Litterfall Deposition and Decomposition in an Atlantic Forest in Southern Goiás

Brazilian biomes have been severely threatened and their conservation depends on the knowledge of intrinsic ecological processes and correct phytophysiognomy identification of legal protection purposes. Litterfall deposition and decomposition patterns associated with taxonomic composition, climate, soil and relief can be considered important tools for vegetation typology. This study contributes to classifying the Natural Park of Morrinhos (Parque Natural de Morrinhos PNM), a forest fragment in the southern region of Goiás, using litterfall deposition and decomposition rates. The results show intense litterfall deposition at the end of the dry period and rapid decomposition in the rainy season (62%), requiring 494 days for total decomposition. The rapid litterfall decomposition rate (k = 1.42 year-1), cumulative annual deposition of 6.43 Mg ha-1 and the taxonomic composition of the vegetation contribute to the classification of PNM as a Semi-deciduous Seasonal Forest.


INTRODUCTION
The great extent and centralized geographic positioning of the Cerrado biome in Brazil resulted in transitional environments of this domain with Amazonian forests to the north and the Atlantic Forest to the south (Méio et al., 2003).However, Cerrado vegetation has been rapidly removed (Sano et al., 2010) to meet the expansion of the agricultural frontier within the country, as initiated by President Vargas government in the 1930s known as 'March to the West' (Marcha para o Oeste) (Pádua, 2007).In the southern region of Goiás and concomitant with Cerrado deforestation, there are only 2.6% of the 1,190,184 ha of Atlantic Forest originally dispersed throughout 26 municipalities, including the municipality of Morrinhos (Fundação SOS Mata Atlântica & INPE, 2013, 2015).Scarano (2014) addresses the scientific, environmental, economic and social relevance of Cerrado and discusses conservation strategies for this biome.
To do so, it is necessary to classify the vegetation to be conserved and/or restored as protected areas, which is carried out through technical and scientific studies.The predominant tropical forest formations in Brazil are classified according to physiognomic and ecological criteria based on four basic interrelated factors: rainfall, altitude, temperature and soil (IBGE, 2012).For example, the duration of the dry period classifies forests into Ombrophilous and Seasonal (Projeto RADAMBRASIL, 1983).Thus, two parameters directly related to climate are considered as classification criteria of the different forest formations: i) production of dead organic matter (litterfall), mainly represented by leaves falling off trees/plants; and ii) speed with which nutrients that compose litterfall become available to plants through the mineralization process that occurs during litterfall decomposition (Olson, 1963;Bray & Gorham, 1964).
Decomposers play a key ecological role within ecosystems, recycling considerable part of the energy and nutrients present in dead plant matter, which was not used by herbivores, and soil fauna and climate are important factors regulating this microbial activity (Lavelle et al., 1993;Pereira et al., 2013).Wieder & Wright (1995) found that the litterfall decomposition rate was accelerated in a Tropical Forest area in "Barro Colorado" through irrigation during the dry period (four months) over five years; however, during the autumn period, the number of fallen leaves was not affected by irrigation, evidencing the evolutionary character of dehiscence to climatic variations.Thus, litterfall production and decomposition can be considered important tools for the description of phytophysiognomies.The aim of this study was to quantify litterfall production and decomposition rates in the Natural Park of Morrinhos (Parque Natural de Morrinhos -PNM), one of the few Atlantic Forest remnants in the state of Goiás, and the only conservation unit recognized in the municipality.

Study site
PNM is located at north of the urban area of Morrinhos in Goiás (17º43'35"S and 49º07'52"W), 790 m above sea level and corresponds to an area of approximately 104 ha.It was created with the main function of protecting the water spring of the "Maria Lucinda" stream.The natural park is composed of perennial and deciduous plant species, with canopy measuring up to 30 m in height.Popular and scientific names of tree species are provided by the PNM administration through signage set up along visitation trails.According to a systematized floristic survey conducted in 2013, 51 species distributed into 40 genera and 26 families were identified, with Siparuna guianensis Aublet, Styrax camporum Pohl, Piper aduncun L., Ocotea aciphylla (Nees & Mart) Mez, Tapirira obtusa (Benth.)D.J.Mitch., Miconia nervosa (Sm.)Triana, Inga marginata Willd, Matayba guianensis Aubl., Casearia sylvestres Swand Copafeira langsdorffi Desf (Junqueira MHM, unpublished data) being the most abundant.The climate of the region is Aw according to the Köppen classification, being seasonal and characterized by cold period from May to August (20.85± 1.10°C), and warm period from September to April (24.33 ± 0.81°C).Rains are intense from November to March (241.07 ± 32.50 mm), reduced from May to August (9.57± 1.44 mm), and intermediate in April, September and October (77.34 ± 13.46 mm) (Pesquero et al., 2012).

Data collection
Total litterfall production (leaves, flowers, fruits and small branches) was monthly estimated from October 2011 to September 2012 using 30 plastic collectors (1 m 2 ) adapted with a drain to avoid rainwater accumulation, elevated 10 cm from the ground and arranged in-line within PNM, with a distance of at least 50 m from each other.All plant material monthly intercepted by collectors was packed in 100-liter plastic bags and kept under refrigeration (10-15°C) until drying in oven at 70°C for 48 h.After drying, leaves were weighed using digital electronic scale (0.01 g accuracy).The rest of the material (small branches, flowers and fruits) was weighed together to verify the proportion of leaves in relation to the total deposited material.
Leaf litterfall decomposition was monthly estimated for one year (October 2011 to September 2012) through the use of 360 nylon canvas bags (30 x 25 cm, 2 mm mesh) numbered according to leaf dry biomass, which varied from 25 to 47 g (26.73 ± 3.46 g), distributed in lots of 12 units located close to the 30 litterfall collectors.Dried leaves placed inside the nylon bags were randomly collected from the upper layer of litterfall collectors arranged above the ground in the PNM in September 2011, and all material was dried in oven at 70ºC and weighed on a digital electronic scale until reaching constant mass (0.01 g accuracy).

Data analysis
The mean monthly litterfall deposition was compared according to the annual deposition seasonality using Student t-test for independent sampling.The annual deposition was compared to results obtained in SSF, SDF and OF (Semi-deciduous Seasonal Forest; Seasonal Deciduous Forest; and Ombrophilous Forest, respectively) studies published in scientific journals in the last 10 years (Scielo and Google Scholar databases); this period was expanded in the case of low n sample by means of the Student t-test for one sample.This test compares the mean values of data obtained in literature with value obtained in the study area (PNM), and is performed by estimating the variance in PNM by the variance of data obtained in literature.This procedure is suitable for small-sized samples (n ≤ 30) (Ayres et al., 2007).
Litterfall decomposition rates (k) were monthly calculated according to Equations 1 and 2 (Olson, 1963): ( ) Where X t is the amount of material decomposed in time interval t, and X 0 is the initial amount of dead organic matter.
Climatic data for precipitation, temperature and air humidity were obtained from the Climatological Station of the State University of Goiás -Campus Morrinhos, in order to verify correlation patterns (Spearman's non-parametric R analysis) with biological deposition and decomposition variables.Simple linear regression analysis was used to estimate loss of litterfall leaf mass over time.Analyses were performed using the BioEstat 5 software (Ayres et al., 2007).The mean values of variables analyzed are accompanied by their respective standard deviations.Data were analyzed through non-parametric tests when they did not meet normality and homoscedasticity.

RESULTS AND DISCUSSION
Litterfall deposition in PNM occurred according to a seasonal pattern with the highest average fall in the period from July to October (peaking in September) compared to the average decomposition in the period from November to June (93.82± 44.13 g m -2 vs. 33.26± 16.71 g m -2 , t = 19.76,df= 358, p < 0.0001, Figure 1).This period of greatest leaf fall corresponds to the dry and cold season of the region (Figure 2) (Pesquero et al., 2012).In fact, dehiscence presented significant negative correlation values with variables temperature, precipitation and air humidity (Table 1).Several studies have demonstrated similar seasonal dehiscence pattern in seasonal forests (Table 2).Leaves accounted for most of the material collected throughout the year (66 ± 11%, n = 12), and within the range expected (60-80%) for Atlantic Forest biome vegetation types (Vieira et al., 2016).
The total litterfall deposited throughout the year in PNM (6.43 Mg ha -1 ) did not differ from the average values for seasonal forests reported in literature (SSF = 8.54 ± 2.03 Mg ha -1 , n = 6, t = 2.54, p = 0.052 and SDF = 8.03 ± 1.71 Mg ha -1 , n = 5, t = 2.09, p = 0.10, respectively, Table 2).However, litterfall deposition in PNM was lower than the average values observed in literature for Ombrophilous Forest (8.31 ± 1.83 Mg ha -1 , n = 11, t = 3.42, p = 0.006, Table 2).Precipitation is considered the main factor that can explain differences in litterfall production in tropical biomes, in which higher rates were found for humid environments (Tonin et al., 2017).In fact, elevated temperatures (average of 25°C) and higher precipitation (<60 days of drought) are climatic factors associated with Dense Ombrophilous Forests (IBGE, 2012).Oliveira & Fontes (2000) point out several factors associated with the floristic composition of the Atlantic Forest such as rainfall, temperature, latitude, altitude, and continentality, and also emphasize its extent in its definition as a biome.Another factor related to the vegetation type of a region is the presence of limestone lithology, which is considered an important feature in identifying SDF (Projeto RADAMBRASIL, 1983).Despite the lack of systematized studies, the PNM soil does not present thin cover or low representativeness of surface drainage, which are characteristics present in calcareous terrain.Although deciduous species such as Copafeira langsdorffii, Hymenaea courbaril L. and Astronium fraxinifolium Schott are found in PNM, together they represent only 6% of the relative density (Junqueira MHM, unpublished data).
The warm and rainy period of the region from October to March (Pesquero et al., 2012) favored the action of decomposing organisms, which consumed 62% of leaves inside the nylon bags, while only 12% were decomposed in the remaining dry months (Figure 3).Thus, 74% of the leaf biomass was decomposed after the period of one year, in which regression analysis predicted approximately 494 days for its total renewal [leaf biomass (g) = 19.271-(0.039 * number of days), adjusted R 2 = 0.89, F(1,10) = 91.031,p < 0.0001, Table 3, Figure 4].Litterfall renewal time in PNM is comprised within values observed by Vital et al. (2004) and Pinto et al. (2009) for SSF; however, it is much shorter than the estimated time of up to 983 days for the disappearance of 95% litterfall in SDF (Cunha et al., 1993, Turchetto & Fortes, 2014).
Results that reinforce the similarity between PNM and SSF were obtained through the litterfall decomposition rate (k), which increased gradually over time until reaching the value of 1.42 in August (Figure 5).Values of k > 1 show rapid cycling of litterfall nutrients, typical of tropical climate biomes (Olson, 1963).The k value of PNM did not differ from the mean k values observed for SSF reported in literature (1.48 ± 0.51, t = 0.29, p = 0.78, n = 7, Table 4).Although the few studies available in literature make comparison impossible, our results suggest higher decomposition rate in SDF and lower in OF (Table 4).

CONCLUSIONS
The dynamics of litterfall deposition and decomposition in PNM shows rapid incorporation of nutrients into the soil, typical of Tropical Forest.The values of these parameters associated with the composition of perennial dominant plant species identify this vegetation fragment as a Semi-deciduous Seasonal Forest.However, further studies should be carried out in the Atlantic Forest (mainly deciduous seasonal rainforest and ombrophilous forest), considering decomposition factor k as an important integrating component for nutrient cycling dynamics and descriptor of the forest typology.

Figure 2 .
Figure 2. Annual variation of climatic parameters and litterfall deposition at the Natural Park of Morrinhos (PNM), Morrinhos (GO).

Figure 4 .
Figure 4. Decomposition rate (litterfall collectors) as a function of the time elapsed in days in the Natural Park of Morrinhos (PNM), Morrinhos (GO).

Table 1 .
Relationship between litterfall deposition and climatic variables at the Natural Park of Morrinhos (PNM), Morrinhos (GO).

Table 2 .
Annual litterfall deposition rates in seasonal and humid forests of the Atlantic Forest biome.

Table 3 .
Regression parameters between leaf biomass losses and the elapsed time at the Natural Park of Morrinhos (PNM), Morrinhos (GO).