Description of the study area

The research was developed in the locality of San Antonio, in the Eastern area of the Municipality of San Miguel Chimalapa, Juchitán, Oaxaca, between the coordinates 16°39’27.56” N and 94°1’34.74” W, at a mean altitude of 1 075 m. The climate is AC(w₂)(w) semiwarm subhumid, with a relatively high humidity, summer rains with a mean annual precipitation of 576 to 3 300 mm, and a mean annual temperature of 20 °C (^{Rodríguez, 2014}).

The vegetation associated with the oak forest is characterized by including specimens of Liquidambar styraciflua L., Pinus pseudostrobus Lindl., Quercus corrugata Hook. and Q. peduncularis, Née with a normal diameter (ND) of 20.0 ± 15.0 cm and a total height (TH) of 14.6 ± 6.2 m. The arboreal stratum of the montane cloud forest has a higher variability in ND (19.0 ± 13.0 cm) and a lower variability in TH (12.1 ± 3.4 m), due to the great diversity of the species it includes, among which the most prominent are: L. styraciflua, Cornus disciflora Moc. & Sessé ex DC., Cinnamomum effusum (Meisn.) Kosterm., Capparis sp, Ficus pertusa L. f., Podocarpus sp, Pinus pseudostrobus, Q. corrugata, Rheedia macrophylla (Mart.) Planch. & Triana, Sweetia panamensis Benth. and Ternstroemia pringlei (Rose) Standl. Finally, the pine forest, characterized by the presence of L. styraciflua, P. oocarpa, P. pseudostrobus, Q. corrugata and Q. crassifolia Humb. et Bonp., has the highest dimensions for ND (24.1 ± 18.5 cm) and TH (14.8.0 ± 6.5 m).

Site selection and biomass sampling

In 2014, several visits were made to the area with the highest frequency of fires, with a surface area of 1 200 ha, where 15 sampling sites were established under supervision based on the variation of the topographic factors and the distribution and vegetation type; contrasts in the tree density (dense, semi-dense and fragmented) were sought in areas with 387, 347 and 261 trees ha^-1, respectively, according to the classification by ^{Castañeda et al. (2015)}.

The sampling units of 0.1 ha were evenly distributed in three types of forests: montane cloud, pine and oak forests, based on the slope, exposure, altitude and topographic position (Figure 1).

Figure 1 Delimitation of the study area and location of the sampling sites. 

In order to estimate the biomass on the forest floor, subsites of 9 and 1 m² were established within the 1.0 ha sites. The site number, exposure, slope (%), geographic coordinates and altitude were registered for each parcel of 0.1 ha (the last two data -coordinates and altitude- were obtained using a Garmin eTrex 20 GPS). The trees were inventoried based on the following data: species, normal diameter (cm) measured using a caliper, and total height (m) measured with a Suunto PM5/66PC clinometer. The diameter of the bushes (<2.5 cm) was measured at soil level using a 1/128” Foy vernier. The topographic data of the sites were also verified.

In both the montane cloud and oak forests a tree of each of the species found in the 15 sites was chosen at random for destructive biomass analysis, based on the mean diameter (20 cm); in the case of the pine forest, trees with a diameter of 25 cm were felled. A slice with a thickness of approximately 5 cm and a sample of a shaving made with a Pressler drill were taken from a height of 1.30 m; both were weighed in field in order to obtain the fresh weight (g) using a CS5000 Ohaus scale with capacity for 5 kg. The samples were identified and taken to the laboratory of Instituto Tecnológico del Valle de Oaxaca (Technological Institute of the Valley of Oaxaca, ITVO) in order to estimate their fresh or green volume (GV, cm³) by the displacement-in-water method; they were then dried in a Memmert Beschickung loading model 100-800 convection oven at a temperature of ±102 °C, until the pieces of interest attained a constant weight and reached their dry weight (DW, g). The dry-weight/green volume ratio gave as a result the specific gravity (SG, g cm^-3) per species, which was used to convert the tree volume, estimated based on the volume tables of the Forest Inventory of the State of Oaxaca, to biomass (^{Conafor- Semarnat, 2013}).

In the 9 m² sampling units, the aerial part of the bushes was cut in order to quantify their total green weight (g); a sample was subsequently selected and weighed in field (GV, g). The weeds, dead leaves and necromass were sampled in the 1 m² site. The total weight of each component and the green weight (g) of the samples stored in paper bags were estimated in field. The data of these samples were recorded for identification purposes, and they were taken to the laboratory, where they were dried. The samples of bushes, necromass, dead leaves and herbs were dried in the same oven at ±70 °C until they reached a constant weight, and then they were weighed again to calculate the dry weight (g) of each component using the scale that was utilized in field.

The conversion factor (CF) had to be estimated, in order to quantify the biomass in the compartments of bushes, dead leaves, weeds and necromass, by dividing the dry weight by the green weight of the samples. The biomass of each of these compartments was estimated by multiplying the fresh weight of the vegetal material in the site by the CF.

Fuel load

The estimated fuel loads are consolidated in two strata: 1) living fuels (weeds and bushes) and 2) dead surface fuels (dead leaves and necromass). This variable was estimated in the sites by compartment, using the equations proposed by ^{Morfín et al. (2012)} (Table 1).

Table 1 Equations utilized to determine fuel loads by forest floor compartment (^{Morfín et al., 2012}). 

L = Load or biomass (t ha^-1); DW = Dry weight (kg); N = Plot size (m²); dl = Depth of dead leaves (mm); d = Apparent density (t ha^-1); k = Constant (1.234); SG = Specific gravity (kg m^-3); QD = Quadratic diameter (cm); L = Transect length (m).

Statistical analysis

The database of the biomass of the wood and of the compartments (dead leaves, necromass, bushes and weeds) was formed. These data of the compartments by vegetation types were subjected to a variance analysis and to Duncan’s mean comparison test (P < 0.05). The SAS software package (^{SAS, 2005}) was used for the analysis routine. Due to the high variability found in the biomass and in the fuel loads, the Log(x) had to be used in order to fulfill the variance normality and homogeneity assumptions and to detect the statistical differences.

Surface and aerial biomass

The assessed vegetation types showed significant differences (p < 0.009) in the biomass of the weeds, which was considerably higher (Duncan, 0.05) in the pine forest (2.11 ± 0.67 t ha^-1) than in the montane cloud forest (0.79 ± 0.41 t ha^-1). The highest wood biomass contents were found in the oak forest (135.5 t ha^-1), and the lowest (43.31 t ha^-1), in the pine forest. Due to the high variability of the wood, it was not possible to detect significant differences using Duncan’s mean comparison test (Table 2).

Table 2 Biomass by surface and aerial compartment in each vegetation type. 

Different letters in the same row indicate statistical differences (Duncan, 0.05). The mean is accompanied by ± standard deviation.

On the other hand, no significant statistical differences (Duncan, p > 0.05) were found between the three vegetation types in the biomass accumulated in the bushes, dead leaves or necromass. Of these, the dead leaves of the oak forest had the highest proportion (7.33 t ha^-1) (Table 2).

The total surface biomass accumulation, excluding the wood of the trees, presented a statistically significant difference (p = 0.08) in each vegetation type. In the oak and pine forests, the figures were 15.248 and 11.192 t ha^-1, respectively, while in the montane cloud forest the total mean value was 8.778 t ha^-1 (Table 2 and Figure 2).

The vertical lines on the bars represent de standard deviation.

Different letters indicate significant statistical differences Duncan, 0.05).

Figure 2 Total average surface biomass accumulation by vegetation type. 

Forest fuel load, total and by compartment

The assessed vegetation types showed differences in the forest fuel load accumulated in the weeds (Duncan, p = 0.009), with higher values in the montane cloud forest. This compartment accumulated a larger amount of fuel in this vegetal association than oak and pine forests; the value for the weeds was 0.56 t ha^-1, and 0.66 t ha^-1 for the necromass. In contrast, the dead leaves and bushes compartments (p ≥ 0.08) did not confirm significant differences, with a fuel average of 1.26 and 0.13 t ha^-1, respectively; the necromass registered the highest variability (Table 3).

Table 3 Surface fuel loads by compartment and vegetation type. 

Different letters in the same row indicate statistical differences (Duncan, 0.05). The mean is accompanied by ± standard deviation. Pr>F= Significance.

When the biomass present in the various compartments (weeds, dead leaves, necromass and bushes) was added, the total biomass turned out to be 2.61 t ha^-1 in the montane cloud forest, and 2.01 t ha^-1 in the oak forest; these values are significantly different (Duncan, 0.05) (Figure 3). The topographic characteristics of the sites and, primarily, their steep slopes, ranging between 28 and 60 %, influence the scarce accumulation of fuel, which, probably during the rainy seasons, cause the dead leaves and the necromass to be dragged downhill, preventing excessive amounts of fuel to accumulate in the sites. The mountainside where the study area is located is oriented mainly toward the northeast, east, north, northwest and southwest, while the latitude ranges from 1 323 to 1 504 masl.

The vertical lines on the bars represent the standard deviation.

Different letters indicate significant statistical differences (Duncan, 0.05).

Figure 3 Total fuel load by vegetation type.