Introduction

Pinus patula Schiede ex Schltdl. et Cham. is a species native to Mexico that grows naturally throughout the states of Nuevo Leon, Tamaulipas, San Luis Potosi, Queretaro, Hidalgo, Estado de Mexico, Tlaxcala, Puebla, Guerrero, Oaxaca, Veracruz and Mexico City (^{Patiño and Kageyama, 1991}; ^{Perry, 1991}). Pinus patula is the Mexican species with the largest plantation area of outside the country (^{Wormald, 1975}; ^{Ladrach, 1985}), mainly because of the shape of their trunk, rapid growth and wood quality (^{Vela, 1980}; ^{Velázquez et al., 2004}). Forest plantations in Mexico, including those of P. patula, are managed with germplasm collected in natural stands without considering the quality of the trees (^{Vargas, 2003}).

In 2004, a trial program of progenies of P. patula started at north of the state of Puebla and east of the state of Hidalgo (^{Salaya et al., 2012}; ^{Morales et al., 2013}). Progeny trials were planted with selected seed from this species in South Africa and Colombia by CAMCORE (International Tree Breeding and Conservation) and germplasm collected in natural stands at the states of Hidalgo, Puebla, and Veracruz selected by the Colegio de Postgraduados (^{Salaya et al., 2012}). Several sexual seed orchards were established once the most fitted trees were determined after the progeny trials (^{Morales et al., 2013}).

A seed orchard is an isolated higher quality plantation, intended to reduce ovule pollination by pollen from trees that grow outside the orchard (^{White et al., 2007}). Seed orchards represent an important relationship between an improvement program and a plantation program (^{Fernandes et al., 2008}).

Environmental conditions, such as temperature and precipitation, can influence the formation of male strobili and therefore pollen production which, in turn, can impact cones and seeds production in Pinus trees (^{Owens et al., 2005}; ^{Owens y Fernando, 2007}). Cones and seeds analysis in a seed orchard determines the quantity and quality of the seeds (^{Bonner, 1993}). This information is used in breeding programs, management strategies of germplasm production units and seed production monitoring of a species. Therefore, the objective of this research was to evaluate the variation of the reproductive characteristics of cones and seeds in three production years of a sexual seed orchard of P. patula. We hypothesized that the variation of the reproductive characteristics is influenced by the genetic constitution of the trees and by environmental variation that leads low pollen production that affects cones and seeds development.

Materials and methods

Results and Discussion

The analysis of variance showed differences (p ≤ 0.01) in the CDW between production years and trees (Table 2). The experimental error contributed with the greatest percentage of the total variance for this and the other variables. The average CDW was 29, 31 and 32 g in 2014, 2012 and 2013. These results (36 and 50 g) were lower than those of P. patula trees in a forest at Pinal de Amoles, Querétaro, Mexico, and at a plantation in Madagascar (^{Patiño and Kageyama, 1991}). In our research, the difference between years could be due to genetic and environmental differences, according to the variance component values for the trees source of variation and experimental error (Table 2). Trees 39, 60 and 62 had the highest CDW in 2012, 2013 and 2014, each (Tables 3, 4 and 5). The CDW of tree 1 was the lowest in 2012 and 2013; only above tree 55 in 2014 (Tables 3, 4 and 5).

The PNS showed differences (p ≤ 0.01) between years and trees (Table 2). The PNS averages ranged between 91 and 117 per cone in 2014 and 2013, each. PNS difference could be due to differences in pollen production between years and to early pollination, since it stimulates fertile scales production that increases PNS (^{Owens et al., 2005}). PNS averages in our study were lower than those reported for P. patula (125 potential seeds per cone, ^{Dvorak, 2002}). Tree 29 had the highest PNS in “n” for 2012 and 2013 and tree 39 in 2014; this tree had a high potential in the three years. In contrast, tree 22 had the minimum seed potential only for 2012, and the seed potential in that year was less than the average of all trees in the three evaluated years.

The analysis of variance showed differences (p ≤ 0.01) in %AO between trees and years. The highest %AO (29 %) was registered in 2012 and 2014 mature cones, the lowest (25%) was found in 2013 mature cones. The ovules abortion may be due to pollen absence, lack of development of the pollen tubes or short pollen tubes, which in turn fails to fertilize the ovule (^{Owens et al., 2005}). The %AO were lower in P. patula in the three evaluated years than in P. leiophylla Schiede ex Schltdl. & Cham. cones (72.9 %) (^{Morales-Velázquez et al., 2010}). In contrast, these were higher in P. patula than in P. engelmannii Carr., P. durangensis Mart. and P. rzedowskii (Madrigal & Caballero) (20, 23 and 22%, each) (^{Bustamante-García et al., 2012}; ^{Bustamante-García et al., 2014}; ^{Castilleja et al., 2016}). Trees had low contribution to the total variation percentage. Trees 2 and 22 had the maximum values for this characteristic in 2012. Trees 50 and 74 showed the maximum value for this characteristic in 2013. Tree 60 reached the maximum value in 2014 and tree 1 showed the lowest value in 2012; trees 26 and 20 and 20 and 39 showed the lowest values during 2013 and 2014.

The analysis of variance showed differences (p ≤ 0.01) in the %FS between years and trees. The average %FS were higher than 50 and 58 % for the three years. These percentages represent 72 and 74 filled seeds per cone in 2012 and 2013 and 55 seeds in 2014. The number of filled seeds per cone per year, over the three years, was greater than the number of filled seeds (22 filled seeds per cone) in cones collected in natural populations of P. patula, and the harvested number (45 to 50 seeds per cone) in a seed orchard of this species in Zimbabwe (^{Geary and Pattinson, 1969}; ^{Barrett, 1972}; ^{Barnes and Mullin, 1974}). The number of filled seeds per cone in the three years in the present research was within the reported values (22 to 93 seeds filled per cone) for P. patula (^{Patiño and Kageyama, 1991}). The number of filled seeds per cone can considerably vary between years due to climate, such as low precipitation, low temperature or damages by insects (^{Andersson, 1965}; ^{Patiño and Kageyama 1991}; ^{De Groot and Schnekenburger, 1996}). Average %FS were higher in P. patula in each of the assessed years at the present research respect to the reported percentages for P. leiophylla (2 %) and P. rzedowskii (17 %) cones (^{Morales-Velázquez et al., 2010}; ^{Castilleja et al., 2016}) and similar to those recorded for P. engelmannii (49 %), P. durangensis (49 %) and P. ayacahuite var. veitchii (Shaw) (55 %) (^{Bustamante-García et al., 2012}; ^{Bustamante-García et al., 2014}; ^{Castilleja et al., 2016}). The %FS among trees considerably varied between years (Tables 3, 4 and 5). Trees 1 and 62 had the highest full seeds percentage in 2012. Number 22 had the lowest percentage in the three years. The high number of filled seeds per cone could be due the high crossing rate between genetically different individuals achieved by leaving individuals of different origins and 20 m spacing between trees of the same family in the orchard.

The analysis of variance showed significant differences (p ≤ 0.01) in the %ES between trees and between years. The %ES was low respect to seed potential during the three years. The %ES average was 15 % in 2012 and 2013 and 17 % in 2014. The %ES average in P. patula in the present research were lower than those reported for P. strobus L. (22 %), P. leiophylla (24 %) and P. rzedowskii (61 %) (^{Rajora et al., 2002}; ^{Morales-Velázquez et al., 2010}; ^{Castilleja et al., 2016}). Also, the averages for this variable were lower than those reported for P. engelmannii (27 %) and P. durangensis (26 %) (^{Bustamante-García et al., 2012}; ^{Bustamante-García et al., 2014}). However, the averages of %ES shown here were similar to those of P. leiophylla (18 % in row 18) in a seed orchard and from natural populations of P. ayacahuite var. veitchii (17 %) (^{Gómez-Jiménez et al., 2010}; ^{Castilleja et al., 2016}). Self-fertilization is the main cause of vain seeds production (^{Dogra, 1967}). In addition, orchard trees hybridization with P. teocote and P. greggii as close relatives of P. patula is likely (^{Hernández-León et al., 2013}) and may also negatively affect seeds quality (^{Wachowiak et al., 2005}; ^{Ávila-Flores et al., 2016}). Tree 68 had the maximum %ES in 2012 and tree 22 following years. Tree 62 had the lowest %ES in 2012 and tree 1 had the lowest values the following years (Tables 3, 4 and 5).

%PS had significant differences between years (p ≤ 0.01) and between trees (p ≤ 0.05). The average %PS was 2.6, 1.4 and 2.3 % from the seed potential in 2012, 2013 and 2014, each. These %PS were higher than those obtained for P. monticola Dougl. ex D. Don in a seed orchard at Kalamanca (1.2 %), Vernon, British Columbia, Canada (^{Owens and Fernando, 2007}) and in a wild population of P. leiophylla (0.9 %) at the State of Michoacán, México (^{Morales-Velázquez et al., 2010}). Trees 40 and 60 had the highest %PS in 2012 (Table 3), trees 50 and 55 had the highest values in 2013 and 2014, each (Tables 4 and 5). The low %PS in the cones throughout the three years confirmed low insects presence in the cones and seeds at the research orchard after three evaluation years.

The II showed differences between years (p ≤ 0.05) and between trees (p ≤ 0.01). The second major variation was with the trees, after the experimental error (Table 2). The II were 0.21, 0.22 and 0.24 in 2012, 2013 and 2014. The inbreeding rate due to tree varied from 0.05 in tree 1 to 0.53 in tree 22 in 2013. II in P. patula were lower in this orchard, in each of the three evaluated years, compared to the reported values for Picea rubens Sarg. (0.61), P. rigida Mill. (0.35), P. leiophylla (0.50) and Picea martinezii T. F. Patterson (0.75) (^{Mosseler et al., 2000}; ^{Mosseler et al., 2004}; ^{Gómez-Jiménez et al., 2010}; ^{Flores-López et al., 2012}). High II is due to high rates of self-pollination, which increases the presence of lethal recessive alleles that cause embryonic death (^{Bramlett, 1977}). Although, based on the average values of II in the three evaluated years, the self-pollination rate and crossbreeding between related individuals was low. This was possibly due to crosses between unrelated trees.

The SWE shows differences (p ≤ 0.01) between production years and between trees. The average SWE during the three evaluated years was 9.5 mg (Tables 3, 4 and 5). This value was higher than that reported (8.7 mg) in P. patula at the state of Veracruz, Mexico, and even higher than the average (8.5 mg) at 16 natural locations of P. patula (^{Barrett, 1972}; ^{Mendizábal et al., 2006}). Seed weight varied from 7 to 13, 7 to 11 and 7 to 12 mg in 2012, 2013 and 2014, each. Trees 62 and 74 had the highest values for this characteristic in 2012 and 2014 (Tables 3, 4 and 5). In contrast, tree 1 had the lowest weights throughout the three production years. The number of seeds per kg varied from 102, 653 to 110, 145. These values are within the average range of 92, 000 to 114, 000) seeds per kg recorded for P. patula in South Africa and Zimbabwe (^{Barnes and Mullin, 1974}).

There were differences (p ≤ 0.01) in the RE between years and between trees (p ≤ 0.05). Trees 1 and 22 had maximum and minimum RE during 2012. The RE varied from 2 to 41 mg of seeds per g of cones in trees 1 and 2 in 2012. The average RE value in our study was similar (16) to that obtained in P. leiophylla, higher (3.3) in P. rigida (^{Mosseler et al., 2004}; ^{Gómez-Jiménez et al., 2010}), and lower values (24 and 30) in Picea mexicana Martínez and Pseudotsuga menziesii Mirb., which are species with light cones (^{Flores-López et al., 2005}; ^{Mápula-Larreta et al., 2007}).

Although cones and seeds production expects to increase with age in P. patula trees, intensive management is recommended to promote cones production. Low pollen amount can lead to cones abortion; these require 80 % or more pollen on the fertile scales to maintain their development (^{Owens et al., 2005}). Therefore, pollen recollection and storage may be a common practice in the P. patula orchards. Pollen availability will allow its immediate dispersion since female strobili would reach receptivity and ensure early pollination. This would stimulate cones development and the formation of a greater number of fertile scales and, therefore, increase the potential seeds number (^{Owens et al., 2005}). Also, controlled breeding can be implemented in the P. patula trees to avoid self-fertilization, which is the main cause of vain seeds production and high inbreeding rates due to the increased probability of deleterious alleles to join in fertilization and form homozygous genotypes (^{Dogra, 1967}, ^{Hartl and Clark, 2007}). Crossing between different individuals promotes full seeds formation (^{Owens et al., 2005}). Irrigation can be part of the management to increase the synchrony between the receptivity of the female strobili and pollen release given that light irrigation can delay pollen dispersion (^{Owens et al., 2005}). Fertilization and hormones application, such as gibberellic acid (AG_4/7), can promote the flowering in Pinus trees (^{Ebell, 1972}; ^{Ross et al., 1985}).

Conclusions

Pinus patula showed high variation in reproductive characteristics. The greatest source of variation in the characteristics was the trees, after the experimental error. The percentage of filled seeds varied between years; but a high number of seeds filled by cones were recorded in the three years. The low percentage of vain seeds and low inbreeding index indicates a low level of self-fertilization in the seed orchard.