Introduction

Scattered trees are predominant components of many anthropogenic landscapes, particularly in livestock grazing systems (^{Rivest, Paquette, Moreno, & Messier, 2013}). In pastures, scattered trees increase soil organic carbon and nutrients (^{Aryal, Gómez-González, Hernández-Nuriasmú, & Morales-Ruiz, 2018}; ^{Avendaño-Yáñez, López-Ortiz, Perroni, & Pérez-Elizalde, 2018}) through litter accumulation. For example, scattered Lysiloma acapulcense (Kunth) Benth. and Vachellia pennatula (Schltdl. & Cham.) Seigler & Ebinger trees increase the nitrogen and phosphorus content in tropical pasture soils (^{Avendaño-Yáñez et al., 2018}). Both species are native to tropical dry forest (TDF), and are used as multipurpose trees in agroecosystems: L. acapulcense is used as shade for livestock, as a source of timber for construction, fencing and firewood; V. pennatula is considered as a weed, even though its forage fruits and firewood are widely used (^{Suárez et al., 2012}; ^{Williams-Linera, Alvarez-Aquino, Hernández-Ascención, & Toledo, 2011}). Yet, the ecological traits and ecosystem services of these two tree species, such as litterfall and its role in soil nutrient inputs, are not fully understood.

Leaf quantity and quality are linked to nutrient cycling in terrestrial ecosystems (^{Dutta & Agrawal, 2001}), because leaves are the primary source of organic matter for the soil (^{Sayer & Tanner 2010}; ^{Vitousek & Sanford, 1986}). Leaf litter decomposition depends on its chemical composition (^{Loranger, Ponge, Imbert, & Lavelle, 2002}), mainly lignin content, because it controls the decomposition rate of plant residues (^{Austin & Ballaré, 2010}; ^{Rahman, Tsukamoto, Rahman, Yoneyama, & Mostafa, 2013}). The lignin-nitrogen ratio (L/N) drives the decomposition process; low ratios accelerate degradation and the release of nutrients to the soil (^{de Oliveira et al., 2016}).

Until now, most studies on litter quantity and quality in tropical environments have been performed in mature and secondary forest or plantations, but the study of the role of scattered trees in pastures has been marginal. The objective of this research was to estimate the litterfall production during the dry season by scattered L. acapulcense and V. pennatula trees in pastures.

Materials and methods

The study was conducted in a pasture within a dry tropical forest ecosystem, in Paso Panal, Paso de Ovejas municipality, Veracruz, Mexico (19° 13' 14'' N and 96° 29' 5'' W at 167 m). The climate type is Aw₀"(w) (i') g, classified as the driest of the warm and humid categories (Köppen classification, adapted by ^{García, 2004}). The average annual temperature is 25.2 °C, with an average annual precipitation of 909 mm. The landscape is highly fragmented due to agriculture and ranching, and only small patches of primary and secondary vegetation from TDF persists in the region.

Litterfall collection

To assess litterfall, 0.25 m² circular traps were handmade with 1.0 mm nylon mesh fixed to iron frames, and protected with a 40 x 40 cm piece of cloth at the bottom to prevent the loss of small leaflets. Three traps were placed under each tree at 1 m aboveground, at midpoint between the base of the shaft and the drip zone (Figure 1). Trapped litter was collected every 30 days over two consecutive periods of natural litterfall in the area (December 2014 - May 2015 and November 2015 - April 2016). Lysiloma acapulcense and V. pennatula lose their leaves during the dry season, but L. acapulcense is partially deciduous (^{Camacho-Moreno et al., 2017}). Litter was dried in a forced air oven at 65 °C for 48 h, and then separated into leaves, twigs, and fruits. Only leaf biomass was evaluated.

Figure 1 Location of litter traps under a Vachellia pennatula tree. 

Results and Discussion

Scattered trees remaining after the fragmentation of TDF play important roles in maintaining ecosystem functions and services in modified environments (^{Manning, Fischer, & Lindenmayer, 2006}) such as pastures. In this study, leaves comprised the largest fraction of the litter for both tree species (53 to 64 %), but fruits were also a large litter component (Figure 2). Although L. acapulcense yielded approximately 4.564 Mg·ha^-1·year^-1 of leaf litter, and V. pennatula 4.202 Mg·ha^-1·year^-1, the leaf litter quantities were similar (t = 0.9, df = 39, P = 0.37). These quantities from dispersed trees are significant because they represent the primary nutrient inputs to pastures. However, other authors have found greater litter quantities for other TDF species. For example, Gliricidia sepium (Jacq.) Kunth produced 10.42 Mg·ha^-1, and Mimosa caesalpiniifolia Benth. 10.79 Mg·ha^-1 in silvopastoral associations (^{de Oliveira et al., 2016}). Pithecellobium dulce (Roxb.) Benth. released 13.83 Mg·ha^-1·year^-1, and Leucaena leucocephala (Lam.) de Wit produced 10.83 Mg·ha^-1·year^-1 in 10-year-old plantations (^{Ceccon, Sánchez, & Powers, 2015}). In late successional forests of the dry tropics, ^{Campo and Vázquez-Yañez (2004)} also reported higher litter biomass (9.20 Mg·ha^-1·year^-1). Although only leaf litter was accounted for, lower amounts of litter assessed in this research may be explained by the fact that litter was collected only during the season of greater leaf-fall, discounting miscellaneous litter released year-round (^{de Queiroz et al., 2019}). In addition, differences could arise from scattered trees being a part of simplified and low-biodiverse agricultural land units, as opposed to higher species-rich stands where different species have temporal and complementary dynamics of litterfall, leading to constant and higher litter supply in time and space (^{Huang, Ma, Niklaus, & Schmid, 2018}). The litter biomass from both evaluated species are more similar to litter from early succession tropical forest stands (4.749 Mg·ha^-1·year^-1) (^{Aryal, De Jong, Ochoa-Gaona, Mendoza-Vega, & Esparza-Olguin, 2015}), but there are few references of similar research to compare to these findings.

Figure 2 Composition of vegetative structures collected from the legume tree species studied in the pasture, during the dry season. 

Regarding leaf-fall timing, the parameters for the beginning and end of leaf-fall as well as maximum peak were significant (P < 0.0001) in the models for both tree species (Figure 3). The beginning of leaf-fall (start of season) was during autumn (on day 110 for V. pennatula, and day 122 for L. acapulcense). The time of maximum leaf-fall (peak of position) for V. pennatula was in December (day 157), and at the end of January (day 180) for L. acapulcense. The end of leaf-fall (end of season) took place towards the end of winter (February-March, day 205) for V. pennatula, and during spring (March-April, day 237) for L. acapulcense.

Figure 3 Annual model output of litterfall for the two legume tree species studied in the pasture, where the beginning, maximum and final point of leaf-fall are observed. 

The quantities of leaf litter produced by both tree species were similar, but the seasonal pattern of leaf-fall varied between the species. Although the start of leaf-fall appeared similar between the species, the maximum leaf-fall for V. pennatula occurred during December, while that for L. acapulcense occurred in January. The end of leaf-fall for V. pennatula was in February, while the leaf-fall for L. acapulcense ended between March and April. Yet, L. acapulcense did not completely lose its leaves because it is a semi-deciduous species (^{Camacho-Moreno et al., 2017}). In environments with long seasonality, deciduous species completely lose their foliage, showing patterns of high leaf deposition towards the middle or end of the dry season (^{Kozovits et al., 2007}). A similar pattern was observed for both species in this study, although the leaf-fall period was longer in L. acapulcense, as observed by Camacho-Moreno et al. (2017).

The concentration of nitrogen (1.49 and 1.94 %) and lignin (47.07 and 40.40 %) in leaf litter did not differ between L. acapulcense and V. pennatula (df = 7, P > 0.05) respectively, although the lignin-nitrogen ratio (L/N) was significantly higher in leaves from L. acapulcense (df = 7, P < 0.0001) than V. pennatula (Table 1). ^{Oyun (2006)} reported that some legume species such as L. leucocephala and G. sepium have leaves with high concentrations of nitrogen (4.5 and 4.8 %), intermediate concentrations of lignin (25.2 and 23.9 %), and low L/N ratios (5.6 and 5.0), respectively. As a result, their litter is considered of high quality. As suggested by the leaf-fall model, L. acapulcense retains its leaves for a long time, and may account for its higher L/N ratio. This would allow trees to mobilize nitrogen from the leaves, causing a greater imbalance in the ratio; however, the lower L/N ratio in V. pennatula litter is indicative of better litter quality and suggests that the leaf litter may be more labile than that from L. acapulcense. Leaf litter with a higher concentration of lignin degrades more slowly, partially reducing the nutrient release rate from the litter and the risk of loss by leaching (^{Krishna & Mohan, 2017}). Thus, the combination of V. pennatula and L. acapulcense can be advantageous under agroforestry management. While V. pennatula leaves degrade faster, L. acapulcense leaves degrade slower, forming a temporary storage of carbon and nitrogen in the litter. However, more studies need to be conducted on other chemical components in the leaves of both species and on their degradation capacities.

Scattered L. acapulcense trees are associated with pastures having long dry seasons, but the number of trees of this species declines due to fire and other management practices. On the other hand, V. pennatula trees are abundant in pasture lands regardless of fire and spraying for weed control performed by ranchers. Further efforts are needed to increase the cultural value of these TDF species for ranchers. Therefore, our research raises awareness of the importance of preserving the two species and enhancing the integration of these native trees into agroforestry management schemes. Still, further studies are required to explore the relationship between scattered tree density and litter production.

Conclusions

Litterfall from scattered trees is the primary source of nutrients for pasture soils. However, leaf litter deposited by scattered trees in pastures has been scarcely studied. This issue is particularly important because nutrient extraction from soils under grazing conditions is constant, with little or no nutrient inputs or amendments. Lysiloma acapulcense and Vachellia pennatula add important leaf litter quantities to pastures even as scattered trees. The different L/N ratios of these species suggest a complementary role in an agroforestry system because litter from V. pennatula has potential for more rapid decomposition and nutrient release while L. acapulcense litter might be a temporary carbon sink. Yet, there are gaps in the knowledge of ecological and biological traits of these species. Filling in these gaps will provide essential knowledge in the management of scattered trees in human modified landscapes; moreover, understanding the role of scattered trees in dry tropical pastures will help to improve agroforestry management.