R E S E A R C H Open Access © The Author(s) 2025. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit ​h​t​t​p​​:​/​/​​c​r​e​a​​t​i​​ v​e​c​​o​m​m​​o​n​s​.​​o​r​​g​/​l​​i​c​e​​n​s​e​s​​/​b​​y​-​n​c​-​n​d​/​4​.​0​/. Kamdem et al. BMC Ecology and Evolution (2025) 25:20 https://doi.org/10.1186/s12862-025-02356-0 BMC Ecology and Evolution *Correspondence: Narcisse Guy Kamdem guynarcissekamdem@yahoo.fr 1Evolutionary Biology and Ecology Unit, CP 160/12, Faculté des Sciences, Université Libre de Bruxelles, Av. F. D. Roosevelt 50, Brussels B-1050, Belgium 2Plant Systematics and Ecology Laboratory, Department of Biology Higher Teachers’ Training College, University of Yaoundé I, P.O. Box 047, Yaoundé, Cameroon 3Center for Tropical Research, Institute of the Environment and Sustainability, University of California, Los Angeles, USA 4International Institute of Tropical Agriculture (IITA), Yaoundé, Cameroon Abstract Background  Mammal-dispersed tropical trees can face regeneration problems due to increasing hunting pressure. We studied the case of Coula edulis Baill. (Coulaceae), an African rainforest tree that produces the ‘African walnut’, an essential food and income resource for rural communities. We compared gene flow and regeneration dynamics in three populations with contrasting levels of human disturbance and mammal abundance. Using 21 nuclear microsatellite markers, we estimated the outcrossing rate and contemporary seed and pollen dispersal distances, and we analyzed the fine-scale spatial genetic structure (FSGS) to infer historical gene dispersal distances. Results  Juveniles were outcrossed while 30% of the seeds from one population were selfed, suggesting the elimination of inbred seeds. The mean dispersal distances were relatively short for seeds (105–219 m) and pollen (173–358 m), both shorter in the most intact forest. Immigration rates were three to four times higher for pollen (33–71%) than for seeds (7–28%), indicating some long-distance pollen dispersal. FSGS was strong in all populations (Sp = 0.023–0.036), suggesting short-range historical gene dispersal distances consistent with contemporary estimates. We detected assortative mating, possibly due to higher flowering synchronicity between related individuals. The most disturbed plots displayed inverted J-shaped trunk diameter structures, typical of continuous regeneration, while the intact forest had diameter structure indicating more limited regeneration. Conclusions  Our results suggest that forest disturbance and mammal hunting do not significantly affect the dispersal distances of seed and pollen for Coula edulis, contrary to other mammals-dispersed trees. We hypothesize that the main dispersers are scatter hoarding rodents that are less impacted, or even facilitated, by hunting pressure. The species appears to regenerate better in disturbed forests, possibly due to a reduction in seed and seedling predators. However, natural populations are threatened by ongoing forest conversion into agriculture. Keywords  African tropical rainforest tree, Gene dispersal, Mating system, Kinship analysis, Regeneration dynamics, Coula edulis Short-distance seed and pollen dispersal in both hunted and intact forests in the lower canopy African rainforest tree, Coula edulis Baill. (Coulaceae) Narcisse Guy Kamdem1,2*, Bonaventure Sonké2, Saskia Sergeant1, Vincent Deblauwe3,4 and Olivier J. Hardy1 http://creativecommons.org/licenses/by-nc-nd/4.0/ http://creativecommons.org/licenses/by-nc-nd/4.0/ http://crossmark.crossref.org/dialog/?doi=10.1186/s12862-025-02356-0&domain=pdf&date_stamp=2025-3-12 Page 2 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 Background Seed and pollen dispersal play a key role in the ecologi- cal and evolutionary dynamics of tree populations and communities [1–6]. It affects reproductive success, a fundamental trait for long-term population viability, and a key feature for species conservation [7]. It is also a determinant of the levels of genetic variation and spatial genetic structure within and between populations [8, 9]. It can also neutralize the potentially deleterious effects of genetic drift and be a source of new alleles within popula- tions [10]. However, dispersal is a highly stochastic pro- cess, determined by the abundance and behavior of seed and pollen dispersal vectors, which can vary between years and populations [11, 12]. In tropical forest ecosystems, wildlife plays a key role in both pollination and seed dispersal [13]. Hence, gene flow in trees depends on this fauna, which is affected by hunting, habitat destruction, and forest fragmentation. Disruption of plant interactions with dispersers or polli- nators can affect genetic variation within species [5, 14, 15], resulting in more structured and less cohesive gene pools, and increased isolation-by-distance over larger areas [16–19]. Although the consequences of human pressures on the gene flow dynamics of tropical forest trees vary across species and contexts [20, 21], they have received a lot of attention in recent years [16–18, 22–27], but they are still very little documented in African forest trees [5, 28–33]. Characterizing gene dispersal requires molecular markers [9, 34–37] such as microsatellites. Gene disper- sal can be characterized by direct approaches identifying parent-offspring pairs through parentage analyses [38], which estimate contemporary seed and pollen dispersal. However, this requires exhaustive sampling of the paren- tal population in the study area [34, 39]. Alternatively, gene dispersal can be evaluated by indirect approaches, for example from the amplitude of fine-scale spatial genetic structure (FSGS) expected under limited seed and/or pollen dispersal in space, providing estimates of historical gene dispersal distance without distinguish- ing the respective roles of seed and pollen dispersal [9, 35, 40, 41]. However, to date, gene dispersal analysis of African rainforest trees has been carried out mainly in canopy dominant and light-demanding species [5, 28, 29, 32, 42] while many tree species are shade tolerant and remain below the canopy, characteristics that can affect seed and pollen dispersal efficiency. In this work, the objective is to characterize gene dis- persal and the FSGS of Coula edulis Baill.(Coulaceae) in Cameroon, comparing three populations under con- trasting anthropogenic pressures. Coula edulis is a her- maphroditic lower canopy tree endemic to the tropical rainforests of Africa. In Cameroon, its natural range is restricted to Atlantic forests [43], although some populations are found in semi-deciduous forests [44]. Its fruits are an important source of food and income for rural populations. The species is most likely shade tolerant [45]. Seedlings occur at low density in natu- ral forests despite abundant fruiting [46, 47]. This could be explained by low germination rates 10–20% [45, 48] and/or predation of seeds and freshly germinated seeds by large and small mammals, including humans [45, 49]. The breeding system and the relative importance of veg- etative propagation in C. edulis are not documented, and we ignore to which extent the natural regeneration of the species and its gene dispersal potential are being affected by human disturbances. It is therefore essen- tial to understand the extent of contemporary and his- torical gene dispersal in different environments. More specifically, by comparing populations undergoing dif- ferent levels of human disturbances, the objectives of the present study were to: (1) compare genetic diversity and inbreeding parameters among cohorts and sites, (2) identify clones to assess vegetative reproduction and characterize fine-scale spatial genetic structure (FSGS) to infer historical gene dispersal distances (σg), (3) char- acterize contemporary seed and pollen dispersal to compare it with estimates of historical gene dispersal, and to assess the impact of disturbance, (4) test for the presence of inbreeding due to selfing and / or assortative mating between adult trees, (5) test whether male and female reproductive success of individual trees is related to trunk diameter, (6) compare and deduce the impact of human disturbances on the dynamics of regeneration between the different populations. Methods Species description Coula edulis Baill., commonly known as ‘African walnut’, is an endemic tree species of the tropical rainforests of the Guineo-Congolian region, mostly found in evergreen forests with occasional occurrences in semideciduous forests, and ranging from West Africa (Sierra Leone) to western Central Africa (southwestern DRC) [44, 50]. It produces edible fruits consumed by rural populations, who sell the kernels on the roadside or in markets in large cities [51, 52]. The wood of Coula edulis is heavy and vir- tually rotproof, with an average density of 1.01 at 12% moisture content [53], and used to make charcoal and as construction material for huts [54]. Its bark is used in traditional medicine as a decoction for the treatment of several diseases [55–57]. It belongs to the Coulaceae fam- ily, a very small pantropical family formerly included in the Olacaceae family, with three monotypic genera dis- tributed on different continents [58]. It is a medium-sized tree, reaching 25 m in height, typically found in the lower forest level and understory of mature forests [52]. The descriptions of the species [44, 59, 60] indicate that the Page 3 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 flowers are hermaphroditic, yellowish-white or slightly reddish, and pollen dispersal is probably ensured by small insects, given the morphology of the flower (personal observation). The species is also known for producing suckers [61]. The fruits are globular or ellipsoid, 3.5 to 5 cm long, yellowish green when ripe with a thin meso- carp (see the photo of the ripe fruit in the related files section). The rounded kernel is made of a hard brown endocarp and the seed is single and spherical, 1.5–2.5 cm in diameter. Coula edulis is a typical climax species according to the definition of Whitemore [62]. The mini- mum diameter at breast height (DBH, i.e. stem diameter at 130  cm above ground) of flowering trees is 10.6  cm, and trees with a DBH above 23  cm fruit regularly [63]. The average annual DBH increment is 0.22  cm/year [63]. In Cameroon, flowering generally occurs between the end of the dry season and the beginning of the short rainy season (February to April), while fruiting occurs during the peak rainy season from July to October [52]. In its natural environment, camera traps revealed that seeds are an important food source for many animal spe- cies such as bush pig (Potamochoerus porcus) and forest elephant (Loxodonta cyclotis), which act as predators, while emin’s rat (Cricetomys emini) and African brush- tailed porcupine (Atherurus africanus) act both as preda- tors and dispersers [49]. Study sites and sampling The study was carried out at three sites in Cameroon (Fig. 1) where we established plots to sample exhaustively adult and juvenile C. edulis trees. (1) The Campo Ma’an National Park (CMNP) site was chosen because it is a protected area. Although the site faces some poaching activities from the population of the surrounding villages, we still find an abundance of large mammals such as for- est elephants (Loxodonta cyclotis), chimpanzees (Pan troglodytes), gorillas (Gorilla gorilla gorilla), and man- drills (Mandrillus sphinx) [64]. (2) The Mbalmayo Forest Reserve (MFR) and (3) Fifinda sites were chosen because they are located in areas subject to strong anthropogenic activities such as illegal logging, the establishment of cocoa and oil palm plantations, and poaching with the following consequences: (i) a dramatic decrease in fauna, especially large mammal and monkey populations; (ii) Fig. 1  Localization of the three Coula edulis populations in Cameroon (top left; the gray shaded area represents the potential rainforest area) and sam- pling scheme in each population. Top right: sampling scheme in Fifinda, with exhaustive sampling in a 18-ha plot delineated by a rectangle. Bottom left: sampling scheme in CMNP, with exhaustive sampling in a 400-ha plot. Bottom right: sampling scheme in MFR, with exhaustive sampling of a 400-ha plot Page 4 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 an increase in the collection of non-timber forest prod- ucts (NTFPs) such as Coula edulis, Irvingia gabonen- sis, Ricinodendron heudelotii, Garcinia kola [65]; (iii) a fragmentation of the forest cover, particularly in the Fifinda site where forest fragments are relatively small. At the CMNP site, we established a 400  ha (2 × 2  km) plot (2.31856°N, 10.17641°E) in a continuous terra firma for- est pocket where C. edulis was widely distributed (Fig. 1). At the MFR site, we established a 400 ha plot (3.43584°N, 11.44020°E) surrounded by the Nyong river on its west- ern, southern and eastern sides (Fig. 1). This plot covered a very heterogeneous area and most C. edulis trees found were located in the northeastern corner of the plot, over an area of approximately 125  ha (Fig.  1). At the Fifinda site, the forest cover was highly fragmented due to the expansion of agriculture, and we established a 18  ha plot (300 × 600  m) (3.19646°N, 9.98077°E) in one forest fragment, where the species tended to occur in several patches (Fig. 1). This plot was surrounded by swamps to the northwest, the Loukoundjé River to the south, and a forest area to the east, part of which had been converted into an agricultural plantation. We also observed the stumbs of three large trees (trunk diameter > 65 cm) that had been cut in the plot to harvest their fruits. In each plot of each site, we opened trails every 100 m, allowing to sample all C. edulis individuals (adults and juveniles). Seeds, however, were collected only at the MFR site. Sampling consisted of taking a piece of leaf or cambium on each individual and dried in silica gel. The diameter at breast height (DBH) was measured 130 cm above ground or above the buttresses, if any. The diameters of stems smaller than 130  cm in height were measured 10 cm above ground. In the rare case of mul- tiple stems at the height of measurement, the largest was selected. The geographical coordinates of each individual have been recorded using a handheld GARMIN GPS 64s and 66  sr in each site with an accuracy of less than 5 m. The individuals were classified as juveniles or adults according to diameter at breast height. Adult trees are defined as all individuals capable of sexual reproduction, which is above 10.6 cm in DBH [63]. Juveniles are defined as individuals with DBH < 10.6  cm or height < 130  cm. Inside the CMNP 400  ha plot, the surveys took place from January to February 2022, collecting 156 juveniles and 646 adults. On the MFR 400 ha plot, inventories were conducted from February to April 2021, during which time we collected 171 juveniles, 220 adults, and 104 seeds corresponding to seven families collected on the ground. We also collected 6 juveniles and 21 adults outside the plot, which will be added to those in the plot as part of the analysis of the FSGS and genetic diversity parameters. Similarly, in June 2021, we collected 72 juveniles and 53 adults in the 18-ha Fifinda plot, and 20 adults were also collected outside the plot. DNA extraction and genotyping DNA was extracted from 25 mg of leaf or 35 mg of cam- bium dried with silicagel, or 25  mg of seed cotyledon using the NucleoSpin 96 Plant Kit (Macherey-Nagel), according to the manufacturer’s instructions. We geno- typed 1469 samples consisting of 960 adult trees, 405 juveniles, and 104 seeds with 21 nuclear microsatel- lites markers, following the protocol developed by [66]. For each sample, 1.3 µL of the PCR product was added directly to 12 µL Hi-Di Formamide (Life Technologies, Carlsbad, California, USA) and 0.3 µL MapMarker® 400 labelled with DY-632 (Eurogentec, Seraing, Belgium) and genotyped on an ABI3730 sequencer (Applied Bio- systems, Lennik, The Netherlands). Genotypes were analyzed using Geneious version 7.1.9. Only samples for which at least seven out of 21 loci were successfully amplified were used for subsequent analyses. The final number of samples used for subsequent analyses was 1457 after eight individuals (four adults and four juveniles at MFR site) with missing data were removed (that is, 0.55% of the samples). Pairwise relationship coefficients [67] were calculated using SPAGeDi v.1–5 [68] to check for the presence of duplicated individuals or clones, as the species is known for producing suckers. Four clones were identified, including two juveniles at CMNP site and another two at MFR site, which had the same genotype as the respective adult tree next to each of them. These samples were interpreted as suckers and were removed for further data analyses. Characterization of genetic diversity and inbreeding We used SPAGeDi v.1–5 [68] to estimate the following genetic parameters for each locus, cohort (adults, juve- niles, and seeds), and population: (i) number of effec- tive alleles (NAE), (ii) allelic richness expressed as the expected number of alleles among k gene copies (AR(k)), (iii) expected heterozygosity (HE), (iv) observed heterozy- gosity (HO), and (v) inbreeding coefficient (FIS). We also used INEst 1.0 [69] to estimate the corrected inbreed- ing coefficient (FIsc), i.e., considering null alleles, for each cohort and population. Analysis of variance in R [70] allowed us to test for significant differences in these parameters of genetic diversity between cohorts and populations. Estimations of Ho and FIS were compared between seeds, juveniles and adults to check whether there was an increase in observed heterozygosity (Ho) or a decrease in heterozygosity deficiency (FIS) with age, a sign of inbreeding depression when inbred individuals are less likely to survive. We also used a method based on identity disequilibrium [71] implemented in SPAGeDi v.1–5 [68] to estimate the selfing rate (S) within each cohort in each site. Page 5 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 Characterization of historical gene dispersal through fine- scale spatial genetic structure (FSGS) At the population level, FSGS was assessed by the rela- tionship between genetic relatedness and spatial distance (kinship-distance curve) in each population. To do this, we used the genotypes of individuals (adults and juve- niles) to estimate the kinship coefficients (Fij) between individuals using the estimator of J. Nason [72] imple- mented in SPAGeDi v.1–5 [68] because of its robust statistical properties [9]. These Fij are then regressed on the logarithm of the distance between individuals (dij), resulting in a regression slope (bLD) [73]. To obtain a graphical representation of the decrease in kinship with spatial distance, means of Fij per spatial distance interval between individuals were also calculated for eight inter- vals (in meters): 0 to 10, 10 to 20, 20 to 40, 40 to 80, 80 to 160, 160 to 320, and 320 to 640 and 640 to 1000. FSGS was assessed in each population, but also at the cohort level, and then tested by randomly swapping the posi- tions of individuals (10,000 randomizations). The statis- tic Sp =-bLD/(1 – F1), which characterizes the strength of FSGS, was obtained for each population and cohort from the observed regression slope (bLD) of Fij over the logarithmic distance dij and the mean kinship coefficient measured in the first distance class (F1) [9]. Assuming drift-dispersal equilibrium, we estimated the historical backward gene dispersal distance (σg) for each population using the method described in [41], based on the kinship-distance curve. We estimated the size of the Wright neighbourhood, defined as Nb = 4π DE.σg 2 where DE represents the effective population density and σg 2 is half the mean squared distance between parents and offspring, using the relationship Nb = (F1-1)/bLD where the regression slope bLD is calculated in a restricted dis- tance interval σg > dij > 20 σg. The dispersal distance of the genes was estimated using SPAGeDi v.1–5 [68] assum- ing a range of effective population density (DE). To this end, DE was estimated knowing the mean population densities (D). These densities (D) were obtained from the inventory data of individuals in the three populations by considering only individuals with DBH ≥ 10.6 cm and dividing their sum by the area of each site. In the case of MFR, since the species was present in a corner of an area of 125 ha of the 400 ha plot, we decided to use this area of 125 ha to calculate the density. We have D = 1.62, 1.76 and 2.94 ind ha-1 in the CMNP, MFR and Fifinda, respec- tively. Assuming that the ratio of effective population sizes to census sizes (Ne / N) generally ranges from 0.1 to 0.5 in plant populations [74], we used three estimates of effective population densities (DE): DE = D/2, D/4, and D/10. These values corresponded to DE = 0.81, 0.41 and 0.16 ind ha-1 for CMNP, DE = 0.88, 0.44 and 0.18 ind ha-1 for MFR, and DE = 1.47, 0.74 and 0.29 ind ha-1 for Fifinda. Characterization of seed and pollen dispersal through parentage analysis and the neighbourhood model The neighbourhood model implemented in NMπ soft- ware using the maximum likelihood approach [36, 75] allowed us to model seed and pollen dispersal kernels, estimate the selfing rate, and infer the impact of DBH on reproductive success. The model was fitted using the spatial locations of the samples, their genotypes, and the standardized DBH values (i.e. after subtracting the mean and dividing by the standard deviation of DBH) of the adult trees. First, an analysis was performed for each population with juveniles and parents (individuals with DBH ≥ 10.6  cm), which contributed to the identification of the most probable mothers and fathers of juveniles with a genealogical probability ≥ 0.8. This confirmed some observations made in the field, where we found fruit remains under some individuals with a DBH < 12 cm. An additional NMπ analysis was performed between all mature trees in the MFR plot and seeds (n = 104). This allowed us to confirm or reject the identity of the most probable mother with a genealogy probability ≥ 0.8. For seeds for which no mother was identified among available adults, we estimated the kinship coefficients (Fij) between them using SPAGeDi v.1–5 and, by reordering the result- ing kinship matrix, we were able to group these seeds into families and manually reconstruct a likely maternal genotype. A third NMπ analysis was then performed for the MFR population, including four reconstructed mater- nal genotypes as potential adults. NMπ analyses between adults and juveniles allowed characterising parameters such as: seed and pollen immi- gration rates (ms/mp), self-pollination rate (s), seed and pollen mean dispersal distance (ds/dp). The immigra- tion rate was estimated by assessing the contribution of parents outside the sampling area (proportion of pollen/ seeds originating from unsampled adults). The dispersal distance parameters are those of the fitted dispersal ker- nels, which describe the probability that an emitted seed or pollen will disperse from a starting position to a final position. The modelled kernels assumed a bidimensional power-exponential distribution coupled with von Misses distribution to account for anisotropy and are character- ized by four parameters: the mean dispersal distance (ds or dp), the shape parameter (bs or bp, equals to 2 for a gaussian, 1 for an exponential, or < 1 for a fat-tailed dis- tribution), a degree of anisotropy (ks or kp, equal to zero under isotropic distribution), and a direction of pre- vailing dispersal (as or ap) [35, 75, 76]. To determine whether the estimated seed and pollen dispersal kernel and the degree of sampling completeness could predict pollen and seed immigration rates, we used an R script described in [5] to simulate the contribution of unsam- pled trees to reproduction. For this aspect, we only used the CMNP population because it was in a continuous Page 6 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 forest where C. edulis was well distributed outside the sampling plot (pers. obs.), while in the MFR and Fifinda sites, the distribution of C. edulis around the sampling plots was very discontinuous (pers. obs.). For parents of offspring detected with a genealogical probability ≥ 0.8, the distribution of their diameter was compared with that of all adult individuals in the plot. This allowed us to see which of the tree diameter classes contributed the most to pollination and established juve- niles. Seed and pollen dispersal kernels were illustrated by showing the position of juveniles with respect to their mother (seed dispersal events) or of mother with respect to the father (pollen dispersal events) on two-dimen- sional maps. Comparison of historical and contemporary gene dispersal estimates To compare contemporary and historical gene dispersal estimates, we need to convert the respective estimates obtained by direct and indirect methods, because con- temporary estimates through NMπ describe pollen and seed dispersal under a power-exponential kernel (param- eters dp, ds, bp, and bs), while historical estimates are expressed in terms of the mean squared parent-offspring distance (σg 2). To convert the d and b parameters into σ, we used the function (1) derived from [35]. σ2 = 0.5 d2 Γ (2/b) Γ (4/b) Γ(3/b)−2� (1) where σ² is half of the mean squared parent–offspring distance; d = dp or ds: mean pollen or seed dispersal dis- tance; b = bp or bs: shape of the pollen or seed dispersal kernel; Г: gamma function. This allowed us to obtain σp 2 and σs 2, which represent the extent of pollen and seed dispersal distances, respec- tively. Equation  (2) then allowed one to estimate the contemporary gene dispersal distance (σg) that can be compared with the corresponding estimates obtained by the method for estimating historical gene dispersal σ2 g = σ2 s + 0.5 σ2 p � (2) Biparental inbreeding and assortative mating When gene flow is limited, biparental inbreeding (mat- ing between relatives) can occur [77, 78], and can be fur- ther enhanced by assortative mating (preferential mating between relatives), for example, when flowering phenol- ogy is heritable. We tested for assortative mating in the MFR population where the parentage analysis identified the two parents of 115 outcrossed offspring (seedlings or seeds), corresponding to 83 unique mating pairs. Using the methodology highlighted by [39], we compared for each unique mating pair the kinship coefficient (Fij) between mates with the one expected based on their spatial distance and the kinship distance curve (i.e., the mean Fij between adults in the same distance interval), using a Student’s t test. If the mean Fij between mating pairs is significantly higher than expected, assortative mating would be inferred, while if it is significantly lower than expected, it could result from inbreeding depression if mating between relatives tends to produce offspring with lower survival rate, or from a mechanism avoiding mating between relatives. Regeneration dynamics The age distribution of natural tree populations provides insights on their regeneration dynamics [79]. Assuming that the DBH distribution is a reasonable proxy of the age distribution in C. edulis, we assessed the diameter distri- bution of individuals (number of stems per 10  cm wide DBH class) to compare the dynamics of regeneration in the different populations. Continuous regeneration through time is evidenced when the number of young individuals is high enough to ensure the renewal of the population, typically leading to a decreasing number of stems with increasing DBH class (“inverted J” distribu- tion) [80, 81]. A deficit of regeneration in recent time appears when there are fewer individuals in the small- diameter classes, leading to a “bell” distribution [82, 83]. Multimodal distribution can reveal pulses of regenera- tion, for example following disturbance events [79]. At the level of each population, we also inspected the distri- bution of cumulative numbers of juveniles according to their diameter to determine which of the diameter classes was the most represented. Results Characterization of clonality, genetic diversity, inbreeding and selfing rate Only 2 (0.25%), 2 (0.51%), and 0 (0%) genotypes were rep- resented by multiple stems in populations CMNP, MFR and Fifinda, respectively, indicating low level of clonal reproduction. For each population, the parameters of genetic diversity did not differ significantly between the adult and juvenile cohorts. However, seeds collected in the MFR population showed significantly lower HO and higher FIS than juveniles and adults (Table 1). The coef- ficient of inbreeding, uncorrected for null alleles (FIS), was significantly greater than zero in all populations and cohorts, except for juveniles in Fifinda (FIS = 0.076). The estimates of inbreeding that account for the presence of null alleles (FISc) were close to zero, except for seeds in MFR, with FISc = 0.145 (Table 1). The inbreeding coefficients (FISc) were consistent with estimates of selfing rate based on identity disequi- librium, which were close to zero in adults and juve- niles, but higher in seeds (S = 0.13) (Table  1). Direct Page 7 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 estimates (NMπ) confirmed the low selfing rate in juve- niles (0 ± 0.004 in CMNP, 0.02 ± 0.01 in MFR, 0 ± 0.01 in Fifinda) and the much higher rate in seeds (0.30 ± 0.05 in MFR) (Table  2). More specifically, of the 104 seeds col- lected from seven trees in the MFR population, we iden- tified the mother and father of 66 of them, of which 17 (26%) were selfed and were present under four trees. About 25% of the seeds collected could not be assigned to any of the seven trees under which they were collected, nor to any other adult tree, but after identifying four families within the latter and adding four reconstructed genotypes of the mothers of these families, a total of 31 seeds (30%) appeared self-fertilized. The strong reduction of selfing rate between the seed and seedling stages indi- cates the expression of inbreeding depression in C. edulis due to high mortality of selfed seedlings, or germination failure of selfed seeds (early-acting inbreeding depres- sion; [84, 85]). Characterization of historical gene dispersal through the fine-scale spatial genetic structure (FSGS) In each population, the kinship coefficients (Fij) decreased fairly linearly with the logarithm of geo- graphic distance, as predicted in the context of isolation by distance (Fig.  2). The mean Fij for the first distance class (< 10  m) ranged from 0.08 to 0.13 and decreased rapidly with distance, giving levels of FSGS for adults and juveniles in the different populations ranging from Sp = 0.036 ± 0.004 in MFR and 0.028 ± 0.003 in Fifinda to 0.023 ± 0.003 in CMNP. Considering that 95% confidence intervals for these Sp estimates can be approximated by estimate ± 2*SE, the three populations show overlap- ping confidence intervals. Similar high Sp values were obtained for juveniles and adults (Table 1). The indirect approach to estimate the gene dispersal parameters from the FSGS converged in each popula- tion under the highest assumed effective density, leading to neighbourhood sizes ranging from Nb = 21 (MFR) and 34 (Fifinda) to 77 (CMNP), and the extent of gene disper- sal ranging from σg = 140 ± 11  m (MFR) and 137 ± 42  m (Fifinda) to 275 ± 63  m (CMNP; Table  3). When the Table 1  Parameters of genetic diversity and consanguinity of the different cohorts of the Coula edulis populations Population Cohort N NAE AR He Ho FIS FISc S (SE) Sp (SE) CMNP Juveniles 154 3.89 10.07 0.695 0.594a 0.146a* 0.037 (0.013–0.061) 0.039 (0.021) 0.021 (0.003) Adults 646 3.81 10.16 0.692 0.608a 0.122a* 0.018 (0.008–0.029) 0.022 (0.01) 0.024 (0.003) MFR Seeds 104 2.88 6.57 0.595 0.406a 0.319a* 0.143 (0.087–0.206) 0.13 (0.06) - Juveniles 171 3.22 6.93 0.633 0.525b 0.171b* 0.031 (0.004–0.061) 0.08 (0.03) 0.047 (0.012) Adults 237 3.07 6.81 0.623 0.548b 0.120b* 0.011 (0.001–0.024) 0.02 (0.03) 0.032 (0.003) Fifinda Juveniles 72 3.46 7.59 0.651 0.602a 0.076a 0.040 (0.009–0.059) 0 (0.01) 0.033 (0.005) Adults 73 3.62 8.46 0.666 0.597a 0.104a* 0.028 (0.001–0.055) 0 (0.01) 0.024 (0.003) N: sample size; NAE: effective number of alleles; AR: allelic richness (k = 130); He: expected heterozygosity (gene diversity corrected for sample size); Ho: observed heterozygosity; FIS: inbreeding coefficient potentially biased by null alleles; FISc: inbreeding coefficient accounting for null alleles (95% posterior range); S: selfing rate based on identity disequilibrium; Sp: degree of fine-scale spatial genetic structure (FSGS); SE: standard error Letters for Ho and FIS: within populations, values that share a common letter do not differ significantly (P > 0.05) in the analysis of variance. * Indicates FIS > 0 at P < 0.01 Table 2  Seed and pollen dispersal parameters (± standard error) of different populations of C. edulis estimated using the neighbourhood model implemented in NMπ Parameter CMNP MFR (Juveniles) MFR (Seeds) Fifinda Density of adults per ha (DBH ≥ 10.6 cm) 1.62 1.76 1.76 2.94 Selfing rate (s) 0 ± 0.004 0.02 ± 0.01 0.30 ± 0.05 0 ± 0.01 Pollen immigra- tion rate (mp) 0.59 ± 0.05 0.33 ± 0.05 0.17 ± 0.04 0.71 ± 0.08 Mean kernel pollen dispersal distance (dp) 173 m [141–223]a 211 m [151–348]a 65 m [48–97]a 358 m [156 - ∞]a Shape of pollen dispersal kernel (bp) 1.68 ± 0.48 0.79 ± 0.18 - 0.5 ± 0.1 Pollen dispersal anisotropy (kp) 0.31 ± 0.34 1.04 ± 0.33 - Pollen dispersal prevailing direc- tion (ap) 0.86 ± 0.15 0.13 ± 0.05 - Seed immigra- tion rate (ms) 0.15 ± 0.03 0.07 ± 0.03 0.2 ± 0.01 0.28 ± 0.07 Mean kernel seed dispersal distance (ds) 105 m [86–140]a 131 m [99–197]a - 219 m [134–599]a Shape of seed dispersal kernel (bs) 0.8 ± 0.16 0.73 ± 0.14 - 0.5 ± 0.1 Seed dispersal anisotropy (ks) 0.30 ± 0.16 0.9 ± 0.19 - Seed dispersal prevailing direc- tion (as) 0.41 ± 0.09 0.57 ± 0.03 - Effect of DBH on female fitness (g) 0.83 ± 0.10 0.41 ± 0.10 - 0.57 ± 0.15 Effect of DBH on male fitness (b) 0.59 ± 0.17 0.61 ± 0.15 - 0.39 ± 0.34 a 95% confidence interval when both the shape and the mean distance of dispersal kernels are estimated - indicates parameters that were not estimated Page 8 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 assumed effective densities were lower, the method did not always converge but led to higher σg estimates in the MFR (200 ± 16 m or 340 ± 57 m) and Fifinda (189 ± 19 m or 276 ± 50 m) populations (Table 3). Characterization of gene dispersal through direct analyses Taking into account only the progeny for which the mother and/or father were identified with probability P ≥ 0.8 following NMπ analyses, we found that moth- ers and fathers were assigned, respectively, to 105 and 44 of the 154 juveniles in CMNP, 103 and 62 of the 171 juveniles in MFR, 32 and 14 of the 72 juveniles in Fifinda. For the 104 seeds sampled in MFR, 76 were mothered by seven trees under which they were harvested, while 66 were fathered by 13 sampled trees. When NMπ analysis was run again after adding four reconstructed maternal genotypes based on the genotypes of seeds not assigned to any adult tree, we found that 96 seeds were mothered by 11 trees and 73 seeds were fathered by 17 trees. The seed immigration rates based on NMπ analy- ses of juveniles ranged from ms = 0.07 ± 0.03 in MFR and 0.15 ± 0.03 in CMNP to 0.28 ± 0.07 in Fifinda (Table  2). Pollen immigration rates were higher but fol- lowed the same trend among populations, ranging from mp = 0.33 ± 0.05 in MFR and 0.59 ± 0.05 in CMNP to 0.71 ± 0.08 in Fifinda (Table  2). For seeds sampled in MFR, ms = 0.20 ± 0.01 and mp = 0.17 ± 0.04 when the reconstructed maternal genotypes were integrated. The mean seed dispersal distances based on the esti- mated kernels were rather short, ranging from ds = 105 m in CMNP and 131  m in MFR to 219  m in Fifinda, but with overlapping confidence intervals (Table  2). There- fore, seed dispersal was certainly not higher in the forest with the most intact fauna. These kernels were moder- ately leptokurtic (bs ranging from 0.5 to 0.8, Table 2) and anisotropic, at least in MFR and CMNP (ks ranging from 0.3 to 0.9), with more dispersal events southward (as Table 3  Parameters for the estimation of historical backward gene dispersal of different populations of Coula edulis Population D (ind ha− 1) DE (ind ha− 1) Nb 𝜎g (m) CMNP 0.81 77 ± 38 275 ± 63 1.62 0.41 NA NA 0.16 NA NA MFR 0.88 21 ± 3 140 ± 11 1.76 0.44 21 ± 3 200 ± 16 0.18 25 ± 8 340 ± 57 Fifinda 1.47 34 ± 17 137 ± 32 2.94 0.74 33 ± 7 189 ± 19 0.29 28 ± 3 276 ± 50 D: density of trees that flower and fruit regularly (DBH ≥ 10.6 cm); DE: assumed effective population density (1/2, 1/4 or 1/10 of D); Nb: Wright’s neighbourhood size ± SE (standard error); σg: gene dispersal distance ± SE; NA: indicates that the estimation procedure did not converge Fig. 2  Comparison of fine-scale spatial genetic structures (FSGS) of C. edulis trees in the three study sites, as assessed by the kinship coefficient (Fij) plotted against geographical distances (in meters, on a logarithmic scale) Page 9 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 ranging from 0.41 to 0.57), a trend also visible when illus- trating inferred seed dispersal events around the mother trees (Fig.  3). Of the 105 seed dispersal events detected in CMNP, 59 (56.2%) occurred within 100  m and only two were beyond 300 m (Fig. 3). Similarly, in MFR, of the 103 seed dispersal events detected, 62 (60.2%) occurred within 100 m and a few beyond 500 m (Fig. 3). In Fifinda, of the 32 seed dispersal events detected, 87.5% occurred within 100 m (Fig. 3) but the small sampling area did not allow detection of long-distance dispersal events. The mean pollen dispersal distances based on the estimated kernels were greater than for seeds, ranging from dp = 173 m in CMNP and 211 m in MFR to 358 m in Fifinda (but note the broad confidence interval for Fifinda, encompassing the estimates of ds of the other populations; Table 2). These kernels were moderately lep- tokurtic to near Gaussian (bp ranging from 0.5 to 1.68, Table  2) and anistropic in MFR (kp = 1.04 ± 0.33), with more dispersal events toward the northeast (ap = 0.13), a trend also visible in Fig. 3 (MFR) but not in the other populations. Of the 44 pollen dispersal events detected in CMNP, 14 (31.8%) occurred within 100 m and 14 (31.8%) beyond 200 m (Fig. 3). Of the 62 pollen dispersal events detected in MFR, 36 (58%) occurred within 100  m and two reached 600 to 700 m (Fig. 3). Of the 14 pollination dispersal events detected in Fifinda, 11 occurred within 100 m (Fig. 3). Although pollen and seed dispersal distances were rather small (dp = 173–358 m; ds = 105–219 m, Table 2), we had a substantial proportion of immigrant pollen (mp = 33–71%) and a small proportion of immigrant seeds (ms = 7–28%) that must originate from trees out- side our sampling areas or from adult trees missed during inventories. When controlling whether dispersal kernels could explain the observed immigration rates in the CMNP population by simulating dispersal events from trees sur- rounding the 400  ha sampling area [5], our simulations predicted seed immigration rates (ms) between 10.5 and 12.5%, a range close to the estimated ms at 15%, suggest- ing that the seed dispersal kernel parameters are prob- ably reliable. On the contrary, for pollen, our simulations predicted a pollen immigration rate (mp) between 16 and 17%, a range far below the estimated mp at 59%, leav- ing a gap of nearly 42% of pollen that is not described by the inferred kernel. When forcing the estimation of pol- len dispersal curve parameters to be more leptokurtic (bp = 0.25) and adjusting dp = 300 m, the predicted immi- gration rate (mp) reached 25%, which remains far from the estimated value. Hence, a significant proportion of pollen disperses over long distances, and the 400 ha sam- pling area remains too small to detect these long-distance dispersal events. Comparison of estimates of historical and contemporary gene dispersal distances Following Eqs.  (1) and (2), the estimated parameters of the pollen and seed dispersal kernel (dp, ds, bp and bs) result into σs = 121, 95, 224  m, σp = 191, 140, 367  m and σg = 181, 137 and 343  m for the populations of CMNP, MFR, and Fifinda, respectively. These contemporary σg estimates tend to be smaller than historical σg estimates in CMNP (275 ± 63) and in MFR (140 ± 11 m to 340 ± 57 m), but with a reverse pattern in Fifinda (137 ± 32  m to 276 ± 50 m; Table 3). However, if we approximate the 95% confidence intervals of historical σg estimates using esti- mate ± 2*SE, the resulting intervals include contempo- rary σg estimates. Moreover, as simulations showed that contemporary pollen dispersal distances were under- estimated in CMNP, observing contemporary σg esti- mate lower than historical one in this population is not Fig. 3  Spatial representation of dispersal events around the source inferred by parentage analyses for pollen (+) and seeds (○) in different populations. Dispersal events inferred with a probability ≥ 0.8 are represented after centring the latitudinal and longitudinal displacements based on the source coor- dinates (0, 0). The circle centered on the source has a radius of 100 m Page 10 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 unexpected. Thus, there is no evidence that contempo- rary gene dispersal distances differ from historical ones. Biparental inbreeding and assortative mating in the MFR population The mean value of the kinship coefficient (Fij) observed between the 83 pairs of mates identified with probabil- ity P ≥ 0.8 in the MFR population, after the exclusion of self-fertilization events, reached Fij = 0.092 ± 0.013, which is a significantly higher than the mean value of Fij = 0.064 ± 0.003 expected based solely on the spatial dis- tances between mates (t test; P = 0.034). This indicates that mating between related individuals occurs more fre- quently than expected by chance, suggesting assortative mating in MFR population. On the contrary, there is no evidence of higher mortality in inbred seedlings through biparental inbreeding depression, which would have led to a lower level of relatedness between mates of estab- lished juveniles than expected by chance. Nevertheless, we cannot exclude that biparental inbreeding depression occurs but manifests itself through lower growth rate or lower reproductive success of adults. Diametric structure and its impact on reproductive success The effect of DBH on reproductive success was inferred through NMπ analyses (Table  2) and by comparing, for each population, the diametric distribution of all trees, inferred mothers, and inferred fathers (Fig. 4). The diam- eter of the trunk positively affected the reproductive suc- cess of both the functions of female (g ranging from 0.41 in MFR and 0.57 in Fifinda to 0.83 in CMNP) and male (b ranging from 0.39 in Fifinda to 0.59 in CMNP and 0.61 in MFR). However, in Fifinda, where we identified the father of only 14 juveniles, the standard error on b (0.34) was as large as the estimate (0.39; Table 2). Whatever the population, both maternity and paternity in C. edulis began at a relatively small diameter (small- est mother or father: 12.6 cm in CMNP, 12.0 cm in MFR, 11.1  cm in Fifinda). Large trees contributed dispropor- tionally to regeneration, particularly in the most intact forests: trees with a DBH ≥ 50  cm mothered or fathered 67% of juveniles in CMNP, 31% in MFR, and 9% in Fifinda. The very low value observed in Fifinda is due to the low proportion of trees > 50 cm (3.5% in Fifinda, com- pared to 15% in MFR and 34% in CMNP). The median DBH of mothers and fathers were, respectively, 53.3 and 53.2  cm in CMNP, 39.5 and 40.1  cm in MFR, 24.7 and 27.6 cm in Fifinda. Diameter distribution analysis and regeneration The diametric distribution of mature trees differed strik- ingly between populations (Fig. 4). In the CMNP popula- tion, C. edulis shows a multimodal distribution of DBH, with high numbers of individuals in the juvenile class but also in the 40–50 cm class, before observing decay- ing numbers with higher diameter classes (Fig.  4). On the contrary, in the MFR and Fifinda populations, the observed diametric distributions follow “inverted J” shapes, indicating a high number of individuals in the juvenile class (n = 169 and 74 in MFR and Fifinda, respec- tively) and a decrease in the number of individuals with higher diameter classes (Fig. 4). Furthermore, among juveniles from MFR (stems with DBH < 10.6 cm), 75% had a diameter between 0 and 2 cm (Fig. 5), suggesting a recent regeneration burst, while in Fifinda and CMNP, the cumulative abundance of stems increased nearly linearly with DBH, indicating a rather uniform distribution within the 0–10  cm DBH class (Fig.  5). Therefore, the most disturbed and defaunated populations (Fifinda, and to a lesser extent MFR) showed a pattern indicative of continuous regeneration (“inverted J” distribution), while the most preserved forest (CMNP) showed less regeneration. However, in Fifinda, no large trees were found (maximum DBH = 58.5  cm compared to 99.5 cm in MFR and 110 cm in CMNP), possibly due to logging by the villagers, as we observed the stumbs of three large trees (DBH > 65 cm) that had been cut in the plot to harvest their fruits. Discussion This study characterized genetic diversity, mating system, historical and contemporary gene flow, and regeneration within three C. edulis populations showing contrasted levels of human disturbances. We now discuss the poten- tial impacts of human disturbances on the dynamics of the C. edulis population. Effects of disturbance on genetic diversity and inbreeding Our results show similar levels of genetic diversity between populations and between cohorts (Table 1). As in most other tropical species, we found that C. edulis is a predominantly outcrossing species, although it has con- siderable potential for seed self-fertilization (30% in the MFR population). This selfing rate is substantially higher than that observed at seed level in several other African tree species (e.g. 4% in Cylicodiscus gabunensis [28]). Our results indicate that self-pollinated seeds rarely pro- duce offspring, as suggested by the decrease in inbreed- ing between the seed (FISc = 0.143) and juvenile (FISc = 0.031) cohorts (Table 1), but also by the very low rate of self-fertilization in juveniles (0 to 2%), inferred through NMπ analyses (Table  2). This high rate of self-fertiliza- tion at the seed level could result from limited pollen dispersal and the absence of efficient prezygotic mecha- nisms avoiding selfing (e.g. self-incompatibility system). We ignore if self-fertilized seeds fail to germinate and/or if the resulting seedlings die early, but this phenomenon reflects early-acting inbreeding depression [85]. Page 11 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 Self-fertilization rates in C. edulis juveniles are close to the 3% obtained at the juvenile stage of C. gabunen- sis [28] and consistent with less than 10% found in other tropical tree species [23]. Higher rates of self-fertilization in juveniles were reported in a few African species 20% in E. suaveolens [31]; 20–40% in B. toxisperma [32]; and 54% in Pericopsis elata [42], although adults were not inbred (except in P. elata). Therefore, inbreeding depres- sion manifested essentially between the seed and juve- nile stages in C. edulis, while it was generally detected between the juvenile and adult stages in other African species [31, 32, 42]. Early-acting inbreeding depression could manifest through high abortion rate [85] and/or low germination rate of self-fertilized embryos, or high mortality rate of the resulting seedlings [87]. Inbreeding depression could explain a large number of regeneration failures during early life stages in tropical tree species that are predominantly outcrossing but have considerable self-fertilization potential [86]. In the MFR population, in addition to selfing, biparen- tal inbreeding results from limited seed and pollen dis- persal but also from some level of assortative mating, possibly due to more synchronous flowering between related than unrelated adults [87]. Assortative mating Fig. 4  Comparison of the DBH structures of all trees with that of inferred mothers and fathers of juveniles in different populations of Coula edulis Page 12 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 has also been observed in other African species such as Erythrophleum suaveolens [5, 88] and Entandrophragma cylindricum [39], and a genetic determinism of flowering phenology was shown in Milicia excelsa [89]. Given that inbreeding patterns did not differ between populations (adults and juveniles show overlapping credible inter- vals of FISc estimates; Table 1), there is no evidence of an impact of human disturbances on inbreeding in C. edulis. Effects of disturbance on fine-scale spatial genetic structure (FSGS) The presence of a FSGS is a common phenomenon in tree species, depending on their reproductive system, population density, seed dispersal vectors, and, to a lesser extent, pollen dispersal [9]. In this study, high lev- els of FSGS were detected in each population and could be characterized by a near-linear decay of relatedness with the logarithm of the distance (Fig.  2). This type of genetic structure results from limited gene dispersal and locally interacting demographic and environmental fac- tors [9, 23, 90]. The strength of FSGS, measured by the Sp statistic, can be compared for some African tropical forest trees reviewed by [23], and [42] for African species and through recent meta-analyses [25, 91]. Values for C. edulis (Sp = 0.023–0.036) were characteristic of trees dis- persed over short distances by wind, gravity, or rodents (mean Sp = 0.023 [23]), which can be explained by the combination of limited seed and pollen dispersal. Coula edulis seeds are dispersed by small mammals such as scatter-hoarding rodents, which are known to be short- distance dispersers [92, 93] and seed predators [94, 95]. For 12 plant species dispersed by animals showing active accumulation of seeds through caching or hoarding and a low mobility (category D in [8]), the mean ± SD of Sp is 0.0178 ± 0.0146, consistenly with our estimate. When comparing different cohorts, high FSGS was found in both juveniles and adults within each population (Table  1). These results contrast with those of [96] and [97] who have shown that human-induced disturbance of habitat and seed dispersal behavior affected the FSGS. Research on Diospyros crassiflora, the ebony tree dis- persed by forest elephants in Central Africa, showed that high FSGS is observed among juveniles in defaunated forests, despite low FSGS among adults, while very low Fig. 5  Cumulative relative frequency of juveniles (DBH < 10.6 cm) with diameter in different populations of Coula edulis Page 13 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 FSGS is found both among juveniles and adults in intact forests [98]. However, for C. edulis, anthropogenic degra- dation of MFR and Fifinda habitats could alter the quality of microhabitats and the conditions for the establishment and survival of juveniles [99–101]. This should affect the strength of FSGS, as observed in other studies [102, 103], but such an effect is currently not visible in the different populations of C. edulis studied, all of which had high but similar Sp values. Thus, there is no evidence of human impact on the FSGS of C. edulis. It must be noted that the FSGS in a population builds up progressively over multiple generations [104]. With a mean annual DBH increment of 0.23 cm [69] and a median DBH of adults approaching 40  cm (in MFR) to 50  cm (in CMNP), the mean generation time of C. edulis is probably of the order of 150–200 years. Therefore, we do not expect that the FSGS observed today in adults was influenced by recent human perturbations, which could impact the FSGS only in the long term. Effects of disturbance on contemporary gene dispersal There was little variation in the estimated seed and pol- len dispersal parameters between the different popula- tions of C. edulis. Pollen dispersal distances were always considerably longer than seed dispersal distances. This is consistent with previous studies that have reported more extensive pollen than seed dispersal distances in most tree species [5, 28, 39, 42, 105]. Consistently, pol- len immigration rates (mp = 33–71%) are substantial, approaching values found in other African trees (e.g. 51% for Distemonanthus benthamianus within an area 6.56  km² [5], 71% for Cylicodiscus gabonensis within an area 839 ha [28]). The higher seed and pollen immigra- tion rates (ms, mp) observed in Fifinda can be explained by the relatively small size of the exhaustively sampled plot (18 ha instead of 400 ha). Similarly, the lower ms and mp values at MFR than at CMNP can be explained by the position of the 400 ha MFR plot, next to a river and inun- dated forests inhospitable for C. edulis, so that C. edulis did not occur in the vicinity of the plots along two of its sides, while the 400  ha CMNP plot was surrounded by forests where C. edulis occurred at similar densities in all directions. The fact that in the CMNP plot the pollen immigration rate predicted from the estimated pollen dispersal kernel (16–25%) was much lower than the measured immigra- tion rate (59%) indicates that a substantial proportion of pollen disperse over longer distances than assumed by the dispersal kernel. Hence, the area of 400 ha delim- ited for estimating pollen dispersal remains too small to capture most dispersal events, a situation reported in other studies [5, 28]. Keeping this caveat in mind, we found that the estimated mean pollen dispersal distances (dp = 173–358 m) in C. edulis are lower than those found for African canopy species: 2500  m in C. gabunensis [28], 942 m in P. elata [42], 506 m in E. cylindricum [39], 294 m in E. suaveolens [5]. Relatively low pollen dispersal distances in C. edulis suggest that it might be pollinated by relatively small insects, as shown for other sub-canopy trees [106], which might result in shorter pollen dispersal distances [107]. Although we inferred relatively low mean seed disper- sal distances (ds = 105–219 m) in C. edulis, they are not so different from those of some large African forest can- opy trees dispersed by wind (ds = 184 m for C. gabunensis [28], 71 m for D. benthamianus [5]) or animals 175 m for E. suaveolens [5]. Camera trap observations on MFR and CMNP plots (unpublished) showed that the main dis- persers are small mammals such as the rodents Criceto- mys emini (Emin’s rat), Atherurus africanus (porcupine), and Heliosciurus rufobrachium (squirrel). This is also confirmed by [49] in the forests of Gabon. According to [94] and [95], rodents are known to be short-distance dis- persers, and the FSGS of plants depend on the behavior of their seed dispersers [8], which can itself be modified by human disturbances [108]. Although the Fifinda and MFR populations are experiencing human disturbances, this has not had a significant impact on seed dispersal patterns, indicating that the dispersal mechanisms have not changed in recent years. Hunting is known to nega- tively affect large wildlife but small mammal populations tend to resist, and sometimes proliferate, in hunted for- ests [109, 110], while they are the main dispersers of C. edulis seeds [49]. The lack of long-distance seed dispersal in CMNP also suggests that forest elephants act only as predators of C. edulis seeds, which is supported by obser- vational studies and monitoring of seed germination in elephant dung [49]. This situation contrasts with the case of ebony trees, D. crassiflora, where long-distance seed dispersal prevails in intact forests hosting forest elephants, while seed dispersal is much more limited in deforested areas [98]. Comparison of historical and contemporary estimates of gene dispersal Our results indicate that a significant fraction of the pollen-mediated gene flow of the C. edulis population of the sampled plot comes from outside. This was underes- timated in the NMπ analyses, so it is not surprising that the direct method produced lower σg values than the indirect method. When this fraction of pollen-mediated gene dispersal is taken into account, the σg values esti- mated by direct and indirect methods converge. The sim- ilarity between the contemporary and historical distances of gene dispersal confirms that the large mammal defau- nation resulting from hunting did not significantly affect gene dispersal in C. edulis. Therefore, by relying on small Page 14 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 mammals for their dispersal, C. edulis populations appear resilient to the hunting effect of human disturbances. Determinants of tree reproductive success Our results show that the trunk diameter (DBH) posi- tively affects the reproductive success of C. edulis trees, for both male and female functions (Table 2), as reported in other tree species [5, 28, 39, 42]. Our results show that all classes of DBH (> 10.6 cm) contribute to reproduction (Fig. 4), which means that both maternity and paternity start early in this species, as the minimum diameter of flowering and fruiting was 11.1 cm. This is in agreement with our observations in the field, where we found some fruit remains under the crown of trees with DBH < 12 cm, and with literature data reporting a minimum flowering diameter of 10.6 cm [63]. The fact that C. edulis flowers at this early stage can be explained by the fact that it is a slower growing species than emergent tree species that have larger flowering diameters [111–113]. Effects of disturbance on the natural regeneration of Coula edulis The DBH structures differed between the C. edulis popu- lations (Fig. 5): the inverted J structure of the MFR and Fifinda populations indicates continuous regeneration, whereas in the CMNP population only 19% of individu- als are juveniles (DBH < 10 cm), which could be enough to regenerate the population only if the mortality rate of established seedlings remains low. The “inverted J” dis- tribution was also reported by [114] in a population of C. edulis from south-eastern Cameroon and by [115] in Gabon. A possible explanation for better regeneration in hunted forests is that C. edulis seed dispersers are resil- ient, or even promoted, by hunting activities [109, 110] while some of their seed predators are selectively hunted. An alternative explanation is that the opening of the for- est due to human disturbances benefits the regeneration of C. edulis. Disturbances in the MFR and Fifinda popu- lations may have altered the microhabitat conditions in favor of the establishment and survival of C. edulis juve- niles compared to the CMNP population (Fig.  5). Since C. edulis is considered as a shade-tolerant species, it was a priori expected that the opening of the forest by human disturbances would lead to higher mortality of juveniles exposed to light, whereas this is not the case here, where we observe a good regeneration as for light-demanding species. It is therefore possible that C. edulis behaves as an intermediate species for shade tolerance and / or that it does not suffer substantially from high irradiance, as observed in other shade-tolerant species [116]. Similar results are reported in D. crassifora, another shade-tol- erant species, by [98] who observed good regeneration in defaunated forests compared to intact forests, despite the much less efficient seed dispersal. On the other hand, our results differ from the conclusions of [117, 118] who showed low regeneration in Leptonychia usambaren- sis and Virola flexuosa, respectively, in fragmented for- ests with high deforestation. Our results also differ from those obtained in Afrotropical forests, where according to [93] hunting-induced defaunation drives increased seed predation and decreased seedling establishment of commercially important tree species. Conclusions Our study shows that the dispersal distances of Coula edulis seed and pollen are limited and do not differ signif- icantly according to the level of human perturbation and defaunation, and did not change over time. This shows that the adverse effects of defaunation cannot be gener- alized to all tree species. Our results also showed higher recruitment in the more disturbed forests, possibly due to better access of the understory to sunlight and / or lower predation of seeds and seedlings. This calls into question the sciaphilic nature of the species. In this study, we showed a high rate of self-fertilization and inbreed- ing in seeds in one population, followed by early-acting inbreeding depression between the seed and juvenile stages that eliminated nearly all self-fertilized individu- als. This, combined with the high predation of seeds and freshly germinated seedlings, may explain the low num- ber of seedlings observed in some populations despite the high fruiting of the species. Despite the apparent resilience of C. edulis in the face of hunting pressures, it is important to conserve the remaining populations of C. edulis to ensure that the current level of genetic diversity is maintained, both for the conservation and domestica- tion of this commercially important species. Abbreviations CMNP � Campo Ma’an National Park MFR � Mbalmayo Forest Reserve SSR � Simple Sequence Repeat DBH � Diameter at Breast Height DRC � Democratic Republic of the Congo Supplementary Information The online version contains supplementary material available at ​h​t​t​p​s​:​​​/​​/​d​o​​i​.​​o​r​​ g​​/​​1​0​​.​1​1​​​8​6​​/​s​1​2​​8​6​2​-​​0​2​5​-​0​​2​3​5​6​-​0. Supplementary Material 1 Acknowledgements We thank the Ministry of Forests and Wildlife of Cameroon for the research authorization N° 4025/L/MINFOF/SETAT/SG/DAG/SDPSP/SP /CBF0RM/ which allowed us to collect the data with the help of the conservation staff of Campo Ma’an National Park. We would also like to thank the Ebony Project, through the Bob Taylor Foundation. Author contributions N.G.K., B.S. and O.J.H. conceived the research. N.G.K. collected the data. N.G.K. and S.S. performed the genotyping. N.G.K and O.J.H. conducted data analyses. https://doi.org/10.1186/s12862-025-02356-0 https://doi.org/10.1186/s12862-025-02356-0 Page 15 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 N.G.K. wrote the first draft and all authors contributed to the final version of the manuscript. Funding PhD grant of NGK was funded by the Université Libre de Bruxelles (ULB, cooperation grant). Laboratory costs were covered by grants T.0119.20 and T.0075.23 from the Belgian Fund for Scientific Research F.R.S.-FNRS, where OJH is a Research Director. Data availability Data available in Supplementary Material. Declarations Consent for publication Not applicable. Competing interests The authors declare that they have no Competing interests. Ethical approval Samples from Cameroon (Mbalmayo, Fifinda) were collected with a research permit granted by MINRESI (000102/MINRESI/B00/C00/C10/C13). Received: 22 October 2024 / Accepted: 18 February 2025 References 1. Charles-Dominique P. Speciation and Coevolution: an interpretation of frugivory phenomena. Frugivory Seed Dispersal Ecol Evol Asp. 1993;:75–84. 2. Charles-Dominique P. Relationships between seed dispersal and behavioural ecology. In: Bongers F, Charles-Dominique P, Forget PM, Théry M, editors. Dynamics and plant-animal interactions in a Neotropical rainforest. Kluwer Academic; 2001. pp. 191–6. 3. Nathan R, Muller-landau HC. Spatial patterns of seed dispersal, their determi- nants and consequences for recruitment. Trends Ecol Evol. 2000;15:278–85. 4. Howe HF, Miriti MN. When seed dispersal matters. Bioscience. 2004;54:651–60. 5. Hardy OJ, Delaide B, Hainaut H, Gillet JF, Gillet P, Kaymak E, et al. Seed and pollen dispersal distances in two African legume timber trees and their reproductive potential under selective logging. Mol Ecol. 2019;28:3119–34. 6. Saastamoinen M, Bocedi G, Cote J, Legrand D, Guillaume F, Wheat CW, et al. Genetics of dispersal. Biol Rev. 2018;93:574–99. 7. Sujii PS, Tambarussi EV, Grando C, de Aguiar Silvestre E, Viana JPG, Brancalion PHS, et al. High gene flow through pollen partially compensates Spatial limited gene flow by seeds for a Neotropical tree in forest conservation and restoration areas. Conserv Genet. 2021;22:383–96. 8. Gelmi-Candusso TA, Heymann EW, Heer K. Effects of zoochory on the Spatial genetic structure of plant populations. Mol Ecol. 2017;26:5896–910. 9. Vekemans X, Hardy OJ. New insights from fine-scale Spatial genetic structure analyses in plant populations. Mol Ecol. 2004;13:921–35. 10. Burczyk J, Lewandowski A, Chalupka W. Local pollen dispersal and distant gene flow in Norway spruce (Picea abies [L.] Karst.). In: Forest Ecology and Management. 2004. pp. 39–48. 11. Nathan R, Safriel UN, Noy-Meir I, Schiller G. Spatiotemporal variation in seed dispersal and recruitment near and Far from Pinus halepensis trees. Ecology. 2000;81:2156–69. 12. Muller-Landau HC, Wright SJ, Calderón O, Condit R, Hubbell SP. Interspecific variation in primary seed dispersal in a tropical forest. J Ecol. 2008;96:653–67. 13. Jordano P. Fruits and frugivory. In: Fenner M, editor. Seeds: the ecology of regeneration in plant communities. CABI; 2000. pp. 125–66. 14. Kremer KN, Bauhus J. Drivers of native species regeneration in the process of restoring natural forests from mono-specific, even‐aged tree plantations: a quantitative review. Restor Ecol. 2020;28:1074–86. 15. Wright SJ, Zeballos H, Dominguez I, Gallardo MM, Moreno MC, Ibáñez R. Poachers alter mammal abundance, seed dispersal, and seed predation in a Neotropical forest. Conserv Biol. 2000;14:227–39. 16. Amorim T, Santos AMM, Almeida NM, Novo RR, Leite AV, de Castro L. CC. Plant traits interplay to balance pollen limitation in the Brazilian seasonal dry forest: A meta-analysis. J Arid Environ. 2021;186 November 2020:104408. 17. Bennett JM, Steets JA, Burns JH, Burkle LA, Vamosi JC, Wolowski M, et al. Land use and pollinator dependency drives global patterns of pollen limitation in the anthropocene. Nat Commun. 2020;11:3999. 18. Jordano P. What is long-distance dispersal? And a taxonomy of dispersal events. J Ecol. 2017;105:75–84. 19. Lowe AJ, Boshier D, Ward M, Bacles CFE, Navarro C. Genetic resource impacts of habitat loss and degradation; reconciling empirical evidence and pre- dicted theory for Neotropical trees. Heredity (Edinb). 2005;95:255–73. 20. Hamrick J. Response of forest trees to global environmental changes. Ecol Manage. 2004;197:323–35. 21. Lowe AJ, Cavers S, Boshier D, Breed MF, Hollingsworth PM. The resilience of forest fragmentation genetics–no longer a paradox–we were just looking in the wrong place. Heredity (Edinb). 2015;115:97–9. 22. Diniz UM, Lima SA, Machado ICS. Short-distance pollen dispersal by bats in an urban setting: monitoring the movement of a vertebrate pollinator through fluorescent dyes. Urban Ecosyst. 2019;22:281–91. 23. Dick CW, Hardy OJ, Jones FA, Petit RJ. Spatial scales of pollen and Seed-Medi- ated gene flow in tropical rain forest trees. Trop Plant Biol. 2008;1:20–33. 24. Grant EL, Conroy GC, Lamont RW, Reddell PW, Wallace HM, Ogbourne SM. Short distance pollen dispersal and low genetic diversity in a subcanopy tropical rainforest tree, Fontainea Picrosperma (Euphorbiaceae). Heredity (Edinb). 2019;123:503–16. 25. Miguel-Peñaloza A, Cultid-Medina CA, Pérez-Alquicira J, Rico Y. Do habitat fragmentation and degradation influence the strength of fine-scale Spatial genetic structure in plants? A global meta-analysis. AoB Plants. 2023;15:1–12. 26. Aguilar R, Cristóbal-Pérez EJ, Balvino-Olvera FJ, de Jesús Aguilar-Aguilar M, Aguirre-Acosta N, Ashworth L, et al. Habitat fragmentation reduces plant progeny quality: a global synthesis. Ecol Lett. 2019;22:1163–73. 27. Kubota TYK, Hallsworth JE, da Silva AM, Moraes MLT, Cambuim J, Corseuil CW, et al. Pollen dispersal and mating patterns determine resilience for a large-yet-fragmented population of Cariniana estrellensis. Conserv Genet. 2023;25:117–32. 28. Bhasin OB, Doucet J-L, Ndonda Makemba R, Gillet J-F, Deblauwe V, Sonké B, et al. Contrasted Spatial, demographic and genetic structures of a light-demanding African timber species, Cylicodiscus gabunensis Harms – Implications for a sustainable management of its populations. Ecol Manage. 2024;551:121527. 29. Bizoux JP, Daïnou K, Bourland N, Hardy OJ, Heuertz M, Mahy G, et al. Spatial genetic structure in Milicia excelsa (Moraceae) indicates extensive gene dispersal in a low-density wind-pollinated tropical tree. Mol Ecol. 2009;18:4398–408. 30. Born C, Hardy OJ, Chevallier MH, Ossari S, Attéké C, Wickings EJ, et al. Small- scale Spatial genetic structure in the central African rainforest tree species Aucoumea klaineana: A Stepwise approach to infer the impact of limited gene dispersal, population history and habitat fragmentation. Mol Ecol. 2008;17:2041–50. 31. Duminil J, Daïnou K, Kaviriri DK, Gillet P, Loo J, Doucet J-L, et al. Relationships between population density, fine-scale genetic structure, mating system and pollen dispersal in a timber tree from African rainforests. Heredity (Edinb). 2016;116:295–303. 32. Duminil J, Mendene Abessolo DT, Ndiade Bourobou D, Doucet J-L, Loo J, Hardy OJ. High selfing rate, limited pollen dispersal and inbreeding depres- sion in the emblematic African rain forest tree Baillonella toxisperma – Man- agement implications. Ecol Manage. 2016;379:20–9. 33. Lompo D, Vinceti B, Konrad H, Duminil J, Geburek T. Fine-scale Spatial genetic structure, mating, and gene dispersal patterns in Parkia biglobosa populations with different levels of habitat fragmentation. Appl Plant Sci. 2020;107:1041–53. 34. Ashley MV. Plant parentage, pollination, and dispersal: how DNA microsatel- lites have altered the landscape. CRC Crit Rev Plant Sci. 2010;29:148–61. 35. Austerlitz F, Dick CW, Dutech C, Klein EK, Oddou-Muratorio S, Smouse PE, et al. Using genetic markers to estimate the pollen dispersal curve. Mol Ecol. 2004;13:937–54. 36. Chybicki IJ, Burczyk J, NM+. Software implementing parentage-based models for estimating gene dispersal and mating patterns in plants. Mol Ecol Resour. 2010;10:1071–5. 37. Robledo-Arnuncio JJ, Austerlitz F, Smouse PE. A new method of estimat- ing the pollen dispersal curve independently of effective density. Genetics. 2006;173:1033–45. Page 16 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 38. Marshall TC, Slate J, Kruuk LEB, Pemberton JM. Statistical confidence for likelihood-based paternity inference in natural populations. Mol Ecol. 1998;7:639–55. 39. Monthe FK, Hardy OJ, Doucet JL, Loo J, Duminil J. Extensive seed and pollen dispersal and assortative mating in the rain forest tree Entandrophragma cylindricum (Meliaceae) inferred from indirect and direct analyses. Mol Ecol. 2017;26:5279–91. 40. Rousset F. Genetic differentiation between individuals. J Evol Biol. 2000;13:58–62. 41. Hardy OJ, Maggia L, Bandou E, Breyne P, Caron H, Chevallier M-H, et al. Fine- scale genetic structure and gene dispersal inferences in 10 Neotropical tree species. Mol Ecol. 2006;15:559–71. 42. Angbonda DMA, Monthe FK, Bourland N, Boyemba F, Hardy OJ. Seed and pollen dispersal and fine-scale Spatial genetic structure of a threatened tree species: Pericopsis elata (HARMS) Meeuwen (Fabaceae). Tree Genet Genomes. 2021;17. 43. Letouzey R. Notice de La Carte phytogéographique du Cameroun Au 1/50000. Domaine de La Forêt dense semi-décidue. Fr: IRA-Yaoundé ICIV Toulouse; 1985. pp. 76–85. 44. Vivien J, Faure JJ. Arbres des forêts denses d’afrique centrale. Nguila Kerou. 2011. 45. Moupela C, Doucet J-L, Daïnou K, Meunier Q, Vermeulen C, Beauchêne J. Essais de propagation par semis Et marcottage Aérien de Coula edulis baill. Et perspectives pour sa domestication. Bois Forets Des Trop. 2013;318:3. 46. Klerk M. Regeneration strategies of some emergent tree species in Côte D’Ivoire. Netherlands: Wageningen Agricultural; 1991. 47. Miquel S. Morphologie fonctionnelle de plantules d’espèces Forestières du Gabon. Bull Mus Hist Nat Paris. 1987;9:101–21. 48. De La Mensbruge G. La germination et les plantules des essences arborées de La Forêt dense humide de La Côte D’Ivoire. Cent Tech For Trop; 1966. 49. Moupela C, Doucet JL, Daïnou K, Tagg N, Bourland N, Vermeulen C. Dispersal and predation of diaspores of Coula edulis baill. In an evergreen forest of Gabon. Afr J Ecol. 2014;52:88–96. 50. White F. The vegetation of Africa, a descriptive memoir to accompany the UNESCO/AETFAT/UNSO vegetation map of Africa. Nat Resour Res. 1983;20:1–356. 51. Moupela C, Vermeulen C, Daïnou K, Doucet JL. Le noisetier d’afrique (Coula edulis Baill.). Un produit forestier non ligneux méconnu. Biotechnol Agron Soc Environ. 2011;15:485–95. 52. Oyog Matig O, Ndoye O, Kengue J, Awono A. Les fruitiers forestiers comes- tibles du Cameroun. Cotonou, Benin. International Plant Genetic Resources Institute (IPGRI) Regional Office for West and Central Africa; 2006. 53. Cirad. Fiche technique de Coula edulis. Fiche N°197. In: Gérard J, Guibal D, Beauchêne J, et al. editors. Tropix 6.0. Caractéristiques technologiques de 245 essences tropicales. France: Montpellier; 2008. 54. Tchatat M, Ndoye O, Nasi R. Produits forestiers autres Que Le bois d’oeuvre (PFAB): place Dans L’amenagement durable des forets humides d’afrique centrale. FORAFRI. 1999;:98. 55. Bukola CA-T, Kola. Antimicrobial activities of Coula edulis. Res J Med Plant. 2008;2:86–91. 56. Téké H, Boesch H. Le savoir de Nos anciens. Parol Forêt. 2005;4:4. 57. Walker RA, Sillans R. Les Plantes utiles du Gabon. Gabon: Libreville; 1995. 58. Kuijt J, Hansen B. Flowering plants. Cham: Springer International Publishing; 2015. 59. Meunier Q, Moumbogou C, Doucet J-L. Les Arbres utiles du Gabon. Gem- bloux, Belgique: Les Presses Agronomiques de Gembloux; 2015. 60. Aubréville A. La flore forestière de la Côte d’Ivoire. Nogent-sur-Marne, Fr Cent Tech For Trop. 1959;2. 61. Alexandre DY. Dynamique de La régénération naturelle En Forêt dense de Côte D’Ivoire: stratégies écologiques des Arbres de La voûte et potentiel floristique. Paris: ORSTOM; 1989. 62. Whitmore TC. An introduction to tropical rain forests. Clarendon; 1990. 63. Moupela C. Ecologie, dynamique des populations et intérêts économiques du noisetier d’Afrique (Coula edulis Baill.) au Gabon. PhD Thesis. Gemboux: Université de Liège – Gembloux Agro-Bio Tech. 2013. 64. Tchamba M, Fondja Hendji P, Weladji R, Etoga GM. Abondance et distribution des grands singes et des activités humaines Dans Le parc National de Campo Ma’an, Sud Cameroun. Tropicultura. 2015;NS:3–11. 65. Peltier R, Temgoua LF. Etude préalable à l’aménagement de la réserve forestière de Mbalmayo (Cameroun): pratiques et modes d’accès des popula- tions locales. cirad-00211794. 2008. 66. Kamdem NG, Sergeant S, Vercruysse C, Deblauwe V, Sonké B, Hardy OJ. Devel- opment and characterization of nuclear microsatellite markers for the African walnut Coula edulis Baill (Coulaceae). Mol Biol Rep. 2024;51:438. 67. Li C, Weeks D, Chakravarti A, Chakravarti A. Similarity of DNA fingerprints due to chance and relatedness. Hum Hered. 1993;43:45–52. 68. Hardy OJ, Vekemans X. Spagedi: a versatile computer program to analyse Spatial. Mol Ecol Notes. 2002;:618–20. 69. Chybicki IJ, Burczyk J. Simultaneous Estimation of null alleles and inbreeding coefficients. J Hered. 2009;100:106–13. 70. R Core Team. A Language and Environment for Statistical Computing. Vienna, Austria. R Found Stat Comput. 2018. 71. David P, Pujol B, Viard F, Castella V, Goudet J. Reliable selfing rate estimates from imperfect population genetic data. Mol Ecol. 2007;16:2474–87. 72. Loiselle Ba, Sork VL, Nason J, Graham C. Spatial genetic structure of a tropical understory shrub. Am J Bot. 1995;82:1420–5. 73. Hardy OJ. Estimation of pairwise relatedness between individuals and characterization of isolation-by-distance processes using dominant genetic markers. Mol Ecol. 2003;12:1577–88. 74. Frankham R. Effective population size/adult population size ratios in wildlife: a review. Genet Res. 1995;66:95–107. 75. Chybicki IJ. NMπ—improved re-implementation of NM+, a software for esti- mating gene dispersal and mating patterns. Mol Ecol Resour. 2018;18:159–68. 76. Oddou-Muratorio S, Klein EK, Austerlitz F. Pollen flow in the wildservice tree, Sorbus torminalis (L.) Crantz. II. Pollen dispersal and heterogeneity in mating success inferred from parent-offspring analysis. Mol Ecol. 2005;14:4441–52. 77. Ellstrand NC, Elam DR. Population genetic consequences of small population size: implications for plant conservation. Annu Rev Ecol Syst. 1993;24:217–42. 78. Castilla AR, Pope NS, O’Connell M, Rodriguez MF, Treviño L, Santos A, et al. Adding landscape genetics and individual traits to the ecosystem function paradigm reveals the importance of species functional breadth. Proc Natl Acad Sci U S A. 2017;114:12761–6. 79. Vlam M, van der Sleen P, Groenendijk P, Zuidema PA. Tree age distributions reveal Large-Scale Disturbance-Recovery cycles in three tropical forests. Front Plant Sci. 2017;7. 80. Hubbell SP, Foster RB. La estructura espacial En Gran Escala de Un Bosque Neotropical. Rev BiolTrop. 1987;35:7–22. 81. Pendje G. Stratégies de régénération de neuf essences commerciales de forêt tropicales (Mayombe, Zaire). 1994. 82. Rollet B. L’architecture des forêts denses humides sempervirentes de plaine. CTFT. 1974;:298. 83. Durrieu de Madron L, Forni E. Aménagement forestier Dans L’Est du Camer- oun. Bois Forêts Des Trop. 1997;254:39–50. 84. Husband CB, Schemske DW. Evolution of the magnitude and timing of inbreeding. Evol (N Y). 1996;50:54–70. 85. Xiong H, Zou F, Guo S, Yuan D, Niu G. Self-sterility May be due to prezygotic Late-acting Self-incompatibility and Early-acting inbreeding depression in Chinese chestnut. J Am Soc Hortic Sci. 2019;144:172–81. 86. Naito Y, Konuma A, Iwata H, Suyama Y, Seiwa K, Okuda T, et al. Selfing and inbreeding depression in seeds and seedlings of Neobalanocarpus heimii (Dipterocarpaceae). J Plant Res. 2005;118:423–30. 87. Waser NM. Population structure, optimal outbreeding, and assortative mating in angiosperms. In N. W. Thornhill, editor, The natural history of inbreeding and outbreeding: Theoretical and empirical perspectives. Univ Chicago Press. 1993;:173–99. 88. Ismail SA, Kokko H. An analysis of mating biases in trees. Mol Ecol. 2020;29:184–98. 89. Daïnou K, Laurenty E, Mahy G, Hardy OJ, Brostaux Y, Tagg N, et al. Phenologi- cal patterns in a natural population of a tropical timber tree species, Milicia excelsa (Moraceae): evidence of isolation by time and its interaction with feeding strategies of dispersers. Am J Bot. 2012;99:1453–63. 90. Heywood JS. Spatial analysis of genetic variation in plant populations. Annu Rev Ecol Syst. 1991;22 see 34:335–55. 91. Goncalves AL, García MV, Barrandeguy ME, González-Martínez SC, Heuertz M. Spatial genetic structure and mating system in forest tree populations from seasonally dry tropical forests: a review. Tree Genet Genomes. 2022;18. 92. Rosin C, Poulsen JR. Telemetric tracking of scatterhoarding and seed fate in a central African forest. Biotropica. 2017;49:170–6. 93. Rosin C, Poulsen JR. Hunting-induced defaunation drives increased seed predation and decreased seedling establishment of commercially important tree species in an Afrotropical forest. Ecol Manage. 2016;382:206–13. Page 17 of 17Kamdem et al. BMC Ecology and Evolution (2025) 25:20 94. Forget PM, Munoz E, Leigh EG. Predation by rodents and bruchid beetles on seeds of scheelea palms on Barro Colorado Island, Panama. Biotropica. 1994;26:420–6. 95. Lazure L, Almeida-Cortez J. Impacts des mammifères néotropicaux Sur les graines. Neotrop Biol Conserv. 2006;1:51–61. 96. Hamrick JL, Murawski DA, Nason JD. The influence of seed dispersal mechanisms on the genetic structure of tropical tree populations. Vegetatio. 1993;107–108:281–97. 97. Hamrick JL, Godt MJW. Effects of life history traits on genetic diversity in plant species. Philos Trans R Soc Lond Ser B Biol Sci. 1996;351:1291–8. 98. Deblauwe V, Luskin MS, Assola S, Hardy OJ, Loubières C, Mempong GG et al. Declines of Ebony and Ivory are inextricably linked in an African rainforest. Science (80-). 2024. 99. De Vere N, Jongejans E, Plowman A, Williams E. Population size and habitat quality affect genetic diversity and fitness in the clonal herb Cirsium dissec- tum. Oecologia. 2009;159:59–68. 100. Filazzola A, Brown C, Dettlaff MA, Batbaatar A, Grenke J, Bao T, et al. The effects of livestock grazing on biodiversity are multi-trophic: a meta-analysis. Ecol Lett. 2020;23:1298–309. 101. Fischer J, Lindenmayer DB. Landscape modification and habitat fragmenta- tion: a synthesis. Glob Ecol Biogeogr. 2007;16:265–80. 102. Alcalá RE, De la Cruz S, Gutiérrez-Granados G. Genetic structure and genetic diversity of Swietenia macrophylla in areas subjected to selective logging in Quintana Roo, Mexico. Bot Sci. 2015;93:819–28. 103. Chiriboga-Arroyo F, Jansen M, Bardales-Lozano R, Ismail SA, Thomas E, García M et al. Genetic threats to the Forest Giants of the Amazon: habitat degrada- tion effects on the socio-economically important Brazil nut tree (Bertholletia excelsa). Plants, People, Planet. 2021;3:194– 210. 104. Hardy OJ, Vekemans X. Isolation by distance in a continuous population: rec- onciliation between Spatial autocorrelation analysis and population genetics models. Heredity (Edinb). 1999;83:145–54. 105. Petit RJ, Duminil J, Fineschi S, Hampe A, Salvini D, Vendramin GG. Compara- tive organization of Chloroplast, mitochondrial and nuclear diversity in plant populations. Mol Ecol. 2005;14:689–701. 106. Dick CW, Etchelecu G, Austerlitz F. Pollen dispersal of tropical trees (Dinizia excelsa: Fabaceae) by native insects and African honeybees in pristine and fragmented Amazonian rainforest. Mol Ecol. 2003;12:753–64. 107. Kettle CJ, Hollingsworth PM, Burslem DFRP, Maycock CR, Khoo E, Ghazoul J. Determinants of fine-scale Spatial genetic structure in three co-occurring rain forest canopy trees in Borneo. Perspect Plant Ecol Evol Syst. 2011;13:47–56. 108. Wang R, Compton SG, Chen XY. Fragmentation can increase Spatial genetic structure without decreasing pollen-mediated gene flow in a wind-polli- nated tree. Mol Ecol. 2011;20:4421–32. 109. Effiom EO, Nuñez-Iturri G, Smith HG, Ottosson U, Olsson O. Bushmeat hunting changes regeneration of African rainforests. Proc R Soc B Biol Sci. 2013;280. 110. Nunez-Iturri G, Olsson O, Howe HF. Hunting reduces recruitment of primate- dispersed trees in Amazonian Peru. Biol Conserv. 2008;141:1536–46. 111. Lourmas M, Kjellberg F, Dessard H, Joly HI, Chevallier MH. Reduced density due to logging and its consequences on mating system and pollen flow in the African Mahogany Entandrophragma cylindricum. Heredity (Edinb). 2007;99:151–60. 112. Bourland N, Kouadio YL, Fétéké F, Lejeune P. Ecology and management of Pericopsis elata (Harms) Meeuwen (Fabaceae) populations: a review. Biotech- nol Agron Soc Envionment. 2012;16:486–98. 113. Kouadio YL. Mesures sylvicoles en vue d’améliorer la gestion des populations d’essences forestières commerciales de l’Est du Cameroun. PhD Thesis. Gem- bloux: Faculté Universitaire des Sciences Agronomique de Gemboux. 2009. 114. Kouob BS. Organisation de la diversité végétale dans les forêts matures de terre ferme du Sud-est Cameroun. PhD Thesis. Brussels: Université Libre de Bruxelles. 2009. 115. Doucet J, Moungazi A, Issembe Y. Etude de La végétation Dans Le lot 32 (Gabon): biodiversité, écologie des espèces, recommandations pour Une gestion durable. Rapp Final Libr Gabon; 1996. 116. Silvestrini M, Válio IFM, de Mattos EA. Photosynthesis and carbon gain under contrasting light levels in seedlings of a pioneer and a climax tree from a Brazilian semideciduous tropical forest. Rev Bras Botânica. 2007;30. 117. Cordeiro NJ, Howe HF. Forest fragmentation severs mutualism between seed dispersers and an endemic African tree. Proc Natl Acad Sci. 2003;100:14052–6. 118. Holbrook KM, Loiselle BA. Dispersal in a Neotropical tree, Virola flexuosa (Myristicaceae): does hunting of large vertebrates limit seed removal? Ecol Soc Am. 2009;90:1449–55. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Short-distance seed and pollen dispersal in both hunted and intact forests in the lower canopy African rainforest tree, Coula edulis Baill. (Coulaceae) Abstract Background Methods Species description Study sites and sampling DNA extraction and genotyping Characterization of genetic diversity and inbreeding Characterization of historical gene dispersal through fine-scale spatial genetic structure (FSGS) Characterization of seed and pollen dispersal through parentage analysis and the neighbourhood model Comparison of historical and contemporary gene dispersal estimates Biparental inbreeding and assortative mating Regeneration dynamics Results Characterization of clonality, genetic diversity, inbreeding and selfing rate Characterization of historical gene dispersal through the fine-scale spatial genetic structure (FSGS) Characterization of gene dispersal through direct analyses Comparison of estimates of historical and contemporary gene dispersal distances Biparental inbreeding and assortative mating in the MFR population Diametric structure and its impact on reproductive success Diameter distribution analysis and regeneration Discussion Effects of disturbance on genetic diversity and inbreeding Effects of disturbance on fine-scale spatial genetic structure (FSGS) Effects of disturbance on contemporary gene dispersal Comparison of historical and contemporary estimates of gene dispersal Determinants of tree reproductive success Effects of disturbance on the natural regeneration of Coula edulis Conclusions References