Materials and methods

Study area. The research was carried out in the Sierra Madre del Sur in the municipality of Chilpancingo, Guerrero (Southern Mexico) (Figure 1). The temperature range between 14 to 28 °C and rainfall between 800-2,500 mm. The weather is warm and sub-humid with rains in summer. The soil type is luvisol and regosol, while dominant vegetation is forest with agricultural areas (^{INEGI 2009}). The study area shows a high biological richness and the presence of endemic species (^{Luna-Vega & Llorente-Bousquets 1993}). Preliminary field surveys determined the main vegetation type in the area as Pinus-Quercus mixed forest. Also, surveys with local people were used to define the common conditions of post-fire regenerated vegetation communities.

Therefore, four areas with different management conditions were chosen: 1) Reforested 2) Fumigated, 3) Regenerated and 4) Control. Excepting the Control area, the reforested, fumigated and regenerated areas showed post-fire plant communities under different management conditions. All the areas are similar in physiographic characteristics (2,100 m asl slope of 20 %, and northeast exposure; ^{Méndez-Osorio et al. 2014}).

The forest fires were originated from prescribed burns that went out of control, the fires had an average severity and occurred between April and May (surveys with local people). The reforested area was burned in 2005 but reforested shortly after with Pinus douglasiana, Martínez. The fumigated area was burned in 2004 and fumigated with a broad spectrum non-selective herbicide that acts by contact (^{Rao 2014}). The regenerated area presented unassisted natural succession, after a fire burned it down in 2005. The control area presented a mature forest community conformed by no-native vegetation species because it was intensively reforested during the 60s and early 70s with Pinus radiata D. Don (^{Arteaga-Martínez 2000}).

Evaluation of the vegetation. The sampling was performed on September of 2012, during the rainy season. Five circular plots of 1,000 m² were established for each area, for a total of 20,000 m². A census of all the tree species was carried out, recording dendrometric measurements of total height (h), normal diameter (DBH) and crown diameter.

Data Analysis. Relative and absolute values of density, dominance (coverage) and frequency and Importance Value Index (IVI) were determined by each species (^{Curtis & Mcintosh 1951}). Species richness (S) was quantified as the total number of species and the Shannon index and Index of entropy (H´) was calculated for each area (^{Magurran 2004}). The equations used to determine the diversity indexes and ecological indicators of the species are showed in Table 1.

In order to draw conclusions about the differences in diversity and structural variables between areas, we analyzed the data using two different approaches. When the data satisfied the parametric assumptions, we used the analysis of variance ANOVA and, as Post Hoc, the Tukey HSD test. However, when data did not satisfy these assumptions, we employed the non-parametric Kruskal-Wallis with the Wilcox test as Post Hoc approximation (^{Zar 2010}). The statistical analysis where carried out using R (^{R Core Team 2016}).

Results

There were 14 species, 12 genera and 11 families. Families having two species were: Betulaceae, Pinaceae, Fabaceae and Fagaceae (Table 2). Alnus acuminata and Carpinus caroliniana (both Betulaceae) were only present in the regenerated area, while Acacia farnesiana and Indigofera palmeri (both Fabaceae) were in the regenerated and fumigated area. Quercus was present in all areas with species such as: Quercus glaucescens and Q. uxoris and Pinaceae was represented by Pinus radiata and P. douglasiana in the regenerated, reforested and control area.

Absolute and relative density (R_den). The control area registered the highest relative density values: 720 ± 15 N ha^-1 (average ± standard deviation) followed by the reforested area with 626 ± 4 N ha^-1, which was different from the other sampled areas (F = 130, df = 3, p < 0.001). The fumigated area showed the lowest relative density (122 ± 12 N ha^-1) (Figure 2A). The species with the highest relative density per area were reforested, Pinus douglasiana (45.1 %); regenerated, P. radiata (52.6 %); fumigated, Wigandia urens (54.1 %) and control: P. radiata (42.8 %; Table 2).

Absolute and relative dominance (R_dom). The contrast of means allows us to detect differences between sampling areas. Regardless of species, the higher canopy cover per area was observed in the control above all the areas (F = 36.7, df = 3, p < 0.001); however, other contrasts showed no significant differences. The control area showed a higher canopy coverage of 8,018 ± 536 m² ha^-1 while the fumigated area showed the lowest (86 ± 30 m² ha^-1). The canopy coverage value per dominant species were as follows: P. douglasiana: 49.5 % in the reforested area; P. radiata: 56.7% in the regenerated area, W. urens: 73.2 % in the fumigated area, while in the control area the dominant P. radiata showed 49.6 % of canopy coverage (Table 2).

Species richness (S). In order to make contrasts, the species richness (S) was computed by area, which allows to compute means, medians and standard errors for each area. Significant differences were found through the Kruskal-Wallis test (χ² = 13.4, df = 3, p = 0.003). The fumigated area showed the lowest average specific richness: 4 ± 1 species (average number of species ± standard deviation), however, significant differences were found in the reforested area (p = 0.016) with the highest number of species (6 ± 1) and in the control area (p < 0.008), (Figure 2C).

Shannon index (H'). Shannon index values showed the same trend as the species richness, registering significant differences (F = 7.37, df = 3, p = 0.002). The index of entropy showed significant differences (F = 7.97, df = 3, p = 0.002) (Figure 3). The area with the highest values of H' was the reforested with H' = 1.46 ± 0.15, which corresponds to 4 equally common species (exp^H'), followed by the regenerated area (H' = 1.11 ± 0.29) which corresponds to 3 equally common species (exp^H'), and the control area (H' = 1.06 ± 0.18), which corresponds to 3 equally common species (exp^H'). The fumigated area had the lowest value of diversity with H' = 0.70 ± 0.34 (Figure 2D), which corresponds to 2 equally common species (exp^H').

Relative frequency (R_fre). In the reforested area, the species with highest relative frequency were: P. radiata, P. douglasiana, Q. uxoris, Byrsonima crassifolia Kunth and Conostegia xalapensis D. Don with 17 % each. In the regenerated area: Pinus. radiata and Quercus glaucescens Bonpl, both species represented 25 % and in the fumigated area W. urens represented 38.5 %. In the control area the highest relative frequency was represented by three species, Q. glaucescens, P. radiata and Arbutus xalapensis Kunth 29.4 % (Table 2).

Importance Value Index (IVI). Pinus radiata and Q. glaucescens were the species with highest IVI (Table 2), these species were recorded in three of the four analyzed areas. Pinus radiata presented the highest IVI values in the regenerated (45 %) and control areas (41 %), and the second in the reforested area (20 %). Quercus glaucescens was the second species with the highest IVI in the regenerated (26 %) and control areas (28 %), and the third species with the highest importance value index in the fumigated area (17 %). In the reforested area, the species with the highest importance value index was P. douglasiana (37%) (Table 2).

Discussion

Alnus acuminata and Carpinus caroliniana belonging to Betulaceae were found only in the regenerated area. Those species are usually present in secondary successional stages of Pine forest and Pine-Oak forests, this can be explained as a direct consequence of the forest destruction and habitat fragmentation (^{CONAFOR 2020}, ^{Luna-Vega 2003}). Acacia farnesiana that was also in the regenerated area is often occurring in disturbed areas as an important element of secondary vegetation (^{CONABIO 2020}). The Fagaceae and Pinaceae families were the most representative in this study, which is similar to previous studies conducted in mixed forests under different conditions in Mexico (^{Alanís et al. 2012}, ^{Almazán-Núñez et al. 2016}, ^{Graciano-Ávila et al. 2017}).

The statistical difference between the relative density in the regenerated and fumigated areas could indicate that the fumigation process affected the development of individuals in the area after the fire. No differences were detected for the reforested and control areas, which is similar to what was found by ^{Almazán-Núñez et al. (2016)}, who evaluated a secondary mixed Pine-Oak forest in the Sierra Madre del Sur at different successional stages. Relative density in post-fire environment species adaptation plays an important role, such is the case of Pinus douglasiana that showed the highest number of individuals (which is expected because P. douglasiana was used for reforestation of the area). This species has the capacity to grow with success in post-fire environments because it requires abundant light when growing (^{CONAFOR 2018a}).

The forest fires promote the dehiscence of the cones of P. radiata individuals, allowing the dispersal of seeds (^{Vallejo et al. 2012}). Also, some factors that could be influencing individuals of P. radiata include orientation (northeast) and slope conditions (^{CONAFOR 2018b}) variables that could be included in further studies. Regarding the adaptation of individuals to the fumigated area we observed that W. urens could be important to agrochemical treatments since it was a species with high density despite being in a highly disturbed area (^{Nash 1979}, ^{Vibrans 2009}). The effect of the forest fire on the canopy coverage is significant since all the areas were compared to the control area, which showed the highest canopy coverage. Previous studies such as ^{Almazán-Núñez et al. (2016)}, reported the highest canopy coverage on a mature mixed forest.

The results regarding richness and diversity of species of this study were similar to some previous studies where it was found that fire do not affect trees diversity and on the contrary enhance species number in some cases, being considered as a significant evolutionary force (^{Keeley et al. 2011}, ^{Gallegos-Rodríguez et al. 2014}, ^{Hedo et al. 2015}, ^{Pereira et al. 2016}, ^{Foster et al. 2017}). The diversity of the fumigated area was similar to the control area but lower to the other areas. The influence of fire and fumigation was negative for plant community restoration, resulting on a lower value of richness and diversity of species for the fumigated area (Figure 2). The frequency was different in species, where W. urens was the most frequent. ^{Rubio-Camacho et al. (2016)} found this species in an area with evidence of fire but without concluding any relation. This opens the opportunity to further research regarding post-fire and species adaptation.

The importance value index of species of Q. glaucescens was second highest in this study for all areas; however, it was reported by ^{Alanís-Rodríguez et al. (2012)} and ^{Rubio-Camacho et al. (2016)} as the highest on post-fire studies; proving that Quercus species have a high cope capacity to fire. The species of Quercus have an evolutionary strategy to re-sprout; thus, individuals of this species are present in the burned area (^{Pausas 2004}, ^{Encina-Domínguez et al. 2009}, ^{Alanís-Rodríguez et al. 2012}).

As general conclusions, we found that the highest IVI on Pinus and Quercus genera, which indicated fire resilience on P. radiata, P. douglasiana, Q. glaucescens and Q. uxoris species. However, the area where the agrochemicals were used showed the lowest values of abundance, dominance, species richness and measured indexes. This suggests that fire associated with fumigation activities on plant communities affects significantly the structure and diversity of entire areas. Further studies on post-fire areas under different treatments are encouraged to increase the knowledge of the restoration of plant communities.