texto en 
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
Introduction
In Mexico, the forests of Pinus hartwegii Lindl. are those that develop at the highest elevations, between 2 900 and 4 000 m, although isolated individuals have been recorded at 4 300 m (Rzedowski, 2010); therefore, the species is considered subalpine. These forests form relatively low stands (5 to 20 m in height) that are moderately to sparsely dense, often pure or nearly pure (Miranda & Hernández-Xolocotzi, 2014; Rzedowski, 2010). At the Iztaccíhuatl-Popocatépetl National Park (PNIP), this forest forms stands of 15 to 20 m in height, which are reduced to 5 or 8 m in the highest altitude areas, around 4 000 m. The species P. hartwegii tolerates low temperatures and frequent snowfall and is associated with dense grasslands and herbaceous plants such as Lupinus spp. (Comisión Nacional de Áreas Naturales Protegidas [CONANP], 2013; Rzedowski, 2010). In Mexico, P. hartwegii is one of the species most adapted to fire, because it regenerates on ash beds, cespitose condition, thick bark, basal sprouts, and recovers foliage in the non-lethal charred area of the canopy (Rodríguez-Trejo, 2014).
The subalpine grassland region of the Valley of Mexico consists of tufted grasses, measure 60 to 120 cm in height and are found at altitudes of 3 000 to 4 300 meters. Below 4 000 m, such grasslands are secondary, derived from the destruction of pine forests, and above 4 000 m there are climax communities. Fire is a determining ecological factor in the existence, dynamics and exploitation of these species (Miranda & Hernández-Xolocotzi, 2014; Rzedowski, 2010). In low-elevation communities with lower fire frequency, succession advances to P. hartwegii forest and the burning of grasses for livestock to take advantage of their resprouts is a cause of forest fire (Rodríguez-Trejo, 2014). In the PNIP, this community is established between 3 700 to 4 350 m. Its dominant species, in different associations, are Calamagrostis tolucensis (Kunth) Trin. ex Steud., Festuca hephaestophila Nees, F. livida Willd. ex Spreng., F. tolucensis Kunth, Muhlenbergia macroura Hitchc. and M. quadridentata Trin. (CONANP, 2013; Rzedowski, 2010).
Studies on forest fuels are important for fire management, because they provide information for estimating forest fire hazards, modeling fire behavior, conducting prescribed burns, managing forest fuels, wildlife and forest pests, as well as estimates of carbon storage, forest ecosystem productivity, and emissions of pollutants and greenhouse gases (Flores-Garnica et al., 2018; Rodríguez-Trejo, 2014; Scott et al., 2014). The above is particularly relevant in high-altitude ecosystems, given their sensitivity to global warming.
The high variability of fuel loads within the same forest has become evident. Chávez-Durán et al. (2021) documented this in one-hectare of a pine-oak forest in the Sierra de Quila Natural Protected Area, Jalisco, with an average of 56.84 Mg∙ha-1 and a variation of 116.61 Mg∙ha-1 over a distance of just 72.11 m. Fieldwork for fuel sampling demands time and human and financial resources; however, few efforts have been made to study variables related to the fuels themselves and environmental factors (e.g., solar radiation) that would allow for simple and quick estimation of fuel loads and other variables. A study of this nature was conducted in the Sierra Norte of Puebla, in a forest of Quercus crassifolia Bonpl., where Rodríguez-Trejo et al. (2021) report that the total fuel load had a significant positive relationship with tree density and basal area; likewise, the understory fuel load (grasses, dicotyledonous herbaceous plants, and shrubs) showed a significant positive relationship with a clearing.
The objective of this study was to establish possible relationships between surface fuel variables (load, cover, height or depth), leaf area index and solar radiation (diffuse, direct, total, clearing) that allow estimating partial (woody materials) and total fuel loads. The study was conducted in a stand of P. hartwegii and in a subalpine grassland at the Iztaccíhuatl-Popocatépetl National Park.
Materials and Methods
Study area
The PNIP is in the northeastern part of the Transversal Volcanic Axis and is part of the Sierra Nevada. The park covers 39 819 ha, mainly in Estado de México, but also in Puebla and Morelos. With rugged topography, the PNIP is located at elevations between 3 000 and 5 480 m. The rocks are mainly basalts and andesites. In the area near the highest elevations, where this study was conducted, Andosol soils are predominant. The climates vary from temperate-humid to cold and very cold, with rainy season mainly in the summer. For example, in Hueyatlaco it is semi-humid with rainfall in summer (annual averages: 7.7 °C and 1 186 mm), cold (between -2.5 and 5 °C) and very cold (<-2 °C) (CONANP, 2013). The park has several vegetation types, two of which were used for this study. For this purpose, an area was selected between the base of the Iztaccíhuatl volcano and the location known as 'Paso de Cortés’ (Table 1; Figure 1).
Table 1 Sampling sites at the Iztaccihuatl-Popocatepetl National Park
| Site | Coordinates | Elevation (m) | Slope (%) | Exposure | |
|---|---|---|---|---|---|
| Latitude | Longitude | ||||
| P. hartwegii forest | |||||
| 1 | 19.10° N | 98.63° O | 3 573 | 19 | South |
| 2 | 19.10° N | 98.64° O | 3 555 | 15 | South-southwest |
| 3 | 19.10° N | 98.64° O | 3 572 | 8 | South |
| 4 | 19.10° N | 98.64° O | 3 581 | 5 | South |
| 5 | 19.13° N | 98.65° O | 3 778 | 55 | West |
| 6 | 19.11° N | 98.64° O | 3 674 | 47 | North-northeast |
| 7 | 19.11° N | 98.65° O | 3 676 | 25 | East |
| Subalpine grassland | |||||
| 1 | 19.123° N | 98.652° O | 3 786 | 13 | Northeast |
| 2 | 19.123° N | 98.652° O | 3 786 | 13 | Northeast |
| 3 | 19.125° N | 98.650° O | 3 781 | 15 | Southeast |
| 4 | 19.125° N | 98.650° O | 3 781 | 15 | Southeast |
| 5 | 19.118° N | 98.646° O | 3 728 | 24 | North-northeast |
| 6 | 19.118° N | 98.646° O | 3 728 | 24 | North-northeast |
| 7 | 19.118° N | 98.646° O | 3 728 | 24 | North-northeast |
Field study
Field study included the survey of 14 composite sampling sites, seven in P. hartwegii forest and seven in subalpine grassland. Each composite site included: surface fuels, trees, solar radiation environment and harvesting of fuels to obtain dry weight. These activities were carried out in May and early June 2021. Mean tree height with diameter at breast height (DBH) > 12.5 cm was 5.3 to 27.2 m and density from 50 to 1 100 individuals∙ha-1.
Site data included: type of vegetation, date, site name, coordinates and altitude (Garmin GPS, model GPSMAP 66s), and exposure (Silva compass, model BK). Slope was measured with a Suunto clinometer (model Pm-5).
Sampling of forest fuels
The following classes of forest fuels were sampled at each site, arranged in a triangular pattern with 12 m per side and randomly interlocked:
1) Leaf litter and fermentation layer (three 0.3 x 0.3 m CPVC -Chlorinated Polyvinyl Chloride- squares at the vertices of the triangles). Coverage (%) was estimated visually and three depths were measured with a steen ruler (cm) at the ends and center of an internal diagonal.
2) Grasses (three 1 x 1 m CPVC squares at the vertices of the triangles). Coverage was estimated visually (%) and three heights were measured with a stadia (m) at the ends and center of an internal diagonal.
3) Dicotyledonous herbaceous plants. Sampling conditions and measurements were the same as for grasses.
4) Woody materials. The Planar Intercept Method (Brown, 1974) and for this, lines marked with string and stakes were established starting from the origin of the sampling site (left side of the baseline, parallel to the contour line, facing the apex of the triangle) to record woody materials oriented east or west, alternately between sites. For materials with 1 to 100 hours of lag time (LT), the number of intersections per line was recorded (with a metal caliper), and for those with 1 000 hours LT, both firm and decayed, the diameter (with a tape measure) was additionally measured at the point of intersection between the central axis of the woody material and the string. Surface materials and those between leaf litter and fermentation layers were included. One line of each length was used per sampling site. Line dimensions for LT class materials were: 1 h (diameter < 0.6 cm, 2 m line), 10 h (0.6 a 2.5 cm, 2 m), 100 h (2.6 a 7.5 cm, 4 m), 1 000 h firm (>7.5 cm, 12 m) and 1 000 h LT decayed (>7.5 cm, 12 m).
Tree sampling
At the origin of each composite site, three square sites for trees, coinciding at one vertex, with the following dimensions, were established:
1) 20 x 20 m for trees with DBH > 12.5 cm. Tree number, species, condition (alive or dead), DBH (diametric tape) and height (Suunto clinometer, m) were recorded.
2) 10 x 10 m for trees with DN < 12.5 cm, height > 1.3 m. Tree number, species, condition, DBH and height (with stadia, m) were recorded
3) 10 x 10 m for sapling (<1.3 m in height). Progressive number, condition (dead or alive), species and height (stadia, m) were recorded.
Solar radiation and leaf area index
Between May and June 2021, a zenith photograph was taken at each of the seven composite sampling sites in the forest. The canopy was photographed from the ground with a digital camera and hemispherical lens (Delta T-Devices). The camera was placed on a leveling structure 95 cm above the ground to avoid including the canopy in the photographs. The images were processed with the Hemiview 2.1 SR4 program (Delta-T Company, 1998), which, from the coordinates, Julian day and altitude, calculates total, direct and diffuse solar radiation, and clearing and leaf area index (LAI). The latter is expressed in units of leaf area surface (one half of the leaf area divided by the surface of the crown area projection). Due to the lack of tree cover, the main factor determining solar radiation in the understory, no radiation data were obtained for the subalpine grassland.
Fuel collection
At the fuel measurement sites, after data collection, grasses, dicotyledonous herbs, shrubs, leaf litter, and fermentation layer were harvested into labeled brown paper bags, which were then placed in plastic bags for transportation to the laboratory. For grasses, the lignified core of the tuft was not harvested, as it is generally not consumed in fires.
Additionally, the following woody materials were collected to determine their basic density: 27 with 1 hour of time-lag (LT), 33 with 10 hours LT, 20 with 100 hours LT, 5 with 1 000 hours LT (firm), and 5 with 1,000 hours LT (decayed). These materials were labeled and placed in brown paper bags and plastic bags. The samples were dried in the laboratory at 90 °C (Ecoshel oven, model 9023A) until constant weight.
Data Analysis
For the statistical analysis, Pearson correlation coefficients (r squared equals the coefficient of determination of simple linear regression R2) and linear regression estimators were obtained with the PROC CORR and PROC REG programs, respectively, from the SAS v. 9.00 ® program (Statistical Analysis System Institute, Cary, NC, 2002). With the second procedure, the coefficients of determination (R2) and normal distribution of errors were determined using the Shapiro-Wilks test (p > 0.05), homoscedasticity with the White test (p > 0.05); variance inflation (no multicollinearity p > 0.05), as well as the Durbin-Watson test. Given the sample size and the number of variables in regressions, low and high Durbin-Watson table values of 1.356 and 2.646, respectively, were used. If the calculated Durbin-Watson value was within this interval, there was no positive or negative correlation. Furthermore, a model of total fuel load explained by the depth of the leaf litter and fermentation layers was fitted, which was linearized exponentially, where the response variable was transformed, in this case total fuel load y i : ln(y i ) = B 0 + B 1 x i where, β 0 is the ordinate to the origin and β 1 is the constant associated with the independent variable x i . For a better interpretation of the model, we returned to the original scale of the (exponential, exp) expression: y i = exp (B 0 + B 1 x i ) + εi. The iterative Gauss-Newton method was used to find the parameter estimators. The graphs were obtained in the Excel program (Microsoft Office).
In the case of woody materials, before placing them in the drying oven, two diameters were measured perpendicularly at both ends, as well as their length, to obtain 'fresh' volume (Fv) with the cylinder formula: Fv = 0.7854 d 2 L; where, d = average diameter (cm) and L = length of the woody material (cm). The basic density (BD, g∙cm-3) was calculated after obtaining the anhydrous weight (Po, g) of each oven sample, with the model: DB = Po / Vf. The model (Brown, 1974; Morfín-Ríos et al., 2012) was used to obtain the loading (C, Mg∙ha-1) of the woody materials at 1, 10 and 100 h LT:
where,
k = constant equal to 1.234
QMD = quadratic mean diameter (cm)
f = number of intersections with the sampling line
c = slope correction factor: 1.00 for 0 and 10 % slope; 1.02, 20 %; 1.04, 30 %; 1.08, 40 %; 1.12, 50 %; 1.17, 60 %; 1.22, 70 %; 1.28, 80 %; 1.35, 90 %; 1.41, 100 %; 1.49, 110 %.
N = number of lines surveyed
L = length of the sampling line (as indicated in “Sampling of forest fuels”).
In the case of woody materials with 1 000 h LT, the model (Brown, 1974; Morfín-Ríos et al., 2012) was used: C = [k DB (∑DC)] / NL; where ΣDC = sum of squared diameters of woody materials.
Results
Pinus hartwegii forest
Tree measurements
The tree stand with DBH > 12.5 cm had means of DBH = 28 cm, height = 12.5 m, basal area = 14.1 m2∙ha-1, and a density of 285.7 individuals∙ha-1; these variables for the tree stand with DBH < 12.5 cm had, respectively, values of 9.4 cm, 3 m, 1.5 m2∙ha-1 and 300 individuals∙ha-1. No regrowth was observed at the sampling sites. The overall averages in the previously mentioned order reached: 26 cm, 12.2 m, 14.6 m2∙ha-1 and 371.4 individuals∙ha-1 (Table 2). The range in dimensions indicates a variety of ages and densities.
Table 2 Tree measurements obtained from Pinus hartwegii forest at the Iztaccíhuatl-Popocatépetl National Park.
| Sitio | Tree stands DBH > 12.5 cm | Tree stands DBH < 12.5 cm | All trees | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DBH (cm) | Height (m) | BA (m2∙ha-1) | Density (Individuals∙ha-1) | DBH (cm) | Height (m) | BA (m2∙ha-1) | Density (Individuals∙ha-1) | DBH (cm) | Height (m) | BA (m2∙ha-1) | Density (Individuals∙ha-1) | |
| 1 | 46.3 | 27.2 | 34.16 | 200 | 0 | 0 | 0 | 0 | 46.3 | 27.2 | 34.17 | 200 |
| 2 | 14.7 | 6.8 | 5.94 | 225 | 0 | 0 | 0 | 0 | 14.7 | 6.8 | 5.94 | 225 |
| 3 | 30.6 | 12.9 | 15.74 | 200 | 6.8 | 0 | 1.95 | 500 | 21.5 | 12.3 | 17.69 | 700 |
| 4 | 15.5 | 5.3 | 0.97 | 50 | 0 | 0 | 0 | 0 | 15.5 | 5.3 | 0.97 | 50 |
| 5 | 36.5 | 9.0 | 11.42 | 125 | 12.0 | 3 | 1.13 | 100 | 31.6 | 7.8 | 12.55 | 225 |
| 6 | 37.5 | 19.0 | 11.21 | 100 | 0 | 0 | 0 | 0 | 37.5 | 19.0 | 11.21 | 100 |
| 7 | 15.0 | 7.0 | 19.44 | 1100 | 0 | 0 | 0 | 0 | 15.0 | 7.0 | 19.44 | 1 100 |
| Means (± SD) | 28.0 (± 13.9) | 12.5 (±8.0) | 14.1 (±10.7) | 285.7 (±364.5) | 9.4 (±4.8) | 3.0 (±1.1) | 1.5 (±0.8) | 300.0 (±186.4) | 26.0 (±12.6) | 12.2 (±8.9) | 14.6 (±10.7) | 371.4 (±384.7) |
DBH = diameter at breast height, AB = basal area, SD = standard deviation.
Solar radiation
The clearing had an average of 0.3 (30 %), ranging from 0.1 to 0.7. The mean LAI reached 1.42 m2∙m-2, with extremes ranging from 0.23 to 3.47 m2∙m-2. The average diffuse solar radiation was 493.3 MJ∙m-2∙yr-1, with extremes from 225.9 to 980.2 MJ∙m-2∙yr-1. The mean direct solar radiation reached 5 285.3 MJ∙m-2∙yr-1, with extremes from 2 711.2 to 9 529.3 MJ∙m-2∙yr-1. Finally, the average total solar radiation had a value of 5 778.7 MJ∙m-2∙yr-1, with extremes from 2 938.4 to 10 510.1 MJ∙m-2∙yr-1 (Table 3).
Table 3 Solar radiation variables and Leaf Area Index of the Pinus hartwegii forest at the Iztaccíhuatl-Popocatépetl National Park.
| Site | Open area | LAI (m2∙m-2) | DSR (MJ∙m-2∙año-1) | DSR (MJ∙m-2∙año-1) | TSR (MJ∙m-2∙año-1) |
|---|---|---|---|---|---|
| 1 | 0.2 | 1.68 | 354.9 | 4 346.7 | 4 701.6 |
| 2 | 0.3 | 1.36 | 433.6 | 4 110.8 | 4 544.4 |
| 3 | 0.1 | 3.47 | 225.9 | 2 792.2 | 3 018.2 |
| 4 | 0.7 | 0.23 | 980.2 | 9 529.8 | 10 510.1 |
| 5 | 0.5 | 0.47 | 750.0 | 8 361.4 | 9 111.4 |
| 6 | 0.3 | 0.97 | 481.5 | 5 145.1 | 5 626.5 |
| 7 | 0.2 | 1.78 | 227.2 | 2 711.2 | 2 938.4 |
| Means ± DE | 0.3 ± 0.2 | 1.42 ± 0.17 | 493.3 ± 279.3 | 5 285.3 ± 2 664.6 | 5 778.7 ± 2 940.7 |
DE = standard deviation, LAI = leaf area index, DSR = diffuse solar radiation, DSR = direct solar radiation, TSR = total solar radiation.
Forest fuel loads
The mean total fuel load was 27.9 Mg∙ha-1 in P. hartwegii forests. Leaf litter and fermentation layer accounted for almost 60 % of the load, followed by woody materials that covered almost a quarter (24 %), particularly those with 1 to 100 h LT (20 %). Grasses reached 18.3 % of the load, while herbaceous dicotyledons and shrubs were not representative (Table 4).
Table 4 Partial and total fuel loads of the forest fuel complex in a Pinus hartwegii forest at the Iztaccihuatl-Popocatepetl National Park.
| Site | Woody materials (Mg∙ha-1) | DH | Grasses | Shrubs | LFL | Total fuel load | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 h LT | 10 h LT | 100 h LT | 1 000 h LT(F) | Woody C | 1 a 100 h LT | (Mg∙ha-1) | (Mg∙ha-1) | (Mg∙ha-1) | (Mg∙ha-1) | (Mg∙ha-1) | |
| 1 | 1.03 | 6.75 | 0 | 0 | 7.78 | 7.78 | 0 | 1.65 | 0.11 | 9.75 | 19.30 |
| 2 | 0 | 2.70 | 0 | 0 | 2.70 | 2.70 | 0 | 1.02 | 0 | 5.55 | 9.26 |
| 3 | 3.60 | 10.80 | 8.31 | 0 | 22.70 | 22.70 | 0.65 | 5.87 | 0 | 12.27 | 41.49 |
| 4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 8.42 | 0 | 0 | 8.42 |
| 5 | 0 | 0 | 0 | 7.1 | 7.1 | 0 | 0 | 3.64 | 0 | 20.04 | 30.69 |
| 6 | 1.68 | 4.41 | 0 | 0 | 6.10 | 6.10 | 0 | 7.57 | 0 | 15.58 | 29.25 |
| 7 | 0.53 | 0 | 0 | 0 | 0.53 | 0.53 | 0 | 7.47 | 0 | 48.72 | 56.72 |
| Mean ± SD | 0.98 (±1.3) | 3.52 (±4.1) | 1.18 (±3.1) | 1.00 (±2.7) | 6.69 (±7.7) | 5.69 (±8.1) | 0.09 (±0.3) | 5.09 (±3.0) | 0.01 (±0.04) | 15.99 (±15.8) | 27.88 (±17.4) |
| Ratio (%) | 3.5 | 12.6 | 4.3 | 3.6 | 24.0 | 20.4 | 0.3 | 18.3 | 0.1 | 57.4 | 100.0 |
LT = Lag time, F = firm woody materials, C = all woody materials. DH = dicotyledonous herbaceous, LFL = leaf litter and fermentation layer.
For the grass component, coverages ranged from 25 to 80 % (mean = 45 %), with heights from 28.3 to 81.7 cm (mean = 67.6 cm). The volume of the imaginary cube (product of coverage and height) ranged from 0.071 to 0.626 m³ (mean = 0.324 m³). The grass load in the pine forest ranged from 1.018 to 8.417 Mg∙ha-1 (mean = 5.091 Mg∙ha-1). No significant correlation was found between the grass load and the variables of height, coverage, or imaginary cube volume, nor with radiation variables or LAI in the P. hartwegii forest.
Regarding the class of woody materials, the results were as follows:
1) 1 h LT. The specific gravity had values from 0.29 to 0.57 g∙cm-3, the mean was 0.38 g∙cm-3 and the coefficient of variation (CV) reached 0.18. These materials had a loading of 0.98 Mg∙ha-1 (Table 4) and showed a significant Pearson correlation with the 10 h LT (r = 0.89, p = 0.0068) and 100 h LT (r = 0.88, p = 0.0096) materials, the same as the 10 with 100 h LT (r = 0.78, p = 0.0394). They also showed a significant Pearson correlation (r = 0.83, p = 0.0205) with the LAI.
2) 10 h LT. Specific gravity ranged from 0.22 to 0.48 g∙cm-3; mean was 0.38 g∙cm-3 and CV = 0.17. The mean load resulted in 3.52 Mg∙ha-1. These materials showed significant Pearson correlation (r = 0.82, p = 0.0254) with LAI.
3) 100 h LT. Specific gravity was recorded between 0.306 and 0.854 g∙cm-3 with a mean of 0.438 g∙cm-3 and CV = 0.33. Also, the material had mean loading of 1.19 Mg∙ha-1 (Table 4). These materials showed significant Pearson correlation (r = 0.84, p = 0.0179) with LAI. Consequently, the sum of woody materials of 1, 10 and 100 h LT yielded significant Pearson correlation (r = 0.88, p = 0.0098) with LAI.
4) 1 000 h LT. Specific gravity of firm woody materials ranged from 0.39 to 0.47 g∙cm-3 with a mean equal to 0.41 g∙cm-3 and CV = 0.07. The mean load was 1.001 Mg∙ha-1. In the case of rotten materials, specific gravities of 0.27 to 0.44 g∙cm-3 and CV = 0.27 were found. This type of material was not observed in the extensive sampling. The total load of woody materials indicated significant Pearson correlation (r = 0.79, p = 0.0363) with LAI.
On the other hand, in the seven sampling sites, the leaf litter and fermentation layer had 0 to 100 % cover (mean = 71.4 %), but excluding one clearing in the forest with 0 %, cover ranged from 70 to 100 % (mean = 83.3 %). The depth of the leaf litter and fermentation layers, outside the referenced clearing (where the mean was 5.8 cm), ranged from 2.4 to 12.6 cm (mean = 6.7 cm). The loading of these fuels ranged from 0 (in the referenced clearing) to 48.72 Mg∙ha-1 (under canopy). The mean reached 15.98 Mg∙ha-1 (with the clearing included) and 18.65 Mg∙ha-1 without such clearing; the CV for the first case reached 0.99, while for the second 0.83. Only Pearson's significant correlation (p < 0.05) was found between the depth of the leaf litter and fermentation layers with the load of the same layers (r = 0.86, p = 0.014).
Relationship between forest variables and partial loads of forest fuels
Six linear models were significant and complied with the parameters of normality, non-heteroscedasticity, non-multicollinearity and non-autocorrelation. Such models allowed estimating the cover of grasses from the clearing; the load of materials with 1 and 10 h LT, load from 1 to 100 h LT and total fuel load as a function of LAI; and the load of leaf litter and fermentation layer from the depth of the same layers (Table 5; Figures 2 and 3). Pearson's correlation was found between total tree stand basal area and the sum of woody fuel loads from 1, 10 and 100 h LT (r = 0.95, p = 0.0043).
Table 5 Linear relationship between forest variables and forest fuel loads in two vegetation types at the Iztaccíhuatl-Popocatépetl National Park.
| Model | R2 | P value | RMSE | p, SW | p, W | VIV, M | Durbin-Watson |
|---|---|---|---|---|---|---|---|
| Subalpine grassland | |||||||
| Gl = - 2.7668 + 0.207 (coz) | 0.95 | 0.0002 | 1.4058 | 0.4651 | 0.2437 | 1.000 | 1.949 |
| Bosque de Pinus hartwegii | |||||||
| Gc = 20.082 + 72.88 (cl) | 0.71 | 0.0179 | 11.1123 | 0.0739 | 0.3787 | 1.000 | 1.708 |
| L1 = - 0.4762 + 1.0231 (LAI) | 0.69 | 0.0205 | 0.8045 | 0.8443 | 0.2603 | 1.000 | 2.473 |
| L10 = - 0.933 + 3.1352 (LAI) | 0.66 | 0.0254 | 2.6160 | 0.2279 | 0.3322 | 1.000 | 2.009 |
| WL = - 1.3399 + 5.649 (LAI) | 0.62 | 0.0363 | 27.3262 | 0.7945 | 0.5964 | 1.000 | 1.796 |
| L1a100 = - 3.7203 + 6.6194 (LAI) | 0.77 | 0.0098 | 4.2923 | 0.1603 | 0.2634 | 1.000 | 1.912 |
| Llfll = - 6.37 + 4.7292 (dlfl) | 0.73 | 0.0140 | 3.1221 | 0.9244 | 0.3131 | 1.000 | 1.863 |
Gl = grassland load, Gc = grassland cover, cl = clearing, LAI = leaf area index, L1 = woody material load with 1 h LT (lag time), L10 = woody material load with 10 h LT, WL = woody material load, L1a100 = woody material load from 1 to 100 h LT, llfll = leaf litter and fermentation layer load, dlfl = depth. R2 = coefficient of determination, RMSE = root mean square error; p, SW = p-value for Shapiro-Wilks test; p, W = p-value for White's test; VIV, M = variance inflation value.
Figure 2 Relationship between grass cover and clearing in Pinus hartwegii forest at the Iztaccíhuatl-Popocatépetl National Park.
Relationship between forest variables and total fuel load
A significant Pearson correlation (r = 0.74, p = 0.0595) was established between the average diameter at breast height of all trees and the total forest fuel load. The following fitted model allows estimating the total forest fuel load (TL), from the depth of leaf litter and fermentation layers (dlfl) (Figure 4): TL = exp (2.2779 + 0.1961 dlfl). With this model we found R2 = 0.95 (p = 0.0004), confidence limits from 1.432 to 3.123 for the constant and from 0.081 to 0.310 for the constant associated to the explanatory variable.
Figure 4 a) Linear model relating leaf litter and fermentation layer load (lfl) according to depth. B) Inverse of the transformed model (non-linear exponential) between total fuel complex loading and leaf litter and fermentation layer depth in a Pinus hartwegii forest at the Iztaccíhuatl-Popocatépetl National Park.
Subalpine grassland
Coverages from 20 to 100 %, mean heights from 0.47 to 0.90 m and individual heights from 20 to 110 cm were determined in the grassland area. Per sampling site, mean loads ranged from 2.182 Mg∙ha-1 to 17.991 Mg∙ha-1 and the following average values were obtained: 6 842 Mg∙ha-1 load, 65.0 cm height and 46.4 % cover (Table 6; Figure 5).
Table 6 Loads, cover and height of subalpine grassland at the Iztaccihuatl-Popocatepetl National Park.
| Site | Cover (%) | Height (m) | Load (Mg∙ha-1) |
|---|---|---|---|
| 1 | 50 | 0.63 | 6.23 |
| 2 | 50 | 0.73 | 9.88 |
| 3 | 30 | 0.65 | 3.25 |
| 4 | 40 | 0.60 | 4.08 |
| 5 | 35 | 0.60 | 4.27 |
| 6 | 20 | 0.47 | 2.18 |
| 7 | 100 | 0.90 | 17.99 |
| MEans ± SD | 46.4 ± 25.9 | 0.65 ± 0.13 | 6.842 ± 5.52 |
DE: standard deviation of the mean.
To estimate the biomass of the grassland, including the portion that is generally not consumed during forest fires, all biomass (including the woody core of clumps of the grasses) was harvested at a single site, recording cover of 80 %, average height of 78.3 cm, and anhydrous biomass of 30.704 Mg∙ha-1. Although radiation was not measured in the subalpine grassland because there was no tree canopy, it is similar to or greater than that of the forest clearing (site 4, Table 3).
The only significant model meeting the parameters referred to in the methodology (normality, no heteroscedasticity, no multicollinearity or autocorrelation) was the one that related the load of grasses and its cover, with R2 = 0.95 and p = 0.0002 (Table 5).
Discussion
Pinus hartwegii forest
Tree measurement characteristics
Due to the high altitude in the study area (3 555 m - 3 786 m) with low temperatures and limitation by humidity, due to water freezing for part of the year and strong winds, P. hartwegii does not reach large dimensions (DBH = 26 cm, height = 12.2 m) as at lower altitudinal levels, as reported by CONANP (2013).
Solar radiation
The radiation levels recorded overlap with the range referred by Islas-Madrid et al. (2013) for P. hartwegii forests in Mexico City (CDMX), with a clearing equal to 0. 39, diffuse solar radiation of 544.3 MJ∙m-2∙yr-1, direct solar radiation of 6 092.3 MJ∙m-2∙yr-1 and total solar radiation of 6 756.6 MJ∙m-2∙yr-1, values similar to the means in this study.
Fuel load
The average total fuel load (27.9 Mg∙ha-1) from this study falls within the range reported (23.95 to 33.81 Mg∙ha-1) for a P. hartwegii forest in Estado de México by Castañeda-Rojas et al. (2015); however, one site was found with a much higher total fuel load of 56.72 Mg∙ha⁻1. This site is productive in both live and dead matter and has not been affected by fire for several years, as fire exclusion increases fuel loads (Scott et al., 2014).
In the present study, a higher grass load was observed as the clearing increased. This is because grasslands and alpine paramos (mostly subalpine grasslands) prefer open areas, which are often the result of forest destruction (where possible at such altitudes), as most grasses follow the C4 photosynthetic pathway and therefore prefer higher levels of solar radiation (Krebs, 2016; Lambers & Oliveira, 2019; Miranda & Hernández-Xolocotzi, 2014). This trend is also consistent with the findings of Islas-Madrid et al. (2013), who report that at higher radiation levels, there is an increase in the cover and dominance of Muhlenbergia quadridentata Trin., a common grass in P. hartwegii forests in Mexico City. Similarly, Caballero-Cruz et al. (2018) found higher grass loads in open stands compared to high-density stands in a temperate forest in Oaxaca.
The LAI is an indicator of productivity in tree species and, consequently, it is also an indicator of the presence of trees. Therefore, a relationship was found between the load of woody materials from 1, 10, 1 to 100 h LT and total with this index. It is evident that with a greater presence of trees, the number of woody materials increases on the forest floor. This was observed despite the disturbance caused by firewood collection and past fires. Likewise, in pine forests in Durango under forest management, through the silvicultural treatments of selection, thinning, and regeneration cutting, in that order, there was a progressively greater load of woody materials (Bautista-Rentería-Ánima et al., 2005). Similarly, in P. sylvestris Lour. stands in Ukraine, the percentages of all woody materials increased with stand age (Hurzhii et al., 2021) and with canopy cover in Chinese forests of Picea schrenkiana Fisch. & C. A. Mey. (Liang et al., 2017).
Regarding leaf litter and fermentation layer, without fire or disturbance from extraction, all types of materials accumulate on the forest floor. Some of the woody materials can be extracted as firewood, but not leaf litter and fermentation layer, which restarts its accumulation after fires, whether ecological succession occurs or remains steady. This latter aspect is observed in the present study, as the grassland can advance to pine forest, but both can persist under a different fire regime, with more frequent fires in the grassland (Rodríguez-Trejo, 2014; Rzedowski, 1981). A study that shows changes in fuel load due to succession was carried out in forests dominated by Pinus douglasiana Martínez, in Jalisco, where Quintero-Gradilla et al. (2019) established that, as ecological succession progressed, leaf litter and fermentation layer increased their load between 8 and 28 years of age of the stand. A similar case was also studied in P. sylvestris forests in Ukraine, where the percentage of the fermentation layer increased from 15 % in young stands to 43 % in mature stands (Hurzhii et al., 2021).
Relationship between forest variables and forest fuel loads
The positive linear relationship between LAI and woody material loads of 1, 10 LT, 1 to 100-hour LT, and total fuel load is explained by the fact that higher values of this index are associated with greater productivity. A similar trend was observed for total fuel load in Quercus crassifolia forests in Puebla, where Rodríguez-Trejo et al. (2021) found a direct positive relationship with the number of trees. The same study reports a positive logarithmic relationship between total fuel load and the basal area of the trees. In the case of the positive relationship between depth and leaf litter load and the fermentation layer, the first variable is a direct indicator of the second.
The implications
There is a range of fuel loads in the area, from light (<10 Mg∙ha-1) to heavy (>50 Mg∙ha-1). The lighter loads indicate open conditions with grassland dominance, as well as some fire recurrence, which contributes to lower fuel accumulation. The heavier loads imply dense tree cover, fire exclusion, and increased fuel load and fire risk, including crown fires, all of which is further intensified by global warming. It is crucial to manage forest fuels in areas with higher accumulation to reduce fire risk, mitigate extreme fire behavior, and lower emissions of pollutants and greenhouse gases.
Subalpine grassland
Total fuel load
The total forest fuel load in the grassland (6.84 Mg∙ha-1) is comparable to the 7.98 Mg∙ha-1 reported by Rodríguez-Trejo and Sierra-Pineda (1996) for grasslands near pine forests or subalpine areas of Mexico City, including Muhlenbergia, Festuca, and Calamagrostis. This load was higher in the forest than in the grassland, consistent with the progression of ecological succession from grasslands to pine forests in Mexico City (Rodríguez-Trejo & Sierra-Pineda, 1996).
Relationship between grassland and fuel load
Grass cover is one of the key characteristics of this fuel complex and is linearly related to the load. As there are more and larger individuals in this type of vegetation, its cover and load increase. This trend was well represented by the significant linear regression (R2 = 0.95; p = 0.0002). The higher the cover, the higher the biomass or load in the grassland (Rodríguez-Trejo, 2014). At the same time, higher values of clearing resulted in higher values of grass cover, since most grasses have a C4 photosynthetic pathway and require direct solar radiation to develop (Krebs, 2016).
The implications
In general, and as is typical, grass loads were not high (3.3 to 9.9 Mg∙ha-1). Steep slopes and strong winds in the fire season, can lead to significant fire speeds and intensities. One site reached 18 Mg∙ha-1, due to its higher productivity and fire exclusion or the age of the grassland. Therefore, it is also necessary to think about fuels management in areas with higher potential for extreme fire behavior. One of the options may be prescribed fire.
Conclusions
In the P. hartwegii forest and the subalpine grassland, statistical models were developed that relate forest fuel loads to light environment variables or to fuel-specific variables. Understanding these models is useful for streamlining estimates and gaining better insight into the characteristics, which helps make more accurate predictions about fire risk, fire behavior, and pollutant emissions, thereby supporting fuel management decisions. In the forest, increased clearings lead to higher grass fuel loads. Conversely, a greater presence of canopies (higher leaf area index values) results in a higher load of woody materials. In the pine forest, total forest fuel load can be estimated using the depth of leaf litter and fermentation layers, while in the subalpine grassland, it can be estimated based on grass cover. As ecological succession progresses from grassland to pine forest, fuel loads increase.
