Location and description of the study area

The study was conducted in the Pueblo Nuevo ejido, located in the Sierra Madre Occidental mountain range in the El Salto region of southwestern Durango state, Mexico (Figure 1). The climate is semi-cold subhumid with rainfall in summer and an average annual temperature between 5 and 12 °C; the temperature of the coldest month ranges between 3 and 18 °C with winter rainfall accounting for between 5 and 10.2 % of the precipitation total (^{García, 1981}). According to the Series ll Edaphological Map, the soil of the study area is classified as Regosol (^{Instituto Nacional de Estadística y Geografía [INEGI], 2010}), while the vegetation type is pine-oak.

Figure 1 Location map of the study area in relation. 

Sampling

Six communities were selected where the seed tree regeneration method was applied in a period prior to 10 years and without subsequent silvicultural treatment, under the following names: Las Ciénegas, Ciénega Grande, La Colmena, El Venado, Los Bajíos and Pino Gordo. Each community was divided into three plots identified as: a) regeneration cut of seed trees (RST), b) regeneration under adult trees (RAT) and c) adult trees (AT). The RST plots are characterized by tree species that were established in the undergrowth after applying the seed tree regeneration treatment; these trees are individuals of Pinus durangensis Martínez and P. cooperi C. E. Blanco, species of major economic importance. In turn, the RAT plots show advanced regeneration, established under the adult tree cover adjacent to the plots where the seed trees were felled. Finally, the AT plots are composed of trees with a diameter larger than 7.5 cm at a height of 1.30 m from the ground and which, like the RAT, are adjacent to the plot where the seed trees were felled.

Sampling sites were located every 75 m within the regeneration area, following the cardinal and intermediate points (north, south, east, west, northeast, southeast, southwest and northwest). By projecting these orientations to adjacent wooded areas, eight sites were located to assess the diversity of tree vegetation and regeneration. The diversity of natural regeneration was evaluated in 5 x 5 m (25 m²) squares, and the diversity of trees greater than 7.5 cm in diameter, in 0.1 ha circles (Figure 2).

Figure 2 Sampling design in the seed tree regeneration areas. The diversity of natural regeneration was evaluated in 5 x 5 m (25 m²) squares, and the diversity of trees greater than 7.5 cm in diameter, in 0.1 ha circles.  

Data analysis

Field, species and number of individuals per species information was used to estimate alpha and beta diversity parameters. The estimators applied to describe the alpha diversity of the RST, RAT and AT plots in each of the communities were species richness (S), defined as the number of species present in each regeneration area; Simpson proportional diversity index ( λ=1-Σpi2) ; Shannon-Wiener index [ H´=Σ piln(pi)] ; and Pielou equity index ( J´=H´H´max ), where p _i is the proportional abundance of species i, and H´max = Ln(S) is the maximum Shannon index.

To estimate beta diversity, the Sørensen qualitative index was applied (^{Moreno, 2001}):

Is=2C(A+B)

To determine whether there is a significant difference in species richness between the RST, RAT and AT plots per community, rarefaction analyses were applied, while for the comparison of Shannon diversity, Simpson diversity and Pielou equity, the t statistical test (α = 0.05) was applied with the support of Past 3.16 software (^{Hammer, Harper, & Ryan, 2001}).

Species

Of the 30 species reported by ^{Medrano-Meraz et al. (2017)} for the El Salto region, P. cooperi, P. durangensis, P. strobiformis Engelm., P. leiophylla Schiede ex Schltdl. & Cham., P. teocote Schltdl. et Cham., P. herrerae Martínez, P. devoniana Lindley, Cupressus lindleyi Klotzsch ex Endl., Quercus rugosa Neé, Q. sideroxyla Bonpl. and Juniperus deppeana Steud. are present in the study area, so Alnus acuminata H. B. K. is added to the list of species present in that region (Table 1).

Description of diversity indices

The number of species present in the RST, RAT and AT plots was 11, 11 and 12, respectively. In the same plot order, the Shannon-Wiener proportional diversity indices were 1.53, 1.71 and 1.70; the Simpson indices were 0.65, 0.72 and 0.73; and the Pielou equity indices were 0.64, 0.71 and 0.69. At community level, the species richness recorded was 7, 4, 9, 9, 7 and 10 species in Las Ciénegas, Ciénega Grande, El Venado, La Colmena, Los Bajíos and Pino Gordo, respectively. In the same community order, the Shannon-Wiener indicators were 1.18, 0.91, 1.31, 1.17, 1.52 and 1.67; the Simpson index had values of 0.55, 0.50, 0.62, 0.52, 0.75 and 0.78; and the Pielou equity index had values of 0.60, 0.66, 0.59, 0.53, 0.78 and 0.73.

Table 2 shows the diversity indicators estimated in the RST, RAT and AT plots in each of the communities. The maximum number of species (S = 10) was recorded in the AT plot in Pino Gordo, while the minimum number (S = 4), which was equal in the three plots evaluated, corresponded to Ciénega Grande. In relation to the Shannon index, the minimum value (0.69) corresponds to the RST plots in Ciénega Grande and the highest (1.68) to the AT plots in Los Bajíos. In turn, the Simpson index ranged between 0.43 and 0.79; the highest value corresponded to the RST plot in Pino Gordo. The Pielou equity index ranged from 0.46 to 0.94; these values were recorded in the communities of El Venado and Los Bajíos, respectively, in the AT plots.

Alpha diversity comparisons

Figure 3 graphically compares the diversity in the plots at study area level. The rarefaction analysis showed that the species richness of the AT plots was significantly higher than that of the RST and RAT ones (P < 0.05). The rejection probabilities estimated with the statistical t-tests and the absence of overlaps between confidence intervals showed that the diversity and proportional abundance of species in the RST plots were significantly lower than in the RAT and AT areas. In turn, the proportional diversities of Shannon-Wiener and Simpson of the RAT and AT plots were statistically similar (P > 0.05) and greater than in the RST ones. Considering the Pielou index, the RST plots were statistically less homogeneous than the RAT and AT ones.

Figure 3 Point estimators and confidence intervals of the diversity indicators. Diversity in the regeneration in seed tree (RST) plots is different from those of regeneration in adult trees (RAT) and adult trees (AT). 

Rarefaction analyses carried out in each of the communities showed that species richness is similar in the RST, RAT and AT areas in the Ciénega Grande, El Venado and La Colmena communities, and that they are different in Las Ciénegas, Los Bajíos and Pino Gordo. In Las Ciénegas and Los Bajíos, species richness was statistically higher (P < 0.05) than in the AT plots (Table 2).

On the other hand, according to Table 3, the values of t and the probability of rejection (P < 0.05) indicate that the RST plots have proportional diversity indicators different from those of RAT in the Ciénega Grande, El Venado, Los Bajíos and Pino Gordo communities. According to the Shannon-Wiener and Simpson indices, the diversity of the RST plots and RAT are statistically different to AT plots.

Figures 4 and 5 show that, according to the Shannon index, the proportional diversity estimated in the RST plots, with respect to the RAT and AT ones, is higher in El Venado, Los Bajíos and Pino Gordo; lower in Ciénega Grande; and similar in Las Ciénegas and La Colmena. On the other hand, the Pielou equity index showed that the number of individuals per species in the RST plots is more homogeneous than that of the RAT and AT plots in the communities of El Venado and Pino Gordo and equally homogeneous in those of Las Ciénegas, La Colmena and Los Bajíos.

Figure 4 Confidence intervals of the Shannon-Wiener and Pielou indices in the Las Ciénegas, Ciénega Grande and El Venado communities. The absence of overlap indicates significant difference (P < 0.05) between plots. RST = regeneration in seed trees, RAT = regeneration in adult trees, AT = adult trees. 

Figure 5 Confidence intervals of the Shannon-Wiener and Pielou indices in the plots of the La Colmena, Los Bajíos and Pino Gordo communities. The absence of overlap indicates significant difference (P < 0.05) between plots. RST = regeneration in seed trees, RAT = regeneration in adult trees, AT = adult trees. 

Index of similarity between communities

Table 4 presents the qualitative Sørensen index that was estimated to assess species similarity between the communities. Considering the RST plots, the index ranged from 0.50 to 1.0. The highest value corresponded to Las Ciénegas and Los Bajíos, which share the same number of species; in contrast, the lowest similarity index was obtained for Ciénega Grande and El Venado, which share only three species. In turn, the indicators of similarity between the communities in the RAT plots ranged from 0.50 to 0.91, where the highest value corresponded to the communities of Las Ciénegas and Los Bajíos and the lowest to El Venado and Los Bajíos. The estimators to assess the similarity between AT communities ranged from 0.33 to 0.95; the lowest value belongs to Ciénega Grande and El Venado, and the highest was estimated in the La Colmena and Pino Gordo pair.

Indices of similarity between plots

According to Table 5, the Sørensen similarity indices between plots ranged from 0.67 to 1.0, indicating high species similarity among the RST, RAT and AT plots. The number of species in the RST plots, with respect to the AT plots, was higher in Las Ciénegas and Los Bajíos; the same in Ciénega Grande, El Venado and La Colmena; and lower in Pino Gordo. These results show that applying seed tree felling in mixed forests positively influences the presence of native species.

Discussion

^{Navar-Cháidez and González-Elizondo (2009)} report that, in plots subject to silvicultural treatments, the Shannon-Wiener and Simpson species richness and diversity indicators were equal to 7, 0.76 and 0.73, respectively; these values are similar to those found in the plots of the present study. Indicators at plot level are lower than those estimated at regional level, as pointed out by ^{Medrano-Meraz at al. (2017)}, who indicate that the Shannon-Wiener and Simpson species richness and diversity indicators were equal to 30, 2.09 and 0.82, respectively.

The evaluation shows that, in five of the six communities studied, tree species richness in the RST was similar to or greater than in the RAT and AT (Table 2). ^{Leyva-López, Velázquez-Martínez, and Ángeles-Pérez (2010)}, in a study conducted in mixed forests in the state of Oaxaca, reported results similar to those of this study. These authors argue that the differentiated presence of the species may be related to their ability to respond to the level of disturbance. ^{Smith et al. (1997)} and ^{Zavala (2001)} state that the number of species is a function of the ability of each of them to reproduce vegetatively or regenerate by seed, as well as the vulnerability to stress caused by the direct incidence of the sun's rays, changes in temperature, water availability and competition. Tree harvesting practices free the soil from adult tree cover and expose it to direct sunlight, limiting the establishment of shade-tolerant species, but favoring the regeneration and development of intolerant species (^{Smith et al., 1997}). ^{Sandor and Chazdon (2014)} report that the remaining trees in the managed areas explain the species composition of the established regeneration.

With respect to the effect that regeneration methods have had on species richness, ^{Chaudhary, Burivalova, Koh, and Hellweg (2016)} state that selection and retention systems, where the protection system and seed trees can be included, are rated as having the least impact on the change in species richness, in contrast to the effect of clear-cuts (^{Návar-Cháidez & González-Elizondo, 2009}). In particular, the selection system by individual trees allows the establishment and development of only shade-tolerant species, the group selection semi-tolerant species (^{Shields, Webster, & Nagel 2007}), the protection cuts to semi-tolerant and shade-intolerant species (^{Nasiri & Parsakhoo, 2012}) and that of seed trees to completely shade-intolerant species (Smith et al., 1996). Considering the above, the equal or greater number of RST species recorded in relation to AT (Table 2) is attributed, in part, to the presence of pine species capable of producing and dispersing seed from the boundaries of the AT plots and of facultative vegetative reproduction species within the RST plots. In this study, the presence of the species of the genera Alnus and Quercus is attributed to their ability to reproduce vegetatively from the roots remaining after felling, and the presence of Pinus is due to their ability to reproduce by seed; furthermore, all of them respond favorably to the direct incidence of the sun's rays. As examples of the response of oaks to vegetative reproduction and of pines to regeneration in open areas, ^{Návar-Cháidez and González-Elizondo (2009)} report that P. cooperi, P. teocote, P. leiophylla, Juniperus spp. and Q. sideroxyla were successfully established after the application of a clear-cut applied in the forests of Durango. ^{Alanís-Rodríguez et al. (2012)} also documented that species of the genus Quercus, in an ecosystem in northwestern Mexico, had high resprouting capacity after being subject to disturbances caused by fires.

The Shannon and Simpson indicators show that the proportional diversity of RST was higher than that of AT in four of the six communities (Figures 4 and 5), which is attributed to the change in the proportional abundance of individuals per species. ^{Hernández-Salas et al. (2013)} and ^{Leyva et al. (2010)} state that applying the seed tree method modifies diversity in regeneration. ^{Hernández-Salas et al. (2013)} report that periodic application of forest management practices, while ensuring the presence of species, changes their proportional abundance, thereby modifying diversity indicators. Changing the proportional abundance of each species modifies the composition and proportional diversity (^{Del Río, Montes, Cañellas, & Montero, 2003}). ^{Pourmajidian et al. (2010)} state that opening the canopy, apart from causing changes in the structure, increases diversity.

Consistent with the alpha diversity results, the beta diversity index indicates that there is high species similarity among the RST, RAT and AT plots (Table 5), demonstrating that the selection of a single species as seed trees in mixed forests does not guarantee their dominance, much less their unique presence, as occurs in forests characterized by a single species (^{Smith et al., 1997}). The species found in the boundaries and within the regeneration area contribute to species richness, under mechanisms of regeneration by seed and reproduction by shoots. In these cases, the number of species can equal or even outnumber the desirable ones, especially when in the regeneration areas there are species that reproduce by shoots. In case of wanting to control the composition, favoring the presence of one or several species, the application of complementary treatments such as thinnings should be considered to manipulate the proportion of desirable species through density control (^{Louman, Quirós, & Nilsson, 2001}). ^{Leyva et al. (2010)} and ^{Hernández-Salas et al. (2013)} affirm that the number of species that develop within the same area decreases when applying silvicultural treatments. Contrary to the above, when the adjacent trees are shade-intolerant, old enough to produce seed and represent the diversity of forest species, the cutting of seed trees will be a good option to take advantage of the timber forest resource while conserving the composition and diversity of the forest.