Is the Genetic Integrity of Wild Coffea Canephora From Ivory Coast Threatened by Hybridization With Introduced Coffee Trees From Central Africa?

Canephora coffee is an important source of income and employment, contributing signicantly to the economies of many developing countries. This species is split into two genetic groups: Guinean and Congolese. Côte-d’Ivoire is the primary diversication center of the Guinean pool, which offers great potential for genetic improvement. However, genetic erosion of this group is currently a threat in this country because of the massive introduction of Congolese accessions in the previous decades. In this work, we analyzed the progenies of isolated population of C. canephora composed of a mix of Guinean and Congolese using SNP markers. The analysis is focus on the male parents’ contribution on open pollination progenies. The result shows that the introduction of cultivars from the Congolese gene pool threat the genetic integrity of wild populations of the Guinean gene pool. We recommend that a conservation program be put in place to ensure the entegrity of wild population of C. canephora through in situ or ex situ conservation. two distant same genetic these two groups different Our study shows differences between the owering peaks of the Guinean and Congolese genetic groups in the years 2018, 2019, 2020 2021. The anthesis period of a species or hybrid is the day when more than 50% of individuals ower We observed a owering lag the Guinean and Congolese genetic groups earliness of the Guineans and a tardiness of Congolese of about 1 day. Our results are consistent lag year.


Introduction
Coffee (Coffea sp) is a perennial plant originated from the tropical and intertropical areas of Africa, Comoros, Mascarene Islands (Mauritius and Reunion Island), Asia and Australia (Davis et al. 2006). Coffee trees are split into 125 species, all belonging to the genus Coffea and the family Rubiaceae (Davis et al. 2011). In all producing countries, the economic and social interest of coffee is no longer in question.
Indeed, the income generated by its cultivation is close to 173,000 million dollars (US) for the entire coffee value chain (ICO 2018). It is a major export product, with about nine million tons of green coffee produced each year (ICO 2018), and its cultivation covers more than 10 million hectares of cultivable land in 80 countries (FAOSTAT 2019). In addition, coffee production and processing employs over 125 million people in the production chain (Van Der Vossen 2016). In Côte d'Ivoire, the coffee cultivation area represents over 400,000 hectares (FAOSTAT 2018). Coffee therefore presents an important economic and social challenge for producing countries.
Among the 125 described species, Coffea canephora is the main species cultivated because of the favorable environment for its cultivation (Coste 1989). This species is diploid (2n = 2x =22), strictly allogamous and self-incompatible (Berthaud 1986;Devreux et al. 1959;Coulibaly 2002). Phylogenetic studies carried out within the species C. canephora, have allowed a structuring into two distinct genetic groups with contrasting morphological traits for architecture, resistance to orange rust, bean size and caffeine content (Berthaud 1986;Leroy et al. 1993 and1997;Montagnon et al. 2008). These are the early owering Guinean pool, originating from West Africa, especially in the mountainous west of Ivory Coast and Guinea as the primary center of diversi cation; and the late owering Congolese pool, originating in Central Africa (essentially the Congo Basin) and Southern Africa. The Congolese pool is subdivided into two subgroups named SG1 and SG2 (Montagnon et al. 1992).
In order to improve the productivity, numerous accessions from SG1 and SG2 were introduced in Côted'Ivoire at the beginning of coffee cultivation (Berthaud 1986;Montagnon et al. 1992b). Thus, due to the contrasting characters of the Guinean and Congolese groups, they were use in a reciprocal recurrent selection breeding scheme 1997;Montagnon 2000). The genetic resources involved in this selection scheme were conserved as ex situ genebanks. Most of these resources are located at the Centre National de Recherche Agronomique (CNRA) station in Divo (Côte d'Ivoire). This gene bank held about 1900 accessions of C. canephora, from various origins (Labouisse et al. 2020).
The geographic range and reproductive system of C canephora, as well as the impact of past climates, have resulted in high genetic variability within the species and the formation of unselected populations obtained from seeds from open pollinations (Gomez et al. 2009;Eskes and Leroy 2012). In addition, other factors such as the lack of pollens and female owers, owering synchronicity, number and density of trees, and their spatial distributions, act to alter the genetic diversity and structure of natural populations Sagnard et al. 2011). In addition, population size and the levels of gene ow in C. canephora, may have long-standing effects on the genetic composition of current gene pools in the wild. These threats are to be consider with the impact of climate change (Bunn et al. 2015), bush res, and heavy deforestation of natural forests in Côte d'Ivoire and Guinea (Brou et al. 2000). Besides, the introduction of the Congolese pool into Côte d'Ivoire has helped foster genetic exchanges with the Guinean group in local plantations . Is the shift between early and late owering a su cient barrier to prevent mass pollination of the wild populations by pollens from the cultivated trees?
The objective of this study is to evaluate the impact of the introduction of the Congolese pool on the genetic integrity of the Guinean pool endemic to Côte d'Ivoire, and mostly distributed in natural areas where coffee growing is important.

Site of study
The study was conducted at the research station of the Centre National de Recherche Agronomique (CNRA) located between 5°46' north latitude and 5°17' west longitude. The relief is characterized by plains with an altitude of between 100 and 200 m. The soils are ferralitic, deep, acidic and desaturated in exchangeable bases (Kassin et al. 2008). Temperatures range from 28 to 32°C, with a daily range of 6 to 7°C. Humidity averages 80% (N'Guessan 2017). The average rainfall is 1400 mm/year.

Plant material
The plant material is composed of an experimental plot named Ci1/11/93 divided into six blocks. This plot is composed of 94 genotypes among which, 34 genotypes of the Guinean group, 44 genotypes of the Congolese group and 16 hybrid genotypes between the Guinean and Congolese groups. All these genotypes were planted in 1993 in plot Ci1/11/93 isolated by an area of primary forest. For the study of owering phenology, only block 1 was considered. The plant material in this block is composed of 62 genotypes including 34 from the Congolese group or 55%, 25 from the Guinean group or 44% and 3

Phenology study
In coffee plants under the climatic conditions of Côte-d'Ivoire, owering is generally triggered by a rainfall of at least 10 mm (Louarn 1992) during the dry season, which extends from December to March in Côte d'Ivoire. The study of owering phenology consisted of noting the time between the rainfall trigger and anthesis on the one hand, and evaluating the owering intensity of the plants on the other, using the following rating scale: 0 : no owering ; 1: very weak owering (at least 20% of the crown covered with owers, about ten); 2: weak (between 20 and 50 % of the crown covered with owers, about twenty); 3: medium (at least 50% of the crown covered with owers). From this score, owering is considered satisfactory (Maalouf et al. 2011); 4: abundant (between 50 and 80% of the crown covered with owers); 5: very abundant (more than 80% of the crown covered with owers).

Paternity study
Discovering SNP markers through data mining SNP data mining was performed using sequence data of 14 C. canephora genotypes which were deposited in the NCBI Sequence Read Archive (SRA) database. These SRA reads were downloaded from the database and mapped on the C. canephora reference genome (Denoeud et al. 2014) using BWA program. The Genome Analysis Toolkit (GATK) package v 3.520 was used for SNP calling using HaplotypeCaller with default parameter.
Then the hard lters (parameters: QD < 2.0 || FS > 60.0 || MQ < 40.0 || MQRankSum < -12.5 || ReadPosRankSum < -8.0) were applied to the lter low-quality alleles.  (Marshall et al. 1998, Kalinowski et al. 2007) was used to assign relatedness with SNPs genotype data. The method used is based on the maximum likelihood approach (Thompson 1976 andMeagher 1986). For each descendant, a measure of the probability that each potential father is the actual father is calculated as a logarithmic score LOD (logarithm of odds). In CERVUS, the male parents of each offspring are determined using the "sex unknown" option. The list of candidate parents is composed of all genotypes in the trial. Paired parent analysis was applied to assign the two parents of each individual. The female parent serves as a control for the applied test. In CERVUS, the proportion of input errors is used as the error rate in the likelihood calculation during the simulation that precedes the parent pair assignment operation. An error rate of 0.2% was used for this analysis. The simulation was done with 10,000 cycles of replicates with discount assuming that 80% of the potential parents were listed. The critical likelihood value (LOD score) was considered with a percentage of 95% in the strict sense and 80% in the broad sense during the simulation, respectively.
The contribution of a male progenitor or group of male progenitors to the pollination of a tree or group of trees in the plot is determined by the percentage of the number of paternal progenies observed for the tree or group of trees. The Mann-Whitney U test was used to compare the contribution values of Guinean and Congolese progeny. This test was performed using the XLSAT 2014 software.

Phenology analysis
Flowering data were analyzed with R v 3.6.2 software. The normality of the data and the homogeneity of the Fisher variances were veri ed by the Shapiro-Wilk test (Shapiro and Wilk, 1965). Thus, non-parametric one-criteria classi cation tests were performed to compare the genetic groups and genotypes of canephora studied. For the twenty-three genotypes, the Kruskal-Wallis test was used and the Wilcoxon-Mann-Whitney test for the 2 genetic groups. These tests allowed multiple comparison of means at the 5% threshold when signi cant differences were observed.

Pollen dispersal in the plot
The distances between pollinated trees and pollinator trees are calculated by the Pythagore theorem. The relationship between the pollination distance and the importance of the contribution to pollination is estimated by linear regression.

Spreading of the owering
Flowering onset was synchronous between the Guinean and Congolese genetic groups over the four years of observation. However, analysis of the owering spread curves shows a shift between the anthesis peaks of each group. Thus : In 2018, the peak of anthesis was observed on the 6th day after which 54.43% of the trees owered in the Guineans. The peak for the Congolese was observed on day 7, with 81.96% of the trees owering. From these two peaks, a continuous in ection of owering in both genetic groups was observed until day 8 (Fig. 1a).
In 2019, 97.68% of the trees in the Guinean group owered on day 6, showing their peak anthesis. The peak anthesis of the Congolese trees was observed one day later with 51.66% of the trees owering. From these two peaks, a progressive decrease in owering of both genetic groups was observed until day 8 (Fig. 1b).
Observations of the February 2020 owering indicate that the peak of anthesis of the Guinean was reached on day 6 with 51.41%. One day later, the peak of the Congolese group was reached with 59.32% of trees in bloom. A continuous in ection of owering in both groups was observed on day 8 (Fig. 1c).
In January 2021, the peak of anthesis of Guinean trees was observed on day 6 with 46.62% of trees in ower. The Congolese peak was observed 2 days later with 91.66% of trees in bloom. From these two peaks, a gradual decrease in owering of the two genetic groups was observed until day 9 (Fig. 1d).

Average owering time in Guinean and Congolese coffee trees
The results of the average owering time recorded over these four years showed that there was a signi cant difference between the owering of the Congolese and Guinean coffee groups after the onset of the rst rainfall following the long dry period (Table 2). Thus, the Guinean genotypes owered earlier on the 6th day after the rst rainfall following the long dry season, while the Congolese cultivars owered one or two days later on the 7th or 8th day in 2018, 2019 and 2021. However, observations in 2020 showed synchronized owering of the two genetic groups on day 7.

Paternity of parents Paternal contributions in the descendants
After open pollination, progeny analysis revealed that Congolese sires were the major contributors to progeny with 48%. Guinean progenitors contributed 37% of the progeny and hybrids contributed 15% (Fig. 2). In the free-living progeny, there were 501 plants of Guinean progenitors, 656 plants of Congolese progenitors and 202 plants of inter-group hybrid progenitors. Thus, there was a very signi cant increase (p<0.001) in the proportion of intergroup hybrids in the progeny, from 1-15%.

Contribution of Guinean, Congolese and hybrid male genotypes in the progeny of Guinean and Congolese feet
The sum of the contributions of the Guinean spawners to the Guinean broodstock set was 52% compared to 40% for the sum of the Congolese spawners (Fig. 3a). The Mann-Whitney test (p = 0.053) showed that there was no signi cant difference between the two proportions.
Among all Congolese mother plants, the sum of the contributions of Congolese broodstock was 58% compared to 18% for the sum of the contributions of Guinean mother plants (Fig. 3b). The Mann-Whitney test (p = 0.002) applied to the contributions of the Guinean and Congolese broodstock showed a signi cant difference.
Hybrids contributed signi cantly to the pollination of Congolese mother trees (24%) and less so to Guinean mother trees (8%). Thus, the hybrids appear to be more compatible with the Congolese than with the Guineans.
Average individual contribution of Congolese, Guinean and hybrid progenitors to the progeny of Congolese and Guinean mother plants The average individual contribution of the progenitors to the progeny of the genetic groups is shown in Fig. 4, ranging from 0 to 5.1% in the Guinean mother trees and from 0 to 4.5% in the Congolese mother trees. A rst group of trees with a low contribution between 0 and 1% can be seen. This group represents 43.52% of the trees that made a low contribution to the pollination of female trees. An identical proportion of 43.52% shows a contribution between 1 and 3%. Finally, 12.94% of the trees are 'superpollinators' that show a contribution higher than 3% and between 3 and 6%.
Pollen dispersal in the plot The pollen dispersal results show that the pollen spread distance of the trees is between 2 and 94.13 m ( Table 3). The maximum dispersal distance of allopollen, which is 94.13 in block1 of the plot, is comparable to the diagonal of block1 of the plot which is 101 m. Thus, the size of the plot could not be a limiting factor for gene exchange between trees. It also allowed us to understand that tree pollination occurs in several directions. No direction, including wind direction, seems to affect pollination (Fig. 5). At the same time, a very weak positive linear regression was found between the contribution of male spawners and the distance between mother plants (Fig. 6).

Discussion
Gene ow between genetic groups of the same species plays a major role in the spatial organization of genetic diversity and is a major evolutionary force (Futuyama 1998). In plants, gene ow results from two major mechanisms, pollination and seed dispersal.
This study contributes to the analysis of gene ow between two genetically distant populations of the same species (C. canephora) (Guinean and Congolese) that co-exist in the same plot. In the case of the two genetic groups studied here, this is particularly relevant because these two groups are known to ower at different dates (Anthony 1992;Louarn 1992 andAkaffou 2013). Our study shows differences between the owering peaks of the Guinean and Congolese genetic groups in the years 2018, 2019, 2020 and 2021. The anthesis period of a species or hybrid is the day when more than 50% of individuals ower (Maalouf et al. 2011). We observed a owering lag between the Guinean and Congolese genetic groups with an earliness of the Guineans and a tardiness of the Congolese of about 1 day. Our results are consistent with those of Berthaud (1986) and Leroy et al. (1993). However, this time lag seems to vary according to the year. In 2020, this time lag does not exist.
The genetic mixing study was conducted on 1361 offspring during 2018, or the average lag between the two groups was greater than ½ day. Despite this owering lag, genetic mixing was observed between the two genetic groups. Our results revealed an increase in the number of hybrids (from 1% in the parental population to 15% in the progeny). This increase is accompanied by a 7% reduction in the frequency of Congolese (55 to 48%) and Guineans (44 to 37%). The signi cant increase in the frequency of hybrids is evidence of inter-group mixing. This inter-group mixing seems to depend mainly on a small number of progenitors (about 12.94%) that contributed to more than 28% of the pollinations. We have veri ed that these "super-spawners" interbreed more with Guineans (19%) than with Congolese (9%) and this even when they are Congolese. This imbalance in the contribution of spawners does not seem to be related to their position in the plot. It does not seem to be clearly related to the number of plants either. For example, the Guinean "super-spawner" 789 is represented only once in the plot.
Assuming that the integrity of the concomitant genetic groups would be maintained, the pollen involved in the fertilization of both groups would have to come exclusively (100%) from the same group. We show in this study that this is not the case. The anthesis mismatch is not a su cient barrier for crosspollination. Paternity analyses showed that the contribution of Guinean and Congolese progenitors in the progeny of Guinean mother trees are equivalent. However, in the progeny of the Congolese mother trees, the contribution of the Congolese gene pool is much greater than that of the Guinean pool (Fig. 3). These differences can be explained by several factors including the founder effect, genetic erosion in the Guinean gene pool (endemic to Ivory Coast) and the type of data generated by the different markers used in this study. On the other hand, preferential fathers were also found for the mother coffee plants of the different genetic groups. This would be the consequence of gametophyte incompatibilities between certain genotypes (Berthaud 1986;Coulibaly 2002), discrepancies between owering periods (Berthaud 1986;Leroy et al. 1993) and the viability of the pollen produced by each progenitor. Furthermore, the introduction of genotypes from the Congolese pool into coffee populations in Ivory Coast most likely represents an evolutionary force. This is supported by the increase of (1-15%) hybrids in the progeny population. This increase is facilitated by gene ow through pollen (Berthaud 1984). Gene ow from these domesticated crops to wild Guinean species represents a threat to the genetic integrity of the Guinean pool. Our results therefore con rm the preliminary ndings of Montagnon et al. 1993. Indeed, repeated occurrences of hybridization can lead to the loss of genetic integrity of the wild species, which becomes assimilated to the cultivar (Aerts et al. 2013). Therefore, to conserve the integrity and genetic diversity of the Guinean pool, in situ and ex situ conservation strategies are essential.

Conclusion And Recommendations
The success of Ivorian coffee growing is largely due to the introduction of the Congolese pool. This pool naturally hybridized with endemic Guinean coffee trees in Ivory Coast to form fertile and sometimes very productive intergroup hybrids that were empirically selected by breeders. However, in addition to deforestation, bush res and climate change, the massive introduction of genes from the Congolese could have consequences for the maintenance of the genetic integrity of wild coffee trees representing the endemic Guinean group in Ivory Coast. Indeed, in our system, owering time differences between the two groups do not constitute a su cient barrier to prevent massive hybridization. It is important to verify if this contamination is already present in the wild coffee trees of Ivory Coast. It is also necessary to verify that seeds could not have been dispersed by wildlife in the forests hosting the Guinean genetic resources. And nally, a comprehensive assessment of existing germplasm should be undertaken, based on which a comprehensive conservation strategy should be developed that includes both in situ and ex situ conservation and addresses the need for duplication of germplasm in multiple locations.

Declarations
Data archiving statement Data presented in this study can be found in online repositories. The name of repository and accessions number(s) can be found below https://tropgenedb.cirad.fr/tropgene/JSP/interface.jsp?module=COFFEE

Ethical Responsibilities of Authors
We declare that this manuscript is original, has not been published before and is not currently being considered for publication elsewhere in any form or language (partially or in full). We ensure that there is no con icts of interest associated with this publication, and there has been no signi cant nancial support for this work that could have in uenced its outcome.
All the authors are contributed as follow:  Contribution of Congolese, Guinean and hybrid progenitors to the progeny of Congolese parent plants