Vegetation composition and structure

Ten 10 m × 10 m random sampling plots were selected in each study site, within a previously established plot (2 500 m²) representative of the site in question. The geographical location was registered using a Garmin Rino 650 GPS navigator. ^{García et al. (2010)} and ^{Alanís-Rodríguez et al. (2011)} used similar sampling dimensions to characterize the tree vegetations in temperate forests of Colombia and in northern Mexico, respectively.

During the summer of 2016, the density was quantified by plant species in each unit (only when at least half of an individual was located within the boundaries of the sampling unit); the following mensuration parameters were measured: normal diameter (d, cm, Haglöf^TM Mantax Blue 950 Mm forest caliper), total height (h, m, Vertex IV Laser Haglöf^TM HS102 hypsometer), and crown canopy (m² ha^-1, CM112 50 m Truper ^TM measuring tape). The crown canopy was estimated based on the length (north-south) and the width (east-west) of each crown, in order to calculate the partial and total cover of all the species present at each site.

The ecological indicators of abundance (A), dominance (D), frequency (F), and importance value (IV) were determined (^{Brower et al., 1997}; ^{Magurran, 2004}). The Shannon-Wiener Diversity Index was utilized to determine the diversity of species (^{Matteucci et al., 1999}). The application of the index requires that the sampling be random, and that all the species of a plant community be present in the sample (^{Mostacedo and Fredericksen, 2000}).

The beta diversity was estimated based on a Bray-Curtis ordination model, which is a graphic representation of the variation in the plant composition. The analysis is based on an algorithm that allows analyzing the similarity between samples through the estimation of the similarity percentage (0-100 %) within a distance measure (^{Beals, 1984}). This is one of the most appropriate methods for the phytosociological, multivariate analysis; it was developed using BioDiversity Pro 2.0 software (^{McAleece et al., 1997}).

The dominance of the species was estimated based on the coverage data (^{Matteucci and Colma, 1982}; ^{Heiseke and Foroughbakhch, 1985}).

Statistical analysis

One way analysis of variance was applied to the Shannon-Wiener diversity index and to the Margalef species richness index (^{Steel and Torrie, 1980}). According to the Kolmogorov-Smirnov and Shapiro-Wilk statistical tests (with the correction by Lilliefors) and Levene statistical tests (Brown and Forsythe, 1974), the data exhibited homogeneity of variances and a normal distribution. They were statistically analyzed using the Tukey mean comparison test with a significance level of p=0.05 (^{Steel and Torrie, 1980}), in order to detect differences between the means of the Shannon-Wiener diversity index and the Margalef species richness index of the sites (^{Steel and Torrie, 1980}; ^{Ott, 1993}). All the statistical procedures were carried out using the Statistical Package for the Social Sciences, standard version 13.0 for Windows, SPSS Inc., Chicago, IL USA) (^{SPSS, 2004}).

Floristic composition

A total of 1 020 individual trees and shrubs were registered, with an average height of 6.2 m for the four sites, and 18 species belonging to six families were identified. Of these, the richest ones were Pinaceae (10), Fagaceae (3), Cupressaceae (2), and Ericaceae (2), which group 94 % of the registered taxa. The Betulaceae family had only one taxon. The families with the largest number of genera were Pinaceae, with 4, and Cupressaceae, with 2, which together account 66 % of the total genera registered at the four study sites. The remaining families included only one genus each. Outstanding species were Pinus (7), Quercus (3), and Arbutus (2); Abies, Alnus, Picea, and Pseudotsuga had one taxon each.

The families Pinaceae, Fagaceae and Ericaceae stood out at the four sites for their number of individuals-393, 486 and 76, respectively-, while Cupressaceae had 47, and Betulaceae, only 18. S1 and S3 had the largest number of families (5), followed by S2 (4) and S4 (3). The sites most abundant in genera (5) were S1 and S3, followed by S2 (4) and S4, where three genera were identified. Nine species were identified at S4, eight at S1 and at S2, and six at S3. Quercus sideroxyla was the best represented species at S3 (323) and S1 (81). Q. durifolia (69) was best represented at S2, and Pinus cooperi (58), at S4. Only one individual was registered for each of the species Alnus acuminate Kunth, Pinus cooperi and Quercus rugosa at S1 and S2; Quercus sideroxyla registered 4 at S4, and Alnus acuminata, nineteen at S3. Of the 1 020 quantified individuals, the site with the highest number of specimens was S3 (512), followed by S1 (206), S4 (168), and S2 (134).

Prominent floristic studies carried out in the WSM are those by ^{De León et al. (2003)}, ^{García-Arévalo (2008)}, ^{Valenzuela and Granados (2009)}, ^{Aragón-Piña et al. (2010)}, ^{Delgado et al.(2016)}, and ^{Graciano-Ávila et al. (2017b)}, according to whom Pinaceae and Fagaceae are the most abundant families; ^{Sánchez-González (2008)} and ^{González-Elizondo et al. (2012)} point out that the high abundance of these families is due to the fact that they include 46 % of the pine species and 34 % of the oak species of Mexico. Likewise, they have been confirmed to be the dominant species in the forests of Durango (^{Graciano-Ávila et al., 2017a}), and ^{Zúñiga et al. (2018)} have recently documented higher importance values for Pinus and Quercus.

Both families are of enormous economic interest in the forest ecosystems of the country, from which, respectively, 129 (80.1 %) and 141 (8.8 %) million m³ of roundwood were extracted in the years 1900-2012, adding up to 94.8 % of the national timber production (^{Semarnat, 2014}). Other studies (^{Mendoza, 1993}; ^{Gavilán and Rubio, 2005}) point out that this dominance may be due to silvicultural treatments geared at creating the adequate conditions -particularly of radiation- for the establishment of the commercially desirable species, whereby the number of species in the ecosystem is minimized.

Under traditional forest management conditions, there is insufficient control over the process or the functioning of the system to maintain the diversity of species -an essential requirement for sustainable forest management (^{Kazana et al., 2005}). For this reason, an adequate silvicultural treatment has an effect over the composition and structure of the forests and produces wood and other environmental services (^{Promis et al., 2010}).

Diversity of species in the study sites

The analysis of variance test detected significant differences in the diversity of species between sites and for the combination of sites (Table 1), due to the absence of similarity, at least in one site, in the species richness and the number of individuals.

Therefore, S3 (512) stands out for the number of individuals contained in it, and is followed by S1 (206), S4 (168) and S2 (134). This tendency coincides with the interval registered by ^{Aragón-Piña et al. (2010)}, ^{Delgado et al. (2016)} and ^{Graciano-Ávila et al. (2017b)}; on the other hand, ^{Návar-Cháidez and González-Elizondo (2009)}, ^{Ramírez et al. (2003)}, and ^{López-Hernández et al. (2017)} estimated a lower average.

Table 2 illustrates the mean comparison test for the Shannon-Wiener diversity index and the Margalef species richness index.

The Shannon-Wiener species diversity index was 0.93 (S2), 1.10 (S3), 1.35 (S1), and 1.51 (S4). The Margalef species richness index was 0.97 (S3), 0.98 (S2), 1.42 (S1), and 1.79 (S4). The average for the four study sites was 1.29 and 1.22, respectively. The main taxa were Pinus leiophylla and Pinus durangensis, which means that they have a greater structural dominance and a higher level of competitiveness.

According to ^{Margalef (1972)}, ^{Magurran (2004)} and ^{Ramírez (2006)}, the diversity is considered to be low when the values are below 2, medium when they are 3 to 5, and high when they are above 5. Consequently, the diversity at the study sites is low, like in the El Salto region of Durango (^{Návar-Cháidez and González-Elizondo, 2009}; ^{Delgado et al., 2016}; ^{Graciano-Ávila et al., 2017a}; ^{Medrano et al., 2017}), with diversity values of 1.33, 1.19, 1.81, and 1.94, respectively.

The variation in the diversity of species is determined by the edaphoclimatic and ecological conditions and by the conversion of the land to extensive agriculture and livestock breeding, as well as to various anthropic activities, all of which lead to the presence of a variety of plant communities (^{Aragón-Piña et al., 2010}; ^{Alanís-Rodríguez et al., 2011}; ^{Ramírez et al., 2013}; ^{López-Hernández et al., 2017}).

Site 4 had the greatest diversity; according to ^{Gordon (1968)} this may be due to its location within a valley that has a permanent current, is protected against dry winds, and is surrounded to the south and the east by a plateau of pine forests, and to the west, by a steep slope, a depression known as Quebrada del Infierno. Picea chihuahuana, Abies durangensis, and Pseudotsuga menziesii, three conifer species occurring at this site, are rare in Mexico in general, and in Durango particularly, where they are distributed as protected remnants with a high conservation status. The presence of Picea chihuahuana is particularly significant.

The proximity of this stand to the Tropic of Cancer provides an infrequent climate for Picea; only another species of this genus grows in Asia, at a locality with similar latitude. Furthermore, Site 4 lacks forest management, as it is located within an ecological reserve. Site 2 was the least diverse, due to the change in soil use from a forest to an extensive livestock breeding area.

Likewise, the variation in the diversity of species is greatly due to the historical precipitation registered at each site.

According to ^{Návar-Cháidez and González-Elizondo (2009)}, the diversity diminishes with the increase in the removed basal area. Habitats with low fertility reduce diversity, through stress caused by nutrients, and the high fertility overcomes the limitations imposed by stress, leading to simplified communities (^{Rosenzweig and Abramsky, 1993}; ^{Tilman and Pacala, 1993}).

Floristic similarity

The floristic composition at all four sites is represented by 18 species, of which they share only one (Juniperus deppeana Steud.), while Arbutus xalapensis Kunth is shared by sites S1, S2 and S3; Pinus cooperi is shared by sites S2, S3 and S4, and Quercus sideroxyla, by sites S1, S3 and S4; the last two stand out for their importance value.

According to the Bray-Curtis ordination dendrogram (Figure 2), the assessed sites have a low similarity. The sites with the highest similarity were between S1 and S3, with 29.81 %, followed by a similarity of 25.59 % in regard to S4; finally, S2 had a similarity of 12.94 % in relation to the next group (Table 3).

Figure 2 Bray-Curtis ordination dendrogram of the study sites. 

^{García et al. (2010)} cite similar results for two natural forests in Colombia, which have a similarity of 37 % as they share 21 out of 57 identified species, and conclude that the low similarity is due to differences in altitude and to certain characteristics that are specific to the area. In relation to four study sites of northeastern Nuevo León, ^{Ramírez et al. (2013)} register that those located within the same altitude gradient were similar in species composition. The authors contend that the dissimilarity between sites is due to the historical precipitation registered in them and to the characteristics of each ecosystem, as well as to differences in the altitudinal gradient.

In their assessment of the effect of silvicultural practices on the composition and diversity of species of a forest, ^{Luna et al. (2015)} registered that the similarity of species between the evaluated treatments increases with the time of application and is influenced by the biotic and abiotic conditions of the ecosystem resulting from the silvicultural treatment utilized.

Although the study sites were located within much the same altitude intervals and had similar mean temperatures in general, they exhibited a low similarity that ranged from 0.13 to 0.40 -an instance of the ecological complexity of the WSM (^{Delgado et al., 2016}). However, the greatest inequality between sites (S2 and S4) = 1.99) may have been caused by the distance between them and by the differences in the characteristics of each site; site (S2) exhibits soil use change from forests to grasslands for extensive livestock breeding, as well as a low precipitation; while at S4 there are steep slopes with a variety of moderate gullies and canyons with water sources, a high humidity, and a higher precipitation than that of the other sites; furthermore, it is located within a protected natural area (PNA) with ecotouristic, scientific, and educational activities.

Ecological indicators at the study sites

Tables 4, 5, 6 and 7 summarize the results corresponding to the ecological variables for each of the four assessed sites.

The four study sites exhibited a coverage of 9 063.5 m² ha^-1. ^{López-Hernández et al. (2017)} observed a similar coverage (8 462.76 m² ha^-1) when assessing a temperate forest in the east of the state of Puebla. Graciano et al. (2009) point out a coverage of 8 838.35 m² ha^-1 in ejido La Campana in the WSM at sites with a similar vegetation to that studied by the present research. The variations in the cover may be greatly due to the historical precipitation received by each site. According to ^{Valenzuela and Granados (2009)}, mountainous areas like those of the WSM remain temperate, with a high precipitation, favoring changes in the structure and the composition of the vegetation.

According to ^{Piedra (2016)}, the changes are due not only to environmental issues but also to management activities such as the distribution of woody material or soil conservation related construction and actions, e.g. cordoning off, which stimulate the accumulation of dead leaves, and consequently hinder seed germination.

^{Challenger et al. (2009)} document that the vegetation cover is affected by various activities, notably: destruction of the habitat, overexploitation, invasive species, pollutants, and climate change.

The estimated IVI indicates that in all four sites, dominance corresponds to Pinus spp., followed by Quercus spp. because their species have economic importance in the WSM. Thus, application of the silvicultural methods favors a higher density and frequency, and a larger coverage and basal area. Nevertheless, generic dominances make it possible to observe pine-oak or oak-pine forests in the same site , as documented in the Environmental Statutes of the state of Durango for the temperate forests of the area (^{Semarnat, 2007}).

The comparative study between adjoining sites made it possible to detect structural and composition variations occurring within a particular ecosystem. Although in general one conifer or broadleaf species may grow at different degrees of slope, other taxa have preferences for particular topoforms (^{Martínez et al., 2013}).

The association of Pinus cooperi var. ornelasi Martínez (Blanco) and Pinus durangensis with other Pinus and Quercus taxa is one of the most representative in the WSM of Durango; they are similar to those registered by ^{Márquez and González (1998)}, ^{Márquez et al. (1999)}, and ^{González et al. (2007)} and ^{González-Elizondo (2012)}.

According to ^{Guzmán (2009)}, when a species reaches the highest importance value, there is a clear ecological dominance, and usually a few species with the highest importance values are representative of the main population. Conversely, when several species compete at the site, none of them has an importance value above 50 %.