Study area

The study was carried out in experimental plots treated with five different thinning intensities and a control, in a community of the Sierra Madre Occidental, particularly at the site named La Guajolota, at ejido La Ciudad, in Pueblo Nuevo, Durango, located at the coordinates 23°41'52.4" N and 105°36'59.5" W and an altitude of 2 800 m. The climate is temperate or semicold-semihumid, and temperate or semidry in the eastern side of the mountain chain. The mean annual temperature ranges is 20 to 22 °C; the mean annual precipitation ranges between 800 and 1 200 mm.

The area has a characteristic relief at its highest point, as well as plateaus aligned from the southwest to the northeast, hills, and undulated areas; in its middle part, where the altitude is lower, moderate to pronounced slopes culminate in canyons, which have a great variation in altitude and a generally steep topography.

Because of its geographical location, it includes various vegetation conditions, from pure oak and pine masses to mixed pine-oak forests. According to the information contained in the edaphological charts (^{INEGI, 1988}), it comprises different types of soils, among which cambisols, lithosols and regosols are prevalent.

Types of treatments utilized in the experimental plot

The research was carried out in a permanent thinning trial, with the application of six different thinning intensities, described below (^{Huizar-Amezcua et al., 2016}): Control: no thinning was performed; T1: Light thinning: removal of 10 % of the basal area per hectare; T2: Moderate thinning: removal of 20 % of the basal area per hectare; T3: Intense thinning: removal of 40 % of the basal area per hectare; T4: Severe thinning: removal of 60 % of the basal area per hectare, and T5: Thinning of Porvenir trees: removal of the trees that competed with the best 300 to 350 individuals per hectare.

The methodology followed in the establishment of the plots based on that developed by the Forestry Commission (^{Hummel et al., 1959}). Their shape was square, covering a surface area of 625 m² (25 m × 25 m), considered sufficient for the eventual permanence of 30 to 40 trees in those sites where the maximum intensity was applied. Furthermore, each plot is surrounded by a 5 to 10 m wide perimeter fringe, with the same treatment as within, in order to avoid the edge effect on the trees located at the boundaries of the plots.

Vegetation of the study area

The main vegetation types located at the study site are pine and pine-oak associations, which are mixes of the genera Pinus and Quercus in different proportions; the dominant species is Pinus cooperi C. E.Blanco.

This community of conifers is characterized by its high density and by the low diversity of its floristic components. From the structural point of view, the dominant elements of its tree stratum have a height that ranges between 15 and 35 m, although some individuals are 40 m tall. It is distributed from the south of the state of Chihuahua to central Durango, at an altitude interval of 2 000 to 2 800 m. It usually forms pure masses; however, it is sometimes associated with Quercus sideroxyla Bonpl., Juniperus deppeana Steud. and Pinus leiophylla Schiede ex Schltdl. ex Cham. Table 1 shows certain characteristics of the dominant vegetation in each treatment; the diameter at normal height (cm), the height (m) and the crown cover (m²) of all the individuals was determined with two measurements using the cardinal directions (north-south and east-west) as reference.

Literfall collection

Three 1.0 m² litter traps ―built with a beveled wood frame and a fine (1 mm) plastic mesh bottom to prevent the accumulation of water during rainy periods― were established with the purpose of estimating the litterfall deposition in each treatment. The baskets or traps were placed at 0.50 m above the ground level in order to intercept the litterfall. Their content was deposited in paper bags labeled with the data for each treatment, date and trap number. This procedure was carried out every 15 days, from April, 2016, to March, 2017. Every month, the two biweekly collections were mixed together and dried at 65 °C during 72 hours in a 16EG GCA Corporation convection oven.

The litterfall content was manually classified into four categories: leaves, reproductive structures (RS: flowers, fruits, and seeds), branches (diameter <2.0 cm), and other components (unidentified tissues, fine structures, bark, insect bodies, insect feces, etc.). The sum of the samplings per month, per repeat, and per treatment determined the annual litterfall production (Mg·ha^-1·year^-1).

Litterfall decomposition

The litterfall decomposition rate in the various treatments was assessed using the litterbag approach (^{Bärlocher, 2005}). The litter was collected a month before the beginning of the experiment (March); it was dried in the open air during one week, and 10 g samples were put into black 10 cm × 10 cm polypropylene bags, with a 1 mm² mesh, allowing access to the interior of the bags by detritivorous invertebrates, but minimizing losses due to fragmentation (^{Douce and Crossley, 1982}).

A mixture reflecting the relative proportion of the structural components of the original matter (leaves, branches, reproductive structure, and other litterfall components) was deposited in each bag.

A total of 300 bags were used during the study period (50 bags per treatment). These were placed on the ground to delimit the boundaries of each plot based on the cardinal directions; 12 bags were placed to the north (0°) and at the center of the plot, 13 were placed to the southeast (135°) and to the southwest (225°). Subsequently, every 30 days for 360 days, five bags per treatment were removed at random and dried at 70 °C in a 16EG GCA Corporation convection oven during 48 hours. The percentage of biomass loss equivalent to the decomposition was estimated based on the differences in relation to the initial weight, for this a C1 Sartorius digital scale was used.

The values for the remaining mass were adjusted to the simple exponential regression model (^{Olson, 1963}) according to the formula:

Where:

X _t = Litterfall mass through time (g)

X ₀ = Initial litterfall mass (g)

β ₀ = Value of the ratio in time zero

e = Napier’s constant

k = Decomposition constant

t = Time (days)

The theoretical time required for 50 %, 95 % and 99 % decomposition of the litter was also estimated, based on the formula:

Environmental variables

Room temperature (°C) and relative humidity (%) were calculated (on an hourly basis) at the study site, with HOBO automated sensors (HOBO U23 Pro v2, Onset, Bourne, MA); the rainfall (mm) was recorded on a daily basis using a similar equipment (Table 2).

Statistical analyses

The data were subjected to a statistical analysis of an experimental design with a classification criterion (^{Ott, 1993}). The mean comparison test was applied according to Tukey’s procedure, with a significance level of 0.05 (^{Ott, 1993}). The loss of litter weight through time was calculated and charted in order to evaluate the decomposition rate. The weight loss curve was then adjusted, using exponential non-linear regressions (_tRatio= β ₀ e^-kt) and the decomposition coefficient k was calculated for each treatment based on the model with the best adjustment. Furthermore, a Pearson’s correlation analysis was performed in order to determine the degree of association between litter degradation with the various treatments and certain environmental variables. All the analyses were performed using SPSS^® (Statistical Package for Social Sciences), standard version 20.0 for Windows (^{SPSS, 2012}).

Literfall production

Figure 1 illustrates the monthly production of litterfall: (a), leaves (b), reproductive structures (RS: c), branches (d) and other components (e). The highest total production occurred in December, with T2 (130.9 g m^-2), and the lowest, with T5 (1.6 g m^-2), in March. In all six sites, the lowest productions were recorded in March, and the highest, in December.

Testigo = Control; Mes de muestreo = Sampling month; Depósito de hojarasca = Deposit of litterfall; Depósito de hojas = Deposit of leaves; Depósito ER = Deposit of RS; Depósito de ramas = Deposit of branches.

Figure 1 Monthly production (mean ± standard error, n=3) of the litter components for the various study treatments. 

The production of leaves was higher than that of the control, with an interval of 3.1 (March) to 101.7 g m^-2 (December); the minimum leaf productions occurred with T5, with 1.5 (March) to 65.8 g m^-2 (December). The collection of reproductive structures was most limited in March (0.03), and much more abundant (2.84 g m^-2) in December, with T5. The maximum branch production occurred in February (27.01), and the minimum, in July (0.10 g m^-2), in the control. As for the “Others” component, the minimum production (0.07) was obtained in March with T4, and the maximum production (9.22 g m^-2), in October, in the control.

According to the results shown in Table 3, litter production was most abundant in the control (4.90 Mg·ha^-1·year^-1), while T5 exhibited the smallest amount (2.58 Mg·ha^-1·year^-1). The leaves were the main source of litter, with 72.5 % (Control) to 89.7 % (T2) of the annual production, followed by the branches, with values ranging between 2.7 % (T2) and 15.6 % (Control), and “Others”, with 5 % (T4) to 11 % (Control), while reproductive structures contributed the least organic matter to the litter, with values under 1 % (T4) to 2.1 % (T5) of the total (Figure 2). There were no significant differences (P>0.05) between treatments in the leaves, reproductive structures, and total components. However, the branches and “Others” components did exhibit such differences (P<0.05).

Tratamientos = Treatments; Producción de hojarasca = Production of litter.

Figure 2 Composition of the litter in percentages for the six treatments. 

Regardless of the components of the total or individual litterfall for each treatment, the values correspond to the productions previously observed in studies of Pinus patula Schiede ex Schltdl. & Cham., Quercus humboldtii Bonpl., Quercus potosina Trel. and Pinus cembroides Zucc. (^{Zapata et al., 2007}; ^{Pérez-Suárez et al., 2009}; ^{Reyes-Carrera et al., 2012}), which recorded productions of 2.43 to 8.30 Mg·ha^-1·year^-1.

The largest contribution to litter was obtained from the leaves from all the plots, with more than 70 % of the total; this result agrees with the findings of ^{Yang et al. (2001)}; ^{Nájera and Hernández (2009)} and ^{López-Hernández et al. (2013)}.

Litterfall degradation

There were two phases of litterfall decomposition in terms of its results: one at the beginning of the process, when the degradation was fastest at 30 to 150 days, and then diminished until it became stable at 180 to 360 days; during this stage, there was a marked drop at 300 days, and, subsequently, the tendency to degradation occurred again in all the treatments. This agrees with the findings cited in the literature (^{Xuluc-Tolosa et al., 2003}; ^{Goma-Tchimbakala and Bernhard-Reversat, 2006}; ^{Villavicencio-Enríquez, 2012}).

The initial exponential loss is due to the leaching of the most soluble substances (^{Bocock et al., 1960}), as well as to the fragmentation by microbial populations, which begin by consuming those substances that are most easily decomposed (^{Aerts and Chapin, 2000}).

The decomposition rates of the second phase are influenced by the content of hard-to-degrade materials, such as hemicelluloses and lignin (^{Bernhard-Reversat, 1993}). The use of newly fallen leaves in this study made it possible to significantly increase the initial loss percentage, as fresh or tender leaves are generally more leachable than ripe or dry leaves (^{Bernhard-Reversat, 1972}; ^{Chara et al., 2007}).

The average weight of litterfall decomposition (Figure 3a) after 360 days was 3.0 g (T1) to 4.2 g (T3), with percentages ranging between 30.2 % (T1) and 42.3 % (T3), while the remaining weight of litter decomposition (Figure 3b) after 360 days was 7.0 g (T1). The results show that slightly more weight was lost with the intense thinning (T3) than with any of the other treatments, while the remaining weight at the end of the experiment was highest with the light thinning (T1).

The p values of the variance analysis for detecting significant differences between treatments for each incubation time appear in bold characters. Control (■), T1 (×), T2 (•), T3 (○), T4 (▲) and T5 (♦).

Figure 3 Decomposition (a) and remaining weight (b) of the litterfall by incubation period for the six treatments. 

As for the decomposition rates (Table 4), the values of k (g g^-1 t) ranged between 0.00129 (Control) and 0.00141 (T3). Apparently, the litter decomposition rate was similar in the six treatments; no significant differences (P>0.05) were registered in the decomposition intervals between the five thinning intensities and the control. However, the values were low, compared to those estimated for temperate ecosystems. For example, ^{Rocha-Loredo and Ramírez-Marcial (2009)} obtained k values of 1.40 to 1.44, with decomposition percentages ranging between 34 % and 52 % for pine and pine-oak forests (Quercus crassifolia Bonpl., Quercus sapotifolia Liebm. and Pinus oocarpa Schiede ex Schltdl.) in Chiapas.

^{Zhang et al. (2008)} registered information for 293 k values, corresponding to 70 studies, and fixed them between 0.006 and 4.993; the lowest values were for the Arctic tundra, and the highest, for the tropical rainforest. ^{Stohlgren (1988)} cites k values of 0.18 to 0.62 for mixed conifer forests in the United States of America, and the time required for a 95 % decay ranged between 11 and 27 years. In the present study, the registered value was 5.8 years (2 129.8 days) to 6.3 (2 313.3 days), which is lower than that of the documented intervals.

The differences in the constants reflect not only those already extant by type of litterfall but also the influence of environmental factors that limit decomposition. In this regard, the temperate area is considered to be geographically critical for the decomposition of matter, since one of the factors that limit decomposition are the low temperatures and low rainfall rates during the study period, which were significantly and negatively correlated with decomposition (P<0.05) (with averages of -0.535 and -0.707, respectively).

The results are consistent with the records of several researchers, according to whom climate factors are more important at the global level, as they account for variations in the decomposition rates (^{Lavelle et al., 1993}; ^{Couteaux et al., 1995}; ^{Hättenschwiler, 2005}). However, when litter decomposition is studied at a local level, as in the present work, climate factors lose relevance, and the chemical quality of the material becomes the main factor, particularly in tropical regions, where high temperatures and humidity levels are relatively constant (^{Didham and Lawton, 1999}; ^{Xuluc-Tolosa et al., 2003}; ^{Vasconcelos and Laurance, 2005}).

As for the structure of the vegetation, there were no significant effects on the decomposition of the litterfall. This agrees with the findings of ^{Barriga (2000)}, who obtained no significant differences in the decomposition rates when comparing fragmented areas of the humid montane forests of Colombia; this author concludes that the decomposition process was not altered after the occurrence of anthropogenic disturbances. Whereas ^{Chacón and Dezzeo (2007)} determined a similar decomposition rate in forests with and without disturbances, and therefore, while the disturbances considered in their study did affect the structure of the forests, they had no significant impact on the littefall decomposition process.