Introduction

The reduction of nutrient and organic matter content in soils is a serious threat for agricultural production and food security in many tropical countries (Hossain et al., 2011). In this sense, agroforestry provides a sustainable opportunity to counteract this threat due to its potential to reestablish degraded or marginal soils. According to Oelbermann et al. (2004) agroforestry systems (AFS) improve soil quality through the input of organic matter from harvest and litter residues maintaining or increasing organic matter in the soil.

Organic matter added through litter, usually improves soil quality as it enhances its water holding capacity, water filtration, biodiversity, soil microorganism activity and nutrient concentration (Ngora et al., 2006; Hossain et al., 2011). Among litter types, leaf litter has comparatively the highest nutrient concentrations and also returns the largest amounts of nutrients to the soil, in essence it is the major source of soil nutrients (Hossain et al., 2011).

In this context, the combination of cacao trees with nontimber species, e.g. plantain and cassava, and with shade trees like Erythrina glauca Willd. and Gliricidia sepium (Jacq.) Kunth ex Walp. are excellent examples of compatibility and complementarity among different species, as well as of sustainability of a multistrata system with significant litter production (Somarriba, 2006; Fontes et al., 2014). However, the litter produced by each species has very particular characteristics influencing the quantity and the time in which the nutrients are released to the soil during the decomposing process (Beer et al., 1990; Schroth et al., 2001; Mendonça and Stott, 2003).

These nutrient deposition, decomposition and release rhythms to the soil must be synchronized with the crop's nutritional needs and the maintenance of the soil's biological functionality and activity (Beer et al., 1990; Schroth et al., 2001; Mendonça and Stott, 2003). For this reason, it is important to know and compare the litter's nutrient and biomass input from each species that compose the AFS with cacao. Therefore, in this study the litter decomposition dynamic from leaves and petioles of the species Gmelina arborea Roxb., Gliricidia sepium (Jacq.) Kunth ex Walp., Cedrela odorata L., and Theobroma cacao L. that comprise an AFS in the municipality of Rionegro, department of Santander was established. Moreover, nutrient input and economic assessment per species was also determined.

Materials and methods

Study site description

The research was carried out in an agroforestry system with cacao at the farm El Encanto located in the municipality of Rionegro, department of Santander, Colombia. The farm is located on western flank of the Eastern Cordillera of Colombia at 07°24'51.0" N and 73°11'33.6" W, and 825 m of altitude. The region has an annual average rainfall of 2,500 mm, a mean temperature of 25°C, 80% relative humidity and an annual solar radiation of 1,700 h. Acording to Holdridge (1967) it is located in the life zone classified as tropical humid forest (bh-T) and also placed in the Kv agroecological zone, characterized by having alluvial soils, clay loam and sandy loam textures. These terrains are distinguished by having a steep relief with slopes higher than 50% that are characteristic of Colombia's eastern mountain range (ICA-IGAC, 1985).

Sample collection and processing

The study was carried out with the species G. arborea, G. sepium, C. odorata and T. cacao that are traditional elements of AFS with cacao in the municipality of Rionegro. The decomposing rate was measured using the decomposing bag technique describe by Bärlocher (2005) during four months. Plant material samples from each species, i.e. fresh leaves and petioles, are taken directly from the plants in order to know the exact time that the decomposing period starts and schedule further monitoring. Each tissue sample (100 g) was placed in a polyethylene bag of 30 χ 20 cm with mesh openings of 1.0 χ 1.0 mm.

The bag method was consisted in close a sample of vegetal tissue with know weight and nutrient content inside of a porous bag that allows the exchanges between the soil and the vegetal tissue. This method was used to determinate the rate of decomposition and nutrient release.

Fifteen samples were placed in the field at a depth of 2 cm from the litter in a completely random design with an area of 0.5 ha. Three bags per species were collected after 8, 15, 23, 84 and 113 d of being placed in the site. The material collected was dried in an oven at 75°C for approximately 72 h until reaching a constant weight. After being dried, the samples were weighed in a precision balance.

Mass loss

The decomposition of the material was evaluated through dry weigh loss per each degradation time. The residual dry weight percentage (%RDW) was calculated considering the dry weight coefficient in the oven of the residual material per period (RMP), over the dry weight in the oven of the initial material (DWI), as follows:

With the RDW value the decomposition rate of each vegetal material was determined. According to Oelbermann et al. (2004), the relative decomposition rate or transfer rate to the soil was calculated as follows:

Where, Y_f is the residual dry weight percentage; Y₀ is the initial dry weight percentage; k is the relative decomposition rate or decomposition speed of vegetal material; and t is the time (days). To evaluate the release or nutrient transfer, a single exponential model was implemented (Oelbermann et al., 2004b; Farfán 2009), as follows:

Where, W_f is the remaining amount of each nutrient; W₀ is the mineral amount of each nutrient; k is the nutrient release constant; and t is the decomposition litter time (days).

Nutrient content measurements

To measure nutrient content, the following methods were used: the organic nitrogen (N) by the Kjeldahl digestion method, based on wet oxidation of organic matter using H₂SO₄ and a digestion catalysis (Horneck and Miller, 1998); the phosphorous (P) by the molybdovanadate colorimetry method using a spectrophotometer; and the potassium (K), calcium (Ca) and magnesium (Mg) by the method of Walsh with atomic absorption spectrophotometry (Unigarro et al., 2009).

Statistical analysis

Litter quality was evaluated through the comparison between residual dry weight percentage and the nitrogen, phosphorous, potassium, calcium and magnesium content measured in percentage. The data obtained were compared and assessed between species using an analysis of variance (ANAVA). A Tukey test was carried out to compare the differences between the species when the F values of the ANAVA showed significant differences (P≤0.05).

Economic analysis

Nutrient input to the soil from different treatments or species used as shade in cacao plantations was assessed in economic terms, considering as a base the equivalent prices of conventional fertilizers in the market according to the availability as commercial formulations or as simple elements.

Results and discussion

Quality of the materials

Regarding the nitrogen levels in plant tissue to be decomposed, C. odorata contributed with the lowest levels (8.2 g kg^-1) and G. sepium with the highest levels (16.2 g kg⁴). Regarding phosphorous, due to the low levels found, this element could have been immobilized as the levels of all the species evaluated showing values beneath 2.5 g kg^-1. The highest levels of potassium were found in G. sepium and in G. arborea (Tab. 1). According to the calcium and magnesium levels found, the contribution per species in ascending order were G. sepium < G. arboreum < T. cacao< C. odorata, and C. odorata < G. sepium < G. arboreum< T. cacao, respectively (Tab. 1). Taking into account the levels established by Palm (1995), all the species evaluated showed a concentration below the critical N and P levels, i.e. 20 to 25 g kg^-1 and less of 2.5 g kg^-1, respectively. According to the same author, this event could have caused an immobilization of these elements.

Decomposition patterns

The residual weight (%) varied significantly in time and within the evaluated species (P≤0.05) (Tab. 2). Eight days after initiating the litter decomposition process, G. sepium showed the highest residual weight loss (31.39%) followed by T. cacao (17.73%) with highly significant differences between the species evaluated (P=0.0007) (Tab. 2; Fig. 1). The species G. sepium maintained the weight loss behavior after 15, 23, 84 and 113 d, reaching 74.80% in weight loss (Fig. 1). Although in the first samples (8 d) G. arborea did not show the highest weight loss values, at day 15, 23, 84 and 113 the weight loss reached considerable values, showing the highest weight loss values in days 84 and 113 (Fig. 1), with significant differences compared with the other species evaluated (P=0.0004 and P<0.0001, respectively) (Tab. 2). Regarding the species with the lowest weight loss, C. odorata and T. cacao showed the lowest values during the assessment period (Fig. 1). In these species, the highest weight loss was found between day 8 and 23, reaching values of 31.28% in C. odorata and 35.23% in T. cacao, showing maximum values in day 113 (41.57% and 43.88%, respectively) (Fig. 1).

Nutrient release percentage

Nitrogen, phosphorous, potassium, calcium and magnesium release from the aerial biomass of the different species evaluated in this study did not show significant differences. However, the highest release was found in G. sepium, followed by G. arborea (Fig. 2a-2e). Due to the fact that nutrients input are measured in terms of N and P mineralization (Mafongoya et al., 1997), in the AFS evaluated in this study G. sepium is considered a high quality source of nutrients as it has the highest content of these elements in its initial plant material content, and a higher nutrient release in the decomposing process (Fig. 2a and 2b). The leaves of G. sepium have a high N content and a low amount of lignin and polyphenols, so these can therefore be decomposed much faster (Mafongoya et al., 1997).

According to Munguía (2003) similar tendencies were found in the Verde Vigor farm, located in Pérez Zeledón canton, Costa Rica, working with litter from Erythrina poeppigiana (Walp.) O. F. Cook, Eucalyptus deglupta Blume and Coffea arabica L. In this study, E. poeppigiana showed to be the shade species that had the fastest N release, although it differs in the percentage and time required (65% in 24 d). Moreover, E. deglupta showed the slowest release rate, i.e. it did not release this element on its own but only when combined with C. arabica (2% in 24 d). Otherwise, Montenegro (2005) found that of the shade species evaluated in a coffee ASF in Turrialba (Costa Rica) the legume species were the ones that generated the highest nutrient release, concurring with the results found in this study. Likewise, both studies found that G. sepium showed the best behavior in relation to nutrient release. Moreover, regarding the low nutrient release showed by T. cacao and C. odorata, this can be due to various factors. Normally, this low nutrient release is related with big quantities of reactive polyphenols or structural lignin in the tissues of species associated to insoluble proanthocyanidins (Ma-fongoya et al., 1997).

The species evaluated showed a quick N, P and K release off in the first lixiviation phase during the decomposition process (Fig. 2a-2c). Anderson and Ingram (1993) mentioned that during the plant tissue decomposition process the release and mineralization of N and other elements frequently show a very fast initial phase. In this phase, the microorganisms that comprise the soil's biomass degrade the litter, and secondary products as cellulose and hemicellulose, constituents of the cellular walls, are obtained. At the same time this new biomass and its metabolic products become substrates for the second phase. This second phase is characterized by being slower due to the increase of recalcitrant fractions and is mainly regulated by the lignin content in decomposing tissue.

Additionally, potassium release was found to occur in all species evaluated between days 8 and 23 with release rate values (k) between -0.051 and -0.116 depending on the species (Fig. 2c, Tab. 3). This is attributed to the high K mobility and its fast lixiviation on the initial phases of the litter decomposition process (Zaharah and Bah, 1999). However, these results are contrary to the ones found by Munguia (2003) regarding K release in litter of E. poeppigiana, where in only 24 d 39% of this nutrient was released, and after 213 d 99% was released with a rate of -1.55 per day, obtaining a higher value compared to other treatments and nutrients. The results in this same study also showed that E. deglupta litter released in 24 d 17% of the K, and after 213 d 88% had been released, showing a k of -0.49 per day, i.e. the lowest value found.

The calcium showed a fast release in the species C. odorata during the period from day 8 to 23 and decreased after 84 and 113 d with a total release of 9.84%. However, the species G. arborea had the highest release constant during all the evaluation period for a total of 10.26% (Fig. 2d, Tab. 3). Munguia (2003) reports for E. poeppigiana a 67% calcium release in 213 d with a rate (k) of -0.5 d^-1, meanwhile a treatment with E. deglupta + C. arabica + E. poeppigiana and another one with a mixture of E. deglupta + E. poeppigiana begin calcium release from day 100 with a release rate (k) of -0.13 and -0.14 per day, respectively (Tab. 3). This slow release is due to the type of link that this element generates with other elements in the cellular wall.

Regarding magnesium, the highest release was shown by T. cacao from day 8 to 113 with a total of 1.82% followed by G. arborea with a release of 1.35%, being the species C. odorata and G. sepium the ones with the lowest release percentages in the evaluation period, i.e. between 0.71% and 1.19%, respectively (Fig. 2e). These results do not coincide with the outcomes reported by Munguia (2003) as the magnesium release values found for E. poeppigiana were 7% in 24 d and 97% in 213 d, showing a release rate (k) of -1.02 per day. Meanwhile, for E. deglupta in only 24 d there was no magnesium release and after 213 d there was a 19% release rate (k) of -0.013 per day.

Economic analysis

According to the nutrient input of each species evaluated and with the traditional fertilization levels used in the municipality of Rionegro, department of Santander, the study showed that to produce 1,000 kg ha^-1 of dry grain of cacao, the species G. arborea contributes with the highest nitrogen and phosphorous quantities in the AFS studied with 9.37% and 1.32%, followed by G. sepium with 8.99% and 0.96%, respectively (Tab. 4). Altogether, the litter produced by the four species that comprise the system contributed with 27.44% of the nitrogen and 3.46% of the phosphorous that is usually employed in an AFS with cacao. Regarding potassium input, the litter produced by G. sepium contributed with 4.97% of the K that was relatively superior to the other species assessed and that ranged from 1.85 to 2.43% (Tab. 4). In total, the litter generated by the four species evaluated contributed with 10.65% of the K that is normally required to fertilize the AFS (Tab. 4).

These contributions from the litter produced by the species that comprise the AFS, allow the optimization of the producer's finances, as currently the main fertilization source for these systems are synthesis fertilizers known as Triple-15 that provide N, P and K to the soil with a market price of75,000 COP (ca. 25 USD) for 50 kg, plus application wage costs of up to 25,000 COP (ca. 8-9 USD). According to the results obtained in this research, with the contribution of the litter produced by the species assessed that constitute the AFS, the fertilization costs can be reduced 10% in average as well as the use of synthesis fertilizers and the amount of labor during fertilization processes.

The amount of nutrients extracted in kilograms to produce and harvest 1000 kg of dry cacao per ha during a year in the municipality of Rionegro are: nitrogen 31-40, phosphorous 5-6, potassium 54-86, calcium 5-8 and magnesium 5-7. Considering that part of the fertilizers are lost due to solubility problems, washing, microorganism absorption and that most of the nutrient content extracted by the plant are not part of the cacao almonds, the application of these element should be higher (Mejia, 2000). In this sense, the litter contributes greatly to the cycling system. This demonstrates the sustainability of the productive potential in agroforestry systems with cacao that is directly linked to the conservation of soil fertility, and is being directly influenced by the nutritional quantity and quality of the plant residues that are annually returned to the soil through litter.

Conclusions

The species G. arborea showed the highest plant material decomposition with an average indicator in residual weight of87.55% after 113 d, meanwhile C. odorata (timber specie) showed the lowest residual weight 40.01%.

G. arborea showed the highest nutrient release followed by G. sepium, and the lowest was shown by T. cacao and C. odorata.

Calcium is the nutrient that shows the highest release percentage followed by nitrogen with a better release behavior compared to potassium and magnesium, meanwhile it also showed to be the nutrient with the slowest release.

Acknowledgments

This study was carried out within the Project "Conventional and organic fertilization schemes for a cacao (Theobroma cacao L.) production system and its effects on grain production and quality in Colombia's main production regions", executed by the Colombian Corporation for Agricultural Research (Corpoica) and financed by the Ministry of Agriculture and Rural Development of Colombia (MADR). The authors wish to thank Karen Amaya Vecht for the revision and translation of the document.

Literature cited

Anderson, J. and J. Ingram. 1993. Tropical soil biology and fertility. 2^nd ed. CAB International, Oxford, UK.

Anderson, J . and M. Swift. 1983. Decomposition in tropical forest. pp. 287-309. In: Sutton, S.L, T.C. Whitmore, and A.C. Chadwick (eds.). Tropical rain forest: Ecology and management. Blackwell Scientific Publications, Oxford, UK.

Bärlocher, F. 2005. Leaf mass loss estimated by litter bag technique. pp. 37-42. In: Methods to study litter decomposition. Springer, The Netherlands.

Beer, J., W. Bonneman, H. Chavez, H. Fassbender, A. Imbach, and I. Martel. 1990. Modelling agroforestry systems of cacao (Theobroma cacao) with laurel (Cordia alliodora) or poro (Erythrina poeppigiana) in Costa Rica. V. Productivity indices, organic material models and sustainability over ten years. Agroforest. Syst. 12, 229-249. Doi: 10.1007/BF00137286.

Farfan V., F. 2009. Producción y descomposición de biomasa seca y transferencia de nutrientes en sistemas agroforestales con café. In: Segundo seminario regional materia orgánica, biología del suelo y productividad agrícola. Sociedad Colombiana de la Ciencia del Suelo, Cenicafé, Federación Naional de Cafeteros, Bogotá, Colombia.

Fontes, A.G., A.C. Gama-Rodrigues, E.F. Gama-Rodrigues, M.V.S. Sales, M.G. Costa, and R.C.R. Machado. 2014. Nutrient stocks in litterfall and litter in cocoa agroforests in Brazil. Plant Soil 383 (1-2), 313-335. Doi: 10.1007/s11104-014-2175-9.

Holdridge, L. 1967. Life zone ecology. IICA, San Jose, Costa Rica.

Horneck, D. A. and R.O. Miller. 1998. Determination of total nitrogen in plant tissue. Handbook of reference methods for plant analysis 2, 75-83.

Hossain, M., M.R.H. Siddique, M.S. Rahman, M.Z. Hossain, and M.M. Hasan. 2011. Nutrient dynamics associated with leaf litter decomposition of three agroforestry tree species (Azadirachta indica, Dalbergia sissoo, and Melia azedarach) of Bangladesh. J. Forestry Res. 22(4), 577-582. Doi: 10.1007/s11676-011-0175-7.

ICA-IGAC. 1985. Zonificación agroecológica de Colombia. Memoria Explicativa. ICA-IGAC, Bogotá, Colombia.

Mafongoya, P.L., K.E. Giller, and C.A. Palm. 1997. Decomposition and nitrogen release patterns of tree prunings and litter. Agroforest. Syst. 38, 77-97. Doi: 10.1023/A:1005978101429.

Mejia, L. 2000. Nutrición del cacao. Tecnología para el mejoramiento del sistema de producción de cacao. Corpoica, Bucaramanga, Colombia.

Mendonça, E.S. and D.E. Stott. 2003. Characteristics and decomposition rates of pruning residues from a shaded coffee system in southeastern Brazil. Agroforest. Syst. 57(2), 117-125.

Montenegro, G. 2005. Efecto del aporte de nutrientes de la biomasa de tres tipos de árboles de sombra en sistemas de manejo de café orgánico y convencional. M.Sc. thesis. Centro Agronómico Tropical de Investigación y Enseñanza (CATIE), Turrialba, Costa Rica.

Munguía, R. 2003. Tasa de descomposición y liberación de nutrientes de hojarasca de Eucalyptus deglupta, Coffea arabica y de hojas verde de Erythrina poeppigiana sola o en mezclas. M.Sc. thesis. CATIE, Turrialba, Costa Rica.

Ngoran, A., N. Zakra, K. Ballo, C. Kouame, F. Zapta, G. Hofman, and O.V. Cleemant. 2006. Litter decomposition of Acacia auriculiformis and Acacia mangium under coconut trees on quaternary sandy soils in Ivory Coast. Biol. Fert. Soils. 43, 102-106. Doi: 10.1007/s00374-005-0065-2.

Oelbermann M., P. Voroney, and A.M. Gordon. 2004. Carbon sequestration in tropical and temperate agroforestry systems: a review with examples from Costa Rica and southern Canada. Agr. Ecosyst. Environ. 104, 359-377. Doi: 10.1016/j.agee.2004.04.001.

Oelbermann M ., R.P. Voroney, A.M. Schlönvoigt, and D.C.L. Kass. 2004b. Decomposition of Erythrina poeppigiana leaves in 3-, 9-, and 18-year-old alleycropping systems in Costa Rica. Agroforest. Syst. 63(1), 27-32. Doi: 10.1023/B:AGFO.0000049430.52250.87.

Palm, C.A. 1995. Contribution of agroforestry trees to nutrient requirements of intercropped plants. Agroforest. Syst. 30 (1-2), 105-124. Doi: 10.1007/BF00708916.

Schroth, G., J. Lehmann, M.R. Rodrigues, E. Barros, and J.L. Macedo. 2001. Plant-soil interactions in multistrata agroforestry in the humid tropics. Agroforest. Syst. 53(2), 85-102. Doi: 10.1023/A:1013360000633.

Somarriba, E. 2006. ¿Cómo analizar y mejorar el dosel de sombra en cacaotales y cafetales? pp. 159-176. In: Gama-Rodrigues A.C., N.F. Barros, and E.F. Gama-Rodrigues (eds.) Sistemas agroflorestais: bases cientificas para o desenvolvimento sustentável. Universidad Estatal del Norte Fluminense, Campos dos Goytacazes, RJ, Brazil.

Unigarro, A., R.L. Insuasty, and G. Chaves. 2009. Manual de prácticas de laboratorio: Suelos generales. Universidad de Nariño, Pasto, Colombia.

Zaharah, A.R. and A.R. Bah. 1999. Patterns of decomposition and nutrient release by fresh Gliricidia (Gliricidia sepium) leaves in an ultisol. Nutr. Cycl. Agroecosyst. 55, 269-277. Doi: 10.1023/A:1009803410654.