Species boundary and extensive hybridization and introgression in Petunia

ABSTRACT

Studying the role of hybridization in the speciation of plants is one of the most thrilling areas of evolutionary biology. Hybridization in natural populations can act in opposition to divergence, contribute to adaptation through introgression or foster the emergence of new lineages via hybrid speciation. Species of the plant genus Petunia grow in open areas in southern South America. Some natural interspecific hybrid events have been described for the genus, such as between the endemic P. exserta and the widespread P. axillaris. Both species occur in sympatry in Serra do Sudeste (Brazil), where they occur in diverse habitats and exhibit floral divergence, which has been related to the attraction of different primary pollinators. The present study evaluates the maintenance of the species boundaries front of hybridization and introgression. Direct and indirect methods of estimating gene exchange employed genotyping 720 reproductive plants and 611 progenies of both species with eight microsatellite loci. Gene exchange was found to be frequent and bidirectional between the species, indicating that introgression changes their genetic constitution in areas of sympatry. Limits of the studied species are being maintained because of the high level of inbreeding and backcrosses that are habitat-dependent.

Keywords:
gene exchange; hybridization; inbreeding; introgression; microsatellite; Petunia; species limits

Introduction

Gene exchange between closely related species is considered an important driver of diversity in angiosperms (Abbott et al. 2016). The consequences of interspecific hybridization range from blurring species’ limits (Soltis & Soltis 2009) to neutral effects (Arnold 2006) in extreme situations, passing for introgression (Kenney & Sweigart 2016) and genetic divergence with new phenotypes and adaptations (Meier et al. 2016). Hybrids’ fates are dependent on their genetic combinations, demographic distribution and population structure, as well as their reproductive ability (Yan et al. 2017).

In Serra do Sudeste, Rio Grande do Sul, Brazil, two closely related species of the Petunia genus are found (Fig. 1). These species share many morphological traits (Stehmann et al. 2009) and display close evolutionary relationships (Reck-Kortmann et al. 2014). Deeper differentiation between P. axillaris subsp. axillaris (Fig. 1C) and P. exserta (Fig. 1E) is translated into ecological aspects, ranging from pollination syndrome to habitat and geographical distribution. Petunia axillaris subsp. axillaris (hereafter P. axillaris) has a very broad distribution, occupying open areas in Pampas grasslands in Brazil, Argentina, and Uruguay (Turchetto et al. 2014a; b). P. exserta is endemic to the Serra do Sudeste, more specifically to the region known as Guaritas (Fig. 1B; Segatto et al. 2014), where both species occupy different microhabitats: P. axillaris occurs on the top and faces of arenitic towers, in sunny and open patches (Fig. 1D), while P. exserta is found inside small cavities (shelters) in those towers (Fig. 1F), with individuals growing totally protected from direct sunlight and rain. Both species present long and salverform (hypocrateriform) corolla tubes, erect growth habit, and yellow pollen (Stehmann et al. 2009), but they differ in morphological characteristics associated with their pollination syndromes, such as corolla color, scent production, and UV response. Petunia axillaris corollas are white, scented, and UV-absorbent, with reproductive organs included on the corolla, traits associated with hawkmoth-pollination (Venail et al. 2010; Klahre et al. 2011); flowers of P. exserta are red, nonfragrant, UV-reflectant and have exserted stamens, characteristics usually attributed to hummingbird-pollination syndrome (Lorenz-Lemke et al. 2006).

Figure 1
A. Landscape of Serra do Sudeste / Guaritas (Brazil). Arrows indicate arenitic towers that characterise the region; B. Schematic representation of 12 towers indicating 30 Petunia exserta (black) and 23 P. axillaris (blue) populations studied; crosses indicate populations from which open-pollination seeds were sampled; C, D. Petunia axillaris morphology and environment; E, F. Petunia exserta corolla flower, individual, and habitat inside shelter.

Despite different floral morphology and pollinators, individuals with intermediary corolla color and other floral traits are found inhabiting the shelter side-by-side with typical P. exserta plants. Several studies have been conducted in Guaritas seeking to understand the origin of those individuals and their impact on typical lineages of P. axillaris and P. exserta. Plastid markers determined via a phylogeographical approach revealed high polymorphism-sharing between both species but pointed to an elevated risk of extinction to P. exserta because of introgression (Lorenz-Lemke et al. 2006). Combining plastid with nuclear markers in an enlarged sample compared to this previous work’s results showed that despite hybridization, the species and especially P. exserta probably preserve their boundaries through a certain limiting of gene exchange or reproductive strategy (Segatto et al. 2014). Interspecific hybridization was confirmed for a small number of intermediary colored individuals (Turchetto et al. 2015b), whereas at least for P. axillaris from the contact zone, high levels of inbreeding were suggested as an effective barrier against gene exchange with P. exserta in Guaritas (Turchetto et al. 2015a).

Moreover, molecular characterization of genes and processes involved in floral syndromes shifts among Petunia species in general and P. axillaris and P. exserta species in particular (Amrad et al. 2016; Sheehan et al. 2016; Esfeld et al. 2018; Rodrigues et al. 2018), demonstrating the importance of a prezygotic barrier to reproductive isolation to maintain the species limits of this group. On the other hand, strong evidence of interspecific hybridization between P. axillaris and P. exserta was also obtained through molecular and morphological characterization, mostly at two sites in Guaritas (Turchetto et al. 2019; LM Caballero-Villalobos unpubl. res.; MC Teixeira unpubl. res.; CK Schnitzler unpubl. res.), confirming introgression in both species.

The role that hybridization plays in the evolutionary history of different taxa is variable and may appear as periods of introgression, hybrid speciation, and/or adaptive introgression (Taylor & Larson 2019). What happens after hybridization occurs is dependent on reproductive barriers between hybrid individuals and their parental taxa. There is widespread evidence for hybridization between species pairs with contrasting mating systems, but the consequences of introgression for each of these members is related to the potential of recurrent backcrossing of hybrids to the parents to allow genetic exchange and, at least for animal-pollinated species, if hybrid self-fertilization increases (Jordan et al. 2017) and stable populations are established.

To evaluate the extension of gene flow and introgression as well as species boundaries between P. axillaris and P. exserta in their cooccurrence area, we sampled populations covering the entire P. exserta distribution and all populations of P. axillaris found in Guaritas, and investigated genetic polymorphism based on microsatellites. First, we ascertained the presence of hybrid plants through indirect methods, then we tested for different hybrid classes and purebred individuals based on Bayesian assignments. We also applied direct methods to estimate gene flow through pollen movement between species in the same reproductive season.

Materials and methods

Plant material

We collected fresh leaves from 720 adult individuals of Petunia axillaris (Lam.) Britton, Sterns and Poggenb (252 individuals from 23 populations) and P. exserta Stehmann (468 individuals from 30 populations) in Guaritas, Serra do Sudeste (Tab. 1; Fig. 1B). We used a Global Positioning System (GPS) to obtain geographic coordinates per population, and because of the proximity among individuals, we recorded the geographic coordinates only per population per species (Tab. 1). All individuals were collected in the same phenophase at the same time during the spring in 2011 (September to December).

Table 1
Sampling information: geographical coordinates of collection sites, sample identity, and number of analysed individuals of P. exserta and P. axillaris. In bold - populations from which mother-plants were collected; N - number of adult individuals; n - number of mother-plants. Population codes follow Figure 1.

Sp	Pop Code	N	n	Geographical Coordinates
Petunia exserta	GE01	22	2	30°49’52.19”S 53°30’11.65”W
	GE02	4		30°49’54.07”S 53°30’12.39”W
	GE03	5	1	30°49’54.11”S 53°30’10.89”W
	GE04	9	1	30°49’53.21”S 53°30’10.87”W
	GE05	9		30°49’51.27”S 53°30’09.02”W
	GE06	21		30°49’50.49”S 53°30’08.53”W
	PF01	17		30°49’55.13”S 53°29’59.78”W
	PF02	14	4	30°50’00.12”S 53°29’52.25”W
	AF01	40	2	30°49’56.23”S 53°29’46.39”W
	ER01	3		30°50’04.22”S 53°30’03.16”W
	ER02	2		30°50’04.89”S 53°30’00.61”W
	ER03	4	1	30°50’06.12”S 53°30’01.85”W
	GRa01	2		30°50’18.43”S 53°29’43.09”W
	GRa02	4		30°50’18.90”S 53°29’42.35”W
	AB01	14		30°50’14.09”S 53°29’54.72”W
	AB02	1		30°50’17.97”S 53°29’57.96”W
	BO01	4		30°50’22.59”S 53°30’07.18”W
	BO02	17	2	30°50’22.92”S 53°30’12.59”W
	BO03	17	1	30°50’25.82”S 53°30’19.19”W
	BO04	4		30°50’23.91”S 53°30’18.14”W
	BO05	17	2	30°50’23.63”S 53°30’23.48”W
	BO06	22	1	30°50’19.00”S 53°30’16.18”W
	ASS01	49		30°50’13.76”S 53°30’15.04”W
	ASS02	22	1	30°50’13.84”S 53°30’16.67”W
	CO1	34		30°50’10.25”S 53°30’16.48”W
	CO2	18		30°50’13.76”S 53°30’15.04”W
	CO3	10		30°50’13.11”S 53°30’19.23”W
	TO01	12		30°50’12.17”S 53°30’22.49”W
	TO02	62	1	30°50’12.42”S 53°30’22.48”W
	TO03	9		30°50’14.61”S 53°30’23.96”W
Total		468	19
Petunia axillaris	CO1	1		30°83’66.41”S 53°50’50.14”W
	TO1	12	2	30°83’74.97”S 53°50’68.97”W
	TO2	8		30°83’76.58”S 53°50’69.69”W
	TO3	14		30°83’72.22”S 53°50’66.67”W
	GA1	8	1	30°83’42.54” 53°50’48.23”W
	GA2	29		30°83’43.92”S 53°50’52.62”W
	BO1	8		30°83’86.12”S 53°50’44.96”W
	BO2	5		30°83’89.70”S 53°50’32.59”W
	BO3	18	1	30°83’97.27”S 53°50’52.92”W
	BO4	4		30°83’82.36”S 53°50’26.69”W
	AF1	3		30°83’10.29”S 53°49’56.35”W
	AF2	1		30°83’22.85”S 53°49’62.20”W
	PF1	1		30°83’28.85”S 53°49’84.07”W
	PF2	25	3	30°83’28.61”S 53°49’83.28”W
	GR1	8		30°83’82.26”S 53°49’50.84”W
	GR2	27	1	30°83’83.71”S 53°49’51.15”W
	ER1	4	1	30°83’42.77”S 53°50’04.51”W
	ER2	45	3	30°83’42.94”S 53°50’02.03”W
	GE1	7		30°83’03.74”S 53°50’21.16”W
	GE2	7	2	30°83’07.88”S 53°50’22.83”W
	GE3	7		30°83’21.89”S 53°50’36.86”W
	GE4	5	1	30°83’12.53”S 53°50’32.77”W
	AB	5		30°83’72.64”S 53°49’93.77”W
Total		252	15

We also collected mature fruits of 15 plants of P. axillaris and 19 plants of P. exserta to sample open-pollinated progeny arrays in that reproductive season for these two species. One to three fruits per mother-plant were collected, and their seeds were pooled. Thirty seeds per mother-plant pool were randomly selected for cultivation. The seeds were pretreated for 24 h in the dark at 4 °C with a solution of 100 µM gibberellic acid (GA4; Sigma-Aldrich Co., St. Louis, USA) dissolved in 1 mL DMSO (dimethyl sulfoxide; Sigma-Aldrich) and diluted in water before planting to break any physiological dormancy associated with seasonal weather changes. The seeds were cultivated in a growth chamber under controlled temperature (25 °C) and luminosity (12-h light:12-h dark). Even after two rounds of germination, some mother-plants produced just a few seedlings that resulted in 285 seedlings of P. axillaris (16-21 per mother plant) and 326 progenies of P. exserta (4-21 per mother-plant).

DNA extraction and microsatellite genotyping

The leaves of each adult individual collected in nature and of progenies obtained in the growth chamber were dried in silica gel, ground in liquid nitrogen, and stored at -20 °C until processing. The genomic DNA was isolated using a CTAB (cetyl-trimethylammonium bromide)-based protocol (Roy et al. 1992). All samples (Tab. S1 in supplementary material) were genotyped with eight polymorphic microsatellite loci (PM8, PM21, PM167, PM173, PM177, PM188, PM192, and PM195) covering five of seven Petunia chromosomes (Bossolini et al. 2011). The selected loci are highly discriminant among Petunia species (Turchetto et al. 2015b) and were successfully used to study the breeding structure in P. axillaris (Turchetto et al. 2015a).

The PCRs were performed according Turchetto et al. (2015b). The DNA fragments were denatured and size-fractionated utilizing capillary electrophoresis on a MegaBACE 1000 automated sequencer (GE Healthcare Biosciences, Pittsburgh, USA). The manufacturer’s software was applied to determine the alleles per locus with manual inspections and reference samples for P. axillaris and P. exserta (Turchetto et al. 2015b) in each run. Genotyping errors from stutter bands, allele dropout, and null alleles were verified with Micro-Checker software (Oosterhout et al. 2004).

Genetic diversity in each species

Considering adult individuals of both species, we estimated the number of alleles, number of private alleles, allele richness, and inbreeding coefficient (F _IS ; Weir & Cockerham 1984) per loci via FSTAT (Goudet 1995). We also estimated observed and expected heterozygosity under Hardy-Weinberg equilibrium after Bonferroni correction and analysis of molecular variance (AMOVA; Excoffier et al. 1992) with 10 000 permutations using ARLEQUIN 3.5 (Excoffier & Lischer 2010).

Admixture analyses though indirect methods

To estimate the levels of hybridization, proportion and direction of introgression, we carried out three complementary analyses. First, we performed model-based Bayesian clustering employing Structure 2.3.2 software (Pritchard et al. 2000) to estimate admixture proportions among P. exserta and P. axillaris adult individuals from the entire sampled area (overall, 720 individuals considering both species). Then, we ran two separated analyses pooling all seedlings of each species (totaling 285 samples in P. axillaris and 326 progenies in P. exserta) and including all adults of both species per comparison to test the admixture proportion in a second generation of each species. Structure analyses were carried out according to an admixture model assuming correlated allelic frequencies, and no a priori population assignment was used in the analyses. Each run was conducted with 2.5 × 10⁵ burn-in periods and 10⁶ Markov chain Monte Carlo (MCMC) repetitions after burn-in. Three independent runs were performed per comparison, and the results were examined for convergence across runs. We identified the number of gene pools through the most likely number of genetic clusters obtained using the maximum value of ΔK (Evanno et al. 2005) as implemented in Structure Harvester 0.6.93 (Earl & Holdt 2012) and the estimate of Pr(X|K) from Structure. Pophelper, an online R package (Francis 2017), was utilized to summarize the output of the optimal K-value from Structure and generate the bar plots. Further, Structure was used to classify individuals as parental species or hybrid. A threshold of Q value > 0.10 (individuals of P. exserta that displayed at least Q = 0.10 of a genetic component found in P. axillaris and individuals of P. axillaris that presented at least Q = 0.10 of a genetic component found in P. exserta) was leveraged to identify hybrid plants. Based on the fact most molecular markers were in HWE disequilibrium (Tab. 2), we also ran multivariate discriminate analyses of principal components (DAPC; Jombart et al. 2010) as implemented in Adegenet of R 3.3.0 (R Development Core Team 2016) based on microsatellite data including all adults of both species and their respective progenies. These analyses were performed to identify genetic clusters within the data set and compare to the Structure results from K = 2 (two species). This method makes no assumption about data structure or underlying population genetic model.

Table 2
Diversity indices based on eight microsatellite per species. N - number of alleles; E - number of private alleles; R - allele richness; H _E - expected heterozygosity; H _O - observed heterozygosity; F _IS - inbreeding coefficient; * - significant HWE values after Bonferroni correction.

Sp		PM8	PM21	PM173	PM167	PM192	PM188	PM195	PM177	Average
P. axillaris	N	6	7	17	11	12	9	7	24	11.63
	E	1	2	10	1	2	0	3	6	3.13
	R	6.00	6.98	16.97	11.00	12.00	9.00	7.00	23.99	11.62
	H E	0.57	0.61	0.56	0.82	0.77	0.79	0.54	0.91	0.70
	H O	0.31*	0.31*	0.32*	0.52*	0.42*	0.48*	0.24*	0.40*	0.38
	F IS	0.46	0.49	0.42	0.36	0.45	0.39	0.57	0.56	0.46
P. exserta	N	6	5	8	11	13	9	8	21	10.13
	E	1	0	1	1	1	0	2	3	1.13
	R	5.52	4.68	7.35	9.74	12.42	8.56	7.12	18.99	9.30
	H E	0.23	0.50	0.75	0.67	0.84	0.35	0.35	0.82	0.57
	H O	0.08*	0.16*	0.27*	0.24*	0.24*	0.12*	0.10*	0.25*	0.18
	F IS	0.72	0.68	0.64	0.65	0.71	0.67	0.73	0.70	0.68

To infer the genetic composition of hybrids, we performed a Bayesian clustering analysis implemented in Newhybrids 1.1 software (Anderson & Thompson 2002), which assigns individuals into different genotypic classes (parental species, F₁, F₂ and back-crosses with each parental species). The analyses were run considering adult individuals and families independently. For the families’ analyses, we included mother-plants as representative of each species’ genetic constitution. For each comparison, we ran two independent analyses using Jeffrey’s priors with uniform priors that included 10⁵ steps as burn-in followed by 10⁶ MCMC interactions performed to assure the convergence of chains and homogeneity across runs. Analyses were performed without previous information on populations or taxonomic identity (Vähä & Primmer 2006).

Direct measures of gene flow between P. exserta and P. axillaris

We performed parentage analyses to estimate the pollen flow between P. exserta and P. axillaris. To determine the most likely pollen donors for seedlings of P. exserta, we included all adult individuals of both species as pollen donor candidates (720 individuals of both species growing at the same site and sampled in the same flowering season) and the families derived from P. exserta mother plants. Additionally, we re-analyzed the paternity assignments of P. axillaris’ 285 seedlings from Turchetto et al. (2015a), including all P. exserta adults, as potential pollen donors.

We employed the Bayesian method to conduct paternity analysis with Cervus 3.0.6 (Marshall et al. 1998; Kalinowski et al. 2007). To determine the likely pollen donors for seedlings of each species, all adult individuals found in the same season were used as a pollen donor candidate for each offspring in equal probability. As selfing is possible, mother plants were also considered father candidates. To define a critical LOD (logarithm of likelihood ratios) score, we ran the simulation of paternity analysis using the allele frequencies calculated for adults as a reference and thus we estimated the Δ statistic (critical Δ; Marshall et al. 1998). Afterwards, we conducted a parentage analysis to assign the father candidate to each offspring. The most likely pollen donor was evaluated relative to the critical Δ scores as determined in previous simulations. The individual with the highest Δ value was accepted as father of the seedling when the difference between its LOD score and the LOD score of the second most likely candidate was above the critical Δ (threshold value). For parentage analysis, the following parameters were used: 10 000 repetitions; 0.9840 proportion of typed loci; and 99 % (strict) and 95 % (relaxed) levels of confidence. We considered 90 % of parents sampled in the area and a 1 % genotyping error. The minimum number of loci required to determine the paternity of a seedling was fixed at six.

Results

Genetic diversity

The eight microsatellite loci produced 93 alleles in P. axillaris and 81 alleles in P. exserta. Most markers presented private alleles in each species. The means of allele richness were higher in P. axillaris than P. exserta (Student’s t-test, P < 0.05). The two species presented high and positive F _IS values. Almost all loci were in Hardy-Weinberg disequilibrium displaying homozygote excess (Tab. 2). AMOVA analyses revealed a genetic differentiation between species (20.34 %, P < 0.001). Both species presented positive and elevated F _IS values, suggesting high levels of inbreeding that were greater in P. exserta (Student’s t-test, P < 0.05).

Genetic composition of species and hybrids

The best K = 2 was obtained in Structure Bayesian clustering analysis according to the criterion of Evanno et al. (2005), which grouped individuals of each species (Fig. 2). Most adult individuals of P. axillaris and P. exserta presented Q values > 0.90, but ca. 2 % of individuals previously assigned as P. axillaris (six among 252) or P. exserta (12 among 468) were now identified as hybrids and had Q values > 0.10. Two populations of P. exserta had the highest number of hybrid individuals (GE01 and CO2, with five and three putative hybrids, respectively), and four populations had only one hybrid individual (GE06, AF01, BO02, and CO3). Population CO2 was the only shelter in which we detected a few individuals with color variation, thereby suggesting hybridization through phenotype (data not shown). One mother plant from GE01 was established as a hybrid based on its genetic profile (E1630). Four populations of P. axillaris had one hybrid individual (GR1, BO3, CO1, and TO1), and the GA2 population had two hybrid individuals. No P. axillaris mother plants were identified as hybrids based on their genotypes, and no indication of hybrid morphological phenotype was observed among adults of this species.

Figure 2
Structure bar plot under admixture coefficients model based on eight microsatellite loci and 720 adult individuals of Petunia exserta and P. axillaris. Each bar represents individuals and black-dotted vertical lines indicate each population; different colours correspond to genetic components (K = 2) and individuals’ membership.

We detected the P. axillaris genetic component in three families of P. exserta (Fig. S1 in supplementary material): two from the GE01 population, from which mother plants also showed the genetic component of P. axillaris, and one from the PF02 population. Among the progenies of P. axillaris (Fig. S2 in supplementary material) we found several individuals presenting the genetic component of P. exserta, especially from two populations (GR2 and GE2).

In the DAPC analysis (Fig. 3), we observed two clusters of individuals as observed in the Structure analyses. Several individuals among adults presented genetic component of both clusters. We observed P. axillaris’ genetic component in progenies of P. exserta from one family (FM_05). However, five progenies of P. axillaris from different families presented mixed genetic components. Several individuals among adults and progenies featured genetic components of both clusters, whereas others had genetic profiles not concordant with their morphological group or spatial distribution (floral morphology corresponding to a given species as well as position relating to shelters, but genetic profile as another species).

Figure 3
Clustering of individuals as revealed in DAPC analysis (K = 2) performed based on microsatellite genotypes of adults and progenies of P. exserta and P. axillaris. Black-dotted lines separate individuals’ groups (each species adults or progenies); each vertical bar corresponds to an individual. Different colours indicate membership probability of each species per individual.

The NewHybrids analyses (Fig. 4) revealed the different classes of crosses among adults as well as progenies, with inbreeding being the most frequent type in both species. Among adults, the second most frequent class was F₂, whereas backcrosses with P. axillaris were high among P. axillaris offspring. Backcrosses with P. axillaris were also observed among P. exserta progenies, and some individuals of this offspring were classified as P. axillaris according their microsatellite profiles. Despite the low frequency, F₁ individuals were observed among adults and progenies. In the family of one P. axillaris mother plant (A1809) from the GRA2 population, we, observed four individuals as a backcross with P. exserta (A2072, A2082, A2084, and A2091).

Figure 4
Posterior probabilities (q) for all analysed plants using NewHybrids assigned to six classes following the legend for pure parental species (P. axillaris or P. exserta), F₁, F₂, backcrosses (BC) with P. exserta, and backcrosses with P. axillaris. (A) Adults and (B) progenies of each species. Vertical black dotted lines separate species or progenies.

Pollen flow between P. exserta and P. axillaris

In the paternity analysis of the 326 P. exserta offspring, considering all P. exserta (468) and P. axillaris (252) adult plants as pollen donor candidates, no P. axillaris individual was assigned as the most likely father (critical value Δ = 2.95; 95 % confidence interval). When we considered a relaxed confidence interval (critical value Δ = 0.69; 85 % confidence interval), two offspring of EM1624 had one P. axillaris plant indicated as pollen donor (from the BO2 population) and one offspring of EM1630 had equal LOD score values for one P. exserta and one P. axillaris individual as the most likely pollen donor (both EM1624 and EM1630 are from the GE01 population of P. exserta). These two mother plants (EM1624 and EM1630) and all their offspring were placed in the P. axillaris cluster through Structure analysis (Fig. 2 and Fig. S1 in supplementary material). For P. axillaris, two of 285 seedlings (mother plant from the GRA2 population) had P. exserta as the most likely pollen donor, considering the strict confidence interval (critical value Δ = 2.95; 95 % confidence interval). Furthermore, considering the relaxed confidence level (critical value Δ = 0.69; 85 % confidence interval), eight seedlings from the same population (GRA2) had one P. exserta as a pollen donor coming from five different populations of P. exserta (CO2, GE06, ASS02, GE01 and PF01). Four of these progenies were also indicated in the NewHybrids analyses as resulting from backcrosses with P. exserta. Unlike P. exserta, the GRA2 mother plant was assigned to the cluster of P. axillaris within the Structure analyses that included all P. axillaris offspring and some of the offspring presenting mixed ancestry (Fig. S2 in supplementary material). These results showed that although pollen flow between P. axillaris and P. exserta is rare, it happens in both directions. In this reproductive season, the interspecific pollen flow occurred more frequently from P. exserta to P. axillaris. However, in the Bayesian analysis of adult individuals, hybrid ancestry was observed more frequently among P. exserta individuals (Fig. 2). These results suggested that fertilization between these species failed when pollen flows from P. exserta to P. axillaris or that the hybrid progeny has a lower success of developing until the adult stage when it is in the P. axillaris microhabitat. Only a few adult individuals were identified as hybrids based on the genetics and no morphological intermediary phenotypes were found among P. axillaris populations. By contrast, some hybrid individuals, as indicated by genetic markers and intermediary corolla color, were collected growing in the P. exserta microhabitat and might effectively reproduce primarily by self-fertilization, as observed with the mother plant from the GE1 population that had a hybrid origin (Fig. 2).

Discussion

Herein, we estimated gene exchange in a contact zone between two closely related Petunia species that, despite interspecific hybridization, are preserving their species boundaries. Gene flow is low but introgression occurs in both directions and, contrary to previously thought (Lorenz-Lemke et al. 2006), introgression is also high among individuals of P. axillaris. Hybrids are predominantly found in the same microenvironment as P. exserta (inside shelters), although individuals with both genetic components exist among P. axillaris plants, in sunny and opened fields. In this latter case, hybrids present the same morphology as canonical P. axillaris, whereas when they are observed inside shelters, they usually exhibit intermediary corolla color. The direct and indirect estimates of gene exchange between these two species revealed high levels of inbreeding in both, with backcrosses with the spatially closer parental being more frequent than other kinds of crossing.

These two Petunia species present suites of floral traits strongly related to their respective floral syndromes, hawkmoth in P. axillaris and hummingbirds in P. exserta (Stehmann et al. 2009). No systematic pollination studies have been conducted in the field to identify which pollinators the intermediary colored plants attract and, except their pink shades that usually attract bees in Petunia species (Rodrigues et al. 2018), other clues on pollinators remain unidentified.

Hummingbirds were observed visiting natural populations of P. exserta (Lorenz-Lemke et al. 2006). Moreover, these birds were also described as visitors in P. axillaris populations in Uruguay (Gübitz et al. 2009). Floral morphology and spatial distribution of individuals of both species in Guaritas suggest that these animals could be responsible for pollen exchange among these individuals. Pollinators tend to forage on nearby flowers when a flower offers rewards (Tremblay et al. 2005) and especially in the absence of other nectar sources. Hummingbirds, in general, can differ in their foraging strategies and degree of specialization regarding floral resources.

Many species of these birds feed along trapping lines following repeated foraging circuits among successive flowers or clumps (Snow & Snow 1972; Colwell 1973; Linhart 1973). Furthermore, the circuit may be long distance and high reward by following a regular route and particular sequence of plants (Feinsinger 1983). Other groups of hummingbirds exhibit greater variety regarding degree of specialization, and most species are territorial only during the bloom peak when flowers are densely packed (Maglianesi et al. 2015b). Although these Petunia species usually bloom in the same period without seasonal isolation (Stehmann et al. 2009), field observations in Serra do Sudeste revealed a certain level of difference in the flowering time between these cooccurring species, with P. axillaris usually blooming earlier than P. exserta, along with some overlaps (MC Teixeira, unpubl. res.). In early spring, and afterwards the reverse, many individuals of P. axillaris can be found with flowers, whereas few individuals of P. exserta are flowering.

The foraging preferences of hummingbird species involve the morphological matching between the bill and floral traits and are also a consequence of resource abundance (Maglianesi et al. 2015a; b). In fact, specialized interactions have been observed in many plant-hummingbird pairs and are implicated as important drivers of plant diversification in the Neotropics (Nunes et al. 2016; Serrano-Serrano et al. 2017). Petunia exserta and P. axillaris present several floral traits in common (Stehmann et al. 2009), such as a long and salverform corolla tube with nectar chambers at the base, which also exist in hybrid individuals. Nectar volume and sugar concentration vary among Petunia species; characteristics of nectar from P. exserta flowers are not known, but this species displays other traits related to hummingbirds (Gübitz et al. 2009; Stehmann et al. 2009). The nectar volume and sugar concentration in P. axillaris (Gleiser et al. 2014) are similar to those described for bird-pollinated species (Baker 1975; Proctor et al. 1996).

Selfing and backcrosses are strategies that can reinforce barriers against the gene flow between species and therefore contribute to species cohesion (Kamran-Disfani & Agrawal 2014; Segatto et al. 2014). Here, we found that pollination was one primary barrier to gene exchange between these two cooccurring species, as inbreeding was the most frequent mating system. This prezygotic barrier can be associated with the specific interaction between pollinators and morphological traits of each species (Huber et al. 2005) and may be implicated in adaptation to microenvironmental conditions (Hu 2015) with selection against hybrids in one or another parental environment (Toews & Brelsford 2012), such as the high amount of backcrosses observed, especially among P. axillaris progenies that grow in sunny and open patches, ecological conditions totally different from the environment inside shelters. Although such prezygotic barriers can prevent interspecific pollen flow, it was permeable, as indicated by the events of interspecific hybridization among progenies of both species, which were also observed among adults of both species and previously hypothesized (Lorenz-Lemke et al. 2006; Segatto et al. 2014; Turchetto et al. 2015a).

Seed dispersal can also influence the genetic structure and pollen movement within populations of these two species because in Petunia species, seeds usually fall near the mother plant (Pijl 1982), and thus each population inside shelters as well as in open areas mostly encompasses related individuals. Self-fertilization and biparental inbreeding rates observed might be a consequence of pollinator behavior, with behavior influenced by the local abundance of flowers of each species in each tower.

Speciation can be thought of as a continuous process (Nosil & Feder 2012). Plant species are typically isolated not by a single factor but by a large number of different barriers in complex interactions (Widmer et al. 2009) that act in a hierarchical order (Dell’Olivo et al. 2011). The Petunia genus is a young group of endemic species that has low genetic differentiation (Reck-Kortmann et al. 2014) and underwent allopatric speciation during the Pleistocene (Lorenz-Lemke et al. 2010); microenvironmental and pollinator diversity have also been suggested as important factors driving their diversification (Fregonezi et al. 2013). In the current work, we put forth several results that support these statements for two of these species, which are found in sympatry, share many morphological and genetic polymorphisms and have interspecific hybridization as a strong driver of diversification.

Acknowledgements

The authors thank J.N. Fregonezi, A.M.C. Ramos-Fregonezi, G. Mäder, and A.L.A. Segatto for assistance during field collection; J.R. Stehmann for help with taxonomic identification; and J. Beduschi for help in DNA extraction. This project was supported by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES). C.T. was supported by PNPD-CAPES/PPGBM, UFRGS.

References

• Abbott RJ, Barton NH, Good JM. 2016. Genomics of hybridization and its evolutionary consequences. Molecular Ecology 25: 2325-2332.
• Amrad A, Moser M, Mandel T, et al 2016. Gain and loss of floral scent production through changes in structural genes during pollinator-mediated speciation. Current Biology 26: 3303-3312.
• Anderson EC, Thompson EA. 2002. A model-based method for identifying species hybrids using multilocus genetic data. Genetics 160: 217-1229.
• Arnold ML. 2006. Evolution through genetic exchange. New York, Oxford University Press.
• Baker H. 1975. Sugar concentration in nectars from hummingbird flowers. Biotropica 7: 37-41.
• Bossolini E, Klahre U, Brandenburg A, Reinhardt D, Kuhlemeier C. 2011. High resolution linkage maps of the model organism Petunia reveal substantial synteny decay with the related genome of tomato. Genome 54: 327-340.
• Colwell RK. 1973. Competition and coexistence in a simple tropical community. The American Naturalist 107: 737-760.
• Dell’Olivo A, Hoballah ME, Gübitz T, Kuhlemeier C. 2011. Isolation barriers between Petunia axillaris and Petunia integrifolia (Solanaceae). Evolution 65: 1979-1991.
• Earl EA, Holdt BM. 2012. Structure Harvester: a website and program for visualizing Structure output and implementing the Evanno method. Conservation Genetics Research 4: 359-361.
• Esfeld K, Berardi AE, Moser M, Bossolini E, Freitas L, Kuhlemeier C. 2018. Pseudogenization and resurrection of a speciation gene. Current Biology 28: 3776-3786.
• Evanno G, Regnaut S, Goudet J. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: A simulation study. Molecular Ecology 14: 2611-2620.
• Excoffier L, Lischer HEL. 2010. Arlequin suite ver 3.5: A new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources 10: 564-567.
• Excoffier L, Smouse PE, Quattro JM. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491.
• Feinsinger P. 1983. Coevolution and pollination. In: Futuyma DJ, Slatkin M. (eds.) Coevolution. Sunderland, Sinauer Associates. p. 282-310.
• Francis RM. 2017. POPHELPER: an R package and web app to analyse and visualize population structure. Molecular Ecology Resources 17: 27-32.
• Fregonezi JN, Turchetto C, Bonatto SL, Freitas LB. 2013. Biogeographical history and diversification of Petunia and Calibrachoa (Solanaceae) in the Neotropical Pampas grassland. Botanical Journal of the Linnean Society 171: 140-153.
• Gleiser G, Internicola AI, Austerlitz F, Bernasconi G. 2014. Stabilizing selection on nectar concentration in wild Petunia axillaris, as revealed by genetic analysis of pollen dispersal. Evolutionary Ecology 28: 869-884.
• Goudet J. 1995. FSTAT version 1.2: A computer program to calculate F-statistics. Journal of Heredity 86: 485-486.
• Gübitz T, Hoballah ME, Dell’Olivo A, Kuhlemeier C. 2009. Petunia as a model system for the genetics and evolution of pollination syndromes. In: Gerats T, Strommer J. (eds) Petunia: evolutionary, developmental and physiological genetics. New York, Springer. p. 29-49.
• Hu X-S. 2015. Mating system as a barrier to gene flow. Evolution 69: 1158-1177.
• Huber FK, Kaiser R, Sauter W, Schiestl FP. 2005. Floral scent emission and pollinator attraction in two species of Gymnadenia (Orchidaceae). Oecologia 142:564-575.
• Jombart T, Devillard S, Balloux F. 2010. Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genetics 11: 94.
• Jordan CY, Lohse K, Turner F, Thomson M, Gharbi K, Ennos RA. 2017. Maintaining their genetic distance: Little evidence for introgression between widely hybridizing species of Geum with contrasting mating systems. Molecular Ecology 27: 1214-1228.
• Kalinowski ST, Taper ML, Marshall TC. 2007. Revising how the computer program CERVUS accommodates genotyping error increases success in paternity assignment. Molecular Ecology 16: 1099-1106.
• Kamran-Disfani A, Agrawal AF. 2014. Selfing, adaptation and background selection in finite populations. Journal of Evolutionary Biology 27: 1360-1371.
• Kenney AM, Sweigart A. 2016. Reproductive isolation and introgression between sympatric Mimulus species. Molecular Ecology 25: 2499-2517.
• Klahre U, Gurba A, Hermann K, et al 2011. Pollinator choice in Petunia depends on two major genetic loci for floral scent production. Current Biology 21: 730-739.
• Linhart YB. 1973. Ecological and behavioral determinants of pollen dispersal in hummingbird-pollinated Heliconia The American Naturalist 107: 511-523.
• Lorenz-Lemke AP, Mäder G, Muschner VC, et al 2006. Diversity and natural hybridization in a highly endemic species of Petunia (Solanaceae): a molecular and ecological analysis. Molecular Ecology 15: 4487-4497.
• Lorenz-Lemke AP, Togni PD, Mäder G, et al 2010. Diversification of plant species in a subtropical region of eastern South American highlands: a phylogeographic perspective on native Petunia (Solanaceae). Molecular Ecology 19: 5240-5251.
• Maglianesi MA, Blüthgen N, Böhning-Gaese K, Schleuning M. 2015a. Functional structure and specialization in three tropical plant-hummingbirds interaction networks across an elevational gradient in Costa Rica. Ecography 38: 1119-1138.
• Maglianesi MA, Böhning-Gaese K, Schleuning M. 2015b. Different foraging preferences of hummingbirds on artificial and natural flowers revels mechanisms structuring plant-pollinator interactions. Journal of Animal Ecology 84: 655-664.
• Marshall TC, Slate J, Kruuk LEB, Pemberton JM. 1998. Statistical confidence for likelihood-based paternity inference in natural populations. Molecular Ecology 7: 639-655.
• Meier JI, Marques DA, Mwaiko S, Wagner CE, Excoffier L, Seehausen O. 2016. Ancient hybridization fuels rapid cichlid fish adaptive radiations. Nature Communications 8: 14363. doi: 10.1038/ncomms14363
  » https://doi.org/10.1038/ncomms14363
• Nosil P, Feder JL. 2012. Genomic divergence during speciation: causes and consequences. Philosophical Transactions of the Royal Society B: Biological Sciences 367: 332-342.
• Nunes CF, Amorim FW, Mayer JL, Sazima M. 2016. Pollination ecology of two species of Elleanthus (Orchidaceae): Novel mechanisms and underlying adaptations to hummingbird pollination. Plant Biology 18: 15-25.
• Oosterhout C, Hutchinson WF, Willis DPM, Shipley P. 2004. Micro-checker: software for identification and correcting genotyping errors in microsatellite data. Molecular Ecology Notes 4: 535-538.
• Pijl L. 1982. Principles of dispersal in higher plants. Berlin, Springer-Verlag.
• Pritchard JK, Stephens M, Donnelly P. 2000. Inference of population structure using multilocus genotype data. Genetics 155: 945-959.
• Proctor M, Yeo P, Lack A. 1996. The natural history of pollination. Portland, Timber Press.
• R Development Core Team. 2016. R: A language and environment for statistical computing. Vienna, R Foundation for Statistical Computing. http://www.R project.org/
  » http://www.R project.org/
• Reck-Kortmann M, Silva-Arias GA, Segatto ALA, Mäder G, Bonatto SL, Freitas LB. 2014. Multilocus phylogeny reconstruction: new insights into the evolutionary history of the genus Petunia Molecular Phylogenetics and Evolution 81: 19-28.
• Rodrigues DM, Caballero-Villalobos LM, Turchetto C, Jacques RA, Kuhlemeier C, Freitas LB. 2018. Do we truly understand pollination syndromes in Petunia as much as we suppose? AoB Plants 10: ply057. doi: 10.1093/aobpla/ply057
  » https://doi.org/10.1093/aobpla/ply057
• Roy A, Frascaria N, MacKay J, Bousquet J. 1992. Segregating Random Amplified Polymorphic DNAs (RAPDs) in Betula alleghaniensis Theoretical and Applied Genetics 85: 173-180.
• Segatto ALA, Cazé ALR, Turchetto C, et al 2014. Nuclear and plastid markers reveal the persistence of genetic identity: A new perspective on the evolutionary history of Petunia exserta Molecular Phylogenetics and Evolution 70: 504-512.
• Serrano-Serrano ML, Rolland J, Clark JL, Salamin N, Perret M. 2017. Hummingbird-pollinated and the diversification of angiosperms: an old and successful association in Gesneriaceae. Proceedings of the Royal Society B-Biological Sciences 284: 20162816. doi: 10.1098/rspb.2016.2816
  » https://doi.org/10.1098/rspb.2016.2816
• Sheehan H, Moser M, Klahre U, et al 2016. MYB-FL controls gain and loss of floral UV absorbance, a key trait affecting pollinator preference and reproductive isolation. Nature Genetics 48: 159-166.
• Snow BK, Snow DW. 1972. Feeding niches of hummingbirds in a Trinidad valley. Journal of Animal Ecology 41: 471-485.
• Soltis PS, Soltis DE. 2009. The role of hybridization in plant speciation. Annual Review in Plant Biology 60: 561-588.
• Stehmann JR, Lorenz-Lemke AP, Freitas LB, Semir J. 2009. The genus Petunia In: Gerats T, Strommer J. (eds) Petunia evolutionary, developmental and physiological genetics. New York, Springer . p. 1-28.
• Taylor SA, Larson EL. 2019. Insights from genomes into the evolutionary importance and prevalence of hybridization in nature. Nature Ecology and Evolution 3: 170-177.
• Toews DPL, Brelsford A. 2012. The biogeography of mitochondrial and nuclear discordance in animals. Molecular Ecology 21: 3907-3930.
• Tremblay RL, Ackerman JD, Zimmerman JK, Calvo RN. 2005. Variation in sexual reproduction in orchids and its evolutionary consequences: a spasmodic journey to diversification. Biological Journal of Linnean Society 84: 1-54.
• Turchetto C, Fagundes NJR, Segatto ALA, et al 2014a. Diversification in the South American Pampas: the genetic and morphological variation of the widespread Petunia axillaris complex (Solanaceae). Molecular Ecology 23: 374-389.
• Turchetto C, Segatto ALA, Telles MPC, Diniz-Filho JAF, Freitas LB. 2014b. Infraspecific classification reflects genetic differentiation in the widespread Petunia axillaris complex: a comparison among morphological, ecological, and genetic patterns of geographic variation. Perspectives in Plant Ecology, Evolution and Systematics 16: 75-82.
• Turchetto C, Lima JS, Rodrigues DM, Bonatto SL, Freitas LB. 2015a. Pollen dispersal and breeding structure in a hawkmoth-pollinated Pampa grasslands species Petunia axillaris (Solanaceae). Annals of Botany 115: 939-948.
• Turchetto C, Segatto ALA, Bedushi J, Bonatto SL, Freitas LB. 2015b. Genetic differentiation and hybrid identification using microsatellite markers in closely related wild species. AoB Plants 7: plv084. doi: 10.1093/aobpla/plv084
  » https://doi.org/10.1093/aobpla/plv084
• Turchetto C, Segatto ALA, Silva-Arias GA, et al 2019. Contact zones and their consequences: Hybridization between two ecologically isolated wild Petunia species. Botanical Journal of the Linnean Society 190: 421-435.
• Vähä JP, Primmer CR. 2006. Efficiency of model-based Bayesian methods for detecting hybrid individuals under different hybridization scenarios and with different numbers of loci. Molecular Ecology 15: 63-72.
• Venail J, Dell'Olivo A, Kuhlemeier C. 2010. Speciation genes in the genus Petunia Philosophical Transactions of the Royal Society B: Biological Sciences 365: 461-468.
• Weir BS, Cockerham CC. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370.
• Widmer A, Lexer C, Cozzolino S. 2009. Evolution of reproductive isolation in plants. Heredity 102: 31-38.
• Yan LJ, Burgess KS, Milne R, Fu CN, Li DZ, Gao LM. 2017. Asymmetrical natural hybridization varies among hybrid swarms between two diploid Rododendron species. Annals of Botany 120: 51-61

Publication Dates

• Publication in this collection
  07 Oct 2019
• Date of issue
  Oct-Dec 2019

History

• Received
  05 Apr 2019
• Accepted
  08 July 2019