Abstracts

Litterfall and litter nutrient content in two Brazilian Tropical Forests

REGINA MARIA DE MORAES¹1. Seção de Ecologia, Instituto de Botânica, Caixa Postal 4005, 01061-970 São Paulo, SP, Brazil. . Seção de Ecologia, Instituto de Botânica, Caixa Postal 4005, 01061-970 São Paulo, SP, Brazil. , WELINGTON BRAZ CARVALHO DELITTI²1. Seção de Ecologia, Instituto de Botânica, Caixa Postal 4005, 01061-970 São Paulo, SP, Brazil. . Seção de Ecologia, Instituto de Botânica, Caixa Postal 4005, 01061-970 São Paulo, SP, Brazil. and YARA STRUFFALDI-DE VUONO¹1. Seção de Ecologia, Instituto de Botânica, Caixa Postal 4005, 01061-970 São Paulo, SP, Brazil. . Seção de Ecologia, Instituto de Botânica, Caixa Postal 4005, 01061-970 São Paulo, SP, Brazil.

(recebido em 03/02/97; aceito em 17/09/98)

ABSTRACT - (Litterfall and litter nutrient content in two brazilian tropical forests). Litterfall and transfer of nutrients was estimated in two tropical coastal forests of Brazil - the Atlantic and the Restinga Forests at Cardoso Island, São Paulo. Samples were collected monthly, from June 1990 to May 1991, using thirty 0.25 m² traps. There were significant differences in litter production between the Atlantic Forest (6.3 t.ha^-1.year^-1)and the Restinga Forest (3.9 t.ha^-1.year^-1). Litterfall was continuous throughout the year with maximum in the beginning of the rainy season in both sites. The annual return of mineral elements through litter in the Atlantic Forest was (kg.ha^-1): 101.8 N, 3.8 P, 20.3 K, 60.0 Ca, 18.0 Mg, and 14.6 S and in the Restinga Forest was: 27.5 N, 1.0 P, 6.5 K, 30.0 Ca, 10.9 Mg, and 6.6 S. The return, although small, is relevant due to the low fertility of the soils in those ecosystems, especially in Restinga. The Restinga Forest seems to be an ecosystem well adapted to oligotrophic conditions, lying among those presenting higher nutrient use efficiency.

RESUMO - (Produção e conteúdo de nutrientes na serapilheira em duas florestas tropicais brasileiras). A produção de serapilheira e o seu conteúdo de nutrientes foram estimados em duas florestas tropicais costeiras do Brasil - Floresta Atlântica e Floresta de Restinga, localizadas na Ilha do Cardoso, São Paulo. As amostras foram coletadas mensalmente de junho de 1990 a maio de 1991, utilizando-se trinta coletores de 0,25 m². A produção anual de serapilheira diferiu significativamente entre a Floresta Atlântica (6,3 t.ha^-1) e a Floresta de Restinga (3,9 t.ha^-1). O processo foi contínuo ao longo do ano com a queda máxima de serapilheira ocorrendo no início da estação chuvosa nas duas áreas. O retorno anual de elementos minerais na Floresta Atlântica foi (kg.ha^-1): 101,8 N; 3,8 P; 20,3 K; 60,0 Ca; 18,0 Mg e 14,6 S e na Floresta de Restinga foi: 27,5 N; 1,0 P; 6,5 K; 30,0 Ca; 10,9 Mg e 6,6 S. Embora pequeno, este retorno é relevante devido à baixa fertilidade dos solos destes ecossistemas, especialmente da restinga. A Floresta de Restinga mostrou-se um ecossistema bem adaptado às condições de oligotrofismo, estando entre os que apresentam maior eficiência na utilização dos nutrientes.

Key words - Litterfall, litter nutrient content, nutrient cycling, tropical rain forest, restinga forest

Introduction

Litterfall is a fundamental process in nutrient cycling, and is the main means of transfer of organic matter and mineral elements from vegetation to the soil surface (Vitousek & Sanford 1986). It is particularly important in tropical rain forest ecosystems, where the trophic chain of detritus predominates (Odum 1969).

Studies on litterfall have focused on obtaining measures of the production, information on decomposition (when associated with estimates of litter standing-crop), data on phenology and nutrient flux (Proctor 1983), as well as indications of nutrient use efficiency (Vitousek 1982) and the stability and capacity of the ecosystem to recover from disturbance (Meguro 1987). Some models have been proposed to predict litter production (Bray & Gorham 1964, Meentemayer et al.1982, Vogt et al. 1986, Londsdale 1988 and Silver 1994).

Quantification of the nutrient flux associated with litterfall is important to the understanding of ecosystem dynamics. The maintenance of natural systems depends on adequate mineral cycling. Nevertheless investigations of this kind are scarce for brazilian coastal ecosystems, including the Atlantic and Restinga Forests. Oliveira & Lacerda (1993) estimated litterfall and litter mineral content in Rio de Janeiro's Tijuca Forest. Similar investigations were conducted in polluted areas of Brazil's Coastal Mountain Range, the Serra do Mar, in the State of São Paulo (Domingos et al. 1990, Leitão Filho et al. 1993). However, there has been only one investigation of a Restinga Forest (Britez 1994).

The present study compares two contrasting tropical forest ecosystems subjected to the same climatic conditions: a tropical rain forest of the Serra do Mar mountains generally referred to as Atlantic Forest, and a coastal plain forest called Restinga Forest, in order to determine how differences in substrate, flora and vegetation could be reflected in different patterns of litterfall and nutrient transfer.

Material and methods

Study area - Study sites were established on Cardoso Island on the south coast of the State of São Paulo, Brazil. Cardoso Island, which is separated from the continent by only a narrow channel, lies between 25°03'-25^o18'S and 47°53'-48°05'W. Mean annual precipitation is 2200 mm, and there is no dry season. The highest rainfall is registered in March (330 mm) and the lowest in August (80 mm). Mean annual temperature is 21.3°C, with the highest monthly mean in February (25.1°C) and the lowest in July (17.8°C). Climatic data were obtained from a meteorological station located 5 km from the study sites, belonging to the Oceanographic Institute of the University of São Paulo. The centre of the island is occupied by a large massif of crystalline pre-Cambrian rocks, reaching 800 m height. Surrounding the massif are the alluvial coastal lowlands typical of the Brazilian meridional coast formed by Holocene and Pleistocene sands (Suguio 1994). These sand deposits are occupied by characteristic animal and plant communities and is referred to as Restinga. The two areas studied, 1 ha each, are located on the NE portion of the island. The Atlantic Forest site (AF) is at 140 m a.s.l. and the Restinga Forest site (RF) is on the coastal lowland at sea level. The vegetation of AF, surveyed by Melo & Mantovani (1994), has 2510 tree individuals per ha (DBH > 8 cm) of 157 species and total basal area of 48 m².ha^-1. The richest families are Myrtaceae, Leguminosae and Rubiaceae. Arecaceae presented the highest IVI (Importance Value Index) value due to the predominance of Euterpe edulis Mart., the most important species in that plant formation. Tree strata cannot be clearly separated, but canopy is more dense at about 10 m height with emergents up to 35 m tall.

In the RF the trees are smaller with irregular crowns forming a single stratum, with canopy around 4-7 m. According to Sugyiama (1993), 867 tree individuals per ha (DBH > 5 cm) were registered, including 31 species and total basal area of 23 m².ha^-1. The richest family is Myrtaceae but higher IVI values are presented by Theaceae, Lauraceae and Guttiferae. The two areas have only six tree species in common. Sorensen's Index of Similarity for the two forest types is therefore very small (6%) showing no floristic similarity between these forests.

Soils of both forests differ sharply in their physical and chemical characteristics (table 1). In the AF the soil is oligotrophic when compared to other tropical forests, but it offers to the vegetation better growing conditions than the restinga.

Methods - Thirty 0.25 m² wooden traps with 1 mm² nylon mesh, were placed in each forest to collect litterfall. They were 20 cm above the ground level and positioned on either side of a 1000 m long transect, at randomly distances. Material was collected monthly between June 1990 and May 1991 and sorted into four categories: leaves (and leaflets and petioles); twigs ( 2 cm diameter following Proctor (1983), and bark fragments); reproductive parts (flowers, fruits and seeds) and miscellaneous material (indeterminate animal and plant material). Sorted material was ovendried at 70°C to a constant weight, grounded and homogenized. The resulting powder was digested with nitric and perchloric acid for determination of P, K, Ca, Mg, S, Fe, Mn, Cu, Zn, B, Al and Na concentrations. Digestion with sulphuric acid and hydrogen peroxide was used to determine N concentrations. Leaf material from the monthly collection of each area was mixed and homogenized from which two replicate determinations were made. The other fractions were each bulked, preserving the monthly percent composition, to constitute a single annual sample. Chemical analyses followed Zagatto et al. (1981).

The t test was used to compare litterfall in both areas. F test was used to analyse monthly litterfall in each area and multiple comparisons Tukey test to determine the production peaks.

Results

Litterfall in AF was higher than at RF during the entire study (t test, p < 0.01), measuring 6.3 and 3.9 t.ha^-1.year^-1 respectively (figure 1, table 2). There was a clear temporal variation in litterfall, with a 3.8 fold difference between maximum and minimum values in AF, and 5.7 fold in RF. This monthly variation were confirmed by F test (p < 0.01 in both areas).

Figure 1. Monthly litterfall (kg.ha^-1, n = 30), and rainfall (mm) in the study areas, Cardoso Island, São Paulo, Brazil. A - Rainfall, B - Litterfall in the Restinga Forest, C - Litterfall in the Atlantic Forest. Total (-_* -); leaves (-¨-); twigs (-n-); reprod. parts (--); miscellaneous (-´-) Significant differences (Tukey test) are indicated by different small letters.

The highest values of litter production were observed at the beginning of the rainy period (Tukey test, figure 1). In AF this occurred in November (978 kg.ha^-1) with 42.7% of total annual production from October to December. Leaf production was higher in October and November (718 kg.ha^-1); the twig fraction reached a peak in November and the other fractions in December.

In RF there was a clear peak of litterfall in December (773 kg.ha^-1, Tukey test, figure 1), corresponding to 20% of total annual production. This peak was determined by leaf production. Twig fall was also more intense in December. Reproductive parts and miscellaneous fractions had higher values in Jan uary.

Table 2 also shows the relative contribution of the various fractions to the litter. Leaves are evidently more important because represent 70-75% of the total litter.

There was little variation in the monthly concentrations of macronutrients in leaf litterfall in both areas. There was no clear evidence of seasonality, although K and S concentrations were lower during the months of higher litter production in AF. N and Ca showed the highest concentrations, and P the lowest. The sequence of concentrations obtained in AF leaf litter was N>Ca>Mg>K>S>P, similar to RF except that in RF litter concentrations of Ca were higher than those of N.

Mean annual concentrations of all macronutrients were lower in RF than in AF litter, except for Mg which in leaves was similar in both ecosystems, and for S in reproductive parts which was lower in AF (table 3). In both forests the miscellaneous material was nutrient rich, mainly in N and P. In AF the highest concentration of K was in the reproductive parts, Ca was highest in twigs, Mg and S were highest in leaves and miscellaneous litter fractions. In RF, Ca and Mg presented the highest concentrations in leaves and K and S in reproductive parts.

Concentrations of trace elements in AF litter were higher, except for Na. The mean annual concentration of Na in the AF litter was higher in leaves. Zn concentration was higher in twigs and Cu, Mn, Al and B concentrations showed higher values in the miscellaneous fraction. Concentrations of Mn and Zn in RF litter were higher in reproductive parts, of Na in leaves and of Fe, Cu, B and Al, in miscellaneous fraction (table 4).

Table 5 shows the return of mineral elements from phytomass to soil through litterfall. The order of the relative annual contribution of the various elements to AF litter was N>Ca>K>Mg>S>P and to RF litter was Ca>N>Mg>S>K>P. Concentrations of Ca were higher than N at RF, a fact that may be related to scleromorphism at the vegetation at RF which in turn is probably related to its oligotrophic soils. The sequence of the relative contribution of Na, Al and micronutrients in AF litter was Na>Al>Mn>Fe>B>Zn>Cu, similar to that of RF litter with higher concentrations of Mn than of Al.

Discussion

The values of litterfall reported here are low when compared to values reported in the literature for tropical rain forests (Proctor 1984). Atlantic Forest ecosystems investigated by Varjabedian & Pagano (1988), Oliveira & Lacerda (1993) and Lopes et al. (1994) at latitudes similar to that of Cardoso Island showed 7.9, 8.9, and 8.3 t.ha^-1.year^-1 of litterfall, respectively (table 6).

Litterfall at AF was lower than that recorded for semi-deciduous forests of south-east Brazil. Although located in the same latitudinal zone as Cardoso Island, these forests lie on more fertile soils (Pagano 1989a) subjected to a typical wet-dry tropical climate which results in the co-existence of deciduous and semi-deciduous species. These ecosystems present different functional patterns as shown by the amount and by the seasonality of litterfall during the dry season (table 6).

Bray & Gorham (1964) found a linear inverse relationship between latitude and litterfall, according to which the expected value for AF would be around 7.5-8.0 t.ha^-1.year^-1. Although low, the results obtained here are still within the range of those reported by those authors. Meentmeyer et al. (1982) propose that litterfall could be determined by a complex of environmental variables such as potential and actual evapotranspiration and latitude. According to this model, the amount of litter to be produced in south-east Brazil would be around 6.0 t.ha^-1.year^-1, a value very similar to that obtained here. Lonsdale (1988) presents a multiple regression model with latitude, altitude and precipitation as predictors (log x = - 0.0012L - 0.000072A + 1.04), which estimates AF litterfall to be 5.4 t.ha^-1.year^-1. However, the model for leaf litterfall (log y = - 0.0090L + 0.80) predicts a full rate of 3.8 t.ha^-1.year^-1 as compared to 3.9 t.ha^-1.year^-1 in this study. This could be related to the high precision of the method in quantifying leaves, although its precision is not very good for reproductive parts and twigs. Another model proposed by Vogt et al. (1986) using latitude (y = 9457 - 120x), shown to be more accurate, resulted in an estimate of 6.5 t.ha^-1.year^-1 for AF litterfall. Local factors, such as nutrient availability (Silver 1994), may explain the results obtained for the Restinga forests. Some of those ecosystems present very low litter production, not fitting well to the values predicted by these models. The present study confirms the observations of Silver (1994) about the availability of nutrients, especially P, as a determinant of litter production.

Peak production occurs at the beginning of rainy season when leaf fall is greater as predicted by the production curve. This pattern differs from most investigations conducted in tropical ecosystems, where the highest deposition of litter occurred in the dry season (Klinge & Rodrigues 1968a, Klinge 1977, Silva 1984, Morellato 1987, Dantas & Phillipson 1989, Scott et al. 1992, César 1993a, Ramos & Pellens 1994). In regions where the dry season is not distinct as in the Atlantic Forest, high deposition is observed in the rainy period (Jackson 1978, Domingos et al. 1990, Leitão Filho et al. 1993, Britez 1994). According to Jackson (1978) leaf fall peaks during the rainy season occur in regions where water stress is moderate and is simultaneous with the production of new leaves. This appears to be the strategy at Cardoso Island where there is no water deficit during the winter and the leaves may be renewed in the summer, when the environmental conditions are more favourable. The fact of the peak of RF leaf litter production take place two months after that of AF may be due to greater water holding capacity in AF soil, which could permit leaf renewal early at the first rains (150 mm).

The relative contribution of the litter components is similar to data reported in the literature for both ecosystems. The contribution of reproductive parts in AF litter is among the highest reported in the literature (table 6) suggesting a vegetation which is vigorous enough to allocate a great part of its energy for reproduction while at the same time requiring frequent replacement.

Mean concentrations of macronutrients in leaf litter at AF are in the range of other rain forests (Proctor 1984). Values at RF were lower, mainly those of N and P, making evident that litter quality is unique to each ecosystem, according to soil properties.

High concentrations of N and P in the miscellaneous material may be explained by the presence of residual material from the other fractions and from animal vestiges. Reproductive parts contained high concentrations of K, probably due to the fact that this element is stored primarily in young and metabolically active tissues while deciduous leaves lose this element by translocation and before and after decay by leaching. Ca was concentrated in leaves and twigs, Mg in leaves and S in leaves and reproductive parts.

The magnitude of mineral flux unequivocally distinguishes the two forests. RF produces lower amounts of low quality litter leading to a smaller return of mineral elements to soil. This becomes evident by comparing the ratio between litterfall at AF and that at RF (1.6) to the ratio between the return of each element in these ecosystems (N: 3.6; P: 4.0; K: 3.0; Ca: 2.0 and Mg: 1.7).

Leaf fall determined the seasonality of mineral return to soil in both areas. A seasonal pattern was therefore derived from curve of litter production and not from concentration values which where quite uniform during the entire period of study.

Table 6 shows annual macronutrients transference obtained in the present study compared with data from other tropical ecosystems. The great variability of results are due to differences in the amount and quality of litter, according to edaphic and climatic conditions found in each ecosystem. Annual amounts of N, P, K, Ca and Mg in AF litterfall are lower than those reported for other tropical ecosystems, and those of RF are lower but similar to tropical savannas.

RF has a reduced stock of available nutrients in the soil and biomass and receives inputs from adjacent systems (Hay & Lacerda 1984). AF, on the other hand, receives proportionally lower additional inputs and its nutrient pools are higher, either in the soil or in biomass. These nutrients are continuously and efficiently recycled, probably due to the better soil condition. The main function of litter at AF is the transference of mineral nutrients from the vegetation to the soil, while at RF litter seems to have also a very important role in improving edaphic conditions, increasing the water and ion retention capacity. Its soils are almost free from clay which makes organic matter from litter the main colloidal source.

Micronutrients are not often quantified in litterfall studies, what restricts the possibilities for comparisons. Micronutrient concentrations were constant across the leaf fall cycle, but the values for Na were greater during the winter. This is probably due to greater accumulation of marine aerosols on the forest canopy, as rainfall is lower at that season. The annual return of micronutrients to the soil was higher in the AF site than in RF site, both within ranges reported by Klinge & Rodrigues (1968a, b), Pagano (1989b), Domingos et al. (1990) and César (1993b) for Brazilian ecosystems.

The ratio between litterfall mass and litterfall nutrient content gives the nutrient use efficiency (NUE) (Vitousek 1982). The mesophyll semi-deciduous forests at São Paulo are inefficient in the use of nitrogen although great amounts of litter are produced (NUE = 51 - Meguro et al. 1979; NUE = 44 - Pagano 1989a, b; NUE = 45 - César 1993a, b). The Atlantic Forest at Cardoso Island presented a NUE of 62, which is equivalent to that reported by Klinge & Rodrigues (1968a, b) for an Amazonian terra-firme forest. Restinga Forest, on the other hand, with a NUE of 143, is very efficient in the use of N, similar to coniferous forests, showing a high level of adaptation to the low nutrient availability. It is therefore an ecosystem with a highly specialized vegetation, able to develop under conditions that would be adverse to the great majority of species from other plant communities.

Acknowledgements - To Kevin Creagan, of Louisiana State University, for the translation, and to Dr. Marcos S. Buckeridge for some suggestions.

LOPES, M.I.M.S., TEIXEIRA, C.B., COMPTE, V.X., LIESS, S. & MAYER, R. 1994. Litter production in the Atlantic forest vegetation of Serra do Mar, Cubatão region, Brazil. Anais do III Simpósio de Ecossistemas da Costa Brasileira, Serra Negra, Publicação ACIESP, n. 87, p.87-94.

2. Departamento de Ecologia Geral, Instituto de Biociências, USP, Caixa Postal 11461, 05422-970 São Paulo, SP, Brazil.

Publication Dates

History