Phenotypically plus tree selection and seed collection areas

During November and December 2017, 42 phenotypically plus trees were selected in stands of P. pseudostrobus var. oaxacana of Santa Catarina Ixtepeji, Teococuilco de Marcos Pérez, Santa María Jaltianguis and San Pedro Yolox municipalities in the Sierra Norte of Oaxaca State using the comparison or control tree methodology (^{Bramlett et al., 1977}) in which the characteristics of an outstanding or plus candidate tree is compared to the best five neighboring trees within a population, with a minimum distance of 25 m and maximum of 50 m. The candidate and control trees were in the reproductive age (51.7 ± 8.0 years), came from the dominant stratum, with a straight stem, not forked or twisted, round crown, insertion of branches at 90° to the stem, crown size 1/3 of the total height of the tree, diameter at breast height (54.5 ± 8.2 cm), height (36.4 ± 4.6 m), total volume (3.8 ± 1.4 m³) and height of clean stem (21.8 ± 4.6 m) higher than the average of control and healthy trees, free of pests and diseases, obtained with a Pressler's drill) (^{Muñoz-Flores et al., 2013}) (Table 1). The heights were measured with a clinometer (Suunto, modelo PM5/1520) and the diameters with a diameter tape (Forestry Suppliers Inc., modelo 283D/20F).

The crown dominance index as a measure of pollen availability for neighboring specimens of the tree selected as superior, was calculated by subtracting the candidate's height minus the average height of the controls; the site quality index was obtained from the quotient between the average height and the average age of the tree selected as plus and its controls; and the site productivity index of the quotient between the average volume and the average age of the tree selected as superior and its controls (^{Vallejos et al., 2010}).

A total of 1 058 cones were collected from the selected trees, which were identified and transferred to the INIFAP forest greenhouse in Valles Centrales de Oaxaca to be analyzed during the 2018-2019 period. The cones were separated by a select tree (42) in identified raffia bags, where they remained between 2 and 3 days; finally the cones were dissected to release the seed, which was stored for 2 to 3 months in paper bags.

Cone and seed analysis

The analysis of the cones was based on the methodology of ^{Bramlett et al. (1977)} and ^{Mosseler et al. (2000)} with modifications according to the study. The indicators or reproductive characteristics that were analyzed were: green weight (PVC, g) of each cone determined by means of an analytical balance with a 200 g ± 0.1 mg capacity (Shimadzu modelo ATY224); length (LC, mm) and cone diameter (DC, mm) were measured with a digital ± 0.2 mm caliper (Titan^® Classic). Afterwards, all the cones separated by tree were exposed to solar radiation in the greenhouse for two to three weeks to favor the opening of scales and extraction of seeds. The dry weight of each cone (PSC, g) was recorded after the opening of the scales at room temperature, two to three weeks after collection. The total number of scales per cone (NE) was counted, the number of seeds developed per cone (NSDC) corresponded to mature seeds (including empty and full); and the weight of the seeds developed by cone (PSDC, g) was determined. With these data, the following relationships were calculated:

Where:

CH = Moisture content (%)

PVC = Green weight of the cone (g)

PSC = Dry weight of the cone (g)

CF = Cone shape coefficient, cone diameter (DC, mm), cone length (LC, mm)

EPSD = Fully-grown seed production efficiency

PSDC = Fully-grown seed weight/cone

PPS = Average weight per seed

NSDC = Number of fully-grown seeds per cone

PEE = Scale efficiency percentage

NE = Total number of scales per cone (as a measurement of effective scales for seed production)

In order to determine the amount of full and empty seeds and thus, to obtain the reproductive efficiency and inbreeding index per cone and tree (^{Bramlett et al., 1977}), The cone and seed variables were taken from a sample of 50 cones.

Afterwards, five reproductive indicators proposed by ^{Mosseler et al. (2000)} were calculated: 1) total weight of cones on the tree (PCA, g) sum of dry weight of cones; 2) number of full-grown seeds per tree seeds (NSDA); and:

Where:

PSDA (g) = Sum of the fully-grown seed weights in the total number of cones

ER (g) = Reproductive efficiency, as a measurement of the energy rate used in the reproductive effort stored in the seed.

IE = Inbreeding index

SV = Empty seeds

SL = Full seeds

Climatic variables

In order to obtain physiographic and climatic variables of the sites of origin of the collected material, the geographic location coordinates of the trees selected as superior were entered on the WorldClim website: Research on climate change in forests: potential effects of Global warming in forests and the relationships between plant climate in western North America and Mexico (^{WorldClim, 2019}).

Data management and analysis

By means of the general linearized model, the trees, their provenances and sub-provenances were differentiated and the were established according to the geographical location of the trees. Ln(x) and square root transformations were performed to reproductive variables of the specimens selected as plus trees for compliance with the normality and homogeneity of the variances.

The model used was:

Where:

Y _ijk = Value of the dependent variable

( = Population mean

R _i = Effect of the i-^th geográfic region

P _j (R _i ) = Provenance or collection location nested in region

A _k (P _j ) = Tree nested in provenance

ε _ijk = Experimental error

The variance components associated with each source of variation (indicated in the model) and their contribution to the total variance were estimated by the VARCOMP procedure, REML option; progenies and provenances were separated by GLM with the Duncan test (( = 0.05); and for the associations between the morphological variables and the reproductive characteristics with climatic variables, a Pearson correlation analysis and the differences estimated with Duncan (SAS^®) were made (^{SAS, 2004}).

Reproductive characteristics of cones and seeds

The analysis of variance shows that the four regions, as well as the interactions provenance × region and tree × provenance had highly significant differences (p ≤ 0.01) in the reproductive characteristics of cones and seeds (Table 2). In other Pinus species such as P. leiophylla Schltdl. & Cham. (^{Gómez et al., 2010}), P. engelmanni Carr. (^{Bustamante-García et al., 2012}), P. greggii Engelm. ex Parl. (^{Rodríguez et al., 2012}) and P. patula Schiede ex Schltdl. & Cham. (^{Salaya-Domínguez et al., 2012}) there have been differences between sites and between trees for the features of these structures, indicating that they are influenced by the environment and that there is probably also genetic differentiation between populations, not being possible to separate the environmental effect from the genetic one (^{Viveros et al., 2013}).

The cones collected in the Yolox provenance had higher values than the general average of the collection in eight variables: length (105.6 mm), diameter (59.2 mm), dry weight (140.9 g), number of scales (138), number of fully-grownseeds (132), weight of fully-grown seeds (3.2 g), average weight per seed (0.024 g) and percentage of scale effectiveness (48 %) (Table 3). The moisture content was lower than the average (19 %), perhaps because when the cones were collected, they showed a greater degree of maturity compared to those from the other three communities. Those from Jaltianguis had the highest averages in shape coefficient (0.57) and green weight (169.3 g), lower average in the number of seeds developed (97) and percentage of scale effectiveness (34 %).

The cones collected in Ixtepeji recorded the lowest averages in eight of 11 evaluated characteristics: length (99.05 mm), diameter (53.5 mm), dry weight (85.2 g), green weight (140.2 g), number of scales (130), shape coefficient (0.54), fully-grown seed weight (2.1 g) and average weight per seed (0.020 g), except in moisture content (0.66 %), which was higher than the general collection average (Table 3).

The length and diameter of the cone for the four provenances are in the range reported in the same species (^{Márquez, 2007}; ^{Espinoza et al., 2009}; ^{Sáenz-Romero et al., 2011}; ^{Domínguez et al., 2016}); on the contrary, they are higher than the average values, for length (79.8 mm) and diameter (38 mm) determined by ^{Espinoza et al. (2009)}. The number of fully-grown seeds is below the range indicated by ^{Domínguez et al. (2016)} and higher in the weight of the fully-grown seeds per cone. Other studies report that P. pseudostrobus seeds weigh a minimum of 0.011 g and a maximum of 0.022 g (^{Hernández et al., 2003}), inclusive values in this study in Oaxaca.

Reproductive Indicators and Climatic Variables

Trees from Ixtepeji produced 35 cones on average, an amount that was significantly (p ≤ 0.05) higher than the 20 and 22 cones that the trees from Teococuilco and Jaltianguis had (Figure 1A). The total cones in each tree weighed from 1 990 to 3 428 g, in which 2 174 to 3 627 seeds were obtained, which together weighed from 49.29 to 78.09 g; the total of filled seeds was from 899 to 1 712 (Figure 1B -D), magnitudes that in each case were not significantly different (Duncan, 0.05). On average, a cone collected from a tree at Yolox weighed 144.83 g, which was 66 % higher than a cone collected from a tree at Ixtepeji; In addition, 71.75 seeds came from each cone collected from a tree in Teococuilco, which was 75 % higher than the number of full seeds from each cone of the Jaltianguis trees. The filled seeds obtained from tree cones in the Yolox locality weighed 24.27 mg (Figure 1F), a magnitude 17 % higher than the weight of the individual seed obtained from trees in Ixtepeji.

A) number of cones per tree; B) cones weight per tree; C) number of full-grown seed per tree; D) weight of full-grown seed per tree; E) vain seed per tree; F) full seed per tree; G) reproductive efficiency and H) inbreeding index. Different letters on the bars represent significant differences (Duncan, 0.05) and the vertical lines on the bars indicate the standard deviation.

Figure 1 Reproductive indicators of Pinus pseudostrobus var. oaxacana (Mirov) S.G.Harrison. 

In P. cembroides Zucc. (^{González-Ávalos et al., 2006}) and P. pinceana Gordon (^{Quiroz-Vásquez et al., 2017}) there were significant differences in cone production between evaluated years, however, not between populations. An average of 2 841.75 full-grown seeds were identified (full and empty), with an average of 1 419.5 (49.97 %) of full seeds with a minimum of 898 and a maximum of 1 712 (Figure 1E). For the same species, ^{Domínguez et al. (2016)} found 44 %, while ^{Bello (1988)} calculated 40 % of full seed; these results are higher than 7.5 % ^{Morales-Velázquez et al. (2010)} obtained in P. leiophylla, but 75.5 % lower than those found in P. pinceana (^{Quiroz-Vásquez et al., 2017}). Of the full-grown seeds, 50.03 % were vain, a lower value than those of other conifers such as Picea mexicana Martínez (78.3 %) (^{Flores-López et al., 2005}) and Picea martinezii T. F. Patterson (80.8 %) (^{Flores-López et al., 2012}).

The ratio of empty seeds to developed ones (inbreeding index) varied from 0.39 to 0.64 (Figure 1H) in this study; values higher than those found in P. caribaea var. caribaea Barret y Golfari (0.26) and P. tropicalis Morelet (0.03) by ^{Márquez (2017)}. This favorable behavior may be due to the fact that it is a source originally designed for seed production, for P. leiophylla the same author found a very high inbreeding index value (0.92); This can be attributed to the reduced size of the populations, which generates self-pollination (^{Morales-Velázquez et al., 2010}). Generally, the inbreeding index is associated with low pollen abundance and quality and increased self-fertilization; in sites with low tree densities, the probability that deleterious alleles form homozygous genotypes and cause abortion of embryos, increasing the amount of vain seed (^{Arista and Talavera, 1996}).

The average value for the reproductive efficiency of the species is 13 mg of seed per gram of dry cone, and in this context, the most efficient trees are from Teococuilco, with 16.2 mg (Figure 1G). The value was higher compared to P. leiophylla, 2.49 mg of g^-1 dry cone seed (^{Morales-Velázquez et al., 2010}), in contrast, the values are lower than those recorded in other conifers, such as P. rigida Miller, Picea mexicana Martínez, Pseudotsuga menziesii (Mirb.) Franco, P. patula, P. pinceana, 55.3, 23.7, 29.6, 16.35 mg g^-1 and 80 mg of dry cone g^-1 seed, respectively (^{Mosseler et al., 2004}; ^{Flores-López et al., 2005}; ^{Mápula-Larreta et al., 2007}; ^{Quiroz-Vásquez et al., 2017}). Reproductive efficiency is a relative indicator of the energy that a tree dedicates to seed production and is influenced by the weight and number of seeds filled per cone (^{Mosseler et al., 2000}; ^{Owens and Fernando, 2007}; ^{Castilleja et al., 2016}); therefore, it could be assumed that the populations of the species studied in Oaxaca show signs of a decrease in their reproductive efficiency.

From significant correlations between environmental variables, characteristics of selected phenotypes and reproductive indicators, the fact that higher sites, with lower temperatures and higher rainfall favor the cones to produce more seeds (r = 0.43), of better quality, stands out. and that reproductive efficiency increases (r = 0.38), and IE decreases (r = -0.38) (^{Sáenz-Romero et al., 2012}). On the other hand, higher values of normal diameter, age, volume and productivity of the site (r> 0.30), are associated with higher quantity and quality of cones per tree (^{Binotto et al., 2010}); old trees produce light cones with few seeds and light ones, and the inbreeding index is favored in the lower altitude places.