Genetic and reproductive evidence of incomplete isolations barriers between Pterodon emarginatus and P. pubescens (Leguminosae, Papilionoideae)

Abstract Pterodon emarginatus Vogel and P. pubescens (Benth.) Benth. are phylogenetically related trees that compose a clade of sister species abundant in the Brazilian Savanna. Despite their morphological differences, some individuals with intermediate morphological characteristics have been reported, indicating the formation of interspecific hybrids. This study proposed to evaluate the genetic structure and controlled pollination of individuals of P. pubescens and P. emarginatus in areas of sympatry with the presence of putative hybrids. For this purpose, we genotyped seven microsatellite loci from 61 individuals collected from four apparent contact zones between P. pubescens and P. emarginatus. Controlled pollination experiments were performed on 4,133 flowers from six trees of P. emarginatus and five of P. pubescens. We observed two genetic clusters (k= 2) that corroborate the divergence between P. pubescens and P. emarginatus. The individual genetic assignment showed evidence of natural hybridization between P. pubescens and P. emarginatus. The genetic assignment did not fully support the visual description of the diagnoses of individual morphological characteristics. Controlled interspecific pollination generated fruit and seed production, possibly indicating the absence of a reproductive barrier at the pollination and ovule fertilization levels between these species. Our results enlarge the understanding of the diversification process of Pterodon species.


Introduction
The evolutionary consequences of hybridization range from the termination of reproductive isolation between species (Servedio & Noor 2003) to the homogenization of both species' gene pools (Currat et al. 2008;Marques et al. 2014).The outcome of these hybrids depends on their genetic combinations, demography, population structure, and reproductive viability (Yan et al. 2017).It is not clear which ecological and evolutionary factors drive one outcome of hybridization and not the other (Turchetto et al. 2022).
The presence of both biotic and abiotic factors suggests that the biomes of Savannas and grasslands might shelter a relatively abundant.Since Savannas are relatively young Neotropical biomes, it is likely that their hybrids reflect a minimal divergence among the endemic species that resulted from recent radiation (Schley et al. 2022;Turchetto et al. 2022).Moreover, these Neotropical biomes have widely distributed groups that present clear genetic evidence of hybridization (e.g., Ribeiro et al. 2018).However, there are only a few studies examining the impacts of hybridization on these biomes (Schley et al. 2022;Turchetto et al. 2022).
Our field observations revealed that both Pterodon species have a disjunct distribution.In addition to occurring in the Cerrado areas further South, the Pterodon pubescens is also present in the states of São Paulo, Mato Grosso do Sul, the south of the states of Mato Grosso, Minas Gerais, Goiás, and in the south of the Federal District.In turn, P. emarginatus occupies the following areas further North: north of the states of Mato Grosso, Federal District, Minas Gerais, and the north center of the states of Goiás, Tocantins, Maranhão, Bahia.Individuals showing intermediate characteristics appeared (SL Lima unpubl.res.) in the regions with overlapping distribution.
In a narrow range in the Distrito Federal (Brazil), Rocha (2006) detected individuals showing intermediate morphological features between both species (Fig. 1CF), in addition to some genetic hybrids (via RAPD).Such a finding reveals the possibility of interspecific permeability of species borders.In turn, a phylogeography study indicated that P. pubescens and P. emarginatus underwent recent divergence; therefore, the barriers of reproductive isolation between them might not be absolute (Lima 2019).Herein, our field observations detected four apparent contact zones between the two species, with flowering adults.Thus, this species pair represents a great study model for reproductive isolation, taxonomic cohesion, and natural hybridization.The reproductive isolation of related species (e.g., Johnson et al. 2015;Kuligowska et al. 2015;Wang et al. 2015) is often associated with post-pollination mechanisms, and might occur in the following situations: if interspecific pollen grains do not germinate on the stigma; pollen tubes do not grow properly in the style, absence of fertilization, and if the embryo and/or endosperm does not develop well (Rieseberg & Carney 1998;Kuligowska et al. 2015).However, such barriers might have occurred if seeds are produced after interspecific cross-pollination.In this sense, controlled crossings might help understand the hybridization between these species.Melo et al. (2022) used P. pubescens sequence data to develop a set of microsatellite markers (SSR) for studying the genetics of both populations.The author's proposal was unprecedented in this research line and indicated a high genetic diversity (higher in P. pubescens) with significant intrapopulation inbreeding.
More recently, methods of hybridization detection have estimated the degree of genetic variation shared between adjacent or co-occurring species, including calculating "hybrid indices" to assess the proportion of ancestry inherited from each parent (Schley et al. 2022).In this regard, population-level sampling strategies and molecular markers approaches (e.g., microsatellites, AFLPs, and SNPs from RADSeq) are used (Schley et al. 2022).Studies using molecular markers that can differentiate between closely related species and their hybrids help enlarge the knowledge on the speciation genetic basis and the effects of interspecific crossing on species integrity.Additionally, these studies provide further information on the evolutionary history of the species involved.
Despite the studies addressing these trees of wide economic and cultural potential (e.g., Mors et al. 1967;Bustamante et al. 2010;Basting et al. 2019;Kleinubing et al. 2022), the understanding of their intra-and interspecific genetic variations is yet to be investigated.Therefore, this study aimed to evaluate the genetic composition and controlled pollination of individuals from apparent contact zones of P. pubescens and P. emarginatus containing putative hybrids.Thereby, we sought to answer the three following main questions: (I) Are there genetic groups that support the divergence between P. pubescens and P. emarginatus?(II) Does the genetic composition of individuals characterized as putative hybrids in the field corroborate the hypothesis of natural hybridization between P. pubescens and P. emarginatus?and (III) Does interspecific crossing produce fruits with well-formed seeds?

Material sampling and identification
We collected young leaves from 61 individuals in four natural sites presenting an apparent sympatry between Pterodon pubescens Benth.(Benth.)and Pterodon emarginatus Vogel (Fig. 2).Each individual collected was identified in the field (Table 1) following the visual description of diagnoses morphological characters (Rocha 2006): P. pubescens (pale pinkish flowers, pubescent leaves, and leaflets with retuse to rounded apex), P. emarginatus (violaceous flowers, glabrous leaves, and leaflets with truncate to strongly emarginate apex), or potentials hybrids (individuals with intermediate or discordant characteristics of Rocha's (2006) classification).
Vouchers were taken to record each phenotype/ species and were deposited at Unidade de conservação/  Herb Nº: 68.411,68.414,68.415,and 68.416, and the potential hybrid, Herb Nº: 68.410.The collected plant material was identified and packed in plastic bags containing spherical silica gel.Subsequently, these packages were transported to the Laboratório de Genética e Biodiversidade (LGBio) of the Universidade Federal de Goiás (Goiânia-GO, Brazil).After separating the material for DNA extraction, the remaining dried leaves were stored in a deep freezer (-80 ºC).

Molecular analysis
We have obtained total genomic DNA from leaf tissue using CTAB (cetyl-trimethyl ammonium bromide) extraction buffer, according to the extraction protocol by Doyle andDoyle (1987), modified by Ferreira andGrattapaglia (1996).DNA quality and concentration were assessed through electrophoresis on an agarose gel of 1%.The genotypes of the individuals were identified through seven nuclear microsatellite markers, developed for P. pubescens and transferred to P. emarginatus (Melo et al. 2022) (Table 2).
Figure 2. Representative map of the four natural sites presenting an apparent sympatry of P. pubescens and P. emarginatus.Table 1 shows the details of geographic coordinates and the number of individuals.
Table 2. List of application multiplexes used for genotyping of the studied individuals using seven microsatellite markers developed by Melo et al. (2022).Forward primers were labeled with specific DYE (fluorochromes) for detection in capillary electrophoresis (genotyping).Potential hybrids were tested at the annealing temperatures of P. pubescens.

Genetic and reproductive evidence of incomplete isolations barriers between
Pterodon emarginatus and P. pubescens (Leguminosae, Papilionoideae) PCR reactions were performed in a Veriti™ 96-Well Fast Thermal Cycler (Applied Biosystems®) under the PCR protocol and the thermal cycling conditions described by Melo et al. (2022): one cycle at 94 °C for five min; 30 cycles of 94 °C for one min, annealing temperature for one min (depending on the locus, Tab. 2), and 72 °C for one min; and 72 °C for 30 min to enforce 3' Taq adenylation.The annealing temperature between P. emarginatus and P. pubescens presents a difference (of 2 ºC) in only one marker (PEM25).Therefore, for the potential hybrid individuals, we tested the PCR reaction with the annealing temperature optimized to P. pubescens.The PCR products were multiplexed, denatured, and size-fractionated using capillary electrophoresis on an ABI 3500 DNA Analyzer (Thermo Fisher Scientific®) with a LIZ (600) molecular size standard (Thermo Fisher Scientific®).The molecular data obtained are available in Table S2 of the Supplementary Material.

Statistical analyses
We estimated the genetic diversity parameters on the FSTAT 2.9.3.2 (Goudet 2002) and GDA Genetic Data Analysis 1.0 (Lewis & Zaykin 2001) programs.The following parameters were estimated: the observed heterozygosity (Ho), the expected heterozygosity under Hardy-Weinberg equilibrium (He), the number of alleles (A), and the average allelic richness based on the minimum sample size (A R ).In addition, we also estimated the probabilities of identity (PI) and paternity exclusion (Q) on the Identity v.1.0software (Wagner & Sefc 1999) to verify the quality of the microsatellite set in individual discrimination.
The genetic structure of the individuals was analyzed based on the Bayesian clustering method implemented on the STRUCTURE 2.3.4 software (Pritchard et al. 2000).We performed 30 independent runs for K values ranging from one (no genetic structure) to four (assuming four natural sites) genetic clusters.Each run covered 1,000.000Monte Carlo simulations through Markov chains (MCMC), with a burn-in of 10%, assuming the mixture model with correlated alleles between individuals (admixture model).We then inferred the appropriate K values to explain the data, as proposed by Evanno et al. (2005) and Puechmaille (2016) in the STRUCTURE SELECTOR platform (Li & Liu 2018).The Structure Selector platform allowed us to obtain the coancestry values of each individual belonging to a given cluster and the assignment probability values were assigned through multiple replicates for the same K value (Kopelman et al. 2015) on the CLUMPAK package, provided by the same platform.
Through simulations on the NEWHYBRIDS 1.1 software (Anderson & Thompson 2002), we also evaluated the posterior probability (PP) of each individual belonging to one of the following probability classes of Mendel's law: Pure Parental A (P. pubescens), Pure Parental B (P. emarginatus), F1 hybrid, F2 hybrid, backcross A (with P. pubescens), and backcross B (with P. emarginatus).The assignment to different hybrid categories was systematized upon the individual belonging to any of the six classes considering a posterior probability ≥ 0.70.Individuals with PP ≥ 0.90 were considered to belong to pure parental lineages, whereas those without probability PP ≥ 0.70 for any of the six Mendel's laws classes were considered of an uncategorized hybrid origin.This test was performed with no prior information on allele frequencies, "Jeffery's like priors", and was based on 100,000 Monte Carlo simulations through Markov chains (MCMC) after a burn-in period of 100,000 to ensure the convergence of chains and homogeneity of runs.

Controlled pollination experiments
We conducted controlled pollination experiments (in 2019 and 2021) in six trees of P. emarginatus from a fragment of Cerrado, at the Planaltina Campus of the Universidade de Brasília (Planaltina-DF, Brazil).For P. pubescens, the experiments used five trees from a fragment of Cerrado at the Campus of Universidade de Brasília (Brasília-DF, Brazil).We performed hand pollinations during the coflowering period to verify whether intra-and heterospecific pollen depositions produced fruits with well-formed seeds, in addition to comparing the seed set rates between intraspecific and interspecific cross-pollinations.We also performed manual self-pollination and checked the occurrence of self-pollination and natural production of fruit and seed (control).The following individuals were chosen for our analyses: a monospecific population of P. pubescens with previously genotyped trees (Rocha 2006) and a monospecific population of P. emarginatus with trees from a contiguous area to that studied by Rocha (2006), identified as P. emarginatus (both by morphological characteristics and RAPD).
The following four treatments were applied to the bagged flowers of each species (see Table 5 for sample size): (1) heterospecific pollination (hybridization): stigmas of several flowers of one of the species were pollinated with pollen from the other species; (2) conspecific cross-pollination: stigmas of several flowers were pollinated with pollen from flowers of another distant plant of the same species (at least 30 m); (3) manual self-pollination: stigmas of several flowers were pollinated with pollen from flowers of the same tree; (4) spontaneous self-pollination: flowers in pre-anthesis and bottom of several inflorescences were counted and the inflorescences were bagged to check whether self-pollination occurs spontaneously; (5) Control: counting of the flowers in several inflorescences of different plants of both species for tagging and fruit counting.From experiments 1 to 4, the inflorescences of the flowers examined were bagged before and after hand pollination.In experiments 1 and 2, the flowers were emasculated before anthers opened, manually pollinated, and bagged again until the fruits were set.Subsequently, the fruits were counted, and the reproductive success rates were compared between treatments.The resulting well-formed fruits were collected and counted during the fruiting period.
We adopted general linear mixed models (GLMM) assuming a binomial error distribution (individual as a random factor) to test whether the fruiting rate differed between species and treatment; also in addition, the interaction between species and treatment was considered.Posthoc comparisons between treatment levels were established through the Tukey's method.All analyses were carried out in R version 4.1.2(R Core Team 2021), using the packages lme4 v4.1-28 (Bates et al. 2014) and Emmeans v1.7.2 (Lenth 2022) for model fit and posthoc tests.

Descriptive estimates of genetic variability
High estimate values of combined paternity exclusion probability (Q ~ 0.99985) were shown by the microsatellite set, thus demonstrating a strong ability of the markers to exclude potentially false paternity.We also found low values of combined probability of identity (I ~ 8.441*10 -11 ), indicating that the microsatellite set presents a strong power of individual discrimination since the estimated PI is almost zero.
The PEM25 locus is the only one that has a different annealing temperature between species.The first PCR test amplified the locus PEM25 in the hybrid individuals at the annealing temperature of P. pubescens (54 ºC).Moreover, all loci were polymorphic, and 96 alleles (Table 3) were obtained, with an average allele richness per locus of 13.446.Higher values of allele number and heterozygosity were observed, thus indicating a greater content of marker information (Table 3).

Genetic structure
The Bayesian analysis using STRUCTURE suggested the following numbers of clusters: k= 2 by the Puechmaille (2016) method and k= 3 by the Evanno et al. (2005) method (Fig. 3).The k= 2 scenario summarized the information better than the k= 3 (Fig. 3).The k= 2 comprises the two parental species as contributors to the genetic structure of our samples, showing distinct cluster assignments between the P. emarginatus (PEM) and P. pubescens (PPU) individuals (coancestry values > 0.82).The k= 3 scenarios present a substructure of P. pubescens into two genetic clusters.
We identified the two parental species in k= 2 as contributors to the total gene pool sampled; therefore, individuals whose coancestry value was above or equal to 0.800 in any of the clusters were considered canonical species (for detailed coancestry values, see Table S1 in Supplementary Material).Thus, we assigned 40 individuals as P. pubescens, 15 as P. emarginatus, and six as hybrids.
The genetic assignment by STRUCTURE showed differences (compared with field record) in the number of individuals assigned, as follows: from 34 to 40 for P. pubescens, from 10 to 15 for P. emarginatus, and from 17 to 6 for the hybrids.Based on six individuals with a clear hybrid identity, this initial evaluation revealed a genetic admixture between the species.

Hybrid assignment
The Bayesian analysis by NEWHYBRIDS provided a clear genetic distinction of the individuals (for detailed posterior probability values, see Table S1 in Supplementary Material).In total, 87% of the individuals (n= 53) had significant values (≥ 0.700) of posterior probability (PP), thus belonging to either of the classes.

Genetic and reproductive evidence of incomplete isolations barriers between
Pterodon emarginatus and P. pubescens (Leguminosae, Papilionoideae) This fine genetic assignment provided 33 P. pubescens individuals, 13 P. emarginatus, and 15 hybrids.The NEWHYBRIDS results show the following differences in the number of individuals (compared with the field record): from 34 to 33 for P. pubescens, from 10 to 13 for P. emarginatus, and from 17 to 15 for the hybrids.

Agreement between hybrid assignment methods
We performed a combined evaluation of the field observation and genetic identification assignments (STRUCTURE and NEWHYBRIDS) (Table 4 and Table S1 in Supplementary Material).Approximately 67% of the  Victor Felipe Mendanha de Lima, Edivani Villaron Franceschinelli, Dulce Maria Sucena da Rocha, et al.

Genetic and reproductive evidence of incomplete isolations barriers between
Pterodon emarginatus and P. pubescens (Leguminosae, Papilionoideae) individuals (Six P. emarginatus, 31 P. pubescens, and four hybrids) were assigned to the same group in both the field and genetic identifications.About ~18% (n= 11) of the individuals showed agreement in the genetic identification but differing in the field records.These individuals presented the following behavior: six were genetically P. emarginatus and potentially hybrid according to the field records, two were genetically P. pubescens and potentially hybrid according to the field records, two were genetically hybrids and P. emarginatus according to the field records, and one was genetically P. emarginatus and P. pubescens according to the field record.The agreement between the field records and STRUCTURE reached ~70% of the individuals, while the field record and NEWHYBRIDS agreed in ~75% of the cases.
For both methods, only ~15% of the individuals (n= 9) showed different genetic assignments (STRUCTURE and NEWHYBRIDS), thus being considered "questionable hybrid" individuals (see Table 4 for detailed data).10% (n= 6) of the total sampled individuals (n= 61) were genetically identified as hybrids by both software programs.

Controlled pollination experiment
All treatments of P. emarginatus generated ripe fruits (Table 5).The rate of fruit and seed set was higher for cross and interspecific pollinations comparing with selfpollination, which produced one manually self-pollinated fruit and one spontaneous fruit.Interspecific and crosspollinations also produced fruits with seeds on Pterodon pubescens (Table 6).Self-pollination could not set any fruits, thus indicating that the species is completely allogamous and self-incompatible.The control treatment had very low rates of fruit set.The final fruiting rate was the same for both species (X 2 = 3.42, p= 0.18) but different between the treatments (X 2 = 44.44,p < 0.0001).Both species had the same outcome in the interspecific cross-pollination treatment (z= 1.33, p= 0.85) and the intraspecific crosspollination treatment did not generate different results either for the P. pubescens (z= 2.12, p= 9.94) or P. emarginatus (z= 0.63, p= 0.99) hybrids.

Discussion
Hybridization plays an important role in plant evolution (Abbott et al. 2016) and varied according to the taxa (Taylor & Larson 2019), presumably due to the biological differences between species pairs, such as timing and speciation mode (Payseur & Rieseberg 2016).Moreover, interspecific gene flow via natural hybrid zones might be a source of genetic variability for adaptation to environmental changes (Janes & Hamilton 2017;Turchetto et al. 2022).
Species that have diverged recently and occur in contact zones provide a unique opportunity to study the evolutionary process involved in speciation.Our study identified sympatric genetic individuals of P. emarginatus and P. pubescens and presenting potential hybrids.Our analysis was based on highly variable and informative microsatellite loci (Melo et al. 2022) with great potential for individual and species discrimination (paternity exclusion probability > 0.99 and almost no identity probability).Overall, the nuclear microsatellite loci demonstrated that the two "sucupira-branca" species are genetically distinct.
Based on our field observations and the literature, P. emarginatus seems to be distributed toward north and P. pubescens toward the south of the Cerrado, with a latitudinal contact range between them, thus forming contact zones (Rocha 2006).Rocha (2006) found individuals living in such a latitudinal range -our sampling site (Figure 1 and 2)presenting morphological features and genetic composition that shown to be intermediary of the two species.
Contact zones opened space for the study of hybrid production and interspecific gene flow.Our results demonstrated that interspecific pollination experiments generated fruits and seeds, thus indicating that the two species have no reproductive barrier for pollen germination on the stigma, pollen tube development in the style, or ovule fertilization levels.Despite the absence of such barriers, the numbers of backcrosses and hybrids are relatively low; therefore, the barriers might be of an ecological nature.Our field observations revealed that some barriers might occur at the floral biology level since flowers of P. pubescens open earlier (around 6:00h) than those of P. emarginatus (around 8:00h), despite the time overlap between 8:00 to 12:00h.In addition, the two species might share pollinators, such as Bombus sp.Furthermore, we did not test whether there is a difference between the seed viability produced during interspecific and intraspecific cross-pollinations.There might be variations in seed germination, as well as seedling and plant development.In this sense, further studies should investigate their floral and pollination biology, seed germination, and seedling development to better understand how the two species avoid intercrossing.
The control test produced fewer fruits and seeds than manual cross-pollinations, which may be explained by a natural pollination deficit.The two cases of seed set after self-pollination of 206 flowers do not indicate the species are self-fertile because they can be a product of cross-pollen contamination.But further study might be conducted to better clarify these species' self-compatibility systems.
Our results corroborate the hypothesis of interspecific genetic mixing between P. emarginatus and P. pubescens at the contact zone, which is often reported in the literature for other phylogenetically closely related species (Lorenz-Lemke et al. 2006;Taylor et al. 2014;Ley & Hardy 2017).Additionally, our analyses confirm a hybrid genetic status in six individuals at the contact zones, out of which four also have intermediate morphology, according to the field records (Table 4: ID 53,54,58,and 59).
We also detected that the morphological and genetic records indicated different assignments, as reported in other studies involving phylogenetically close plants (Teixeira et al. 2019;Schnitzler et al. 2020).In fact, the unpredictable phenotype expression in hybrids hampers the morphological hybrid diagnosis (Rieseberg et al. 1993;Teixeira et al. 2019).Since several genetic and environmental factors influence the hybrid phenotypic expression, genetic analyses using molecular markers should provide more robust diagnoses for hybrids presenting the phenotypic expression of a parental species, as our results show (López-Caamal & Tovar-Sánchez 2014).
We found a uniform hybridization pattern at the four sampling sites, given that they were limited to F2 hybrids (see NEWHYBRIDS results), indicating a cross between F1 hybrids.However, NEWHYBRIDS analyses may underestimate backcrosses when evaluating parental species that have recently diverged, thus classifying individuals as belonging to the pure lineage of either parent (Vähä & Primmer 2006).Such a scenario could indicate a lower frequency of backcrosses in our observations.The genetic composition of each studied individual, with clear genetic distinction, confirm that P. emarginatus and P. pubescens should be considered independent species.However, there are no definite reproductive barriers between the two species, thus promoting typical hybrid genetic mixing.We suggest that the small genetic admixture in the canonical species (see STRUCTURE results) might be associated with size homoplasy since the alleles have the same size but the sequence is different (Estoup et al. 2002).Nonetheless, such a hypothesis should be verified through genotyping by sequencing the microsatellites.
An easy and successful interspecific crossing depends directly on the phylogenetic relationship between the species involved in hybridization, in addition to chromosomal homology.Otherwise, there might be incongruities/ incompatibilities between the species' genomes.Pterodon emarginatus and P. pubescens have the same karyotypic number 2n= 16, with small and morphologically similar chromosomes (Bandel 1974;Coleman & Demenezes 1980;Albernaz 2020).More recently, Albernaz (2020) studied the genomes of P. pubescens and P. emarginatus and found highly similar repetitive fractions.Thus, there are several elements in P. pubescens and P. emarginatus that facilitate interspecific hybridization, such as phylogenetic proximity, recent divergence, and vast cytomolecular similarity (both structural and numerical).
It is difficult to detect the processes that lead to species hybridization, even more so for those with a history of introgression and ancestral polymorphisms persisting in the speciation process, such as "sucupira-branca" (Lima 2019).The function of these hybrids in diversifying the studied species are unknown, as well as whether they reinforce reproductive isolation (due to the low viability of hybrids).However, hybrid formation does not necessarily have a specific function, it might simply be result from incomplete species barriers, such as in genetic material exchange (e.g., Zhang et al. 2016).In this sense, further studies should detail the reproductive biology and viability of hybrids of P. emarginatus and P. pubescens to better understand the role of this phenomenon in their evolutionary history.
Our study enlarges the knowledge on the diversification process of the "sucupira-branca" species.The microsatellite set indicated genetic groups that distinguish the P. pubescens and P. emarginatus individuals.Although approximately 10% of the studied individuals were genetically identified as hybrids, the intermediate morphological features per se do not allow us to identify an individual as hybrid.These contrasts found between the genetic and morphological attributions of P. pubescens, P. emarginatus, and their potential hybrids, highlight that integrative studies must investigate the ecological and evolutionary scenarios of these species deeper.
Although Rocha (2006) and Sonsin-Oliveira et al. (2022) has suggested the existence of hybridization between P. emarginatus and P. pubescens, our study is the first to reveal the reproductive viability of hybrid formation, in addition to genetic evidence of natural hybrids between these species.On the other hand, considering their evolutionary proximity, studies adopting more specific approaches, including the genetic identification of hybrid seeds or broad genomic analyses, should differentiate recent hybridization from the retention of ancestral polymorphism/ ancient hybridization.

Figure 1 .
Figure 1.Trees of Pterodon emarginatus (A), P. pubescens (B), and potential interspecific hybrid (C) in the natural environment, followed by their respective flowers (D, E, and F).Elaborated by the authors, DMS Rocha & VFM Lima.

Figure 3 .
Figure 3. Bayesian clustering based on coancestry values, as follows: A) the formation of two genetic groups (k= 2) highlighted in dark pink and purple by the Puechmaille (2016) method and B) the formation of three genetic groups (k= 3) highlighted in dark pink, light pink, and purple by the Evanno et al. (2005) method.All data were grouped according to the field records, based on Rocha (2006), as follows: PEM -P.emarginatus, PPU -P.pubescens, and PSP -potential hybrids.

Figure 4 .
Figure 4. Posterior probability (PP) according to the category analyses on NEWHYBRIDS to verify hybridization between "sucupirabranca" species.Pure parent A (P. pubescens) is in pink; Pure parent B (P. emarginatus) is in purple.The size proportion of the categories in the individual bars indicates a higher assignment probability per class.All individuals were grouped according to the field records, based on Rocha (2006).

Table 1 .
Rocha (2006)of the four natural sites in Brazil where apparent sympatry of Pterodon pubescens and P. emarginatus was reported, as well as the presence of potential hybrid (except in Itacaiú/GO).Field recording based onRocha (2006).

Table 3 .
The estimates of genetic variability parameters were analyzed from seven microsatellite loci for the sample of individuals registered as P. pubescens, P. emarginatus, and their potential hybrids.A: number of alleles; AR: average allelic richness based on minimum sample size (n= 53); He: expected heterozygosity under HWE; Ho: observed heterozygosity; PI: identity probability; Q: paternity exclusion probability; IC: combined probability of genetic identity; and QC: combined probability of paternity exclusion.All values are rounded to three decimal places.

Table 4 .
Rocha (2006)cation was compiled from the field record according to the morphological diagnosis byRocha (2006)and the genetic identification obtained by STRUCTURE's coancestry and posterior probability (PP) in NEWHYBRIDS.Discrepancies in genetic identification between STRUCTURE and NEWHYBRIDS were classified as "Questionable hybrid" individuals.*Discrepancies between field records and genetic identification.

Table 5 .
Results of controlled pollination experiments conducted during the flowering of P. emarginatus in 2019 and 2021.

Table 6 .
Results of controlled pollination experiments conducted during the flowering of P. pubescens in 2019 and 2021.