The study was carried out from April to July of 2009 in the Brega and Santuário limestone caves (Brega - 419447.15E, 7742205.17S, elevation 725 m and Santuário 419312.50E, 7742176.54S, elevation 722 m), located in the municipal district of Pains, Minas Gerais, Brazil. The caves have permanent groundwater and only Brega cave receives also water from rain floods.

The Brega Cave had predominantly vadose genesis, presenting a horizontal development of approximately 760 meters and two opposite entrances. Santuário Cave had also predominantly vadose genesis and presents a horizontal development of approximately 700 meters and two entrances. At the end of the main conduits of the two caves are small perennial water holes feed by upwelling groundwater.

According to the classification of Köppen (Menegasse et al. 2002), the climate of the external area is the Cwa type: mild climate with hot, humid summer and dry winter. The average annual temperature is 20.7°C, July being the coldest month, with an average temperature of 16.3°C, and January the hottest month, with an average of 23.3°C. The local annual average precipitation is 1344 mm. The Brega and Santuário caves belong to the hydrographic sub-basin known as ribeirão dos Patos (530 km²), favored by the direct infiltration in the limestone fissures and the karstic absorption features such as dolines and sinkholes.

The local tropical climate is responsible for the seasonal variation of the hydric dynamics of the ribeirão dos Patos, significantly altering the surrounding landscapes. The wet season lasts from October to March, which according to Menegasse et al. (2002) corresponds to 81% of the annual precipitation. The great hydric volume of the Patos stream makes the increase of the water level present inside the caves possible, leaving a large part of their floors flooded.

Rates of litter breakdown in lentic environments were measured using bags containing 50 intact plant leaf disks (area = 63.6 mm²/each disk, ± 2 grams). The bags with 0.04 mm² and ⁹ mm² meshes, measuring 10 × 10 cm² on the sides, enable the microinvertebrates colonization (Bärlocher 2005; Venarsky et al. 2012). 42 bags were used in each site, 21 with the smaller mesh size and 21 with the larger mesh size, totaling 84 litter bags in the two caves.

The leaves were collected from trees (Moraceae: Ficus sp.) located in the surrounding area before the fall of leaves, thereby reducing the possibility that the experimental material had been previously exposed to decomposition after falling on the ground.

Triplicates of each mesh size were removed (after intervals of 14, 35, 50, 64, 78, 85 and 99 days) to quantify the decomposition through the weight loss and to determine the composition of the associated invertebrate fauna. Before the incubation of the litter bags in the cave water, the initial weight and the initial nitrate and phosphate levels present in the foliar portions were measured. After each collection, the material was washed with distilled water (Bärlocher 2005), the wash water and the removed disks were conditioned separately in Petri dishes and observed with the aid of a magnifying glass, to verify the presence of associated invertebrates.

The remaining plant material and the wash water were oven-dried at 100° C for 48 hours and later weighed. After the material was dried and weighed, it was sent to the Plant Nutrition Laboratory of the Federal University of Paraná, Brazil, where the analysis of remaining nitrogen and phosphorous in the fractions under decomposition was conducted. Total leaf nitrogen and phosphorus were also measured, using the microKjeldahl method on another portion of the leaf (Varley 1966).

To evaluate the influence of the moisture of the green leaves on the weight loss measurements, 50 green leaves were weighed (2.03 gr), oven-dried and reweighed (0.48 gr). Thus, the 76.3% plant disk weight loss is due to the moisture loss and not the mass loss. Water samples were collected from each cave on the sampling start date and analyzed for turbidity, dissolved oxygen, phosphorus and nitrogen.

The amount of organic matter remaining after exposure was expressed in percentage (%) of remaining dry weight [ (FW × 100) /IW]. Where IW is the initial weight of the sample at time zero, FW is the final weight of the sample at time t+1. The decay processing rate (k^-day) in each cave was described by the model Mt = M₀ e^-Kt (Oslon 1963; Bärlocher 2005), Where Mt is the weight at time t, M₀ is the initial weight; t is the exposure time of the sample.

To determine the processing speed (k/day) of plant debris we compared our results to data from the literature for aquatic environments and defined slow (≤ 0.005), moderate (0.006–0.10) and fast (0.10–0.15) processing (Table 1). To evaluate differences in the average final weight values among sites and among the dates of collection and in the same different mesh types, the t Test was used (Zar 1984).

The plant disks exposed to decomposition presented slow processing speed, but varied with time and between the two exposure caves (Table 1, Figs 1 and 2). However, significant differences were observed in the average final weight between both the two caves and the two mesh sizes (Table 2).

In Brega Cave, during the first 14 days of exposure of the 0.04 mm² mesh an average remaining weight of 86% was observed. After 85 days of exposure the plant debris had an average remaining weight of 82% in relation to the initial mass (Fig. 1). In the same cave, during the first 14 days of the 9 mm² mesh plant disk exposures, an average remaining weight of 88% was observed. After 85 days of exposure of the plant disks, the average remaining weight corresponded to 88% (Fig. 2).

In Santuário Cave, during the 14 days of exposure in the 0.04 mm² mesh, an average remaining weight of 83% was observed. After 85 days of exposure of the plant disks, the average remaining weight was 84% (Fig. 1).

In Santuário Cave from the 14^th to the 35^th days of exposure of the 9 mm² mesh plant disks, an average remaining weight of 80% was observed. From the 78^th to 85^th days of exposure the average remaining weight was 78%. At the end of the 99 days of exposure, the average remaining weight corresponded to 82% (Fig. 2).

We can assume that for all of the mesh sizes a slow weight loss rate was registered. As such, we can also consider the low decay (K/day) values (average 0.003 K^-day) for the water of these two caves (Table 1).

There were no macroinvertebrates associated with the debris.

The water quality parameters evaluated for the two caves showed quite different values. In Brega Cave the values were as follows: turbidity (6.97 unt), dissolved oxygen (5 mg/l), phosphorus (1.2 mg/l) and nitrogen (0.8 mg/l). In Santuário Cave the values were as follows: turbidity (39.5 unt), dissolved oxygen (4.5 mg/l), phosphorous (1.8 mg/l) and nitrogen (1.4 mg/l).

The values for nitrogen and phosphorous in the plant disks in the two caves showed different values during the experiment. However, significant differences were not found in the average amount of nitrogen and phosphorous present in the debris of the litterbags of the two caves.

In Santuário Cave, the initial nitrogen value was 17.1 μg/kg. After 14 days of plant disk exposure (April 18) this value reduced to 10.9 μg/kg. After 99 days of exposure (July 12) the lowest value (7.2 μg/kg) of the remaining nitrogen was registered (Fig. 3). In Brega Cave, the initial nitrogen value was also of 17.1 μg/kg. With 14 days of exposure (April 18), the amount of nitrogen in the plant disks increased (28.4 μg/kg). On the 99^th day of exposure of the plant debris (July 12), the nitrogen value was of 26.9 μg/kg (Fig. 3).

In Santuário Cave, the initial value of phosphorous was 1.44 μg/kg. After 14 days of exposure of the plant debris (April 18) the lowest amount of this nutrient was registered, corresponding to 0.76 μg/kg. After 85 days (June 28) there was a small reduction in the amount of phosphorous in the plant debris (Fig. 3). In Brega Cave, the initial phosphorous value was also of 1.44 μg/kg. After 14 days of exposure (April 18) the lowest amount of this nutrient was registered, 0.92 μg/kg. On the 99^th of exposure (July 12), the amount of the remaining phosphorous in the plant debris showed a reduction to 1.20 μg/kg (Fig. 3).