Clonality Strongly Affects the Spatial Genetic Structure of the Nurse Species Aechmea Nudicaulis (L.) Griseb. (Bromeliaceae)

bs_bs_banner

Botanical Journal of the Linnean Society, 2015, 178, 329–341. With 4 ﬁgures

Clonality strongly affects the spatial genetic structure of the nurse species Aechmea nudicaulis (L.) Griseb. (Bromeliaceae)

ROBERTA LOH1,2,3, FABIO RUBIO SCARANO2,4, MARCIO ALVES-FERREIRA1 and Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 FABIANO SALGUEIRO3*

1Laboratório de Genética Molecular Vegetal, Departamento de Genética, Centro de Ciências da Saúde, Instituto de Biologia, Universidade Federal do Rio de Janeiro, Av. Prof. Rodolpho Paulo Rocco s/n, prédio do CCS, Sala A2-93, Ilha do Fundão, Rio de Janeiro CEP.: 21949-900, RJ, Brazil 2Laboratório de Ecologia Vegetal, Departamento de Ecologia, Centro de Ciências da Saúde, Instituto de Biologia, Universidade Federal do Rio de Janeiro, Av. Prof. Rodolpho Paulo Rocco s/n, Ilha do Fundão, Rio de Janeiro CEP.: 21941-590, RJ, Brazil 3Grupo de Pesquisa em Biodiversidade Molecular Vegetal, Departamento de Botânica, Instituto de Biociências, Universidade Federal do Estado do Rio de Janeiro, Av. Pasteur n°458, sala 512, Urca, Rio de Janeiro CEP.: 22290-240, RJ, Brazil 4Conservation International, Rua Buenos Aires n°68, 26° andar, Centro, Rio de Janeiro CEP.: 20070-022, RJ, Brazil

Received 9 September 2014; revised 3 February 2015; accepted for publication 15 March 2015

Aechmea nudicaulis is a clonal bromeliad common to the Brazilian Atlantic forest complex and is found abundantly in the sandy coastal plain vegetation (restinga) on the north coast of Rio de Janeiro state, Brazil. This restinga site is structured in vegetation islands, and the species plays a key role as a nurse plant, much favoured by its clonality. We studied the clonal structure and consequences of clonality on the population spatial genetic structure (SGS) of this species using six nuclear microsatellites. Spatial autocorrelation analysis was performed to study the effects of sexual and clonal reproduction on the dispersal of A. nudicaulis. Analyses were performed at the genet (i.e. excluding clonal repeats) and ramet levels. Genotypic richness was moderate (R = 0.32), mostly as a result of the dominance of a few clones. The spatial distribution of genets was moderately intermingled, the mean clone size was 4.9 clonal fragments per genet and the maximum clonal spread was 25 m. Expected heterozygosities were high and comparable with those found in other clonal plants. SGS analyses at the genet level revealed signiﬁcantly restricted gene dispersal (Sp = 0.074), a strong SGS compared with other herbaceous species. The clonal subrange extended across 23 m where clonality had a signiﬁcant effect on SGS. The restricted dispersal and SGS pattern in A. nudicaulis, coupled with high levels of genetic diversity, indicated a recruitment at windows of opportunity (RWO) strategy. Moreover, the spatial distribution of genetic variation and the habitat occupation pattern in A. nudicaulis were dependent not only on the intrinsic biological traits of the species (such as spacer size and mating system), but also on biotic interactions with neighbouring species that determined suitable habitats for germination and the establishment of new genets. © 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 178, 329–342.

ADDITIONAL KEYWORDS: Atlantic forest – clonal growth – gene ﬂow – microsatellite – restinga – SGS.

INTRODUCTION development of a single zygote and consists of a group of genetically identical semi-autonomous units, Clonal plants are characterized by a hierarchical called ramets (Harper, 1977). Life history traits are organization, in which each genet is the product of the markedly different between clonal and non-clonal plants (Honnay & Jacquemyn, 2008). For instance, *Corresponding author. E-mail: mobility through clonal growth is limited in com- [email protected]. parison with seed dispersal (Eriksson, 1993), and

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 178, 329–341 329 330 R. LOH ET AL. horizontal spread by ramet propagation affects the clonal structure and clonal subrange, which represents spatial distribution of genets, influencing levels of the spatial scale beyond which clonality does not affect dispersal and population structure. It follows that the genetic structure (Alberto et al., 2005; Vallejo- spatial genetic structure (SGS) in clonal plants results Marín et al., 2010). from the combined effects of pollen and seed dispersal, Clonal growth is a common feature among brome- and of clonal growth. Clonality strongly influences liad species (Murawski & Hamrick, 1990; Sampaio spatial structure, not only because of the potential et al., 2004; Sampaio, Pico & Scarano, 2005; Barbará aggregation of clones, leading to a spatial concentra- et al., 2009). However, there are few studies describ- tion of genetic variation, but also because clonal ing clonal structure in bromeliads using molecular growth itself is a component of dispersal (Alberto et al., markers (e.g. Murawski & Hamrick, 1990; Izquierdo 2005; Arnaud-Haond et al., 2007). Moreover, patterns & Piñero, 2000). Aechmea nudicaulis (L.) Griseb. is a Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 of aggregation also vary into guerrilla (genets distrib- nurse bromeliad common to the Brazilian Atlantic uted in an irregular fashion and widely spread ramets) forest complex. This species has several features or phalanx (regularly distributed genets and clonal which make it an interesting model species for SGS copies densely aggregated near the parental ramet) studies in clonal and non-clonal bromeliads. First, (Lovett-Doust, 1981); although guerrilla growth leads being a self-incompatible species (Matallana et al., to high genet intermingling and clonal dispersal, 2010), the issue of selfing resulting in identical geno- phalanx growth results in spatially aggregated clones types can be disregarded, making it easier to compute and an increase in SGS (Vallejo-Marín, Dorken & the amount of sexual reproduction vs. clonal growth. Barrett, 2010). Seedling recruitment dynamics is also Moreover, in certain habitats, the species presents a important in determining SGS in clonal plants, and terrestrial growth habit, which means that individu- contemplates a spectrum of strategies between two als can be considered to be distributed in a two- endpoints: initial seedling recruitment (ISR), charac- dimensional space. In addition, it is a ubiquitous terized by a colonization by a single sexual recruitment species, which makes it easier to obtain large sample event and associated with long-distance dispersal sizes suited for SGS studies. leading to a decrease in SGS; and repeated seedling This study aims to provide information on the recruitment (RSR), characterized by the continuous clonal and genetic structure of this nurse bromeliad. establishment of new genets through seed germination Thus we: (1) evaluated the clonal structure of A. nudi- and associated with short-distance dispersal leading to caulis by describing the diversity components and the an increase in SGS (Eriksson, 1993). In between, two spatial components of clonal growth; (2) characterized more categories were later described to elaborate this the clonal subrange of A. nudicaulis in the population dichotomy further: RWO (recruitment at windows of studied; and (3) evaluated the small-scale SGS in opportunity) and RSR/ISR (systematic large-scale the population studied. For this purpose, a detailed variation between RSR and ISR) (Eriksson, 1993, analysis of the distribution of genetic variation among 1997, 2011). In the RWO model, there is continuous ramets of A. nudicaulis was conducted in the field site seedling recruitment, but only under specific condi- using a set of six polymorphic microsatellite markers. tions that are spatially unpredictable. In the RSR/ISR category, there is a source/sink dynamic between a MATERIAL AND METHODS group of local populations, in which some are RSR, acting as seed sources, and some are ISR, acting as STUDY SITE seed receptors. The Brazilian Atlantic forest complex is one of the most In addition, the amount and distribution of genetic threatened ecosystems on the planet (Ribeiro et al., diversity within clonal populations may have an 2011). It is composed of a core rain forest surrounded impact on the evolvability of these populations by a number of marginal ecosystems that range from (Eriksson, 1993; Dorken & Eckert, 2001; Honnay & open to forest formations. One such peripheral habitat Jacquemyn, 2008). In addition, a small effective popu- is the so-called ‘restinga’, which is a coastal sandy lation size, which can be expected in clonal popula- plain vegetation, separating the rain forest from the tions, may lead to high levels of inbreeding and thus a sea (Scarano, 2009). On the northern coast of Rio de strong genetic drift. Therefore, the characterization of Janeiro state, restinga is typically structured in veg- levels of genetic variability and SGS in clonal plants is etation islands separated by a matrix of bare sand, and essential for understanding the evolutionary dynamics its dynamics are intimately dependent on nurse plant and conservation status of these species. species, such as the tree Clusia hilariana Schltdl. and When describing SGS in clonal plants, it is essential the clonal bromeliad Aechmea nudicaulis (Scarano, to account for the effects of clonality on the estimate. 2009). Therefore, an ideal approach is that the characteriza- Our 0.5-ha plot was located inside the Restinga de tion of the SGS be preceded by a characterization of the Jurubatiba National Park (22°08′–22°19′S; 41°17′–

41°43′W), on the north coast of the state of Rio de Janeiro, south-eastern Brazil, which protects an area of 14 838 ha of sandy plains and coastal lagoons. The climate type is Köppen’s Aw tropical rainy, with a hot and rainy summer and a dry winter (Scarano, 2002). Rainfall is concentrated in the summer months (November–February) and temperatures are high throughout the year (mean, 23 °C; average maximum, 30 °C; average minimum, 20 °C). The sandy soil hinders the germination of most plant species, because of the high temperatures and low water and Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 nutrient storage capacity.

STUDY SPECIES Bromeliads occur in a myriad of tropical habitats from coastal to rocky outcrops, growing as epiphytes and/or terrestrials (growing directly on the soil surface instead of tree trunks) (Givnish et al., 2011). Aechmea nudicau- Figure 1. Clonal fragment of Aechmea nudicaulis. Each lis is distributed across Central and South America and clonal fragment consists of a group of connected live can be found in different vegetation types, such as rain ramets and thus each fragment represents a single genet. forests, ‘cerrados’ (Brazilian savannas) and restingas Only one live ramet of each clonal fragment was sampled, (Wendt, 1997). It has a semalparous life cycle; flowering as described in the Material and methods section. On the left side of the image is a group of four senescent ramets occurs mainly between May and October with inflores- and, on the right side, a group of four live ramets in the cences that are 60–100 m in height, and panicles with same fragment. several yellow hermaphrodite flowers (Wendt, 1997). The species is self-incompatible (Matallana et al., 2010) and is pollinated by hummingbirds and bees (Schmid surface and last longer than individual ramets. In the et al., 2011). Fruits are oval shaped (10 × 7 mm) and open area of the studied site, ramets of A. nudicaulis orange when mature, carrying several small seeds occur in groups of three or four physically connected (2 × 1 mm). Fruits do not open spontaneously after live ramets, called clonal fragments (Fig. 1), that typi- ripening (R. Loh, pers. observ.), which means that cally grow in a zigzag pattern. The distance between the liberation of seeds depends on the intervention of a ramets in a clonal fragment is approximately 10 cm dispersal agent. So far, no information is available on (Sampaio et al., 2004, 2005). seed dispersers, but fruit characteristics indicate that dispersal is possibly performed by animals (Charles-Dominique, 1993). SAMPLING DESIGN In Atlantic forest, the species often grows as an Sampling was performed within a 0.5-ha rectangular epiphyte, whereas, in restingas, it is mostly terres- plot (100m×50m).Genets of A. nudicaulis are made trial (growing on the sand surface). Seeds are heat up of connected ramets in a clonal fragment. All sensitive and do not germinate after being exposed to connected ramets of the same clonal fragment char- temperatures above 50 °C (Pinheiro & Borghetti, acterize a genet, but independent fragments can also 2003). As the sandy soil may easily reach tempera- be part of the same genet which may have been tures above that limit (Sampaio et al., 2004, 2005), divided. As it is impossible to know this prior to germination of A. nudicaulis possibly occurs only in genotyping, we sampled one ramet of each existing specific places sheltered from intense heat and solar clonal fragment within the plot, for a total of 187 radiation (inside the vegetation islands). sampling units. For each ramet, a portion of leaf was In A. nudicaulis, clonal growth takes place through removed and stored in silica gel before transportation the formation of sympodial rhizomes initiated from to the laboratory, where it was stored in a freezer at axillary meristems around the shoot base, which give −80 °C until DNA isolation. rise to new ramets that are able to grow in all directions (Sampaio et al., 2004). Each plant produces from one to two new ramets every year that typically DNA ISOLATION, MICROSATELLITE AMPLIFICATION remain connected to the mother plant during the first AND GENOTYPING year of development (Sampaio et al., 2004, 2005). Rhi- DNA was isolated using the cetyltrimethylammonium zomes are usually buried within 5 cm from the bromide (CTAB) extraction method adapted from

Doyle & Doyle (1987) and modiﬁed by Margis et al. veriﬁed using Micro-Checker software v. 2.2.3 (Van (2002). Samples were genotyped for six microsatellite Oosterhout et al., 2004). loci previously described for other bromeliad species: CT5 (Boneh, Kuperus & Van Tiederen, 2003), Pit5 and Pit8 (Sarthou et al., 2003), ACPCT138A (Kinsuat & CLONE IDENTIFICATION Kumar, 2007), VgC01 (Palma-Silva et al., 2007) and To account for the clonal replicates in the sample, we

PaZ01 (Paggi et al., 2008). Polymerase chain reactions calculated the probability Psex that multilocus geno- (PCRs) were performed on an MJ96+ thermocycler types (MLGs) repeated n times in the sample origi- (Biocycler), according to the author’s recommendations nated from distinct sexual reproductive events and for each locus, with some modifications as follows. not clonal growth. This probability is based on the Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 Approximately 15 ng of nuclear DNA were amplified in probability Pgen of a given genotype being assembled a final volume of 10 μL containing 200 μM of each considering the allele frequencies and number of loci deoxynucleoside triphosphate (dNTP), 2 mM MgCl2, used to discriminate between genotypes. As sug- 200 mM Tris-HCl (pH 8.4), 500 mM KCl, 1 μM each gested by Arnaud-Haond et al. (2007), we used the primer (forward and reverse) and 0.5 U Taq DNA inbreeding coefficient (Fis) as described by Weir & polymerase (Thermo Scientific). Cycling conditions for Cockerham (1984) to account for possible departures primers CT5, ACPCT138A, VgC01 and PaZ01 con- from Hardy–Weinberg equilibrium when calculating sisted of an initial denaturing step of 5 min at 94 °C, Pgen. Psex estimates were calculated for all groups of followed by 10 cycles of touchdown with 30 s at 94 °C, repeated genotypes with the software GENCLONE 30 s at 58 °C (reduced by 1 °C in each cycle) and 30 s at version 2.0 (Arnaud-Haond & Belkhir, 2007) using 72 °C, 30 additional cycles of 30 s at 94 °C, 30 s at the round-robin method (Arnaud-Haond et al., 2005), 48 °C and 30 s at 72 °C, and a final elongation step of a subsampling approach that avoids the overestima- 5 min at 72 °C. For primers Pit5 and Pit8, cycling tion of rare allele frequencies (Arnaud-Haond et al., conditions were denaturation for 5 min at 94 °C, 40 2007). cycles of 30 s at 94 °C, 30 s at 53 °C and 30 s at 72 °C, To evaluate the power of the set of loci to discrimi- and a final elongation step of 5 min at 72 °C. nate all MLGs present in the sample, the Monte Carlo PCR products were mixed with 1 vol of formamide procedure available in the software GENCLONE was buffer [98% formamide, 10 mM ethylenediaminetet- used to analyse the total number of MLGs encoun- raacetic acid (EDTA) (pH 8.0), 0.1% bromophenol tered by all possible combinations of K loci (from blue and xylene cyanol] and denatured at 95 °C for K =1 to K = 6 loci). The total number of MLGs dis- 5 min prior to electrophoresis. PCR products for each criminated in the sample should stabilize before the marker were separated by electrophoresis in 4.5% whole set of loci is used, and thus the great majority large sequencing polyacrylamide gels (acrylamide– of genotypes was probably discriminated. We consid- bis-acrylamide, 19 : 1; urea, 7 M) and genotyping ered the set of six loci to be fairly adequate as the was achieved through silver staining (Creste, rarefaction plot seems to be reaching an asymptote Tulmann-Neto & Figueira, 2001). Molecular sizes in (Fig. S1, see Supporting Information). base pairs were estimated by comparison with 10-bp ladder size standards (Invitrogen). Genotyping errors were monitored by repeating a small set of five CLONAL DIVERSITY COMPONENTS samples containing all the alleles of a given locus in All statistics of clonal diversity were calculated using every gel as a positive control. Possible somatic the program GENCLONE. Genotypic richness was mutations were not accounted for because of their calculated according to Dorken & Eckert (2001) as low frequency in natural populations (Ally, Ritland & R =(G – 1)/N – 1, where G is the number of distinct Otto, 2008). genotypes and N is the sample size. Clonal diversity was given by the complement of Simpson’s index (D) that expresses the probability of two units taken at GENETIC DIVERSITY STATISTICS random from the sample being different MLGs (GEN-

Allele frequencies, expected (He) and observed (Ho) CLONE calculates an unbiased estimate D* inde- heterozygosities and ﬁxation index (F) were estimated pendent of sample size). Simpson’s evenness index as described by Weir & Cockerham (1984) using the was calculated as described by Arnaud-Haond et al. software GenAlEx version 6.3 (Peakall & Smouse, (2007) to estimate the equitability in the distribution 2006) considering the genet and ramet level. Hardy– of ramets among genets. Weinberg equilibrium and genotypic linkage disequi- The Pareto distribution was also used to describe librium (using a single copy per genet) were tested clonal diversity and evenness as the relationship using GENEPOP version 4.2 software (Raymond & between genet sizes (number of replicates) and their Rousset, 1995). The presence of null alleles was cumulative frequencies (Arnaud-Haond et al., 2007).

The Pareto index is a good option for comparing The slope of the regression of the mean pairwise clonal diversity among studies as it is less sensitive to kinship coefficients on the logarithm of spatial dis- sample size and also reflects a combination of clonal tance between sampling units (blog) was estimated to richness and equitability (Arnaud-Haond et al., 2007). characterize the strength of SGS, and significance A linear relationship is expected if the distribution of tests were performed with the software SPAGeDI. clonal membership conforms to a Pareto distribution. The strength of SGS was also described by the Sp A steep slope of the Pareto distribution (β) represents statistic, which represents the rate of decrease of high diversity and evenness, whereas low values of β pairwise kinship with spatial distance (Vekemans & (shallow slope) reflect a heterogeneous distribution Hardy, 2004). The Sp statistic is calculated as: with few large genets and many small genets. Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 Sp=− blog()1 − F()1 SPATIAL GENETIC STRUCTURE where F(1) is the mean pairwise kinship coefficient in Two sets of analyses were carried out separately to the first distance class. determine the SGS of A. nudicaulis in the Restinga de Jurubatiba population. First, a ramet-level analysis was performed with the whole dataset (clonal repli- RESULTS cates included) to define the clonal subrange, which was inferred by estimating the probability of clonal GENETIC DIVERSITY identity at different spatial scales. The clonal sub- Thirty-six alleles were amplified from the six micro- range is defined as the spatial distance at which the satellite loci (Table 1). Genetic diversity given by the probability of clonal identity (fraction of all pairs of expected heterozygosity (He) was high, both including ramets of the same genet separated by a given dis- and excluding clonal replicates (Table 1). Significant tance) reaches zero (Harada, Kawano & Iwasa, 1997; departures from Hardy–Weinberg equilibrium were Alberto et al., 2005). The clonal subrange was also observed for both levels (ramet and genet) in most loci inferred by comparing spatial autocorrelograms because of heterozygote deficiency. However, a few including the whole dataset and correlograms includ- cases presented heterozygote excess (Table 1). Tests ing only one copy of each MLG. for genotypic linkage disequilibrium at the genet level Second, a genet-level analysis was carried out using accounted for three pairs of loci that rejected the null a single copy of each MLG and aimed at characteriz- hypothesis of genotypes at one locus being independ- ing the SGS of A. nudicaulis, i.e. the formation of ent from genotypes at the other locus after Bonferroni local pedigree structures as a result of limited gene correction. No evidence of null alleles was observed dispersal. The genet-level analysis was carried out in for the six loci. two different ways as described in Alberto et al. (2005) by a procedure available in GENCLONE. First, the central coordinates of the set of ramets of each genet GENOTYPIC DIVERSITY AND CLONAL STRUCTURE (average X and Y) were used as the spatial represen- Sixty-two unique MLGs were identified within 187 tation of the genet, hereafter referred to as the Genet sampling units. All replicates of the same MLG had −4 Central analysis. Second, the Genet Random analysis Psex < 0.01 (mean value, 2.02 × 10 ). Pgen values for all was performed by a re-sampling procedure in which a genotypes were carefully checked to detect potential random ramet of each genotype with multiple copies errors in genotype assignment, and all values were less was drawn to create a matrix with a single copy of than 1.27 × 10−5. Therefore, all identical MLGs were each genet. The procedure was repeated 1000 times to considered to be the result of clonal growth. In one calculate the dispersion of the estimates. case, two identical MLGs were separated by 25 m, For both datasets, we calculated the average raising the possibility of two distinct genotypes being genetic kinship coefficient described in Loiselle et al. wrongly assigned as the same MLG, but Pgen and Psex (1995) between pairs of individuals separated by 3-m values for this pair of ramets were 3.6 × 10−7 and intervals in the range 0–60 m using the software 6.7×10−5, respectively, and so they were considered as GENCLONE and SPAGeDI version 1.3 (Hardy & replicates of the same MLG. The set of six microsatel- Vekemans, 2002). To test the significance of kinship lite loci was considered to be adequate to discriminate values, spatial locations among individuals were ran- all the MLGs in the sample as the rarefaction plot domly permuted 10 000 times in order to test whether calculated with the Monte Carlo procedure in GEN- the observed mean kinship values were different from CLONE presented an asymptotic behaviour (Fig. S1). those expected under a random distribution of geno- Genotypic richness was moderate (R = 0.32) with types for each spatial distance class. Correlograms < 50% of the sampled ramets being unique MLGs. The were constructed by plotting mean pairwise kinship complement of Simpson’s index presented a high value coefficients as a function of spatial distance class. (D* = 0.96), indicating a high clonal diversity, and the

Table 1. Size range of microsatellite alleles, number of alleles (A), number of effective alleles (Ae), expected heterozygosity

(He), observed heterozygosity (Ho) and ﬁxation index (F) (Weir & Cockerham, 1984) for genet-level and ramet-level analyses of Aechmea nudicaulis at Jurubatiba restinga. Signiﬁcant departures from Hardy–Weinberg equilibrium are coded: **P < 0.001; *0.001 < P < 0.01. N is the number of sampling units

Genet Ramet N =62 N = 187

Locus Size range AAe He Ho FHe Ho F

138A 180–190 5 3.74 0.73 0.64 0.13 0.73 0.62 0.14** Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 CT5 160–172 4 1.79 0.44 0.38 0.13 0.43 0.34 0.21** PaZ01 150–180 8 6.42 0.84 0.67 0.21** 0.84 0.68 0.18** Pit5 330–360 6 4.06 0.75 0.82 −0.08 0.74 0.86 −0.15** Pit8 268–282 6 4.15 0.76 0.95 −0.24** 0.76 0.98 −0.28** VgC01 160–180 7 4.79 0.79 0.64 0.19* 0.79 0.70 0.12** All 36 0.72 0.68 0.05 0.72 0.70 0.03 equivalent evenness index was also high (V = 0.94). Clonal heterogeneity described by the Pareto distribution (Fig. 2A) was high (β = −0.63). This result may reﬂect the presence of many small genets and few large ones (Fig. 2B). For example, 48% of the genotypes are unique MLGs, 30% correspond to genotypes with two or three clonal repetitions and the remaining 22% include genotypes with more than four clonal repetitions (Fig. 2B). Mean clone size was 4.9 clonal fragments per genet. There is a moderate aggregation of clonal fragments belonging to the same genet, as seen in Figure 3 (each sampling unit corresponds to a clonal fragment consisting of three or four ramets). Some genotypes, such as clone 33, present all the copies aggregated, whereas genotype 13 is intermingled with other genotypes. The average clonal dimension given by the average distance between ramets of the same genet was 3.71 m, and most clonal fragments showed a maximum distance between replicates of 5 m or less (19 of 32 genets, or 59% of the clones). The maximum distance found between ramets of the same genet was 25 m.

SPATIAL GENETIC STRUCTURE Figure 2. Clonal heterogeneity describing the distribu- Ramet-level analysis tion of ramets among genets. A, Pareto plot describing the Spatial autocorrelation analysis was conducted for distribution of ramets into groups of genets of different the ramet-level analysis, and mean kinship values for sizes. This plot should display a straight line if the distri- each distance class were plotted against spatial dis- bution of clonal membership conforms to a Pareto distri- tance (Fig. 4A). The mean kinship coefficient in the bution. The coefficient of linear regression (β) is given in ﬁrst distance class was fairly high (F(3 m) = 0.405) and the plot. *P < 0.05. B, Histogram of the frequency of the the slope of the linear regression of kinship values number of clonal replicates. against distance was shallow (Table 2, Fig. 4A). The shape of the correlogram shows that the rate of decrease of mean kinship value with distance is very the Sp statistics, which show a high value for the steep up to 18 m, when it stabilizes near zero for the ramet-level analysis, reﬂecting a strong spatial struc- remaining distance classes (Fig. 4A). A more precise ture (Table 2). The clonal subrange, i.e. the distance description of the shape of the correlogram is given by range beyond which clonality has negligible effects on

© 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 178, 329–341 CLONALITY AND SGS IN AECHMEA NUDICAULIS 335 Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021

Figure 3. Sampling plot of Aechmea nudicaulis at Jurubatiba restinga representing the map of genotypes identiﬁed in the area. All clonal fragments present in the area were sampled. The numbers identify different genets (62 in total), and repeated numbers represent copies of the same genet. Open circles represent existing vegetation islands.

genetic structure, was estimated by plotting the prob- with the random method (F(3 m) = 0.19; P < 0.05). A ability of clonal identity against distance (Fig. 4B). significant SGS was detected for the genet-level analy- The probability of clonal identity declined with sis (for both methods), but the slope of the regressions increasing distance, from 60% in the first distance on the log of spatial distance was always shallow class (3 m) to zero at 23 m, characterizing the clonal (blog = −0.06; P < 0.01 for both methods). A lower Sp subrange (Fig. 4B). statistic was also observed for the genet level in comparison with the ramet level (Table 2), with the Genet-level analysis central method presenting a slightly higher value Mean kinship values at the genet level are reasonably (Sp = 0.079) than the random method (Sp = 0.074). lower than those at the ramet level because of the There was no marked difference between methods for inflating effect of clone mates in the kinship values. defining the geographical position of each genet. Both genet-level analyses (central and random coordi- Therefore, the comparison between ramet- and genet- nates) showed significant positive mean kinship coef- level correlograms (Fig. 4A) was presented only for the ficient values for the first distance class (Table 2), but central method (i.e. considering the central coordinate the central method showed a higher value in the first of the clonal fragment as its origin). Although the distance class (F(3 m) = 0.24; P < 0.05) in comparison genet-level analysis revealed a markedly weaker SGS

it is now known that clonal species can show high levels of genetic diversity (Widén, Cronberg & Widén, 1994; Honnay & Jacquemyn, 2008; Vallejo-Marín et al., 2010) if seedling recruitment is continuous. In addition, the long life span of clonal plants, which is a consequence of clonal growth itself, contributes to the production of multiple copies of a single genotype that may persist for many generations (Eriksson, 1993). As a result, the erosion of genetic variability through drift would be slow. Even low rates of sexual input would generate new genets that, in addition to Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 the existing ones, would result in the buildup of genetic variation with a minimum loss, given that the growth rate of different clones is similar. Otherwise, an increase in drift is expected as a result of an increase in the variance of reproductive success among different genotypes, i.e. if some clones are predominant, they will have a greater contribution to the sexual reproduction events and this would inflate drift (Eriksson, 1993). Self-incompatibility (SI) also contributes to a high genetic diversity, despite clonality, by limiting geito- nogamy. However, SI species, such as A. nudicaulis, Figure 4. Spatial autocorrelation analysis with the can suffer from pollen wastage because of high levels kinship coefficient described in Loiselle et al. (1995). A, of self-pollination, especially in animal-pollinated Ramet- and genet-level analyses using the central coordi- species, which may incur high fitness costs and low nates of the clone. Significant values do not extend beyond fertility (Honnay & Jacquemyn, 2008; Vallejo-Marín 9 m. B, Probability of clonal identity. The spatial distance et al., 2010). In the end, antagonistic mechanisms that at which the probability of clonal identity reaches zero increase diversity vs. those that reduce it will deter- represents the clonal subrange. mine the genetic variation levels of a given population. Therefore, Eriksson (2011) advised that inferences of RSR based solely on levels of genet diversity should be approached with caution and field-based information pattern, the ramet- and genet-level analyses showed on the biology of the species should be considered. A similar correlogram shapes, with a steeper decrease up scenario for A. nudicaulis is discussed in the last to 20 m and further stabilization near zero (Fig. 4A). In section. addition, the genet and ramet curves joined each other near the 20 m value, which also characterizes this distance as the clonal subrange. CLONAL DIVERSITY COMPONENTS Genotypic richness The estimate of genotypic richness was moderate DISCUSSION (R = 0.32) because of the presence of large clones in the Aechmea nudicaulis showed high levels of genetic sample. About half of the genets in the sample have a diversity, as observed previously for other Brazilian single copy, suggesting a significant contribution of clonal bromeliads (Cavallari et al., 2006; Barbará sexual reproduction, as single copy MLGs are likely to et al., 2009), and several clonal species (Alberto et al., be young genets. Further, the presence of large clones, 2005; Jacquemyn et al., 2006; McGlaughlin & Friar, with more than 10 replicates, suggests a great longev- 2007). Thus, clonality does not seem to be affecting ity of some genets. However, this value of genotypic genetic diversity levels in A. nudicaulis. In addition, richness is based on the number of clonal fragments, the presence of large genets (with more than ten which results in an overestimation of the value based copies) suggests high genet longevity. The combina- on the total number of ramets in the sampling area. tion of persistence and repeated sexual recruitment The clonal fragment was chosen as the sampling unit is essential for evolvability and resilience of the because it was considered unnecessary to genotype all species. Despite early suggestions that clonal growth ramets of a single fragment, once the physical connec- would reduce genetic variability within populations tion between them was easily verified. In addition, the (Murawski & Hamrick, 1990; Sydes & Peakall, 1998), number of ramets in each fragment is fairly constant,

Table 2. Characterization of the spatial genetic structure of Aechmea nudicaulis at two different levels (ramet and genet).

Mean kinship coefficients in the ﬁrst distance interval (F(3 m)), coefficient of linear regression of kinship values on log distance (blog) and Sp statistics. *P < 0.01, **P < 0.001

F(3 m) (± SE) blog (± SE) Sp(0–60 m)

Ramet level 0.405** ± 0.045 −0.123* ± 0.001 0.206 Genet level (central coordinates) 0.248** ± 0.042 −0.061* ± 0.004 0.079 Genet level (random coordinates) 0.195** ± 0.043 −0.058* ± 0.003 0.074 Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 with three or four live ramets per fragment (Sampaio within genets. The Pareto index seems to be more et al., 2004). Considering a constant number of three sensitive to the presence of large genets of different ramets per fragment, the estimate of genotypic rich- sizes (number of clonal replicates). Even at low fre- ness would be R = 0.11. However, as the genotypic quencies, large genets of different sizes considerably richness (R) is a measure of the contribution of sexual reduce the slope of the Pareto regression. A close input and, within each fragment, only one ramet is inspection of the number of clonal replicates plotted sexually mature in each generation (Sampaio et al., against the number of genotypes (Fig. 2B) reveals that 2004), the estimate of R = 0.32 is appropriate. the Pareto distribution provides a better reflection Genotypic richness (R) for A. nudicaulis is lower than Simpson’s measure of clonal equitability, as half than the average value for clonal species calculated of the genets appear as a single fragment, and the for 195 taxa (G/N = 0.42 ± 0.02; Vallejo-Marín et al., other half are distributed in different size classes (from 2010), indicating a moderate sexual input and also two to 25 copies). This pattern of distribution of ramets a smaller clone size. However, previous surveys within genets, with few large and many small clones, have shown a great variation in genotypic diversity implies that few genets have a large potential contri- within clonal species (Widén et al., 1994; Honnay & bution to sexual reproduction (Vallejo-Marín et al., Jacquemyn, 2008; Vallejo-Marín et al., 2010), with 2010). some populations being composed of a single clone and others being composed entirely of unique MLGs Small-scale SGS – a scenario for A. nudicaulis (Pfeiffer et al., 2011). In that sense, the estimated Before the analysis of SGS in a clonal species, it is diversity for A. nudicaulis depicts a considerable con- necessary to define the extension of the effect of tribution of sexual reproduction. clonality on SGS (i.e. clonal subrange). Our results show that the clonal subrange for A. nudicaulis is Clonal diversity and equitability 23 m. Thus, within this spatial scale, clonal growth Clonal diversity measured by the complement of Simp- significantly affects SGS, inflating kinship values as a son’s diversity index was high (D* = 0.96) when com- result of the presence of several clonal copies of the pared with other studies of clonal plants (McGlaughlin same genet. Several studies have not measured the & Friar, 2007; Honnay & Jacquemyn, 2008), but clonal subrange per se, but, in a certain way, they have similar values have been observed for other bromeliad estimated the extension of clonal spread based on the species from Atlantic forest areas in Brazil (Barbará extension of genets. Reported values vary greatly et al., 2009). However, the high values observed for among species (from 0.59 m to 35.00 m) and certainly these bromeliads might be a reflection of the sampling show a strong relation to intrinsic developmental strategy that favoured a large number of genotypes traits, occurrence of agamospermy and the size scale of and may be an overestimation of the unique MLGs. the plant (see Alberto et al., 2005; McGlaughlin & The high D* value observed for A. nudicaulis is related Friar, 2007; Barbará et al., 2009; Pfeiffer et al., 2011). to the measure of genotypic richness being based on The influence of clonality also varies greatly among the number of clonal fragments (R = 0.32) instead of clonal plants, with some species showing little influ- the total number of ramets (R = 0.11). However, as ence (Alberto et al., 2005) and species, such as A. nudi- discussed previously, the estimate of R = 0.32 is more caulis, showing a great difference between kinship adequate for the present study. values in genet- and ramet-level analyses. Comparison However, the Pareto index (β = −0.63) gave a mod- between the two types of analysis (ramet vs. genet erate regression slope, reflecting a low clonal diversity level) reveals that the presence of clones inflates in A. nudicaulis when compared with 13 other clonal kinship values in c. 50% in the first distance class. species (0.93; Ohsako, 2010). However, Simpson’s Actually, the kinship value obtained for the first dis- index indicated a high equitability. This is mostly a tance class in the ramet-level analysis is close to the result of the heterogeneity in the distribution of ramets maximum possible kinship value between full-sibs.

Architecture in the context of the clonal fragment of germination sites and clonal propagation), with the A. nudicaulis is typically phalanx, resulting in short first group probably predominating over the second spacers and obtuse ramification (Sampaio et al., 2005), group as a strong SGS is observed. This scenario is but our results showed that different clonal fragments consistent with the dynamics of RWO described by are sometimes part of the same genet. In this context, Eriksson (1993, 2011) and, in the case of our study the degree of intermingling among genets is higher, as species, the vegetation islands characteristic of the in a guerrilla growth habit. The degree of intermin- landscape play a major role in determining the gling reflected by clonal architecture between different amount of successful reproduction in each generation genets is an important parameter in terms of SGS, as (as they represent adequate germination sites). it may promote reproduction between genetically dif- Previous studies have shown that the vegetation ferent individuals that were originally distant from islands at the Jurubatiba restinga are cyclic (Scarano, Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 each other, thus reducing the SGS. 2002). Most islands are characterized by the presence The data revealed a significant SGS for A. nudicau- of a central woody species that provides suitable lis in the studied population as a consequence of conditions for the establishment and survival of other limited gene dispersal. Significant values of kinship species. When this central plant perishes, the island extended up to 9 m, which characterizes the spatial disappears, because most species cannot stand the distance beyond which genetic variation is randomly harsh environment of the open restinga, and thus distributed in space. only a few, such as cacti and the bromeliad A. nudi- The parameters measuring the intensity of SGS caulis, remain. Furthermore, it has been documented presented high values when compared with other that A. nudicaulis individuals inside vegetation clonal and non-clonal plant species (Vekemans & islands tend to send out new shoots towards the open Hardy, 2004; Alberto et al., 2005; McGlaughlin & areas and ‘walk’ from inside to outside in each clonal Friar, 2007) and also with other bromeliad species generation (Sampaio et al., 2004, 2005). Therefore, (Barbará et al., 2009). Under a homogeneous seed bromeliads outside vegetation islands would be either dispersal model, SGS is expected to be lower in self- remnants from old islands or bromeliads that have incompatible species, such as A. nudicaulis, as pollen walked to the open areas, which explains the large flow is not possible between genetically identical number of individuals outside the islands. Once new individuals that are normally aggregated in space ramets reach the open area (the sand matrix), further (Honnay & Jacquemyn, 2008). However, in the clonal spread can potentially start new islands in a present study, the strong territorial behaviour of the fairly random fashion (Sampaio et al., 2004). putative pollinator (a hummingbird species; Schmid Thus, the spatial distribution of genetic variation et al., 2011) might restrict pollen dispersal to short and the habitat occupation pattern in A. nudicaulis distances between neighbouring flowering plants, are dependent not only on the species intrinsic bio- leading to a strong SGS pattern. Studies of pollinator logical traits (such as spacer size and mating system), behaviour in clonal plants have indicated that the but also on biotic interactions with neighbouring majority of pollen movement occurs between neigh- species that determine suitable habitats for germina- bouring inflorescences independent of pollinator tion and establishment of new genets. Such a depend- group (Peakall & Beattie, 1991). However, the limi- ence on the surrounding biotic environment has tation of appropriate germination sites decreases SGS already been demonstrated for other clonal species because seeds falling close to the mother plant in (Matesanz et al., 2011; Benot et al., 2013). open areas (where the sand reaches high tempera- The strong SGS found in the Jurubatiba population tures) do not germinate. Only seeds that are trans- has significant implications for conservation pur- ported to vegetation islands may successfully poses. As the genetic variation is not randomly dis- establish new genets. In addition, the presence of tributed in space, size limitations of protected areas predominant clones and the similar growth rate of might result in the loss of significant amounts of genets (Sampaio, 2004) indicate that genet longevity genetic diversity. Thus, SGS and clonal structure data might be a possible mechanism of preservation of should always be taken into consideration in conser- genetic diversity. Thus, some successful sexual repro- vation programmes for clonal plants. duction might occur only in rare good years, restricted to vegetation islands, but still be sufficient to sustain high genetic diversity because genet longevity con- ACKNOWLEDGEMENTS tributes to the persistence of the existing genotypes until sexual reproduction is possible. The overall sce- This study was supported by grants from the Brazil- nario results from a balance between processes that ian Research Council (CNPq: n°306025/2010-8), Edu- increase SGS (e.g. territorial pollinator and clonal cation Council (CAPES) and Fundação de Amparo habit) and processes that decrease SGS (scarcity of à Pesquisa do Rio de Janeiro (FAPERJ: n° E-26/

170.599/2007 and n° E-26/102.861/2008). We thank Dorken ME, Eckert CG. 2001. Severely reduced sexual Instituto Brasileiro do Meio Ambiente e dos Recursos reproduction in northern populations of a clonal plant Naturais Renováveis for granting access to the con- Decodon verticillatus (Lythraceae). Journal of Ecology 89: servation unit, Durvalina B. Felix for technical 339–350. support, and Joanito and Erick Grigorovski for help Doyle J, Doyle J. 1987. A rapid DNA isolation procedure for with ﬁeldwork. small quantities of fresh leaf tissue. Phytochemical Bulletin, Botanical Society of America 19: 1–15. Eriksson O. 1993. Dynamics of genets in clonal plants. REFERENCES Trends in Ecology and Evolution 8: 313–316. Alberto F, Gouveia L, Arnaud-Haond S, Pérez-Lloréns Eriksson O. 1997. Clonal life histories and the evolution of

JL, Duarte CM, Serrão EA. 2005. Within-population seed recruitment. In: de Kroon H, van Groenendael J, eds. Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 spatial genetic structure, neighbourhood size and clonal The ecology and evolution of clonal plants. Leiden: Back- subrange in the seagrass Cymodocea nodosa. Molecular huys Publishers, 211–226. Ecology 14: 2669–2681. Eriksson O. 2011. Niche shifts and seed limitation as mecha- Ally D, Ritland K, Otto SP. 2008. Can clone size serve as a nisms determining seedling recruitment in clonal plants. proxy for clone age? An exploration using microsatellite Preslia 83: 301–314. divergence in Populus tremuloides. Molecular Ecology 17: Givnish TJ, Barfuss HJ, Van EB, Riina R, Schulte K, 4897–4911. Horres R, Gonsiska P, Jabaily R, Crayn DM, Arnaud-Haond S, Alberto F, Teixeira S, Procaccini G, Smith JA, Winter K, Brown GK, Evans TM, Holst B, Serrao EA, Duarte CM. 2005. Assessing genetic diversity Luther H, Till W, Zizka G, Berry PE, Sytsma KJ. 2011. in clonal organisms: low diversity or low resolution? Com- Phylogeny, adaptive radiation, and historical biogeography bining power and cost efficiency in selecting markers. in Bromeliaceae: insights from an eight-locus plastid phy- Journal of Heredity 96: 434–440. logeny. American Journal of Botany 98: 1–24. Arnaud-Haond S, Belkhir K. 2007. GENCLONE: a com- Harada Y, Kawano S, Iwasa Y. 1997. Probability of clonal puter program to analyze genotypic data, test for clonality identity: inferring the relative success of sexual versus and describe spatial clonal organization. Molecular Ecology clonal reproduction from spatial genetic patterns. Journal of Notes 7: 15–17. Ecology 85: 591–600. Arnaud-Haond S, Duarte C, Alberto F, Serrão E. 2007. Hardy OJ, Vekemans X. 2002. SPAGeDI: a versatile com- Standardizing methods to address clonality in population puter program to analyse spatial genetic structure at the studies. Molecular Ecology 16: 5115–5139. individual or population levels. Molecular Ecology Notes 2: Barbará T, Martinelli G, Palma-Silva C, Fay MF, Mayo 618–620. S, Lexer C. 2009. Genetic relationships and variation in Harper JL. 1977. Population biology of plants. London: Aca- reproductive strategies in four closely related bromeliads demic Press. adapted to neotropical ‘inselbergs’: Alcantarea glaziouana, Honnay O, Jacquemyn H. 2008. A meta-analysis of the A. regina, A. geniculata and A. imperialis (Bromeliaceae). relation between mating system, growth form and genotypic Annals of Botany 103: 65–77. diversity in clonal plant species. Evolutionary Ecology 22: Benot M, Bittebiere A, Ernoult A, Clément B, Mony C. 299–312. 2013. Fine-scale spatial patterns in grassland communities Izquierdo LY, Piñero D. 2000. High genetic diversity in the depend on species clonal dispersal ability and interactions only known population of Aechmea tuitensis. Australian with neighbours. Journal of Ecology 101: 626–636. Journal of Botany 48: 645–650. Boneh L, Kuperus P, Van Tiederen PH. 2003. Microsat- Jacquemyn H, Brys R, Vandepitte K, Honnay O, ellites in the bromeliads Tillandsia fasciculata and Roldán-Ruiz I. 2006. Fine-scale genetic structure of life Guzmania monostachya. Molecular Ecology Notes 3: 302– history stages in the food deceptive orchid Orchis purpurea. 330. Molecular Ecology 15: 2801–2808. Cavallari M, Forzza RC, Veasey E, Zucchi MI, Oliveira Kinsuat MJ, Kumar SV. 2007. Polymorphic microsatellite G. 2006. Genetic variation in three endangered species of and cryptic simple repeat sequence markers in pineapples Encholirium (Bromeliaceae) from Cadeia do Espinhaço, (Ananas comosus var. comosus). Molecular Ecology Notes 7: Brazil, detected using RAPD markers. Biodiversity and 1032–1035. Conservation 15: 4357–4373. Loiselle BA, Sork VL, Nason J, Graham C. 1995. Spatial Charles-Dominique P. 1993. Speciation and coevolution: an genetic structure of a tropical understorey shrub Psychotria interpretation of frugivory phenomena. In: Fleming TH, officinalis (Rubiaceae). American Journal of Botany 82: Estrada A, eds. Frugivory and seed dispersal: ecological and 1420–1425. evolutionary aspects Dordrecht: Kluwer Academic Publish- Lovett-Doust L. 1981. Population dynamics and local spe- ers, 75–84. cialization in a clonal plant Ranunculus repens.I.The Creste S, Tulmann-Neto A, Figueira A. 2001. Detection of dynamics of ramets in contrasting habitats. Journal of simple sequence repeat polymorphisms in denaturing poly- Ecology 69: 743–755. acrylamide sequencing gels by silver staining. Plant Molecu- Margis R, Felix D, Caldas JF, Salgueiro F, Araujo DSD, lar Biology Report 19: 299–306. Breyne P, Montagu MV, De Oliveira D, Margis-

Pinheiro M. 2002. Genetic differentiation among three Habel JC, eds. Biodiversity hotspots: distribution and protec- neighboring Brazil-cherry (Eugenia uniﬂora L.) populations tion of conservation priority areas. Heidelberg: Springer, within the Brazilian Atlantic rain forest. Biodiversity and 405–434. Conservation 11: 149–163. Sampaio MC. 2004. Respostas da bromélia clonal Aechmea Matallana G, Godinho MAS, Guilherme FAG, Belisario nudicaulis à heterogeneidade ambiental da restinga aberta M, Coser TS, Wendt T. 2010. Breeding systems of Brome- de Clusia. Ph.D. Thesis. Rio de Janeiro – RJ, Brazil: Uni- liaceae species: evolution of selﬁng in the context of sym- versidade Federal do Rio de Janeiro. patric occurrence. Plant Systematics and Evolution 289: Sampaio MC, Araujo T, Scarano FR, Stuefer J. 2004. 57–65. Directional growth of a clonal bromeliad species in response Matesanz S, Gimeno T, Cruz M, Escudero A, Valladares to spatial habitat heterogeneity. Evolutionary Ecology 18:

F. 2011. Competition may explain the fine-scale spatial 429–442. Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021 patterns and genetic structure of two co-occurring plant Sampaio MC, Pico FX, Scarano FR. 2005. Ramet demog- congeners. Journal of Ecology 99: 838–848. raphy of a nurse bromeliad in Brazilian restingas. American McGlaughlin ME, Friar EA. 2007. Clonality in the endan- Journal of Botany 92: 674–681. gered Ambrosia pumila (Asteraceae) inferred from RAPD Sarthou C, Boisselier-Dubayle MC, Lambourdiere J, markers, implications for conservation and management. Samadi S. 2003. Polymorphic microsatellites for the Conservation Genetics 8: 319–330. study of fragmented populations of Pitcairnia geyskesii L. B. Murawski DA, Hamrick JL. 1990. Local genetic and clonal Smith (Bromeliaceae), a specific saxicolous species of structure in the tropical terrestrial bromeliad Aechmea inselbergs in French Guiana. Molecular Ecology Notes 3: magdalenae. American Journal of Botany 77: 1201– 221–223. 1208. Scarano FR. 2002. Structure, function and floristic relation- Ohsako T. 2010. Clonal and spatial genetic structure ships of plant communities in stressful habitats marginal to within populations of a coastal plant, Carex kobomugi the Brazilian Atlantic Rainforest. Annals of Botany 90: (Cyperaceae). American Journal of Botany 97: 458– 517–524. 470. Scarano FR. 2009. Plant communities at the periphery Paggi GM, Palma-Silva C, Bered F, Cidade FW, Sousa of the Atlantic forest: rare-species bias and its risk CB, Souza AP. 2008. Isolation and characterization of for conservation. Biological Conservation 142: 1201– microsatellite loci in Pitcairnia albiflos (Bromeliaceae), an 1208. endemic bromeliad from the Atlantic Rainforest, and cross- Schmid S, Schmid VS, Zillikens A, Harter-Marques B, amplification in other species. Molecular Ecology Resources Steiner J. 2011. Bimodal pollination system of the brome- 8: 980–982. liad Aechmea nudicaulis involving hummingbirds and bees. Palma-Silva C, Cavallari MM, Barbará T, Lexer C, Plant Biology 13: 41–50. Gimenes MA, Bered F. 2007. A set of polymorphic micro- Sydes M, Peakall R. 1998. Extensive clonality in the endan- satellite loci for Vriesea gigantea and Alcantarea imperialis gered shrub Haloragodendron licasii (Haloragaceae) (Bromeliaceae) and crossamplification in other bromeliad revealed by allozymes and RAPDs. Molecular Ecology 7: species. Molecular Ecology Notes 7: 654–657. 87–93. Peakall R, Beattie AJ. 1991. The genetic consequences of Vallejo-Marín M, Dorken M, Barrett S. 2010. The ecologi- worker ant pollination in a self-compatible clonal orchid. cal and evolutionary consequences of clonality for plant Evolution 45: 1837–1848. mating. Annual Review of Ecology Evolution and Systemat- Peakall R, Smouse PE. 2006. GenAlEx 6: genetic analysis in ics 41: 193–213. Excel. Population genetic software for teaching and Van Oosterhout C, Hutchinson WF, Wills DP, Shipley P. research. Molecular Ecology Notes 6: 288–295. 2004. MICRO-CHECKER: software for identifying and cor- Pfeiffer T, Klahr A, Heinrich A, Schnittler M. 2011. Does recting genotyping errors in microsatellite data. Molecular sex make a difference? Genetic diversity and spatial genetic Ecology Notes 4: 535–538. structure in two co-occurring species of Gagea (Liliaceae) Vekemans X, Hardy OJ. 2004. New insights from fine-scale with contrasting reproductive strategies. Plant Systematics spatial genetic structure analyses in plant populations. and Evolution 292: 189–201. Molecular Ecology 13: 921–935. Pinheiro F, Borghetti F. 2003. Light and temperature Weir BS, Cockerham CC. 1984. Estimating F-statistics for requirements for germination of seeds of Aechmea nudicau- the analysis of population structure. Evolution 38: 1358– lis (L.) Griesebach and Streptocalyx floribundus (Martius ex 1370. Schultes F.) Mez (Bromeliaceae). Acta Botanica Brasilica Wendt T. 1997. A review of the subgenus Pothuava 17: 27–35. (Baker) Baker of Aechmea Ruiz and Pav. (Bromeliaceae) in Raymond M, Rousset F. 1995. GENEPOP (version 1.2): Brazil. Botanical Journal of the Linnean Society 125: 245– population genetics software for exact tests and ecumeni- 271. cism. Journal of Heredity 86: 248–249. Widén B, Cronberg N, Widén M. 1994. Genotypic diversity, Ribeiro MC, Martensen AC, Metzger JP, Tabarelli M, molecular markers and spatial distribution of genets in Scarano FR, Fortin MJ. 2011. The Brazilian Atlantic clonal plants, a literature survey. Folia Geobotanica et Phy- Forest: a shrinking biodiversity hotspot. In: Zachos FE, totaxonomica 29: 245–263.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s website: Figure S1. Box plot describing the genotypic resolution of microsatellites in the sample for all possible combinations of loci from one to six. The edges of the boxes show the minimum and maximum number of genotypes, respectively, and the central line shows the average number of genotypes identiﬁed in the sample using N microsatellite loci. Downloaded from https://academic.oup.com/botlinnean/article/178/2/329/2416415 by guest on 02 October 2021