Plant material

P. hartwegii seed collection was carried out between 2017 and 2018, in two natural populations in the State of Mexico (Mount Tlaloc and Nevado de Toluca) and Veracruz (Cofre de Perote). In each population, several sites with different altitudinal levels were considered for the collection (Table 1). Cones (mature strobili) from 5 to 15 trees were collected at the sites, depending on the density, spatial distribution, and distance between them (100 m when possible). The selection of individuals was based on the presence of cones and the absence of pests.

The cones collected from each tree were taken to the laboratory for seed extraction, where they were then placed in an air column to separate them by weight. The empty seeds (without megagametophyte or embryo) floated and were removed, whereas the filled seeds remained at the base of the column and were refrigerated (at 4 to 6 °C).

Germination

The study was conducted during the first week of February 2019, in the greenhouse of the Graduate Program in Forest Sciences of the Colegio de Postgraduados. The experiment was established in a completely randomized design, in which for each altitudinal level (nine in total) four replications of 100 seeds of P. hartwegii were used, for a total of 3 600 seeds.

First, the seed was germinated in 310 mL forest tubes (one seed per tube), previously filled with a mixture of peat moss, agrolite, and vermiculite (3:1:1, respectively). The seeds were kept on the surface of the substrate, inside a small hole, without covering them completely. Then, the tubers were placed in 54-cavity trays and maintained in greenhouse conditions, with an average temperature of 24±12 °C. The humidity of the substrate was kept constant by irrigation. Germination was recorded daily for a period of 30 days. Once the seed germinated, it was buried in the substrate.

The germination data were used to calculate: 1) The germination percentage (germination capacity (GC)=(germinated seeds/total seeds) 100), 2) The peak value (PV), i.e., the maximum value of the sum of the germination percentage divided by the number of days in the evaluation and the number of days to reach 50 % of the total germination achieved, and 3) the germination value (GV) as the peak value multiplied by the average germination (^{Capilla-Dinorin et al., 2021}).

The seedlings produced remained in greenhouse conditions, and irrigation was provided every four days or earlier when required.

Assessment of plant growth and development

For this phase of the experiment, in the first week of March 2019, a sample of 60 plants per altitudinal level was taken and distributed in four replications of 15 plants under a completely randomized design. The diameter at the base (DB) and the total height (TH) of each plant were measured. Subsequently, during the last week of October 2019, the number of cotyledonary leaves (NCL) was recorded; the breakage of the grass stage, that is, the emission and elongation of epicotyl, which was evaluated as the percentage of plants with (PWCS) and without (PWOCS) grass stage, and the diameter of the apical bud (ABD). The DB and TH were obtained with a Mitutoyo^® digital vernier model 500-702 (±0.01mm) and Abacus^® graduated ruler (±0.1 cm), respectively. The growth in DB and TH was calculated from the differences between the initial and final measurements.

Statistical analysis

The variable data were subjected to Shapiro-Wilk and Bartlett's tests in order to verify the assumptions of normality and homogeneity of variances, respectively, using the UNIVARIATE and GLM procedures, with the SAS/PC for Windows version 9.4 statistical package (^{SAS Institute, 2003}). Since the GC data did not meet the assumptions of normality and homogeneity of variances, we chose to use the nonparametric Kruskal-Wallis test.

The other germination parameters met normality (p≥0.0714) and homogeneity of variances (p≤0.3356); therefore, they were analyzed using the GLM procedure of the SAS^® statistical package (^{SAS Institute, 2003}) with the model:

Where:

Y _ijk = Feature value of the k ^th altitude nested in the j ^th population in the i ^th repetition

µ = Effect of the overall mean

R _i = Random effect of the i ^th repetition

P _j = Effect of the j ^th population

A _k = Effect of k ^th altitude within j ^th population

e _ijk = Experimental error

The data for DB, TH, NCL, ABD, and PWOCS after checking the assumptions of normality (p≥0.772) and the homogeneity of variances (p≥0.0994) were analyzed with the MIXED procedure of the SAS^® statistical package (^{SAS Institute, 2003}) using the model:

Where:

Y _i = Value of the characteristic of the k ^th altitude nested in the i ^th population

µ = Effect of the overall mean, P _i is the effect of the i ^th population

A _k = Fixed effect of the K ^th altitude within the i ^th population

e _ijk = Experimental error

Finally, the separation of means was performed with the Tukey test (p=0.05).

Germination

The analysis of variance showed no significant differences between the three populations for the germination parameters evaluated, except for the peak value, while there were differences between altitudes in all parameters (Table 2).

In general, the average percentage of germination capacity for the seeds of the three populations studied was above 85 %. Although there were no significant differences among the populations, those of Nevado de Toluca registered approximately 10 % less germination in relation to the other two populations (Table 3). The peak value showed differences between populations: the PV of the seeds from the Cofre de Perote were 11 % higher than those of the other two populations (Table 3). In any case, the favorable germination parameters with no statistically significant differences in the three populations suggest good seed quality. Similar results were recorded by ^{Ortega-Mata et al. (2003)} with germination values above 90 % for several populations of P. hartwegii located in the State of Mexico.

In the Cofre de Perote, there are germination percentages of 80 to 95 % (^{Tejeda-Landero et al., 2019}). Germination percentages above 75 % indicate good seed quality (^{Capilla-Dinorin et al., 2021}). In this regard, studies on seeds from a seed orchard of Pinus leiophylla Schiede ex Schltdl. & Cham. showed that both the peak and germination values are good indicators of the quality of the seed produced (^{Gómez et al., 2010}). In the present investigation, the germination capacity recorded in the three populations may reflect good site quality. Nutrient-rich sites favor seed production by providing the necessary resources for seed development (^{Lauder et al., 2019}).

In the selection and establishment of seed areas, the quality of the site is a factor to be considered, since it ensures the production, in quantity and quality, of the seeds of the trees conserved in these areas (^{Manzanilla-Quiñones et al., 2019}). In addition to the site, certain environmental variables ―particularly precipitation― influence the reproductive processes that define seed formation and production (^{Andrade-Gómez et al., 2021}). Therefore, it will be necessary to monitor these factors in order to obtain more information regarding their impact on the reproduction and conservation of the species.

Although the results obtained could rule out any problem affecting seed quality by reduced populations of P. hartwegii (^{Capilla-Dinorin et al., 2021}), a more detailed study of various forest census aspects is required in the three populations in order to analyze the status of reproductive indices and their relationship with natural regeneration processes.

Regarding the altitudinal levels, all the values of the parameters studied were also high and indicative of a good quality of the seeds produced in stands located at different altitudes, except for the Nevado de Toluca stand, in the highest region (4 100 m). At this site, the germination capacity, the peak value, and the germination value of seeds decreased by more than 42, 36, and 63 %, respectively, compared to the values recorded at the other altitudes (Table 3).

A study on the relationship between environmental variations at different altitudes and the reproductive attributes of Pinus pseudostrobus Lindl. determined that seed germination of populations located at their upper altitudinal limits decreased in a ratio of approximately 3:1 with respect to those at intermediate limits; drought stress was identified as one of the causes of the decrease (^{López-Toledo et al., 2017}).

Poor seed quality has been explained by poor soil moisture (^{Lauder et al., 2019}). At higher altitudinal gradients, such as at 4 100 m, soils are stony and have low water retention capacity; in addition, at these altitudes, frost and cold winds in the off-season cause damage and loss of reproductive buds (^{Trant et al., 2018}). Therefore, these abiotic stress factors could cause the poor quality of seed produced at the 4 100 masl site.

These results show that the altitude influences the reproductive processes of P. hartwegii, which must be associated with the environmental variables that occur at each altitude and that cause stress directly impacting the populations. At altitudinal limits, environmental stress is more extreme, in addition, populations are more exposed (to snow, frost, and wind), and soils are generally poor in nutrients. Therefore, the conservation of P. hartwegii populations occurring near the alpine grassland, such as the one located at 4 100 m, requires implementing, in the medium and long term, management programs that will include prevention of logging, programmed fertilization in reproductive trees to improve the quality and quantity of the seed production, a study of the stand seed bank dynamics, and the protection of natural regeneration, as well as considering assisted regeneration.

Growth in Pinus hartwegii plants

The growth characteristics evaluated in P. hartwegii plants, except for the number of cotyledonary leaves, indicated significant differences between populations and altitudes (Table 4).

The number of cotyledonary leaves, with an average of 5 cotyledons per seedling, did not vary between populations or altitudes. The lower altitudes (3 400 and 3 600 m) had slightly lower average values of cotyledons (Table 5). The lack of variation in the number of cotyledons is unusual, because in conifer species, during the early stages of development there is a large variation in the number of cotyledons: this is considered an important characteristic for the detection of genetic variation and it is related to the initial growth of the seedling (^{Romanovskii and Morozov, 2019}).

In general, plants started to break the grass stage (epicotyl elongation) when they reached five months. At the end of the experiment, the percentage of breakage differed between populations and altitudes. Among the populations, the highest breakage percentage (approximately 70 %) was obtained with plants from the Cofre de Perote, exceeding by over 30 % those from Mount Tlaloc, which exhibited the lowest breakage percentage (Table 5). On the other hand, in regard to the altitudes, there was no trend in breakage between plants from lower and those located at higher altitudes; however, there was a significant difference between them. The lowest breakage percentages occurred at 3 600 masl on Mount Tlaloc and at 4 100 m on Nevado de Toluca. At both altitudes, the percentage of breakage was reduced by more than 70 %, compared to those of the Cofre de Perote (3 800 and 4 000 m), which had the highest values (Table 5).

Thus, it can be inferred that site conditions may have a greater influence than altitude on the breaking of the cespitose state of P. hartwegii. The cespitose state is considered to be an adaptive feature, acquired to survive fires and drought conditions (^{Robles-Gutiérrez et al., 2016}). The variation observed in the breaking of the cespitose state between populations and altitudes evidences the existence of different levels of genetic control in each population, probably influenced by the different conditions (temperature, humidity, fire incidence) prevailing at each altitude (^{Harsch and Bader, 2011}).

The advantages and disadvantages of the rapid breaking of the grass stage should be evaluated under field conditions, where the vulnerable P. hartwegii cespitose plants are exposed to changes in temperature, drought and competition between species in the ecosystem. Although, the cespitose condition confers resistance to abiotic stress factors (^{Dixit et al., 2020}), it has also been documented that, if the grass stage is prolonged for a few years, the survival of the plants will be at risk due to ecological competition from fast-growing species (^{Jin et al., 2019}). It was also demonstrated that under drought conditions, Pinus palustris Mill. plants at the grass stage were very vulnerable because their water potential, cell turgor and photosynthetic capacity declined rapidly when the water stress was prolonged (^{Hart et al., 2020}). Plants with a prolonged grass stage at 3 600 m on Mount Tlaloc and at 4 100 m on Nevado de Toluca could be in trouble if exposed to frequent drought and competition events.

P. hartwegii plants from the Cofre de Perote stood out for their height growth. These were 25 % taller, and their apical bud size, over 85 % larger, than those of Mount Tlaloc (the shortest among the populations), but their diameter was smaller by over 20 % compared to those of Nevado de Toluca (Table 5). ^{Benavides et al. (2011)} indicate a similar difference in diameter and height growth in P. hartwegii plants from the Cofre de Perote with respect to populations of the State of Mexico and Mexico City. This is explained by the rapid breaking of the grass state of the plants of the Cofre de Perote, which allowed them to grow faster and take better advantage of their resources. In plants that emerge more quickly from the grass stage, most of the reserve substances are destined for growth in length (epicotyl elongation) and apical bud development, while plants at the grass stage with little or no growth in height allocate their resources to the growth of the diameter (at stem base) and root (^{O’Brien et al., 2008}).

A study on the grass stage of P. palustris seedlings concluded that the plants decrease in vigor if the grass stage remains for several years (due to competition with other species and to diseases), with the consequent threat to their survival; in addition, stress factors such as drought prolong the cespitose condition for several years (^{Jin et al., 2019}).

As for the diameter of the apical bud, the plants from Mount Tlaloc were slightly smaller (3.9 mm) than those of the other two populations (>4.0 mm). This coincides with the grass state breakage parameter, which was lowest in the plants of Mount Tlaloc. Apical bud size may indicate increased mitotic activity prior to hypocotyl elongation, both of which are processes influenced by the environment. In particular, conifer populations tend to be genetically differentiated, and their growth patterns, such as bud formation and elongation, adapt to changes in ambient temperature ranges to avoid severe damage and cell death (^{Dixit and Kolb, 2020}).

Therefore, the populations of Nevado de Toluca, located at higher extreme altitudes (4 100 m), would be at greater risk, not only because of poor germination, low cespitose state breakage, and slow growth, but also because of the predictions of the climatological models for this location, which forecast low rainfall and high temperature conditions (^{Manzanilla et al., 2019}).