Study area
This study was undertaken in a natural population of A. aculeatum, located in the lot number 12 of the Manaus Agrarian Project, in a rural property named “Natajuba” (latitude -02°53’27.9”S and longitude -60°06’08.2”W) in the state of Amazonas, Brazil (Fig. 3). The region is characterized by a tropical forest climate (Af type), according to the Koppen-Geiger world map of climate classification [39]. The population is bordered to the East by the hydrographic basin of the Tarumã-Açu river, and to the West by a native forest that is part of the legal reserve of the “Natajuba” property and other properties within the Manaus Agrarian Project. North and South boarders are delimited by the rivers “Cuieiras” and “Jacaré”, respectively. Each side of the streams has a riparian forest of width larger than 200 m. Two other natural populations of A. aculeatum are located 400 m from the northern and more than 1,000 m from the southern boarders of the population studied here. Additional A. aculeatum populations are located at distances higher than 3,000 m. The population used in this study has been exploited continuously since 1996 to provide A. aculeatum fruits to the Manaus market. This population is also part of the in situ conservation of the superior germplams program from Embrapa Western Amazonia, within the project “Research, development, and innovation in oil producing palm plants and the economical use of by-products and residues”, PROPALMA (Embrapa-Propalma).
Sampling
In March 2011, 12 mother plants of A. aculeatum with maturing fruits generated by open-pollination were identified in the population area [4]. Twenty-five fruits were collected from each mother plant and placed within properly identified polyethylene bags and taken to the seed laboratory of the Western Amazonia Embrapa, in Manaus (Brazil). The pulp of each fruit was removed and the seeds obtained were dried at a temperature of 30 °C up to the point where the moisture content was 14.5 %, allowing the separation of the seed tegument by mechanical breakage [3]. Soaking and germination processes were immediately carried out [3]. The offspring germination phase was conducted during three months in the greenhouse. A total of 120 offspring from 12 mother plants of A. aculeatum was obtained. The number of offspring obtained per plant varied from 2 to 15. The distance between the mother plants in the area ranged from 6.2 to 83.2 m, with a mean of 37.7 m and a median of 37.1 m. Based on the geographic distribution of the mother plants, the center point between them was identified. From this central point four transects were established to conduct sampling of adult plants that are potential parents (pollen donors) of the offspring obtained from the mother plants. The end of each transect was established when there were no more plants within a distance of 100 m. Each transect was 10 m wide. These transects were extended to the Northeast, Southeast, Northwest and Southwest directions. In the four transects, 112 possible pollen donor plants were sampled. The distance between possible pollen donors varied from 0.5 to 351.4 m, with a mean of 89.6 m and a median of 78.9 m. The geographic position of each adult plant (pollen donor and mother plant) was determined by using a global positioning system (GPSmap 60CSx - GARMIN).
Leaf samples from 120 offspring, 12 mother plants and 112 potentials parent plants (pollen donors), were collected and stored in silica gel at -20 °C in the Laboratory of Molecular Biology at the National Institute of Amazonian Research (LTBM-INPA). Total DNA was extracted according to the cationic detergent protocol CTABLE 2× (Cationic Hexadecyl Trimethyl Ammonium Bromide) [40] and quantification was performed according to Ramos et al. [20].
This research was supported by the norms of Resolution 21 - 31 August 2006 - of CGEN (Conselho de Gestão Do Patrimônio Genético - Ministério do Meio Ambiente) (http://www.mma.gov.br/estruturas/sbf_dpg/_arquivos/res21cons.pdf). Material collection was registered in SISBIO (Sistema de Autorização e Informação em Biodiversidade/Instituto Chico Mendes de Conservação da Biodiversidade - ICMBio/Ministério do Meio Ambiente - MMA), voucher number 39950-4. A sample of the taxon Astrocaryum aculeatum G. Meyer (tucumã-do-Amazonas) was deposited in the INPA herbarium under No. 246369.
Microsatellites amplification
In this study, 12 microsatellite loci developed for A. aculeatum (Aac01, Aac02, Aac03, Aac04, Aac06, Aac07, Aac09, Aac10, Aac11, Aac12, Aac13 and Aac14) were used [41]. These microsatellites were amplified by polymerase chain reaction (PCR) using the Veriti Thermal Cycler (Applied Biosystems) in a total reaction volume of 10 μL, containing 10 ng genomic DNA, 1× buffer (10× standard Taq reaction buffer), 210 μM of each dNTP, 1.5 mM MgCl2, 0.16 μM of forward and M13 labeled primers (FAM or NED dyes) [42], 0.32 μM of reverse primers, 1.05 U Taq DNA polymerase (Invitrogen), and 3.49 μL of ultra pure water. The amplifications via PCR occurred in two phases, the first being specific for the primers and the second to connect the M13. The first stage began by stabilizing the temperature at 68 °C for 2 min and at 92 °C for 30 s, followed by 30 cycles (30 s at 92 °C for denaturation process, 35 s at the primer-specific annealing temperature {Table 1 of [41]}, and 30 s at 68 °C {72 °C for Aac07 and Aac11} for extension); the second step consisted of 15 cycles (30 s at 92 °C, 30 s at 53 °C, 30 s at 72 °C) and a final extension at 72 °C for 15 min followed by a period of 15 min at 68 °C [20, 41].
Amplification products were checked by electrophoresis on 1.5 % agarose gels stained with GelRed (Biotium) in 1× TBE buffer (pH 8.0). Amplified products of the PCR were submitted to an automatic DNA analyzer by capillary electrophoresis in the ABI 3130XL Genetic Analyzer (Applied Biosystems). The ET-550 ROX size standard (GE Healthcare) was used to determine the size of the alleles. Amplified fragments were observed and analyzed with the GENEMAPPER v4.0 software (Applied Biosystems).
Statistical analysis
Analysis of genetic diversity and fixation index
Genetic diversity was determined to compare adults and offspring, using the indexes total number of alleles over loci (k), average number of alleles per locus (A), number of private alleles in each generation (A
p
), and the observed (H
o
) and expected (H
e
) heterozygosities. These indexes were estimated using the GDA program [43]. Inbreeding was estimated using the fixation index (F). To test whether the F values were statistically different from zero, 1,000 Monte Carlo permutations of alleles among individuals, associated to a Bonferroni correction (95 %, α = 0.05), were obtained using SPAGeDi 1.3 [44]. To investigate if the mean values of A, H
o
, H
e
and F were significantly different between adults and offspring, the Student t-test was used, with a prior verification of the homogeneity of variances of the two groups, using a Fisher’s F-test. These analyzes were performed using the var.test and t.test functions of R package from the R project [45].
Analysis of the spatial genetic structure
The intrapopulation spatial genetic structure was studied using the mean coancestry coefficient (θ
ij
) between pairs of adult plants, calculated according to Loiselle et al. [46] and using the SPAGeDI program. To visualize the spatial genetic structure, values of θ
xy
were plotted against ten distance classes with the same number of pairwise individuals. In order to verify whether the spatial genetic structure had a significant deviation from a random structure, the CI of 95 % was calculated for each θ
ij
observed value and each distance class, using 10,000 Monte Carlo permutations of individuals among different distance classes. To compare the spatial genetic structure with other studies we estimated the S
p
statistic: Sp = − b
k
/(1 − θ
1) [23], where θ
1
is the average coancestry coefficient calculated in the first distance class (0 to 21 m), and b
k
is the slope of the regression curve in relation to the logarithm of the spatial distance (up to 361 m). To test the intensity of SGS, the spatial position of the individual was permutated 10,000 times to obtain the distribution frequency of b
k
where the null hypothesis states that θ
1
and ln d
xy
are not correlated (d
xy
is the spatial distance between individuals x and y). These analyses were run using SPAGeDI 1.3 program.
Analysis of the group coancestry and population effective size
The group coancestry (Θ) [47] was estimated for the adult plants from pairwise coancestry coefficient between all pairs of individuals (θ
ij
), using the estimator described in Loiselle et al. [46], implemented in the SPAGeDI program: Θ = [0.5n(1 + F
p
) + ∑
n
i = 1
∑
n
j ≠ i
θ
ij
]/n
2, where n is the number of sampled individuals, F
p
is the inbreeding coefficient of the population, estimated from the fixation index (negative value are assumed as zero). The effective population size (N
e
was calculated following Cockerham [48] from the variance of gene frequencies due to genetic drift (σ
2
p
= [(n − 1)/n)Θ + (1 + F)/2n]p(1 − p), where n is the sample size, p is frequency for a given neutral allele and F is the average inbreeding coefficient. In an idealized population under random mating, σ
2
p
value for a group of n individuals is σ
2
p
= p(1 − p)/2n and as in a idealized population there is not related and inbred individuals, the term n can be substituted by N
e
: σ
2
p
= p(1 − p)/2N
e
. Thus, we can equate both σ
2
p
expression and derive the variance effective population size, \( {N}_e=\frac{0.5}{\varTheta \left(\frac{n-1}{n}\right)+\frac{1+F}{2n}} \).
Analysis of the pollen flow, dispersal patterns and dispersal kernel estimation
For the paternity analysis, the CERVUS 3.0.3 program [49] was used, based on a categorical maximum likelihood method. The offspring paternity was determined by the Δ estimated statistic, calculated using simulations, considering 10,000 repetitions (simulated for the offspring), zero error rate at the loci (0.00) and all the 124 reproductive palm trees (112 adults + 12 mother plants) as pollen candidates for the offspring (60 % of sampled pollen donors collected in the study area). We adopted the confidence levels of 80 % as suggested by Marshall et al. [49] for the paternity assigned. Self-fertilization was also considered as a possibility and was estimated. The pollen immigration rate (m
p
) within the area was estimated as the number of offspring for which no father was assigned in the sampled area, divided by the total number of sampled offspring. The pollen dispersal distance for each progeny was calculated as the distance between the seed trees and the putative pollen donors by the Euclidian distance between two points. To verify whether the reproduction patterns were due to the distance between plants, the frequency of pollen dispersal curve was compared with the spatial distance among all plants using the Kolmogorov-Smirnov test [50]. The effective pollination neighboring area (A
ep
) was calculated assuming a circular area around a central seed tree, by A
ep
= 2πσ
2
p
[51], where σ
2
p
is the axial variance of the pollen dispersal.
The combined probability of exclusion of second parent, P
2p
[52], was estimated with the NM+ program [53]. We also estimated pollen flow, selfing and pollen dispersal distance assuming an exponential power dispersal kernel [37], implemented in the NM+ program [53]. This program is based on neighborhood model [54]. In this model, the pollen dispersal distance and patterns are not derived from individual paternity assignments, as in the case of Cervus, but indirectly from a spatial explicit mating model. The model considers that paternity of an offspring may result from: i) self-fertilization with probability s; ii) migrant pollen from outside the plot, with probability m
p
, or iii) outcrossing with a male located within the plot, with probability 1-s-m
p
[54]. The NM+ was matched with initial settings using categorical paternity assignment for our study plot. The neighborhood parameter was set to ‘infinite’ to include all sampled adults in our study plot as the neighborhood size [53]. Pollen dispersal was modeled using the exponential-power family parameter [37, 53] with estimates given of the scale (a) and shape (b) parameters from which the average distance of pollen dispersal (δ) is estimated.
Ethics
Not applicable.
Consent to publish
Not applicable.
Availability of supporting data
The data sets supporting the results of this article are included within the article.