Study area

The research was carried out at the Forest School Ecological Campus of the Universidad Autónoma de Nuevo León (UANL), in the Iturbide municipality, Nuevo León, which is located between 24°42´28” N and 99°51´43” W, at 1 600 m high (Figure 1). Pine-oak forests are the predominant vegetation type; soils are Leptosol (35.5 %), Calcisol (2.8 %) and Kastanozem (1.7 %). The average annual temperature is 14 °C and the average annual rainfall, 635 mm (^{Inegi, 2009}).

Figure 1 Location of the study area. 

Environmental variables

Relative humidity (%), air temperature (°C) and soil temperature (°C) at 5 cm deep were recorded using the HOBO automated sensors every hour; rainfall (mm) was recorded by event.

Leaf litter decomposition and collection frequency

The litter bag method (^{Bocock and Gilbert, 1957}) was used; 5 g of freshly fallen litter were collected, which were dried in the open and deposited in 25 cm x 20 cm bags of 1 mm nylon mesh fabric. This opening allows access inside the bags to certain invertebrate detritivorous, but minimizes fragmentation losses (^{Douce and Crossley, 1982}). Due to the effect of time, the difference in weight allows the decomposition of the collected material to be assessed.

In the study area, three conditions were selected by type of temperate forest vegetation: BE for the oak forest; BPE for the pine-oak and BP for the pine forest. The experiment began on December 15^th, 2016 and ended on May 9^th, 2018.

A homogeneous mixture of litter components (leaves, branches, reproductive structures and other litter components) were deposited in each bag. A total of 110 bags were placed in five experimental plots (20 m × 20 m) per study site, at a rate of 22 bags per plot distributed at random. Each of them settled on the floor surface that was previously cleaned so that the bags were in contact with the substrate.

Litter decomposition was assessed on 12 dates, collecting all the material from two bags per experimental plot every 0, 15, 30, 60, 90, 150, 210, 270, 330, 390, 450 and 510 days, which were they took to the Chemistry Laboratory of the School of Forest Sciences of the UANL, where they were dried at 65 °C in a Felisa oven, model FE-292AD, until they reached a constant weight. With these data, the percentage of weight loss was calculated.

The decomposition rate was determined through ^{Olsons’s (1963)} Negative Exponential Simple Model:

Where:

k = Decompositon constant (year^-1)

X = Litter mass in a specific time (g)

X₀ = Initial litter mass (g)

t = Time expressed in years (years)

The necessary time to decompose 50 and 99% of litter was calculated with the following equations (^{Olson, 1963}):

For decomposition at 50% the following formula was used:

For decomposition at 99% the following formula was used:

Statistical analyses

An analysis of variance was applied to the leaf litter decomposition data; To verify the assumptions of homogeneity of variances, the Kolmogorov-Smirnov and Shapiro Wilk tests (^{Steel and Torrie, 1980}) were used. The non-parametric Kruskal-Wallis test (^{Ott, 1993}) was used because the data did not show normal distribution.

To determine the leaf litter decomposition rate, four models were used: the Simple Negative Exponential proposed by ^{Olson (1963)} and three based on the recommendations of ^{Moreno et al. (2018)}; the root of the mean square error (RECM) and the Pearson Correlation Coefficient were selected, based on the fit of the models and the coefficient of determination (R²).

Statistical analyzes were done with the Statistical Package for the Social Sciences (^{SPSS, 2016}) version 22.0 for Windows.

Environmental Variables

Figure 2 shows the mean air temperature, the mean soil temperature and the mean relative humidity, as well as the monthly precipitation recorded during the study period. Rainfall was 719 mm in 2017 and 184 mm at the end of the experiment, with maximum peaks in September and November, with a total of 903 mm. The air temperature ranged from 14 to 23 °C and there was no frost, while the relative humidity was 48 to 85 %. The soil temperature followed the same trend as the environmental one, with variations from 12 to 20 °C.

Figure 2 Air temperature (°C), Soil temperature (°C), relative humidity (%) and rainfall (mm) recorded during the experiment. 

Litter decomposition

A phase of slow decomposition from day 15 to 90 and greater decomposition until day 510 is observed (Figure 3); the tendency of the process in the three types of forest remains constant. As the time of the test lasted, the loss of biomass increased. At the end of the essay (510 days after it started), decomposition was 27 % for the oak forest, 23 % for pine-oak forest and 17 % for pine forest. According to the Kruskal-Wallis test, the values did not show significant differences on days 0, 15 and 450.

Be = Oak Forest; BPE = Pine-Oak Forest; BE = Oak Forest.

The p-value of the Kruskal-Wallis test to detect differences between forest types at each time of the trial is shown in the graph. Probabilities in bold (upper part) indicate significant differences.

Figure 3 Leaf litter decomposition (%) of the three types of forests in northeastern Mexico. Each value plotted represents the mean (n = 10 ± standard error). 

When evaluating the loss of biomass in a temperate forest for one year in two locations in El Salto, Durango, ^{Moreno et al. (2018)} calculated decomposition rates of 31 % and 35 %. Likewise, ^{López et al. (2018)} reported that in the first 30 and 150 days the greatest loss of biomass of Pinus cooperi C. E. Blanco occured in plots with different intensities of thinning in El Salto, Durango; while ^{Rocha and Ramírez (2009)}, obtained decomposition results that vary from 34 to 52 % and found no significant differences, for pine-oak forest in Chiapas, Mexico.

^{Sierra et al. (2009)} mention that the initial phase of decomposition presents a more pronounced curve in the loss of biomass and, in the end, the slope is significantly reduced, which means that the slow decomposition of leaf litter is controlled by environmental factors and by litter quality (chemical composition of litter, variables not analyzed in the present study). ^{Ibarra et al. (2011)} indicate that the decomposition behavior is due to the fact that all the components of the litter are used, including those of rapid decomposition (leaves and reproductive structures) and those of slow decomposition (twigs and bark).

It also influences the nature of the litter, the C / N ratio of the litter, as well as the content of lignin and tannins, the greater the degree of lignification of the needle the longer the decomposition time.

Dry pine needles have a high lignin content (25.93-30.05 %) (^{Díaz et al., 2007}) and tannins (0.07-0.12 %), precursors of humic substances in organic matter, high C / N ratio (around to 150), also have a low Ca (39.5-54.3 %) and Mg (10.8-18.8 %) content (^{Bernabé et al., 2013}).

Decomposition models

In order to assess decomposition, linear, potential and exponential mathematical models, among which is the Negative Exponential Simple Model by ^{Olson (1963)}, as well as some other evaluated by ^{Moreno et al. (2018)}.

The models that were used in the actual research study were:

Where:

H= Part of the remaining litter in t years

t= Time expressed in years.

β₀ = Initial weight percentage

β₁ = Relative decomposition rate

*Proposed by ^{Moreno et al. (2018)}.

Table 1 shows the models obtained for the three experimental sites; it shows that the values of the determination coefficient varied from 0.766 to 0.899 and are similar to those recorded by ^{Moreno et al. (2018)} with 0.965 to 0.969, as well as those of ^{Del Valle (2003)} with 0.966, as well as those of ^{Ibarra et al. (2011)} with 0.946 and ^{Martínez et al. (2007)} with 0.698 to 0.821. For this reason, the Negative Simple Exponential Model (1) is the best fit according to the determination coefficient and the Pearson correlation coefficient for the three types of forests.

The correlation coefficient was distributed in a range between 0.875 and 0.948 for the model (1). It can be seen that the Negative Simple Exponential Model is the best fit based on the highest value of the Pearson determination coefficient and correlation coefficient, as well as the lowest RECM for the three types of forest.

Litter decomposition rate

Table 2 shows the different values of k, which corresponds to the decomposition rate in the three types of forest: for BE -0.259, for BPE -0.209 and for BP -0.158. The time required for decomposition at 50 and 99 % was 2.67 to 17.74 years for BE and 4.38 to 29.11 years for BP. The decomposition rate mentioned by ^{Moreno et al. (2018)} in Durango was from -0.424 to -0.425, during a year of study in a temperate forest and the time required for decomposition was 1.6 (50 %) to 10.8 years (99 %).

The prediction of leaf litter decomposition (Figure 4) indicates the distribution of data (relative weight of leaf litter) and the prediction of values for each type of forest (Olson model), during the time of the test. It is possible to notice the dispersion of the data in each type of forest and the prediction that the data should have during the experiment, according to the referred model.

Figure 4 Prediction of litter decomposition by duration of the test in the three types of forests.