Study area

The La Lluvia mountain is located in the community of La Ciénaga, in Malinaltepec, Guerrero, at the coordinates 17º13’42.99’’ N and 98º38’3.01’’ W. Its altitude ranges between 2 120 and 2 440 masl (Figure 1). The climate is semi-warm subhumid A(C) w2 (w) (^{INEGI, 2008}); the mean annual precipitation is 2000 mm (^{INEGI, 2006}), and the mean annual temperature, 18 °C (^{INEGI, 2007}). The mountain is located in the southern Pacific hydrological-administrative Region, in the upper part of the basin of the Omitlán river. This is not only a strategic conservation area but also a source of water supply for the community of La Ciénaga, Malinaltepec, Guerrero.

Simbología = Simbology; Área de studio = Study area; Muestreo = Sampling; Curvas de nivel = Contour line.

Figure 1 Location of the study area and distribution of the sampling sites. 

Vegetation is made up of an oak forest (^{Inegi, 2017}), where the tree stratum forms a semi-closed canopy. The main species belong to the genus Quercus (Quercus elliptica Née, Quercus acutifolia Née, Quercus candicans Née, Quercus martinezii C. H. Mull., Quercus obtusata Bonpl., Quercus gentryi C. H. Mull., Quercus peduncularis Née); other species present in the area are Arbutus xalapensis Kunth, Befaria leavis Benth, Bejaria aestuans Mutis ex L., Vaccinium leucanthum Schltdl., Alnus acuminata Kunth, Ostrya virginiana (Mill) K.Koch, Clethra kenoyeri Lundell, Clethra hartwegii Britton, Licaria aff. capitata (Chamisso et Schlechtendal) Kosterm., Persea chrysantha Lorea-Hern., Clusia multiflora Kunth, Magnolia schiedeana Schltdl., Miconia glaberrima (Schlechtendal) Naudin., Fraxinus uhdei (Wenz.) Lingelsh., Phyllonoma laticuspis (Turcz.) Engl., Styrax argenteus C. Presl. and Daphnopsis nevlingii J. Jiménez Ram.

In order to determine the surface area, the vertices of the polygon comprising the mountain were georeferenced using a Global Positioning System (GPS, Garmin, eTrex 20x); the total surface area, of 44 ha, was subsequently determined using the ArcGis software, version 10.2.

Sample size and sampling design

The distribution of the sampling sites across the 44 ha was random; the Create Random Points tool of ArcGis 10.2 was used to adequately establish the sites (Figure 1). The amount of fuel per category -woody and fine litter- was expressed by size classes that determine the time that the fuel takes to reach a balance with the atmospheric moisture, known as delay time (^{Xelhuantzi et al., 2011}). The center of the sampling site was determined in the field with the aid of a GPS (UGarmin, eTrex 20x) and marked with a stake; the slope (%) was determined in each site using a Brunton^TM clinometer. Three planar interception lines oriented at 0°, 120° and 240° of azimuth with a Brunton^TM compass; the lines were 15 m long.

The woody fuels were quantified according to the planar intersections technique based on a count of branches or stems that intersect the vertical plane defined by a transect (^{Brown, 1971}); the 1 h fuels were counted at 2 m (0-0.6 cm); 10 h fuels were counted at 4 m, (0.6-2.5 cm), and 100 h fuels were registered at 10 m (2.6-7.5 cm), with the measures pre-established in a handmade caliper, and the 1 000 h fuel was registered along the whole 15 m line (>7.5 cm) -divided between healthy and rotting-, whose diameter was measured directly with a Truper T-3ME flexometer (Figure 2).

Línea = Line; Línea de intersección = Intersection line; Calibrador = Caliper; Cuadrante = Square; Centro del sitio = Site center.

Figure 2 Sampling site for quantifying woody fuels and leaf litter. 

The leaf litter was quantified in a 30 × 30 cm square at the end of each planar interception line, resulting in a total of 45 samples; the thickness of the leaf litter layer was measured using a Barrilito^TM ruler calibrated in cm. All the leaf litter collected in the 0.09 m² was introduced into brown paper bags; the bags were labeled and transported to the laboratory in order to obtain their dry weight.

Ten pre-sampling sites were surveyed in order to estimate the sample size (n); the sum of the 1 and 10 h fuel loads and the leaf litter per site were used for estimating the coefficient of variation (𝑆𝑥%) when occurring at all the sites; the final sample was determined using the following formula (^{Ancira-Sánchez and Treviño, 2015}):

Where:

n= Sample size

t= t Student value (with degree of freedom and probability value of 90 %)

Sx-%= Coefficient of variation

Sx-%= Standard error in perentage

Based on the previous formula, 15 sampling sites were determined for the study area, distributed across the 44 ha, with a coefficient of variation 𝑆𝑥% = 54.86 %.

Laboratory work

The leaf litter samples collected in the field were taken to the Biological Sciences Laboratory of the Universidad Autónoma de Guerrero. The dry weight was determined from the drying process in a Felisa^® oven at 70 ^oC; the leaf litter was monitored periodically until a constant weight was obtained, measured with a NOVAL TH-II scale with 0.1 g accuracy. The woody fuels were estimated using the formulas proposed by ^{Brown, 1971} (Table 1). The leaf litter load was calculated using the dry weight, which was extrapolated to t ha^-1 using the following formula:

Where:

LLL = Leaf litter load (t ha^-1)

LDW = Litter dry weight (g) in 0.09 m² (30 × 30 cm)

0.1111 = Factor of conversion from g in 0.09 m² into t ha^-1

Statistical analysis

The loads per site (15 sites, 3 lines per site) were estimated using the data of 1, 10, 100, 1000 h fuel and leaf litter; an analysis of variance (ANOVA) at 95 % confidence level was subsequently carried out in order to detect statistical differences between the sites for a given fuel; when significant differences were found, a Tukey mean comparison test (α = 0.05) was performed.

Analyses were also made to find statistical differences between fuel types (1, 10, 100, 1 000 h and leaf litter; n = 45); since the data do not follow a normal distribution, the Kruskal-Wallis non parametric mean comparison test was applied (^{Kruskal and Wallis, 1952}). Finally, correlations were made between the 1 h, 10 h and 100 h fuels and dead litter (t ha^-1) with the land slope (%) and the thickness of the litter layer (cm). All the statistical procedures were made by using the IBM SPSS Statistics software, 20 version (^{SPSS, 2011}).

Total fuel load

The average fuel load at the 15 sampling sites was 11.11 t ha^-1; the maximum value was found at site 12 (36.90 t ha^-1), and the minimum value, at site 7 (3.82 t ha^-1), without significant differences P = 0.36 (Figure 3). The average obtained, is below that reported by ^{Rodríguez and Sierra (1995)}, of 13.33 t ha^-1, in a broadleaf forest in the State of Mexico; ^{Xelhuantzi et al. (2011)} obtained 17.90 t ha^-1 in pine-oak forests of the states of Coahuila, Puebla and Jalisco, and ^{Rubio et al., (2016)} registered up to 36.6 t ha^-1 in pine-oak forests of a fire-free area in Nuevo León.

The means were statistically equal (Tukey mean comparison test; n = 3, F = 1.14, and W = 0.36). The bars represent the standard deviation.

Figure 3 Total forest fuel load by site in an oak forest of the La Lluvia mountain. 

The mean litter load was 7.28 t ha^-1 (65.53 %); this was the most representative material in the study area, where the values ranged between 97.72 and 18.10 % at sites 10 and 12, respectively (Figure 4). Similar values, of with 7.33 t ha^-1, were reported by ^{Hernández et al. (2016)}. ^{Rubio et al. (2016)} registered similar values to those of this research, namely 8.52 and 8.64 t ha^-1 of litter in pine-oak forests with and without burn, respectively. ^{López et al. (2015)} recorded high litter values, of 14 t ha^-1, in a 0.5 ha plot dominated by Q. magnoliifolia and Q. conspersa; however, they never reported the woody component, which leads to the assumption that it was absent or contributed little to the total fuel load of that area.

Componente = Component; Sitio = Site; Hojarasca = Litter; Leñoso = Woody component.

Figure 4 Percentage of woody fuel and leaf litter per site. 

The woody fuel load registered exhibited low values, amounting to merely 3.87 t ha^-1 (34.47 %) of the total load; the highest value occurred at site 12 (81.90 %), and the lowest, at site 10 (2.28 %); the result was above the figures registered by ^{Hernández et al. (2016)}, with 2.32 t ha^-1 for the oak ecosystem in Juchitán, Oaxaca; ^{Rubio et al. (2016)} registered values of 18.31 and 15.84 t ha^-1 when comparing between the fuel loads of pine-oak forests with and without burn, respectively.

The variation in the fuel loads differs at a spatial scale within an ecosystem, as suggested by the results of ^{Chávez et al. (2016)}, who registered up to 92.49 t ha^-1 for the oak forest in the state of Jalisco; conversely, ^{Villers and López (2004)} quantified fuel loads in oak forests of the La Malinche National Park in Tlaxcala, as 16.3 t ha^-1 of woody material and 0.27 m of thickness for the topsoil.

Fuel loads per site

According to the results of the ANOVA performed for the various fuel categories per site, leaf litter exhibited no significant differences (p = 0.36); the maximum value of the litter fuel of 12.77 t ha^-1, was found at site 2; the lowest values were registered at sites 7 and 14, at 2.63 t ha^-1 (Table 2). In this regard, ^{Martínez et al. (1990)} consider that a high accumulation of litter contributes to a higher incidence of fires, and its presence and thickness will determine the magnitude of the fire, as it can be easily ignited.

The 1 h fuels were in average 0.50 t ha^-1 (4.50 %) of the total quantified fuel, with maximum and minimum values of 0.10 to 0.91 t ha^-1 (Table 2). This load is below the values reported by ^{Muñoz et al. (2005)}, of 0.70 t ha^-1, but higher than that reported by ^{Villers and López (2004)}, of 0.42 t ha^-1. The contribution of this category to the total load was low.

The variance analysis showed significant differences in this category (p = 0.001). The Tukey mean comparison test indicated low values at sites 10 and 1, of 0.10 to 0.18 t ha^-1; conversely, the highest values, of 0.88 y 0.91 t ha^-1, were found at sites 8 and 14. ^{Barrios-Calderón et al. (2018)} registered differences between mangrove sites, with values of 5.39 and 2.85 t ha^-1; ^{Castañeda et al. (2015)} reported differences in high mountain forests dominated by Pinus hartweggi (p = 0.0399) in forests with a dense cover and forests with semi-dense and fragmented covers.

The 10 h fuel was quantified at 0.63 t ha^-1 (5.67 %); the maximum and minimum values ranged between 0.35 and 0.05 t ha^-1 at sites 8 and 10 (Table 2); no significant differences were found (p = 0.17). In contrast, ^{Castañeda et al. (2015)} quantified high values, of 3.82, 4.48 and 5.18 t ha^-1, for dense, semi-dense and fragmented covers (p = 0.23).

Non significant differences were found in the 100 h fuel (p = 0.61); the average value was 0.77 t ha^-1 (6.93 %), while the highest and lowest values, of 3.05 and 0.38 t ha^-1, were registered at sites 12 and 1 (Table 2). ^{Xelhuantzi et al. (2011)} registered lower values, of 0.3 t ha^-1, in forests dominated by pine and oak species in the states of Coahuila, Puebla and Jalisco.

Finally, 1 000 h fuels exhibited values of 0.61 to 26.22 t ha^-1 at sites 11 and 12, respectively (Table 2); these fuels were registered only at three of the 15 sites, with an average of 1.93 t ha^-1 (17.37 %), and without significant differences (p = 0.42). This is a higher value than the 0.43 t ha^-1 found by ^{Chávez et al. (2016)} for 1 000 h fuel in a rotten condition for a pine-oak forest of the state of Jalisco. Likewise, in this study, all the material quantified in this category was found in a state of decay, and therefore it entails little danger, as it will soon be incorporated into the organic layer of the soil.

Forest fuels by type

Regarding the loads of the various types of forest fuels, the highest value is for leaf litter, i.e. 65.53 % (7.28 t ha^-1), followed by the 1 000 h fuels, with 17.37 % (1.93 t ha^-1); 100 h fuel, with 6.93 % (0.57 t ha^-1); 10 h fuel, with 5.67 % (0.169 t ha^-1), and 1 h fuel, with 4.50 % (0.067 t ha^-1) (Figure 5). These differ from the loads reported by Xeluantzi et al. (2011), who estimated 6.19 t ha^-1 for leaf litter; 0.30 t ha^-1, for 10 h; 0.12 t ha^-1 for 100 h, and 0.03 t ha^-1 for 1 h fuel in clusters of temperate forests distributed between Coahuila, Puebla and Jalisco.

Carga de combustible = Fuel load; Tipo de combustible = Type of fuel; Hojarasca = Litter.

Vertical lines represent the standard deviation.

Figure 5 Load by type of fuel in the oak forest in Guerrero, Mexico. 

The Kruskal-Wallis mean comparison test exhibited significant differences between leaf litter and the 1, 10, 100 and 1 000 h fuel types (p <0.001), and between 1000 h fuel and 1, 10 and 100 h fuels (p = 0.002, 0.002 and 0.023); the rest of the comparisons did not show significant differences, with p > 0.05 (Table 3). Given the higher litter fuel load, a superficial fire may be expected; although the component of the 1 000 h fuel registered the second most important load, its distribution was incipient, since it was registered only at 3 of the 15 sites.

Slope and depth of leaf litter with fuels

The correlation between the thickness and the load of the litter layer was positive, with r = 0.773 and p = 0.001 (Table 4); the thickness of the layer was 3.18 ± 1.28 cm, i.e. lower than the thickness registered by ^{Estrada and Ángeles (2007)} for the oak forests of the El Chico National Park in Hidalgo. Likewise, the relationship between the slope of the terrain and the 1 h fuel was positive and significant, with r = 0.639 and p = 0.010 (Table 4); the largest loads (0.91 t ha^-1) occurred at the most pronounced slope (75 %). This result contradicts those of ^{Villers et al. (2012)}, who registered higher fuel loads at sites with the least slope as a result of dragging due to gravity or runoff. For their part, ^{Castañeda et al. (2015)} found a positive correlation between 1 000 h fuels in dense forests with slopes ranging between 7 and 12° (p = 0.049).

Among the various fuel types, a positive and significant relationship was found only between the loads of the 1 and 10 h fuels, with r = 0.650 and p = 0.009 (Table 4), which is reasonable, as the wind and the rainfall naturally cause the detachment of small branches and twigs. For the rest of the fuels, as assumed by ^{Xelhuantzi et al. (2011)}, increases in any type of fuel are independent from the rest of the categories and may be the result of other disturbance factors, such as fires, excessive felling, and grazing, which was not observed at the 15 sampling sites.