These native species were planted in Quincunx (4 pioneers, with one climax in the center) at a spacing of 2.0 × 1.5 m, using seedlings produced from seeds collected in fragments of Atlantic Forest (Woodland). For Euc and Ap, the adopted spacing was 3 × 2 m.

To evaluate the soil CO₂ efflux we installed chloro-polyvinyl chambers (0.20 m x 0.30 m) in triplicate 42 months after planting the Eu, Ap and Nat, as well as in the Woodland and NCov. We installed chambers beneath and outside the litter collectors, allowing to evaluate the soil CO₂ efflux with and without the input of litter from the aerial part of Euc, Ap and Nat. We collected the CO₂ using syringes in the periods of 0, 10, 20 and 40 minutes after closing the chambers. We collected soil CO₂ every two months between September 2014 and July 2015 and we analysed it in a cavity resonance spectrometer (CRDS - cavity ring-down spectroscopy, G2131-i, Picarro, Sunnyvale, CA). During the soil CO₂ efflux measurements, we also assessed soil temperature with a digital thermometer inserted at a depth of 5 cm, and soil moisture by the gravimetric method. The CO₂ efflux was calculated using the equation:

where Flux: CO₂ (mg m^-2h^-1); ΔQ/ΔT: the angular coefficient of the adjusted straight line (ppm/s) obtained by readjustment of the gas concentrations during the time (T) of collection; M: the molar mass of CO₂ (g mol^-1); P: internal pressure of the chamber, assumed to be 1 atmosphere (atm); V: volume of the chamber (L); R: universal gas constant (0.08205 L atm K^-1mol^-1); T: soil temperature (K); A: base area of the chamber (m²).

Aiming to identify the contribution of autotrophic (AR) and heterotrophic respiration (HR) to the total soil respiration (TR), we used the partitioning method that was carried in areas with Euc, Ap and Nat. We also analysed the relative contribution of soil microbiota respiration under the influence of litter (HR_litter) and SOM (HR_SOM).

Where, TR_nolitter and TR_withlitter: total soil CO₂ efflux in the treatment with the absence and presence of litter, respectively; TR_Ncover: total soil respiration in the minded area without cover

Aiming to identify other variables that may influence the soil CO₂ efflux, we determined the main soil soil properties and roots density. Soil samples from the 0-10 layer of each plot, ground in a mortar and passed through a 60-mesh sieve (0.250 mm), were used to determine total organic carbon (TC) and total nitrogen (TN). Total organic carbon was quantified by wet oxidation (YEOMANS; BREMNER, 1988) and TN was determined by the Kjeldahl method, modified by Tedesco (1985). In addition, we analysed the soil bulk density (Ds), by the volumetric ring method; soil particle density (Dp) by the volumetric balloon method and total porosity (Pr). The microbial biomass carbon (CMB) and nitrogen (NMB) of the soil were determined in the 0-10 cm layer only. Samples (<2 mm) were placed in lidded plastic flasks, incubated for 10 days at 25°C, with moisture corresponding to 80% of the moisture equivalent, in order to re-establish the microbial community. After incubation, the CMB and NMB content were determined by irradiation-extraction (ISLAM; WEIL, 1998).

To deternine the roots density, we collected blocks of soil with dimensions of 20 x 20 x 20 cm in triplicate in all plots. After collection, the roots were manually separated from the soil, washed and divided into two classes, with a diameter smaller or larger than 2 mm. After determining the wet weight, the roots were placed in plastic pots with 25% alcohol and stored in a refrigerator for later evaluation. An Epson XL 10000 scanner, equipped with an additional light unit (TPU), together the WinRHIZO Pro 2009 software, was used to determine the following morphological characteristics: the biomass of roots smaller than 2 mm (BioRSm2), length of roots smaller than 2 mm (LengRSm2), specific root length (SRL), sectional area of roots smaller than 2 mm (SARSm2), mean root diameter and volume of roots smaller than 2 mm (VRSm2). Following these evaluations, the roots were placed in paper bags and dried in an oven at 60°C to obtain the dry weight. All results of soil properties and roots area presented in the Appendix 1.

The data for soil CO₂ efflux were submitted to analysis of variance (ANOVA) and the mean values compared by Tukey’s (10%). The influence of biotic and abiotic variables on the CO₂efflux was evaluated using multivariate regression analysis.

The soil presented the highest moisture content during November 2014, March and May 2015 for most of the types of cover, while the highest soil temperatures occurred in September 2014, and January and March 2015 (Fig. 1). Woodland showed the highest soil moisture and NCov the highest soil temperature for the studied period.

The five treatments showed significant differences in CO₂ efflux (p <0.1) (Fig. 2), with NCov presenting the lowest soil CO₂ values. Woodland had the highest values for CO₂ efflux during January and March 2015, while the soil CO₂ efflux in Euc, Ap and Nat did not vary during the study period. Euc and Nat had a similar soil CO₂ efflux pattern, with a tendency for higher CO₂ efflux during November, March and May. The increase of CO₂ efflux in Euc, Ap and Nat showed high percentage values compared to Woodland for all measurements, except in May 2015, when the soil CO₂ efflux presented the lowest percentages (< 33%) (Fig. 2a). In November, the three forest covers showed the same soil CO₂ efflux compared to the Woodland.

The highest CO₂ effluxes were recorded in the months of higher soil moisture (Fig. 3), i.e. November 2014, March and May 2015, with Woodland presenting the highest and NCov the lowest values. Although Euc showed a tendency with higher soil CO₂efflux during the months of higher soil moisture when compared to Ap and Nat, there were no significant differences between the three types of cover (p>0,1). In the driest months (September 2014, January and July 2015) Woodland, Euc, Ap and Nat showed no differences for CO₂ efflux, differing only from NCov.

All covers types showed similar soil moisture in the drier months (September 2014, January and July 2015), except Woodland, which had higher soil moisture (Fig. 4). In the rainy months (November 2014, March and May 2015), the lowest soil moisture was found for Euc and the highest for Woodland. Soil temperature was similar for Euc, Ap and Nat in the months of lowest soil moisture, while the lowest and the highest soil temperature were measured for Woodland and NCov respectively. In the rainy season, the highest soil temperature was recorded for NCov.

The contributions of AR and HR to soil TR were statically significant over the months (p <0.1), with HR presenting the highest values (Fig. 5), These differences can be seen in November 2014, and March and May of 2015, when HR contributed more to TR than AR. In September 2014, and January and July 2015, there was no difference between the contributions of AR and HR to soil TR. AR and HR did not differ in the areas of Euc, Ap or Nat.

Grouping the values for AR and HR during the months of highest and lowest soil moisture, (i.e. in the months of the highest and lowest precipitation) for the moments before gas collection it was seen that during the drier months there was no difference between the contributions of AR and HR to TR, with HR accounting for around 55% of TR. However, during the wettest months, HR contributed more to TR (71%). The forest plantations did not differ during the dry months concerning AR and HR, and during the rainy months, differences were only seen between Ap and Nat for AR (Fig. 6).

The litter contributed significantly more than the SOM to HR for the cover’s types (p <0.1; Fig. 7). The greatest contribution to HR by the litter was recorded in November 2014 for the three types of forest cover. The contribution of litter to HR does not differ between Euc, Ap and Nat, except for November 2015, when the HR was higher in Euc than Nat.

The multivariate model with two principal components (PCA) explained 70.7% of the variation of the data set: PC1 (60.3%) and PC2 (10.1%) (Fig. 8). The variables TC, CMB and LC together contributed 43.4%, TN and NMB 25.9%, and DRsm and DRgr 19.7%. From the PCA analysis, we identified three groups, with NCov and Woodland at the two extremes of the scatter plot and the other forest types are allocated between in the middle. In general, soil CO₂ efflux (TR) was positively correlated with soil moisture (Ms), total TC, TN and CMB.