Consequences of removal of exotic species (eucalyptus) on carbon and nitrogen cycles in the soil-plant system in a secondary tropical Atlantic forest in Brazil with a dual-isotope approach

Site of study, species and sampling period

This study was held in the União Biological Reserve (Rebio União), in Rio de Janeiro State, Brazil (22°27′S; 42°02′W). Field experiments were approved by SISBIO/ICMBio/MMA, to develop the project “Capacidade aclimatativa de espécies da Mata Atlântica após manejo de eucalipto na Reserva Biológica da União” at the Rebio União, Rio de Janeiro, Brazil, under the licence number 37996-3.

The climate in the region is humid tropical Aw (Alvares et al., 2013), with a yearly average temperature of 25 °C and an annual rainfall of 1,920 mm (85% being between October and April). This Reserve comprises around 7,800 ha of ombrophilous dense Atlantic forest, of which around 220 ha had the native tree vegetation suppressed and replaced by eucalyptus crops (Corymbia citriodora) (Hook.) K.D. Hill & L.A.S. Johnson. The eucalyptus tree plantations ceased to receive silvicultural treatments related to understory removal in 1996. The eucalyptus was clear-cut between 2009 and 2015.

We realized this research in three latosolic dystrophic red–yellow argisoil sites, classified according to Empresa Brasileira de Pesquisa Agropecuária (EMBRAPA) (1999), Miranda, Canellas & Nascimento (2007) and Dos Santos et al. (2018). One site is a well-preserved secondary forest; whereas the other two are eucalyptus-harvested sites, employing clear-cutting, with a single difference related to after-handling time: the M12 (where eucalyptus had been removed 12 months before our evaluation); and the M3 (where eucalyptus removal had happened three months before assessment) (Fig. 1; Table 1; Data S1). The post eucalyptus removal times were chosen according to the management activities in the Rebio União, and is sufficient to detect processes related to the decomposition of organic matter. For each of these sites, three ecosystem compartments were studied: photosynthetically active leaves, organic litter fractions, and the soil, due to its relation with organic matter decomposition processes.

For the elemental and isotopic analyses of C and N, we considered the most abundant tree species within the eucalyptus removal sites M12 and M3 (Evaristo, Braga & Nascimento, 2011), being all of those from the initial succession stratum (Lorenzi, 2000): Siparuna guianensis Aubl. (Siparunaceae), Xylopia sericea A. St.-Hill (Annonaceae), and Byrsonima sericea D.C. (Malpighiaceae). We collected the leaves from the eucalyptus removal sites with individuals from 2 to 5 m high, and diameter at breast height (DBH) >5 cm. For a better diversity-related sampling within the secondary forest site, we collected samples from 20 species (Data S2) according to their importance value index (IVI), all of those ranging between 10 and 15 m tall, and with a DBH >10 cm. The leaves, plant litter, and soils were collected in December 2013 during the rainy season.

Determination of the elemental and isotopic composition of C and N within the leaves, litter and soil compartments and sample number

We standardized collected samples of the third sun-exposed photosynthetically active pair of leaves, once photosynthetic activity and irradiance variation are important factors to determine δ¹³C (Vitória et al., 2016). Young leaves are not photosynthetically active and enriched in ¹³C (Cernusak et al., 2009), similar to shadow leaves (Vitória et al., 2016). Besides, great variation can be found in δ¹³C within a single tree crown on account of the variations in environmental conditions (Le Roux et al., 2001; McDowell et al., 2011). Thus, sun-exposed branches of individuals of the secondary forest were collected from the canopy by using high pruning shears. Sun-leaves from eucalyptus removal areas (M3 and M12 sites) were manually collected. For each species, 5–7 leaves per individual were sampled and then dried in an oven at 50 °C for 48 h (Vitória et al., 2016). Afterward the leaves were macerated in liquid N and homogenized per individual. Regarding the eucalyptus removal sites, the samples collected comprise 5–15 individuals (n = 5–15) of the three species from the regenerating understory in both the M12 and the M3 sites. For a better diversity-related sampling within the secondary forest site, we collected samples from 20 species (Data S2). For all of these 20 species, 5 adult individuals had their leaves sampled, except for the following species: Ficus gomelleira (n = 4), Cupania racemosa (n = 4), Brosimum glazioui (n = 4), Virola gardneri (n = 2), Micropholis guianensis (n = 2), Ocotea diospyrifolia (n = 2). Thus, for the secondary forest, we collected samples from 2 to 5 individuals (n = 2–5).

The litter layer on the soil was collected via a 20 × 20 cm metal frame randomly arranged 25 times in each of the three study sites, totaling 75 samples with 400 cm² of litter. The collect points were chosen at random. The collected material was dried at constant weight in an oven at 50 °C for 48 h (Mayor et al., 2014; Camara et al., 2018). These samples were screen-sorted by the following fractions: leaves (leaves and leaflets, whole or fragmented); wood (branch or bark pieces with a diameter smaller than 2 cm, and fragments), flower + fruits (flowers, inflorescence or organ, and fragments recognized as such, fruits and seeds, whole or fragmented), and the rest (residues smaller than 2 mm, animal-origin and unknown matter) (Proctor et al., 1983). The 25 dried litter fractions were macerated and grouped in 5 composed samples for each soil-litter fraction per site (n = 5).

The superficial soil (0–10 cm) was collected from below the litter layer withdrawn, totalizing 75 samples with 400 cm² of soil. The 75 soil samples were dried at constant weight in an oven at 60 °C for 48 h (Mayor et al., 2014; Severo et al., 2017, Camara et al., 2018). No decarbonation was performed once previous tests verified the absence of inorganic carbon in the soil (Camara et al., 2018). A soil sub-sample was used to determine granulometric fractions (item 3), and another sub-sample was sieved in a 2 mm mesh and homogenized for isotopic and elemental determination of C and N (n = 25) (Telles et al., 2003).

Leaves, organic litter fractions, and soils were dried, ground, weighed (1 mg approximately), and inserted in tin capsules for C and N elemental and isotopic determination. All samples were combusted in an elemental analyzer (Flash 2000 Organic Elemental Analyser) which measured elemental concentrations of C and N; coupled to a stable isotope ratio mass spectrometer (IR-MS Delta V Advantage; Thermo Scientific, Waltham, MA, USA); and combined with a ConFloIV Interface (Thermo Scientific, Waltham, MA, USA), which measured the isotopic composition of C and N. Pee Dee Belemnite (PDB) and atmospheric N were used as standard values for C and N analyses, respectively. The values of PDB δ¹³C and atmospheric δ¹⁵N are respectively, 0.0111802‰ and 0.0036‰. The analytical precision was ±0.1 for δ¹³C and ±0.2 for δ¹⁵N, and the accuracy for elemental and isotopic compositions were determined by certified Organic Analytical Standard (OAS): Wheat Flour Standard OAS/Isotope Cert. 114858 and Low Organic Content Soil Standard OAS/Isotope Cert. 201613. The values of OAS 114858 were 39.38 ± 0.19% for C; 1.36 ± 0.23% for N; −27.21 ± 0.13‰ for δ¹³C; 2.85 ± 0.17‰ for δ¹⁵N. The values of OAS 201613 were 1.52 ± 0.02% for C; 0.13 ± 0.02% for N; −27.46 ± 0.11‰ for δ¹³C; 6.70 ± 0.15‰ for δ¹⁵N.

Carbon and nitrogen stable isotope values are reported in ‘‘delta” notation as δ values in parts per thousand (‰).

Determination of the soil granulometric fractions

After soil preparation procedures, as aforementioned, sub-samples were sieved, separating >2 mm fractions by sifting in successive intervals. Soil texture (granulometry) was determined in the soil fraction <2 mm. The methodology used here followed Wentworth (1922) by using a laser diffraction particle analyzer (SALD-3101; Shimadzu, Kyoto, Japan).

For precision analytical control, we measured the analytical variation among the analytical triplicates in every 20 samples, with an acceptable variation coefficient inferior to 10%. The accuracy surpassed 90%. Three certified samples supplied by the manufacturer of the equipment (Shimadzu, Kyoto, Japan) with differentiated particle size range (JIS class 11, Licopodium and glass beads, Data S3) were used to evaluate the accuracy of the soil granulometric fractions. The limit of detection herein was 0.1%.

The granulometric distribution was compiled by the software SysGran (3.0), and the data was grouped in sand, silt, and clay. The data are expressed in mean and standard error values.

Statistical analysis

In order to assess the normality in the data, the Shapiro Wilk (P ≤ 5%) test was conducted. After assessing the normality and homogeneity, the data underwent parametrical analyses. We analyzed differences between means via ANOVA (two-way), followed by Tukey test (P ≤ 5%). The factors considered herein were the compartments (leaves, litter fractions, and soil), sites, and granulometric fractions; it all aims to understand the interference of eucalyptus permanence time in the δ¹³C and δ¹⁵N in leaves, soil and organic litter fractions. The variance analysis was realized in the software Statistica 7.0. We calculated the sand regression coefficient correlations with Sigma Plot 11.0 software package Statistical Package for the Social Sciences (SPSS).

The C isotopic fractionation during organic matter decomposition was calculated from the difference between the δ¹³C in the soil and the litter fractions. The same was employed to N, by using δ¹⁵N. We inferred the soil organic matter decomposition based on these calculations.

The relationship between C and N in the soil-plant system

The values of δ¹³C and δ¹⁵N were higher in the soil > litter > leaves for all studied sites (Fig. 2; Data S4–S6). We observed a direct relationship between δ¹³C and δ¹⁵N (Figs. 2A, 2C and 2E), and an inverse one between δ¹⁵N and N in all three sites (Figs. 2B, 2D and 2F).

Regardless of the site, the elemental concentration values for C and for the C/N ratio were respectively distributed: litter fractions > leaves > soil (Table 2). For N, leaves had higher values than litter, being its concentration thus distributed: leaves > litter fractions > soil (Table 2). Significant higher concentrations of C and N in the soil were found in the M12 site, when compared with the secondary forest and to the M3 site (Table 2). The N concentrations of all plant litter fractions were significantly smaller in the two eucalyptus removal sites (M12 and M3, Table 2). The C/N ratio in the leaves was higher in the M12 site. Higher C/N ratio values in wood fractions were obtained in the eucalyptus removal sites, in comparison with those of the secondary forest (Table 2).

Decomposition of organic matter in the litter fractions

There was an enrichment in the ¹³C between litter fractions and soil in all sites (except for flower and fruit fractions from the secondary forest and the fraction of the rest in the M3 site) (Table 3). On the other hand, there was no enrichment in ¹⁵N in any of the studied sites, being observed only negative values at the time of calculating differences between litter fractions to the soil, for each site (Table 3).

There was a negative relationship between δ¹⁵N and the C/N ratio in the soil (Fig. 3), having the secondary forest the highest value for δ¹⁵N in comparison to the eucalyptus removal sites, which did not vary between one another. The soil C/N ratio was higher in the most recent eucalyptus removal site (M3) than in the M12 site and the secondary forest (Fig. 3).

The secondary forest soil, where greater values of δ¹⁵N were found, presented higher clay content, whereas in the M3 site, which presented a higher C/N ratio, had also featured a more significant amount of sand (Fig. 4; Data S7).

Comparison of δ¹³C and δ¹⁵N between the compartments and sites

The δ¹³C values of soil varied from −26.7‰ and −29.6‰ and were consistently higher than those found in leaves and leaf litter fractions (Fig. 5A). The mean soil δ¹³C in the M3 site was higher than those from both the secondary forest and the M12 site. Significant differences were observed in δ¹³C values of leaves for the M12 site, in comparison to the other two sites; and also with foliar δ¹³C values ranging from −31.6‰ (secondary forest) to −32.2‰ (M3 site). The δ¹³C value of the litter rest fraction in the M3 site was higher than those from the secondary forest and the M12 site (Fig. 5A). In the organic litter, we found that the rest fraction presented the highest δ¹³C value (−23.8‰, M3 site), whereas the lowest value was found in the leaf fraction (−31.3‰, M12 site).

The δ¹⁵N values have significantly been higher in the secondary forest samples, when compared with those from the eucalyptus removal sites, regardless of the compartment under analysis (Fig. 5B). Concerning the soil, only positive values were verified, which varied from 2.6‰ to 6.0‰. Positive values were also observed for leaves, going from 1.6‰ to 2.9‰. Negative values of δ¹⁵N were registered only in a few litter fractions (rest and wood) from the M12 site.

Variation of carbon isotopic composition (δ¹³C) in all compartments in the eucalyptus removal sites

The different δ¹³C variations of compartments between the eucalyptus removal sites and the secondary forest are mainly caused by the difference in the floristic composition, the organic litter input, and the irradiance, which consequently promotes variations in temperature and humidity, thus affecting processes such as assimilation of C and decomposition. The eucalyptus removal has altered the δ¹³C in every compartment of the managed sites, in comparison with the secondary forest. The higher δ¹³C values of the residue litter fraction and the soil in the M3 site reflect the presence, and carbon soil incorporation originated from the photosynthesis of Poaceae species (grassy), with photosynthetic C₄ syndrome (Farquhar, 1983). The regenerating understory of tree species in the Atlantic forest in the M3 site was less developed than in the M12 site, possibly on account of the less time for native vegetation to grow and the competition with grassy that is abundant in the M3 site. C₄ species present a greater δ¹³C (ranging from −15‰ to −9‰) than that of the tree species, which feature photosynthetic C₃ syndrome (whose δ¹³C values may vary from −35‰ to −23‰) (Farquhar, Ehleringer & Hubick, 1989; Martinelli et al., 2009). The grassy species contributed to the organic matter onto the soil, raising the δ¹³C of soil from the M3 site. In the secondary forest, δ¹³C values were higher in the sequence soil > litter fractions > leaves. A similar pattern was observed in the eucalyptus removal sites, except for the rest litter fraction in the M3 site given the reason aforementioned. The highest soil δ¹³C value, in comparison with the remaining analyzed compartments, may be explained by the fast mineralization of the components with lower δ¹³C values (Gerschlauer et al., 2018). Differences between the δ¹³C from the distinct litter fractions and the δ¹³C from the soil are interpreted as consequences of the litter organic matter decomposition (Gerschlauer et al., 2018). During decomposition of litter, fractionation occurs due to the decomposers that preferentially use ¹²C resulting in ¹³C-enrichment in the remaining soil organic matter (Lichtfouse et al., 1995). The difference in δ¹³C value (δ¹³C litter − δ¹³C soil) and the low C/N ratio in the soil, observed in the M3 site, agree with the idea about altering the dynamics of this soil organic matter.

Our results suggest there was organic matter decomposition in every litter fraction and in every site. However, a higher decomposition of the organic matter determined from the δ¹³C in the M3 site was also observed. This result possibly comes from the more significant recent input of vegetal matter into the M3 site soil, as well as the presence of grassy (herbaceous species of rapid decomposition) exclusively in this site. Despite the soil humidity reduction caused by microclimatic alterations from recent handling in the M3 site (such as less humidity, higher irradiance, temperature, and winds), the input increase was enough to evidence more significant decomposition in this particular site.

The highest δ¹³C values found in the M12 site leaves, compared to both secondary forest and the M3 sites, are explained by the fact that this vegetation was formed under greater irradiance, for 1 year after handling. Sites exposed to more intense irradiance provide the environmental conditions which increase the δ¹³C values within the photosynthesizing tissues (Ehleringer et al., 1986; Domingues et al., 2005; Van der Sleen et al., 2014; Teixeira et al., 2015; Vitória et al., 2016).

In all sites, the δ¹³C values from the litter leaf fractions were a reflex of the δ¹³C living leaves. However, despite the previous discussion, the influence of the irradiance variation upon δ¹³C values in leaves was not the single cause for δ¹³C values in the litter rest fraction, mainly in the M3 site, where grassy (C₄ species) enrich the soil with ¹³C. Compared with litter leaves, the remaining fractions (“flowers + fruits” and wood) presented higher δ¹³C values. Variations between the δ¹³C from the source and sink organs have been widely reported, is the enrichment extension variation of the sink organs δ¹³C dependent on the tissue, species, and environmental conditions (Cernusak et al., 2009; Vitória et al., 2016).

Among the causes of isotopic variation between source and sink is the variability in the biochemical tissue composition (Cernusak et al., 2009). The leaves present higher cellulose concentrations than do stems and roots which possess more lignin. After eucalyptus removal in the M3 and the M12 sites, the cellulose from leaves decomposed faster than did the lignin from stems, thus changing the soil δ¹³C. Cellulose is more enriched with ¹³C than lignin and lipids (Badeck et al., 2005; Cernusak et al., 2009). Another isotopic difference-contributing factor between source and sink is the isotopic fractionation during night breathing. The CO₂ respired in the dark by the aerial part is significantly enriched with ¹³C when compared with the respiratory substrate pool, whereas the CO₂ aspired by the roots does not present enrichment (Klumpp et al., 2005). Moreover, stems grow preferentially at night (Steppe et al., 2005; Saveyn, Steppe & Lemeur, 2007). This is when the exported carbohydrates present higher δ¹³C values. As a consequence, the wood features enrichment of ¹³C when compared with the leaves. Also, carbohydrates exported during the day present lower δ¹³C values, and supply preferentially for the leaf growth, which occurs faster during the day (Walter & Schurr, 2005).

However, it is essential to consider that the nature of the site sampling, which occurred under a unique field experimental condition, prevents the replication so as to increase the robustness of this data. In this way, although the data indicate that the presence of grassy and the higher irradiance was determining factors for the alteration of δ¹³C in the analyzed compartments, these data should be regarded cautiously.

Differences in nitrogen isotopic composition (δ¹⁵N) in the soil-plant system related to eucalyptus removal

The higher δ¹⁵N values within the soil-plant system in the secondary forest, compared to the eucalyptus removal sites, can relate to differences in the composition of vegetal species, and in the strategies employed by them regarding the use of the soil nutrients (Viani, 2010). Depending on the variation in the functional diversity among coexisting species, one expects a more efficient use of resources and, in parallel, a reduction in the interspecific competition which, consequently, may increase the ecosystem productivity (Naeem et al., 2009). Some species may also disproportionately contribute to the ecosystem-related processes on account of a high specialization level, as the symbiotic nitrogen fixation (Reed, Cleveland & Townsend, 2008). Although tropical forests have an immense nutrient stock, forest sites under management, in general, may lack nutrients, thus presenting a decrease in these stocks (Bellote, Dedecek & Silva, 2008). Besides, other factors such as the dominant species, the management period and activity type, the rain cycles and intensity, and the soil properties interfere in the dynamics of nutrients (Bellote, Dedecek & Silva, 2008).

Our data shows the influence of eucalyptus removal on the superficial soil layer, being the eucalyptus removal sites (M3 and M12), the ones to present lower δ¹⁵N in comparison with the secondary forest. Similar δ¹⁵N values, to those from the secondary forest, were also found in: other Atlantic forest soils (between −1.2‰ and 6.0‰) (Viani, 2010; Gragnani, 2014); in the litter (between −1.8‰ and 1.8‰) (Owen, 2013); and in tree leaves from tropical forests (between −3.0‰ and 12.0‰) (Davidson et al., 2007; Nardoto et al., 2008, Vitória et al., 2018).

Other aspects regarding the retention of N and the difference in δ¹⁵N values in the soil-plant system between the sites should be taken into account. First, there is the N mobility before the foliar senescence, being its concentration thus distributed: leaves > litter fractions > soil in all the three sites. Second, the differences in soil textures, where more clayey soil was found in the secondary forest than in the eucalyptus removal sites. Clayey soils have greater retaining capacity when compared with sandy ones, especially the N ammoniacal form (Trumbore & Camargo, 2009). Clay soils present more intense processes of mineralization, nitrification, and denitrification than do sandy soils in the same region, and that influences the forest productivity and nitrogen availability (Keller et al., 2005). This suggests that N and δ¹⁵N are relatively higher in sites with greater clay concentration (Martinelli et al., 1999; Davidson et al., 2007; Houlton et al., 2007; Craine et al., 2009). Nonetheless, greater water availability also tends to increase N concentration and δ¹⁵N values in leaves and soils (Austin & Vitousek, 1998; Handley et al., 1999; Amundson et al., 2003; Houlton et al., 2007; Nardoto et al., 2008).

Even though the relationship between clay and soil δ¹⁵N might be promote higher N fractionation (via nitrification, denitrification, and volatilization), in clay-rich soils the relative proportion of N fractionation might not be directly influenced by clays, but by the stability of clay particles containing ¹⁵N-enriched organic matter. As a result of the higher organic matter decomposition and/or higher clay concentrations, it is possible for soils from hotter and/or drier ecosystems to have a greater proportion of N in the organic matter associated with minerals, as opposed to having a higher proportion of N that is lost by fractionation processes (Craine et al., 2015).

The lower C/N litter ratio limits the N in the soil, thus raising the organic matter decomposition rates by the biota (Berhe, 2012). The C/N ratio in the leaves was greater in the M12 site. Higher C/N ratio values in wood fractions were obtained in the eucalyptus removal sites, in comparison with those of the secondary forest. That demonstrates the slow eucalyptus branch wood decomposition compared to that of the native species. Therefore, the secondary forest site, with less C/N ratio in the litter, has its N-rich organic compounds consumed rapidly; while the excess of N is mineralized, increasing hence the δ¹⁵N value in the soil. That explains the negative relationship between δ¹⁵N and the C/N ratio, which is also reported by big-scale evaluations (Amundson et al., 2003; Craine et al., 2015).

Whenever both soil and plant contain high δ¹⁵N values, one may infer the presence of organic matter and decomposition activity in the soil, aside from variables such as temperature and gas flow into the atmosphere, which also interfere in the system (Craine et al., 2015). More humid soils (clayey) usually have more nitrogen gases emissions than those sandy soils (Davidson et al., 2000; Keller et al., 2005). Soils with larger quantities of silt and clay present a lower C/N ratio (Stemmer, Gerzabek & Kandeler, 1998; Christensen, 2001), as observed in the study herein.

The ¹⁵N-rich organic matter decomposition, in general, has smaller δ¹⁵N values in mountainous ecosystems from tropical forests than do a wide range of tropical forests in low lands worldwide with average values ranging from 3‰ to 14‰ (Martinelli et al., 1999; Nardoto et al., 2014). Such low δ¹⁵N values were reported in hilly ecosystems and were related to both high concentration of NO₃⁻ in soil solution and significant N lixiviation rates (10–15 kg N ha⁻¹ year⁻¹) (Gütlein et al., 2016).

A significant increase in soil δ¹⁵N has also been observed in forests where recent fire events took place (Hemp, 2005; Zech et al., 2011). Vegetation burning may cause loss of gaseous NO₂ resulting in depleted values of δ¹⁵N (Zech et al., 2011). Some disturbances in the ecosystem (forest fire, selective cutting, and others) trigger changes in the vegetal coat which affects the organic matter dynamics, and which may be regarded as variations in the C/N ratio in the soil (Saiz et al., 2016). Hence, this work suggests that the high δ¹⁵N values in leaves and soil from the secondary forest are a consequence of relatively clay-rich soil combined with a moderate water deficit (Ometto et al., 2006; Nardoto et al., 2008; Quesada et al., 2012) which, consequently, would then lead to greater availability and loss of N (Ometto et al., 2006; Nardoto et al., 2008).

The alteration in vegetation after eucalyptus removal modified the soil properties and furthered the decomposition processes of soil organic matter

The higher concentration of C and N in the soils of the M12 site is associated with two factors mainly: (1) greater eucalyptus litter entry in the system regarding eucalyptus logging, once only eucalyptus tree trunks were removed from the site in order to be commercialized; and (2) longer period since eucalyptus removal in this site, in comparison with the M3 site, which enabled the decomposition and incorporation of the M12 site litter-origin nutrients by the soil. However, the soil C concentration (3.3%) in the M12 site is among the lowest values reported for tropical forest soils, which vary from 3.0% (Mazurec, 2003) to 9.0% (Clevelario Júnior, 1996). A similar concentration of elemental N in the soil, found in the M12 site (0.3%) had been reported by Villela et al. (2001) in superficial soils of eucalyptus crops (0.31%) in the same studied site (Rebio União). The concentration of elemental N is greater in the leaves than it is in the soil and the litter fractions, which may be explained by the translocation of this mobile element in the plant before the foliar senescence (Aerts & Chapin, 2000; Satti et al., 2003; Kuang et al., 2010). The great necessity of N by this organ is due to the photosynthetic process in which many enzymes are entailed, as protein molecules with a catalytic function that contain N in their structures (Paul & Pellny, 2003).

The biotic and abiotic conditions fostered the decomposition of ¹³C-enriched compounds in the M3 site. In more sandy-textured soils (M3 site), the organic matter tends to decompose more rapidly, which has less capacity to retain organic matter (Rosell, Galantini & Iglesias, 1996). The more intense presence of sand also favours the lixiviation of more soluble organic compounds, which might dissipate after the litter deposition. Moreover, the greater presence of grassy (C₄ plants) in the M3 site soil, compared with that of the M12 site, presented a higher value for δ¹³C in the soil because the C₄ species showed less isotopic discrimination than did C₃ species. From that, the M3 site soil had a higher δ¹³C value (Farquhar, 1983). Studies on the Atlantic forest soils (Vitorello et al., 1989; Bonde, Christensen & Cerri, 1992) and the Amazon forest soils (Desjardins et al., 1996) stated that soils with greater clay quantity had higher values for δ¹³C. Even so, it was not confirmed by this research, where the δ¹³C value was higher in the M3 site soil, which had a more considerable amount of sand. Certainly, the presence of both grassy species and sandy soil in the M3 site was an influencing factor upon δ¹³C values in this soil.

The relationship between the C/N ratio and the δ¹⁵N may be used as a tool for verifying degradation and stabilization of the soil organic matter in disturbed systems, in comparison with the secondary forest, which has already been observed (Conen et al., 2008). In the present study, the soil δ¹⁵N increased along with the decrease in the soil C/N ratio soil. Craine et al. (2015), studying 6,000 soil samples from 910 sites, had also noted the increase in soil δ¹⁵N associated with a decrease in the soil C/N ratio, and as a consequence of greater stabilization of the soil organic matter. Soils with high clay content, as it is the case of the secondary forest, also have high soil δ¹⁵N. The clay and silt fractions are frequently associated with this soil organic matter, thus hampering the decomposition process (Liao, Boutton & Jastrow, 2006).

The organic matter with a longer incorporation time into the soil should be a better indicator of the isotopic signature of the decomposer than of the initial input from the fall of branches and plant leaves (Craine et al., 2015). For this reason, the soil organic matter fractions generally show higher δ¹⁵N values in humidified soils (Liao, Boutton & Jastrow, 2006; Marin-Spiotta et al., 2009). Liao, Boutton & Jastrow (2006) quantified C and N stable isotopes in the soil, as well as their granulometry, in sites where tree species and bushes (C₃ plants) were replaced by pastures (C₄ plants). According to these authors, the soils with higher silt and clay fractions were those with higher δ¹⁵N values, thus suggesting organic matter stabilization (Marin-Spiotta et al., 2009).

The data presented herein indicate the alteration of C and N dynamics in the ecosystem considering land use (presence and subsequent removal of the exotic eucalyptus species from the Atlantic forest fragment). It is important to emphasize the need for caution in the interpretation of these data since there are limitations around this type of experimental design. In studies that make use of real situations in the field, true replicates (sites) are difficult to work. In addition, this study addressed isotopic and soil granulometry analysis. More robust information can be obtained when other indicators and characteristics of vegetation and soil are also analyzed, including functionality attributes, photosynthetic performance, details of soil organic matter composition, among others. However, it is of utmost importance to understand how this exotic species affects the functioning of the ecosystem since eucalyptus has been planted all over the world for commercial or forest restoration purposes, either as monoculture or in symbiotic consortia with native tropical forests (Guariguata, Rheingans & Montagnini, 1995; Parrotta, Turnbull & Jones, 1997; Feyera, Beck & Lüttge, 2002; Baptista & Rudel, 2006). The substitution of native forest cover by the exotic eucalyptus species has been widely used in order to give financial returns to the landowners (Baptista & Rudel, 2006). In some cases, the economic and environmental purposes are enhanced simultaneously, being the eucalyptus used for its rapid growth in deforested areas (Feyera, Beck & Lüttge, 2002), facilitating later forest restoration, as seen in plantations of this exotic species in Central America and the Caribbean (Guariguata, Rheingans & Montagnini, 1995; Parrotta, Turnbull & Jones, 1997). The fact that tropical forests are being the focus of recovery efforts after disturbances and degradation is mainly due to the urgent need to mitigate the consequences of climate change. In this sense, the data presented here are important to improve the understanding of the beneficial effect of the eucalyptus removal in the ecosystem.