit the effective population size of some populations, as these habitats 805 are not abundant in Moroccan mountain systems. Together with the jagged orography and 806 geomorphology of the majority of rivers in the High Atlas, these conditions favor the 807 geographical isolation of Moroccan trout populations. Moreover, the ever-increasing 808 exploitation of water resources by people living in High Atlas settlements as well as the 809 different trout stocking programs that do not take into account the genetic origin of 810 populations have led to the poor conservation status of Moroccan trout populations.
811 The findings presented here also highlight the need to consider the influence of contemporary 812 processes related to anthropogenic activity on overlapping historical evolutionary processes. 813 The genetic analyses of trout populations from Morocco using mitochondrial and 814 microsatellite markers have distinguished between introgressed (Lakhdar, Moulouya, Tamda, 815 some tributaries of the Oum er Rbia basin, Sidi Rachid and Farda) and non-introgressed (Isli, 816 Ifni, Tifnoute, Kannar, Tessaout, Ourika and Drâa Basin) populations. These analyses have 817 revealed the low genetic diversity of the majority of native populations and a high level of
26 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
818 inbreeding of some populations such as in Isli Lake. These analyses have also allowed 819 evaluating the impact of the trout stocking program implemented in Morocco over the last 820 several decades that have not stringently considered the genetic origins of populations. The 821 high genetic diversity values of, for instance, Moulouya and Tamda populations is likely an 822 artifact, a consequence of being a mixed population of different origins as a result of human- 823 mediated processes. Other native populations reinforced with trout individuals from the Oum 824 er Rbia Basin also showed high levels of genetic diversity. However, the conservation status 825 of these highly diverse populations cannot be considered satisfactory as other stocking-related 826 problems, such as loss of local adapted genotypes, may be affecting these populations. Hence, 827 management programs intended to reinforce Moroccan trout populations should be revised.
828 Acknowledgments
829 We thank P. Garzón, A. Doadrio and I. Doadrio Jr. for their help in field sampling. A. 830 Machordom gave us helpful suggestions during microsatellite performance and I. Hortelano 831 provide some help in laboratory procedure. Melinda Modrell helped with the English editing 832 of this manuscript. This study was supported by the MENFPESRS and CNRST from 833 Morocco under grant N° PPR1/2015/2 for the project “Impact des changements climatiques 834 sur la diversité génétique des poissons des eaux douces du Maroc”. The High Commissioner 835 for Water, Forests and Fight Against Dessertification of Morocco provided permission for fish 836 collection.
837 6. References
838 Albrecht, C., & Wilke, T. (2008). Ancient Lake Ohrid: biodiversity and evolution. 839 Hydrobiologia, 615, 103-140.
840 Alexiades, A.V., Flecker, A.S., Kraft, C.E. 2017. Nonnative fish stocking alters stream 841 ecosystem nutrient dynamics. Ecological Applications, 27(3), 956-965.
842 Allendorf, F.W. (2017). Genetics and the conservation of natural populations: allozymes to 843 genomes. Molecular Ecology, 26, 420-430.
844 Allendorf, F.W., England, P.R., Luikart, G., Ritchie, P.A., & Ryman, N. (2008). Genetic 845 effects of harvest on wild animal populations. Trends in Ecology and Evolution, 23(6), 327- 846 337.
27 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
847 Almodovar, A., & Nicola, G.G. (2004). Angling impact on conservation of Spanish stream- 848 dwellling brown trout Salmo trutta. Fisheries Management and Ecology, 11, 173-182.
849 Almodovar, A., Nicola, G.G., Ayllón, D. & Elvira, B. (2012). Globlal warming threatens the 850 persistence of Mediterranean brown trout. Global Change Biology, 18, 1549-1560.
851 Almodovar, A., Nicola, G.G., Elvira, B., & García‐Marín, J.L. (2006). Introgression 852 variability among Iberian brown trout evolutionary significant units: The influence of local 853 management and environmental features. Freshwater Biology, 51, 1175-1187.
854 Araguas, R.M., Vera, M., Aparicio, E., Sanz, N., Fernández-Cebrián, R., Marchante, C., & 855 García-Marín, J.L. (2017). Current status of the brown trout (Salmo trutta) populations 856 withine Eastern Pyrenees genetic refuges. Ecology of Freshwater Fish, 26(1): 120-132.
857 Arthington, A.H., Dulvy, N.K., Gladstone, W., & Winfield, I.J. (2016). Fish conservation in 858 freshwater and marine realms: Status, threats and management. Aquatic Conservation Marine 859 and Freshwater Ecosystems, 26, 838-857.
860 Babault, J., Van Den Driessche, J. & Teixell, A. (2012). Longitudinal to transverse drainage 861 network evolution in the High Atlas (Morocco): the role of tectonics. Tectonics, 31, 1-15.
862 Bandelt H-J, Forster P., & Röhl A. 1999. Median-Joining network for inferring intraspecific 863 phylogenies. Molecular Biology and Evolution16: 37-48.
864 Bardakci, F., Degerli, N., Ozdemir, O., & Basibuyuk, H.H. (2006). Phylogeography of the 865 Turkish brown trout Salmo trutta L.: mitochondrial DNA PCR-RFLP variation. Journal of 866 Fish Biology, 68(Suppl.A), 36-55.
867 Belkhir, K., Borsa, P., Chikhi, L., Raufaste, N., & Bonhomme, F. (2004). GENETIX 4.05. 868 Logiciel sous Windows TM, pour Genetique des Populations. Laboratoire Génome, 869 Populations, Interactions, CNRS UMR 5171, Université de Montpellier II, Montpellier.
870 Benito, G., Macklin, M.G., Panin, A., Rossato, S., Fontana, A., Jones, … Zielhofer, C. (2015). 871 Recurring flood distribution patterns related to short-term Holocene climatic variability. 872 Scientific Reports, 5, 16398.
873 Bernatchez, L. (2001). The evolutionary history of brown trout (Salmo trutta L.) inferred from 874 phylogeographic, nested clade, and mismatch analyses of mitochondrial DNA variation. 875 Evolution 55(2), 351–379.
28 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
876 Bernatchez, L., Guyomard, R., & Bonhomme, F. (1992). DNA sequence variation of the 877 mitochondrial control region among geographically and morphologically remote European 878 populations. Molecular Ecology, 1, 161-173.
879 Bernatchez, L., & Danzmann, R.G. (1993). Congruence in control-region sequence and 880 restriction-site variation in mitochondrial DNA of brook charr (Salvelinus fontinalis Mitchill). 881 Molecular Biology and Evolution, 10(5), 1002-1014.
882 Berrebi, P., Caputo Barucchi, V., Splendiani, A., Muracciole, S., Sabatini, A., Palmas, F., … 883 Maric, S. (2018). Brown trout (Salmo trutta L.) high genetic diversity around the Tyrrhenian 884 Sea as revealed by nuclear and mitochondrial markers. Hydrobiologia, 826(1), 209-231.
885 Bourret, V., O’Reilly, P.T., Carr, J.W., Berg, P.R., & Bernatchez, L. (2011). Temporal change 886 in genetic integrity suggests loss of local adaptation in a wild Atlantic salmon (Salmo salar) 887 population following introgression by farmed escapees. Heredity, 106(3), 500-510.
888 Boulton, S.J., Stokes, M., & Mather, A.E. (2014). Transient fluvial incision as an indicator of 889 active faulting and Plio-Quaternary uplift of the Moroccan High Atlas. Tectonophysics, 633, 890 16-33.
891 Carim, K.J., Eby, L.A., Barfoot, C.A., & Boyer, M.C. (2017). Consistent loss of genetic 892 diversity in isolated cutthroat trout populations independent of habitat size and quality. 893 Conservation Genetics, 17, 1363-1376.
894 Casal-López, M., & Doadrio, I. (2018). The Messinian imprint on the evolution of freshwater 895 fishes of the genus Luciobarbus Heckel, 1843 (Teleostei, Cyprinidae) in the western 896 Mediterranean. Journal of Biogeography, 45(7), 1593-1603.
897 Caye, K., Deist, T.M., Martins, H., Michel, O., & François, O. (2016). TESS3: Fast inference 898 of spatial population structure and genome scans for selection. Molecular Ecology Resources, 899 16, 540-548.
900 Chaabout, S., Chennaoui-Aoudjehane, H., Reimold, W.U., Aboulahris, M., & Aoudjehane 901 M.M. (2013). Evidence of non-impact cratering origin of Imilchil (Morocco) lakes (Isli and 902 Tislit). In: Large Meteorite Impacts and Planetary Evolution V, 3047.
903 Chapuis, M-P., & Estoup, A. (2007). Microsatellite null alleles and estimation of population 904 differentiation. Molecular Biology and Evolution, 24, 621-631.
29 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
905 Clavero, M., Calzada, J., Esquivias, J., Verissimo, A., Hermoso, V., Qninba, A., & Delibes, 906 M. (2017). Nowhere to swim to: climate change and conservation of the relict Dades trout 907 Salmo multipunctatus in the High Atlas Mountains, Morocco. Oryx, 52(4), 627-635.
908 Delling, B., & Doadrio, I. (2005). Systematics of the trout´s endemic to Moroccan lakes, with 909 description of a new species (Teleostei: Salmonidae). Ichthyological Exploration of 910 Freshwaters, 16, 49-64.
911 Doadrio, I. (1994). Freshwater fish fauna of North Africa and its biogeography. Ann. Mus. r. 912 Afr. Centr., Zoology, 275, 21-34.
913 Doadrio, I., Casal-López, M., Perea, S., & Yahyaoui, A. (2016). Taxonomy of rheophilic 914 Luciobarbus Heckel, 1842 (Actinopterygii, Cyprinidae) from Morocco with the description of 915 two new species. Graellsia, 72(1), e039.
916 Doadrio, I., Perea, S., & Yahyaoui, A. (2015). Two new species of Atlantic trout 917 (Actinopterygii, Salmonidae) from Morocco. Graellsia, 71(2), e031.
918 Doukkali, M.R. (2005). Water institutional reforms in Morocco. Water Policy, 7, 71-88.
919 Drummond, A.J., Suchard, M.A., Xie, D., & Rambaut, A. (2012). Bayesian Phylogenetics 920 with BEAUti and the BEAST v.1.7. Molecular Biology and Evolution, 29(8), 1969-1973.
921 Dudgeon, D., Arthington, A.H., Gessner, M.O., Kawabata, Z-I., Knowler, D.J, Lévêque, C., 922 … Sullivan, C.A. (2006). Freshwater Biodiversity: importance, threats, status and 923 conservation. Biological Reviews, 81(2), 163-182.
924 Duglosh, K.M., Anderson, S.R., Braasch, J., Cang, F.A., & Gillette, H.D. (2015). The devil is 925 in the details: genetic variation in introduced populations and its contributions to invasión. 926 Molecular Ecology, 24, 2095-2111.
927 Durand, E., Jay, F., Gaggiotti, O.E., & François, O. (2009). Spatial inference of admixture 928 proportions and secondary contact zones. Molecular Biology and Evolution, 26, 1963-1973.
929 Earl, D.A., & vonHoldt, B.M. (2012). STRUCTURE HARVESTER: a website and program 930 for visualizing STRUCTURE output and implementing the Evanno method. Conservation 931 Genetics Resources, 4(2), 359-361.
30 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
932 Eckert, C.G., Samis, K.E., & Lougheed, C. (2008). Genetic variation across species’ 933 geographical ranges: the central-marginal hypothesis and beyond. Molecular Ecology, 17, 934 1179-1188.
935 El Fels, A.E.A., Alaa, N., Bachnou, A., & Rachidi, S. (2018). Flood frequency analysis and 936 generation of flood hazard indicator maps in semi-arid environment, case of the Ourika 937 watershed (western High Atlas). Journal of African Earth Sciences, 141, 94-106.
938 El Harfi, A., Guiraud, M., & Lang, J. (2006). Deep‐ rooted “thick skinned” model for the 939 High Atlas Mountains (Morocco): Implications for the structural inheritance of the southern 940 Tethys passive margin, Journal of Structural Geology, 28, 1958-1976.
941 El Hassen, A., Driss, B., Mohamed, B., Hamid, E., Eby, F.M., & Mohamadou, O. (2011). 942 Physicochemical topology of water of a middle atlas river (Morocco) where common trout 943 (Salmo trutta macrostigma, Dumeril, 1858) live: Oued Sidi Rachid. African Journal of 944 Environmental Science and Technology, 5(5), 348-354.
945 Estevez, E., Rodríguez-Castillo, T., Álvarez-Cabría, M., Peñas, F.J., González-Ferreras, A.M., 946 Lezcano, M., & Barquín, J. (2017). Analysis of structural and functional indicators for 947 assessing the health state of mountains rivers. Ecological indicators, 72, 553-564.
948 Estoup, A., Presa, P., Krieg, F., & Guyomard, R. (1993). (CT)n and (GT)n microsatellites: a 949 new class of genetic markers for Salmo trutta L. (brown trout). Heredity 71, 488-496.
950 Estoup, A., Rousset, F., Michalakis, Y., Cornuet, J.M., Adriamanga, M., & Guyomard, R. 951 (1998). Comparative analysis of microsatellite and allozyme markers: a case study 952 investigating microgeographic differentiation in brown trout (Salmo trutta). Molecular 953 Ecology, 7 (3), 339-353.
954 Estoup, A., Jarne, P., & Cornuet, J.M. (2002). Homoplasy and mutation model at 955 microsatellite loci and their consequence for population genetics analysis. Molecular Ecology, 956 11(9), 1591-1604.
957 Evanno, G., Regnaut, S., & Goudet, J. (2005). Detecting the number of clusters of individuals 958 using the software STRUCTURE: a simulation study. Molecular Ecology, 14: 2611-2620.
31 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
959 Excoffier, L., & Lischer, H.E. (2010). Arlequin suite v. 3.5: a new series of programs to 960 perform population genetic analyses under Linux and Windows. Molecular Ecology 961 Resources, 10(3), 564-567.
962 Faulks, L.K., Kerezsy, A., Unmack, P.J., Johnson, J.B., & Hughes, J.M. (2017). Going, going, 963 gone? Loss of genetic diversity in two critically endangered Australian freshwater fishes, 964 Scaturiginichthys vermeilipinnis and Chlamydogobius squamigenus, from Great Artesian 965 Basin springs at Edgbaston, Queensland, Australia. Aquatic Conservation: Marine and 966 Freshwater Ecosystems, 27(1), 39-50.
967 Fekhaoui, M., Yahyaoui, A., Perea, S., & Doadrio, I. (2016). Fragilité hydrologique et 968 biodiversité piscicole et aquacole des plans d’eau des Parcs Nationaux de la Cédraie de l’Atlas 969 (Maroc). Travaux de l’Institut Scientifique, Série Zoologie, 50.
970 Frankham, R., Bradshaw, C.J.A., & Brook, B.B. (2014). Genetics in conservation 971 management: Revised recommendations for the 50/500 rules, Red List criteria and population 972 viability analyses. Biological Conservation, 170, 56-63.
973 Fraser, D.J., Weir, L.K., Bernatchez, L., Hansen, M.M., & Taylor, E.B. (2011). Extent and 974 scale of local adaptation in salmonid fishes: review and meta-analysis. Heredity, 106, 404- 975 420.
976 Fu, Y.X. (1997). New statistical tests of neutrality for DNA samples from a population. 977 Genetics, 143, 557-570.
978 García de Leaniz, C., Flemming, I.A., Einum, S., Verspoor, E., Jordan, W.C., Consuegra, S., 979 … Quinn, T.P. (2007). A critical review of adaptive genetic variation in Atlantic salmon: 980 implications for conservation. Biological Reviews, 82, 173-211.
981 García-Ruiz, J.M., López-Moreno, J.I., Vicente-Serraron, S.M., Lasanta-Martínez, T., & 982 Beguería, S. (2011). Mediterranean water resources in a global change scenario. Earth- 983 Science Review, 105(3-4): 121-139.
984 Gaume, E., Borga, M., Llassat, M.C., Maouche, S., Lang, M., & Diakakis, M. (2016). 985 Mediterranean extreme floods and flash floods. The Mediterranean Region under Climate 986 Change. A Scientific Update. IRD Editions, Coll. Synthèses.
32 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
987 George, S.D., Baldigo, B.P., Smith, A.J., Robinson, G.R. (2015). Effects of extreme floods on 988 trout populations and fish communities in a Catskill Mountain river. Freshwater Biology, 989 60(12), 2511-2522.
990 Gibson, S.Y., Van der Marel, R.C., & Starzomski, B. M. (2009). Climate change and 991 conservation of leading-edge peripheral populations. Conservation Biology, 23(6), 1369- 992 1373.
993 Giger, T., Excoffier, L., Day, P.J., Champigneulle, A., Hansen, M.M., Powell, R., & 994 Largiadèr, C.R. (2006). Life history shape expression in salmonids. Current Biology, 16(8), 995 R281-282.
996 Goudet, J. (2001). FSTAT, a program to estimate and test gene diversities and fixation 997 indices, Version 2.9.3.
998 Guiot, J., & Cramer, W. (2016). Climate change: The 2015 Paris Agreement thresholds and 999 Mediterranean basin ecosystems. Science, 354(6311), 465-468.
1000 Guo, S.W., & Thompson, E.A. (1992). Performing the exact test of Hardy-Weinberg 1001 proportion for multiple alleles. Biometrics 48, 361-372.
1002 Hare, M.P., Nunney, L., Schwartz, M.K., Ruzzante, D.E., Buford, M., Waples, R.S., … 1003 Palstra, F. (2011). Understanding and estimating effective population size for practical 1004 application in marine species management. Conservation Biology, 25, 438-449.
1005 Heggenes, J., Roed, K.H., Hoyheim, B., Rosef, L. 2002. Microsatellite diversity assessment of 1006 brown trout (Salmo triutta) population structure indicate limited genetic impact of strockin in 1007 a Norwegian alpine lake. Ecology of Freshwater Fish, 11, 93-100.
1008 Horreo, J.L. (2017). Revisiting the mitogenomic phylogeny of Salmoninae: new insights 1009 thanks to recent sequencing advances. PeerJ, 5, e3828.
1010 Horreo, J.L., Abad, D., Dopico, E., Oberlin, M., & García-Vázquez, E. (2015). Expansion of 1011 non-native brown trout in South Europe may be inadvertently driven by stocking: molecular 1012 and social survey in the North Iberian Narcea River. International Journal of Molecular 1013 Sciences, 16(7), 15546-15559.
1014 Hudson, R.R., Slatkin, M., & Maddison, W.P. (1992). Estimations of levels of gene flow from 1015 DNA sequence data. Genetics, 132: 585-589.
33 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1016 Hughes, P.D., Fenton, C.R, Gibbard, P.L. (2011). Quaternary glaciations of the Atlas 1017 Mountains, North Africa. In: Ehlers, J., Gibbard, P.L. & Hughes, P.D. (Eds.). Quaternary 1018 Glaciations - Extent and Chronology, Part IV – A closer Look. Developments in Quaternary 1019 Science, 15. (pp 1071-1080). Eselvier. Amsterdam
1020 Ibhi, A., Nachit, H., Abia, E.H., Ait Touchnt, A., & Vaccaro, C. (2013). Isli and Tislit: the 1021 first dual impact crater discovered in Morocco. International Journal of Astronomy and 1022 Astrophysics, 3(2A), 1-4.
1023 Ibouh, H., Michard, A., Charrière, A., Benkkadour, A., & Rhoujjati, A. (2014). Tectonic- 1024 karstic origin of the alleged “impact crater” of lake Isli (Imilchil district, High Atlas, 1025 Morocco). Comptes Rendus Geoscience, 346(3-4), 82-89.
1026 Kearse, M., Moir, R., Wilson, A., Stone-Havas, S., Cheung, M., …, Drummond, A. (2012). 1027 Geneious Basic: an integrated and extendable desktop software platform for the organization 1028 and analysis of sequence data. Bioinformatics, 28(12), 1647-1649.
1029 Kliman, R., Sheehy, B., & Schultz, J. (2008). Genetic Drift and Effective Population Size. 1030 Nature Education, 1(3), 3.
1031 Kohout, J., Papousek, I., Sediva, A., Slechta, V. (2012). Effects of stocking on the genetic 1032 structure of brown trout, Salmo trutta, in Central Europe inferred from mitochondrial and 1033 nuclear DNA markers. Fisheries Management and Ecology, 19, 252-263.
1034 Kopelman, N.M., Mayzel, J., Jakobsson, M., Rosenberg, N.A., Mayrose, I. (2015). 1035 CLUMPAK: a program for identifying clustering modes and packaging population structure 1036 inferences across K. Molecular Ecology Resources, 15(5), 1179-1191.
1037 Krijgsman, W., Capella, W., Simon, D., Hilgen, F.J., Kouwenhoven, T.J., Meijer, PTh., … 1038 Flecker, R. (2018). The Gibraltar corridor: Watergate of the Messinian Salinity Crisis. Marine 1039 Biology, 403, 238-246.
1040 Kumar, S, Stecher, G., & Tamura, K. (2016). MEGA7: Molecular Evolutionary Genetics 1041 Analysis version 7.0 for Bigger Datasets. Molecular Biology and Evolution, 33(7), 1870- 1042 1874.
34 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1043 Lanfear, R., Calcott, B., Ho, S.Y.W., & Guindon, S. (20129. PartitionFinder: combined 1044 selection of partitioning schemes and substitution models for phylogenetic analyses. 1045 Molecular Biology and Evolution, 29(6), 1695-1701.
1046 La Sorte, F.A., & Jetz, W. (20109. Projected range contractions of montane biodiversity under 1047 global warming. Proceedings of the Royal Society B: Biological Sciences, 277(1699), 3401- 1048 10.
1049 Leigh, J.W., Bryant, D. 2015. PopART: Full-feature software for haplotype network 1050 construction. Methods in Ecology and Evolution 6(9), 1110–1116.
1051 Li, L., Zheng, B., & Liu, L. (2010). Biomonitoring and Bioindicators used for river 1052 ecosystems: definitions, approaches and tools. Procedia Environmental Sciences, 2, 1510- 1053 1524.
1054 Librado, P., & Rozas, J. (2009). DnaSP v5: A software for comprehensive analysis of DNA 1055 polymorphism data. Bioinformatics 25, 1451-1452.
1056 Luikart, G., Allendorf, F.W., Cornuet, J.M., & Sherwin, W.B. (1998). Distortion of allele 1057 frequency distributions provides a test for recent population bottlenecks. Journal of Heredity, 1058 89(3), 238-247.
1059 Machordom, A., Doadrio, I. (2001). Evidence of a Cenozoic Betic-Kabilian connection based 1060 on freshwater fish phylogeography (Luciobarbus, Cyprinidae). Molecular Phylogenetics and 1061 Evolution. 18, 252-263.
1062 McGowan, P.J.K., Traylor-Holzer, K., & Leus, K. (2017). IUCN Guidelines for Determining 1063 When and How Ex Situ Management Should Be Used in Species Conservation. Conservation 1064 Letters, 10(3), 361-366.
1065 McMeel, O.M., Hoey, E.M., & Ferguson, A. (2001). Partial nucleotide sequences, and routine 1066 typing by polymerase chain reaction-restriction fragment length polymorphis, of the brown 1067 trout (Salmo trutta) lactate deshidrogenase, LDH-C1*90 and *100 alleles. Molecular Ecology, 1068 10(1), 29-34.
1069 Merriam, E.R., Fernandez, R., Petty, J.T., & Zegre N. (2017). Can brook trout survive climate 1070 change in large rivers? If it rains. Science of the Total Environment, 607, 1225-1236.
35 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1071 Michard, A., Frizon de Lamotte, D., Saddiqi, O., & Chalouan, A. (2008). An outline of the 1072 Geology of Morocco. Lecture notes in Earth Sciences, 116.
1073 Molle, F., & Tanouti, O. (2017). Squaring the circle: Agricultural intensification vs. water 1074 conservation in Morocco. Agricultural Water Management, 192, 170-179.
1075 Muhlfed, C.C., Kovach, R.P., Al-Chokhachy, R., Amish, S.J., Kershner, J.L., Leary, R.F., … 1076 Allendorf, F.W. (2017). Legacy introduction and climatic variation explain spatiotemporal 1077 patterns of invasive hybridization in a native trout. Global Change Biology, 23, 4663-4674.
1078 Nachit, H., Ibhi, A., & Vaccaro, C. (2013). The Imilchil meteorite strewn field Isli-Agoudal 1079 craters. International Letters of Chemistry, Physics and Astronomy, 11, 65,71.
1080 Nei, M. (1978). Estimation of average heterozygosity and genetic distance from small number 1081 of individuals. Genetics 89(3), 583-590.
1082 Nei, M., & Chesser, R. (1983). Estimation of fixation indices and gene diversities. Annals of 1083 Human Genetics, 47: 253-259.
1084 Ninua, L., Tarkhnishvili, D., & Gvazava, E. (2018). Phylogeography and taxonomic status of 1085 trout and salmon from the Ponto-Caspian drainages, with inferences on European Brown 1086 Trout evolution and taxonomy. Ecology and Evolution, 8: 2645-2658.
1087 Oosterhout, van C., Weetman, D., & Hutchinson, W. (2006). Estimation and adjustment of 1088 microsatellites null alleles in non-equilibrium populations. Molecular Ecology Notes, 6(1), 1089 255-256.
1090 Pavlova, A., Beheregaray, L.B., Coleman, R., Gilligan, D., Harrisson, K.A., Ingram, B.A., … 1091 Sunnucks, P. (2017). Severe consequences of hábitat fragmentation on genetic diversity of an 1092 endangered Australian freshwater fish: a call for assisted gene flow. Evolutionary 1093 Applications, 10, 531-550.
1094 Pellegrin, J. (1924). Les Salmonides du Maroc. Comptes rendus de l'Académie des Sciences. 1095 Paris 178, 970-972.
1096 Petereit, C., Bekkevold, D., Nickel, S., Dierking, J., Hantke, H., Hahn, A., Puebla, O. (2018). 1097 Population genetic structure after 125 years of stocking in sea trout (Salmo trutta L.). 1098 Conservation Genetics, 19(5): 1123-1136.
36 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1099 Piry, S., Luikart, G., & Cornuet, J-M. (1999). BOTTLENECK: A Computer Program for 1100 Detecting Recent Reductions in the Effective Population Size Using Allele Frequency Data. 1101 The Journal of Heredity, 90, 502-503.
1102 Presa, P., & Guyomard, R. (1996). Conservation of microsatellites in three species of 1103 salmonids. Journal of Fish Biology. 49(6), 1326-1329.
1104 Pritchard, J.K., Stephens, M., & Donnely P. (2000). Inference of populations structure using 1105 multilocus genotype data. Genetics, 155, 945-959.
1106 Prodöhl, P.A., Ferguson, A., Bradley, C.R., Ade, R., Roberts, C., Keay, E.J., … Hynes, R. 1107 (2019). Impacts of acidification on brown trout Salmo trutta populations and the contribution 1108 of stocking to population recovery and genetic diversity. Journal of Fish Biology, 95(3), 719- 1109 742.
1110 Pujolar, J.M., Vincenzi, S., Zane, L., Crivelli, A.J. (2016). Temporal changes in allele 1111 frequencies in a small marble trout Salmo marmoratus population threatened by extreme 1112 flood events. Journal of Fish Biology, 88(3), 1175-1190.
1113 Rambaut, A., Drummond, A.J., Xie, D., Baele, G., & Suchard, M.A. (2018). Posterior 1114 summarisation in Bayesian phylogenetics using Tracer 1.7. Systematic Biology, 67(5), 901- 1115 904.
1116 Ramos-Osins, S.E., & Rozas, J. (2002). Statistical properties of new neutrality tests against 1117 population growth. Molecular Biology and Evolution, 19(12), 2092-2100.
1118 Rice W.R. (1989). Analyzing tables of statistical tests. Evolution, 43, 223-225.
1119 Rogers, A.R., Harpending, H. (1992). Population growth makes waves in the distribution of 1120 pairwise genetic differences. Molecular Biology and Evolution, 9(3), 552-569.
1121 Roman, A., & Ait Hssaine, A. (2014). Causes and Consequences of the November 2014 1122 torrential rains in the center and south Morocco. International Journal of Environment, 5(3), 1123 2019-2854.
1124 Rominguer, J., Gayral, P., Ballenghien, M., Bernard, A., Cahais, V., Chenuil, A., … Galtier, 1125 N. (2014). Comparative population genomics in animals uncovers the determinants of genetic 1126 diversity. Nature, 515, 261-263.
37 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1127 Ronquist, F., Teslenko, M., Van der Mark, P., Ayres, D.L., Darling, A., Höhna, S., … 1128 Huelsenbeck JP. (2012). MrBayes 3.2: Efficient Bayesian phylogenetic inference and model 1129 choice across a large model space. Systematic Biology, 61(3), 539-542.
1130 Rosenberg, N. (2004). DISTRUCT: a program for the graphical display of population 1131 structure. Molecular Ecology Notes, 4(1), 137-138.
1132 Rousset F. (2008). Genepop’007: a complete reimplementation of the Genepop software for 1133 Windows and Linux. Molecular Ecology Resources, 8, 103-106.
1134 Rychlik, W. (2007). OLIGO 7 Primer Analysis Software. In Yuryev, A. (Ed.). Methods in 1135 Molecular Biology, 402: pp. 35-59. PCR Primer Design; Humana Press Inc., Totowa, NJ.
1136 Saint-Pé, K., Blanchet, S., Tissot, L., Poulet, N., Plasseround, O., Loot, G., … Prunier, G.J. 1137 (2018). Genetic admixture between captive-bred and wild individuals affects patterns of 1138 dispersal in a brown trout (Salmo trutta) population. Conservation Genetics, 19, 1269-1279.
1139 Sánchez-Hernández, J., Shaw, S.L., Cobo, F., Allen, M.S. (2016). Influence of a Minimum- 1140 Length Limit Regulation on Wild Brown Trout: an Example of Recruitment and Growth 1141 Overfishing. North American Journal of Fisheries Management, 5, 1024-1035.
1142 Sanz, N. (2017). Phylogeographic history of brown trout: a review. In: Lobón-Cerviá, J. & 1143 Sanz, N. (Eds.). Brown Trout. Biology, Ecology and Management (pp. 15-63). John Wiley & 1144 Sons.
1145 Skaala, Ø., Wennevik, V., & Glover, K.A. (2006). Evidence of temporal genetic change in 1146 wild Atlantic salmon (Salmo salar L.) populations affected by farmed escapees. ICES Journal 1147 of Marine Science, 63, 1224-1233.
1148 Škraba, D., Bećiraj, A., Šarić, I., Ićanović, I., Džaferović, A., Piria, M., … Simonović, P. 1149 (20179. Haplotype Diversity of Brown Trout (Salmo trutta L.) Populations from Una River 1150 Drainage Area in Bosnia and Herzegovina: Implications for Conservation and Fishery 1151 Management. Acta Zoologica Bulgarica, 69(1), 25-30.
1152 Slatkin, M., & Barton, H.H. (19899. A comparison of three indirect methods for estimating 1153 average levels of gene flow. Evolution, 43(7), 1349-1368.
1154 Siler, C.D., Oaks, J.R., Cobb, K., Ota, H., & Brown, R.F. (20149. Critically endangered island 1155 endemic or peripheral population of a widespread species? Conservation genetics of Kikuchi's
38 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1156 gecko and the global challenge of protecting peripheral oceanic island endemic vertebrates. 1157 Diversity and Distributions, 20(7), 756-772.
1158 Snoj, A., Maric, S., Sušnik Bajec S., Berrebi, P., Janjani, S., Schöffmann, J. (2011). 1159 Phylogeographic structure and demographic patterns of Brown trout in North-West Africa. 1160 Molecular Phylogenetics and Evolution, 61(1), 203-211.
1161 Stamatakis, A. (2006). RaxML-VI-HPC: maximum likelihood based phylogenetic analyses 1162 with thousands of taxa and mixed models. Bioinformatics, 22(21), 2688-2690.
1163 Stamatakis, A., Blagojevic, F., Nikolopoulos, D., & Antonopoulos, C. (2007). Exploring new 1164 search algorithms and hardware for phylogenetics: RaxML meets the IBM cell. J. VLSI Signal 1165 Process, 48: 271-286.
1166 Strayer, D.L., & Dudgeon, D. (2010). Freshwater biodiversity conservation: recent progress 1167 and future challenges. Journal of North American Benthological Society, 29, 344-358.
1168 Suárez, J., Bautista, J., Almodóvar, A., & Machordom, A. (2001). Evolution of the 1169 mitochondrial control region in Palaeartic brown trout (Salmo trutta) populations: the 1170 biogeographical role of Iberian Peninsula. Heredity, 87, 198-206.
1171 Sušnik, S., Pustovrh, G., Jesenšek, D., & Snoj, A. (2015). Population genetic SNP analysis of 1172 marble and brown trout in a hybridization zone of the Adriatic watershed in Slovenia. 1173 Biological Conservation, 184, 239-250.
1174 Szucs, M., Melbourne, B.A., Tuff, T., & Hufbauer, R.A. (2017). Genetic and demographic 1175 founder effects have long-term fitness consequences for colonising populations. Ecology 1176 Letters, 20(4), 436-444.
1177 Tajima, F. (1989). Statistical method for testing the neutral mutation hypothesis by DNA 1178 polymorphism. Genetics, 123, 585-595.
1179 Tekken, V., Kropp, J.P. (2015). Sustainable water management-perspectives for tourism 1180 development in north-eastern Morocco. Tourism Management Perspectives, 16, 325-334.
1181 Thorsen, J., Zhu, B., Frengen, E., Osoegawa, K., de Jong, P.J., Koop, B.F., Davidson, W.S., & 1182 Hoyheim. B. (2005). A highly redundant BAC library of Atlantic salmon (Salmo salar): an 1183 important tool for salmon projects. BMC Genomics, 6(1), 50.
39 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1184 Thornton, D.H., Wirsing, A.J., Lopez-Gonzalez, C., Squires, J.R., Fisher, S., Larsen, K.W., … 1185 Murray, D.L. (2017). Asymmetric cross-border protection of peripheral transboundary 1186 species. Conservation Letters, 11, e12430.
1187 Tougard, C., Justy, F., Guinand, B., Douzery, E.J.P., Berrebi, P. (2018). Salmo macrostigma 1188 (Teleostei, Salmonidae): nothing more than a brown trout (S. trutta) lineage?. Journal of Fish 1189 Bioloy, 93(2), 302-310.
1190 Valiquette, E., Perrier, C., Thibault, I., & Bernatchez, L. (2014). Loss of genetic integrity in 1191 wild lake trout populations following stocking: insights from an extensive study of 72 lakes 1192 from Quebec, Canada. Evolutionary Applications, 7, 625-644.
1193 Vera, M., Garcia‐Marin, J.L., Martinez, P., Araguas, R.M., & Bouza, C. (2013). Identification 1194 and conservation of remnant genetic resources of brown trout in relict populations from 1195 Western Mediterranean streams. Hydrobiologia, 707, 29-45.
1196 Vera, M., Martinez, P., & Bouza, C. (2018). Stocking impact, population structure and 1197 conservation of wild brown trout populations in inner Galicia (NW Spain), an unstable 1198 hydrologic region. Aquatic Conservation: Marine and Freshwater Ecosystems, 28(2), 435- 1199 443.
1200 Vucetich, J.A., & Waite, T.A. (2003). Spatial patterns of demography and genetic processes 1201 across the species’ range: Null hypotheses for landscape conservation genetics. Conservation 1202 Genetics, 4(5), 639-645.
1203 Weir, B.S., Cockerham, C. (1984). Estimating F-Statistics for the analysis of population 1204 structure. Evolution, 38(6), 1358-1370.
1205 White, S.L., Miller, W.L., Dowell, S.A., Bartron, M.L., & Wagner, T. (2018). Limited 1206 hatchery introgression into wild brook trout (Salvelinus fontinalis) populations despite 1207 reoccurring stocking. Evolutionary Applications, 11(9), 1567-1581.
1208 Young MK, Isaak DJ., Spaulding S, Thomas CA, Barndt SA, Groce MC, Nagel DE. 2018. 1209 Effects of Climate Change on Cold-Water Fish in the Northern Rockies. In Climate Change 1210 and Rocky Mountain Ecosystems (pp. 37-58). Springer, Cham.
40 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1211 Zardoya, R., & Doadrio, I. (1998). Molecular evidence on the evolutionary and 1212 biogeographical patterns of European cyprinids. Journal of molecular evolution 49 (2), 227- 1213 237.
1214 Závorka, L., Koeck, B., Cucherousset, J., Brijs, J., Näslund, J., Aldvén, D., … Johnsson, J.I. 1215 (2017). Co-existence with non-native brook trout breaks down the integration of phenotypic 1216 traits in brown trout parr. Functional Ecology, 31(8), 1582-1591.
1217 Zkhiri, W., Tramblay. Y., Hanich. L., Berjamy. B. (2017). Regional flood frequency analysis 1218 in the High Atlas mountainous catchments of Morocco. Natural Hazards, 86(2), 953-967.
1219
41 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1220 Table 1. Populations and sampling site locations, including the names of the corresponding 1221 river and lake basins from which samples were collected. The number (N) of individual Salmo 1222 trutta samples included in the mitochondrial (MT-CYB and D-loop) and nuclear microsatellite 1223 DNA analyses are also indicated. GeneBank accesion numbers are provided for the MT-CYB 1224 and D-loop sequences. (Note: awaiting for Genbank accession numbers)
POPULATION LOCALITY N (MT-CYB / D- MT-CYB D-LOOP (NUMBER IN MAP) LOOP / STR) ACCESION ACCESION NUMBERS NUMBERS Moroccan populations
FARDA (1) Laou Basin / 10 / 10 / 15 Mediterranean slope
KANNAR (2) Kannar Basin / 12 / 13 / 16 Mediterranean slope
MOULOUYA (4) Moulouya Basin / 20 / 20 / 20 Mediterranean slope
SIDI HAMZA (5) Ziz Basin / flowing 40 / 23 / 50 to Sahara
SIDI RACHID (17) Sidi Rachid R. 34 / 34 / 32 Sebou Basin/ Atlantic Slope
MIAAMI (3) Oum er Rbia Basin/ 16 / 15 / 14 Atlantic slope
MELLOUL (7) Oum er Rbia Basin/ 30 / 32 / 29 Atlantic slope
LAKHDAR (10) Oum er Rbia Basin/ 24 / 26 / 26 Atlantic slope
TESSAOUT (11) Oum er Rbia Basin / 20 / 20 / 15 Atlantic slope
AMENGOUSS (18) Oum er Rbia Basin / 4 / 4 / 4 Atlantic slope
AIT NACER (19) Oum er Rbia Basin / 3 / 3 / 0 Atlantic slope
OURIKA (13) Tensift Basin / 28 / 26 / 28 Atlantic slope
RHERAYA (14) Tensift Basin / 14 / 14 / 13 Atlantic slope
TIFNOUTE (16) Souss Basin/ 26 / 29 / 29 Atlantic slope
M’GOUN (9) Drâa Basin / 27 / 29 / 29 Atlantic slope
42 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
DADES (8) Drâa Basin / 35 / 34 / 34 Atlantic slope
ISLI (6) Endorheic lake 24 / 23 / 26
IFNI (15) Endorheic lake 17 / 21 / 23
TAMDA (12) Endorheic lake 40 / 43 / 35
Other populations
GARONA Garona Basin / 6 / 9 / 0 Atlantic slope (Iberian Peninsula)
MANDEO Mandeo Basin / 4 / 4 / 0 Cantabrian slope (Iberian Peninsula)
ARANEA Aranea Basin / 2 / 2 / 0 Cantabrian slope (Iberian Peninsula)
DUERO Duero Basin / 6 / 9 / 0 Atlantic slope (Iberian Peninsula)
TURIA Turia Basin / 5 / 9 / 0 Mediterranean slope (Iberian Peninsula)
GUADALQUIVIR Guadalquivir Basin / 21 / 25 / 0 Atlantic slope (Iberian Peninsula)
SICILY Sicily 3 / 3 / 0 LT617555- LT617612- LT617557 LT617614
ADRIATIC 3 / 3 / 0 LT617523, LT617590, LINEAGE LT617528- LT617595- LT617529 LT617596
MARMORATUS 2 / 2 / 0 LT617574- LT617616- LINEAGE (Salmo LT617575 LT617617 marmoratus)
MEDITERRANEAN 2 / 2 / 0 LT617581- LT617623- LINEAGE LT617582 LT617624
DANUBIAN 3 / 3 / 0 LT617546- LT617608- LINEAGE LT617548 LT617610
Salmo macrostigma 3 / 3 / 0 LT617630- LT617630- LT617632 LT617632
Salmo ohridanus 2 / 2 / 0 AY926568 AY926569
Salmo obtusirostris 2/ 2 / 0 JX960841 EF469833
1225
43 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1226 Table 2. Genetic hierarchical population structure (AMOVA) for both mitochondrial markers. 1227 Analyzed basins (genetic pools): Drâa, Farda, Ifni, Isli, Kannar, Moulouya, Oum er Rbia, 1228 Sebou, Tamda, Tensift, Souss and Ziz.
MYT-CB D-LOOP
Structure tested % Variance P-value % Variance statistics P-value statistics
Global structure (one gene pool)
Among basins 84.15 ΦST = 0.840 < 0.0001 82.23 ΦST = 0.822 < 0.0001
Within basins 15.93 17.77
Independent basins (twelve gene pools)
Among basins 51.08 ΦCT =0.511 0.0498 61.10 ΦCT = 0.611 < 0.0001
Among rivers 33.47 ΦSC = 0.684 < 0.0001 21.82 ΦSC = 0.561 < 0.0001 within basins
Within rivers 15.45 ΦST =0.845 < 0.0001 17.08 ΦST = 0.892 < 0.0001
1229
44 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1230 Table 3. Genetic hierarchical population structure (AMOVA) for the nine microsatellite 1231 markers. Analyzed basins (genetic pools): Drâa, Farda, Ifni, Isli, Kannar, Moulouya, Oum er 1232 Rbia, Sebou, Tamda, Tensift, Souss and Ziz.
Structured tests % variance Statistics p-value Global structure (one gene pool)
Among basin 61.95 FST = 0.6195 < 0.0001
Within rivers 38.05
Independent basins (twelve gene pools)
Among basins 46.01 FCT = 0.4601 < 0.0001
Among rivers within 16.73 FSC = 0.3098 < 0.0001 basins
Within rivers 37.27 FST = 0.6273 < 0.0001
1233
45 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1234 Figure legends
1235 Figure 1. Sampling localities of S. trutta in Morocco. 1236 Figure 2 Bayesian phylogenetic tree of Moroccan trout populations based on the analysis of 1237 mtDNA (MT-CYB and D-loop) sequences. Numbers above branches indicate posterior 1238 probability and bootstrap values, respectively. Phylogenetic relationship of Sidi Rachid 1239 population was different depending on Bayesian (A) of ML (B) topology.
1240 Figure 3. Haplotype network based on mitochondrial MT-CYB sequences. Dashes in branches 1241 represent the number of mutational steps.
1242 Figure 4. Haplotype network based on mitochondrial D-loop sequences. Dashes in 1243 branches represent the number of mutational steps.
1244 Figure 5. Divergence time estimations of Moroccan trout populations. Divergence time 1245 estimations and corresponding confidence intervals (HPD 95%) are indicated above 1246 branches. Posterior probability (pp) values are indicated below branches. ** pp=1; * 1247 pp>0.95.
1248 Figure 6. Clustering analyses of microsatellite data using STRUCTURE (A) or TESS (B). 1249 Bar plot of estimated membership of each individual in K=7 and K=17 clusters. In (A), the 1250 number of Moroccan trout populations with the highest posterior probability expressed as the 1251 ΔK is represented after the bar plot.
1252 Figure 7. Factorial correspondence analyses (FCA) of Moroccan trout populations based of 1253 the allelic frequency of nine microsatellites. Each point represents one individual from the 16 1254 sampling sites analyzed. The first two axes resulting from the analysis are shown. A) 1255 considering all populations. B) excluding the genetic groups of Dades-M’Goun, Farda-Kannar 1256 and Ifni-Tifnoute. C) excluding the genetic groups of Ourika and Rheraya.
1257
46 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1258 Figure 1.
1259
1260
47 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1261 Figure 2.
1262
1263
48 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1264 Figure 3.
1265
1266
49 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1267 Figure 4.
1268
1269
50 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1270 Figure 5.
1271
1272
1273
51 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1274 Figure 6.
1275
1276
52 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.06.027219; this version posted April 7, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1277 Figure 7.
1278
1279
53