Luara Castilho Pereira · Leonardo Balbinot ·Gregorio Nolazco Matus · Herly Carlos Teixeira Dias ·Kelly Cristina Tonello

Abstract Although the scientif ic knowledge about some associations between physical parameters of the soil is consolidated, studies focused on investigating the dynamics of forest restoration processes and the structure of soils remain incipient. The study area is located in Cerrado and is called as Private Natural Heritage Reserve (RPPN) Águas Perenes Forest (perennial water forest) and was selected as a “High Conservation Value Forest”, provides scope to investigate the soil physical properties recovery on passive restoration sites, with dif ferent ages but with the same land use history.So, the aim was to investigate the soil density, moisture, penetration resistance and inf iltration rate with long-term from old (FA, 46 year-old), medium (FB, 11 year-old) and young(FC, 8 year-old) passive restoration sites. It was observed differences in soil density, as well as in soil moisture, resistance to penetration and inf iltration rate. Mean densities recorded for FA, FB and FC were 1.38, 1.48 and 1.53 g cm −3 , respectively. FA recorded higher mean soil moisture (14.9%) than FB (11.5%) and FC (10.0%), whereas mean soil resistance to penetration was increased from FA (0.87 MPa) to FB(0.91 MPa) and FC (1.10 MPa). The average inf iltration rate was almost similar in FA and FB (FA = 76.43 mm h −1 ;FB = 77.48 mm h −1 ) and approximately 39% higher than FC(55.79 mm h −1 ). Thus, although soil density and moisture were dif ferent in each passive restoration site, 11 year-old passive restoration site showed similar resistance to penetration and inf iltration rate to the old (46 years) passive restoration site. These features allowed seeing improvements in hydrological maintenance in the soil and revealed that passive restoration in Cerrado is able to improve the physical features on a soil presenting similar texture, climate and management history.

Keywords Ecological restoration · Ecosystems service ·Forest management · Forest restoration · Soil conservation ·Tropical forest

Introduction

The restoration of forest ecosystems has been a major concern worldwide. Understanding the ef fectiveness of restoration models, their costs and time necessary to achieve tropical forest restoration success remains a challenge(Crouzeilles et al. 2016, 2017; Holl 2017; Meli et al. 2017).Research over the past two decades has shown that a host of abiotic and biotic factors can slow tropical forest recovery, but that the specif ic barriers to and rate of recovery are site-specif ic (Holl 2017). Although several factors can inf luence the restoration success the scientif ic literature shows that most of them are related to ecological indicators as the diversity of species, vegetation structure (cover, density, biomass, litter) and fauna (invertebrates, birds and mammals)on dif ferent restoration approach: passive restoration (ending the prior anthropogenic disturbance to allow natural or unassisted forest recovery) or active restoration–(range of human interventions that aim to accelerate and inf luence the successional trajectory of recovery) (Holl and Aide 2011).Tropical restoration projects have been motivated by improving carbon sequestration, biodiversity conservation and water quality and supply, although, the ef fect of forest restoration on hydrology services has been poorly documented(Murcia et al. 2016; Zhang et al. 2016; Bessi et al. 2018b;Vilarinho et al. 2019).

Evaluations on changes in soil physical properties are considered necessary for understanding the ecological consequences of vegetation recovery and can be used as indicators of soil quality and ecosystem functioning. After all,forest restoration (passive or active) involves the establishment of forests, the recovery of previously degraded soils,as well as other processes taking place in the scope of ecosystem services. Root penetration, the amount of gas in the soil, af fect the growth and distribution of tree species and microfauna interactions thus can inf luence physicochemical properties, the availability of water and nutrients (Safar et al.2019). Yet, promotes ground cover, increasing rain interception (Bessi et al. 2018a; Yang et al. 2019), retaining water in forests, reducing runof f, erosion and nutrient loss, improving the inf iltration (Jian et al. 2018) and recharge of aquifers.Information on the dynamics of soil water following longterm vegetation restoration is essential for managing water resources and would be helpful for adjusting relevant government policies (Zhang et al. 2016).

Thus, as well as assessing ecological restoration indicators over time it is necessary to include soil and hydrology indicators in forest restoration projects and understand how these processes interact in water conservation: linking ecology restoration, soil recovery and hydrological processes and answer the questions: How long is necessary for forest restoration improves the soil’s physical properties? What is the time required for the restoration to promote greater water conservation?

Cerrado is the second largest biome in South America, as well as the main biome connecting four of the f ive Brazilian biomes. It hosts the springs of three important hydrographic basins in South America (Araguaia/Tocantins, São Francisco and Prata) that largely contribute to water recharge in Guarani Aquifer, that is one of the largest and most important drinking water reserves in the Southern portion of South America (Leite and Ribeiro 2018; Sindico et al. 2018; Mendonça and Costa 2018), thus it has strategic value for several countries, mainly for the ones facing increasing water scarcity—which is a challenge to water management processes(Leite and Ribeiro 2018). The studied area is located in Cerrado − called as Private Natural Heritage Reserve (RPPN)Águas Perenes Forest (perennial water forest)—and was selected as a “High Conservation Value Forest”, as it provides basic environmental services in critical situations, i.e.,watershed protection. It comprises one of the main tributaries of the Brotas municipality main river, the Jacaré-Pepira,with a signif icant water supply coming from the watershed,where the present study site is located. AfterEucalyptussp.removal in 2006, silvicultural interventions and the establishment of Cerrado passive restoration have been carried out and the area is dedicated exclusively to nature conservation and to protection of the watershed. Thus, this site allowed us to investigate the soil physical properties recovery on passive restoration sites, with dif ferent ages but with the same land use history. The collected data were used to investigate the overarching question: how does soil density,moisture and penetration resistance dynamics with longterm from young to old passive restoration sites of Cerrado?Is the inf iltration process improved along the passive restoration ages? We hypothesized that soil density, soil moisture, soil penetration and inf iltration increase with passive restoration ages. The present study can help to improve the current understanding of passive restoration and soil–water conservation in disturbed areas in Cerrado.

Materials and methods

Study area

With 812 ha, the Floresta das Águas Perenes is situated in Brotas − São Paulo State and is characterized as secondary vegetation of the Cerrado and Cerradão. The Köppen climate-type of the region is Cwa (Dubreuil et al. 2019), which corresponds to a subtropical climate (C), characterized by warm summers and dry, cool winters (w), such that the average temperature in the hottest month (January) is more than 22 °C (a). Based on the meteorological data recorded from 2018, the annual average rainfall was 1337 mm, and the annual average temperature was 20 °C. The predominant soil type is quartzarenic neosol (Santos et al. 2018).The study was carried out in three forest sites undergoing three distinct passive restoration stages: site A (FA) referred to the tall fragment subjected to 43-year restoration, site B(FB) concerned the medium fragment subjected to 11-year restoration and site C (FC) corresponded to the young fragment subjected to 8-year restoration. The study was carried out in 3 sample units (400 m 2 each) located 10 m away from each other, in each site. Thus, the total sampled site in each fragment was 1200 m 2 . Rainfall, Net precipitation (throughfall + stemf low), soil density, soil moisture, soil resistance to penetration and infiltration rate were measured from May/2018 to Apr/2019. The structural attributes of vegetation in the stand are shown in Table 1 .

Soil physical properties

Soil density (SD) and soil moisture (SM) were obtained from three random samples in each plot (three plots per fragment),with 100 cm 3 metallic volumetric rings to obtain undistortedsamples and their weight obtained on a precision scale. The determination of these attributes was through the thermogravimetric method (Guariz et al. 2009; Embrapa 2017) which consists of weighing the freshly collected and dried fresh mass after 24 h in a forced circulation oven of 105 °C. Soil density(SD, g cm −3 ) was determined by the ratio between dry soil mass (Ms, g) and ring volume (V, m 3 ). Soil moisture (SM, %)was measured gravimetrically and expressed as a percentage of soil water to dry soil weight (g). Soil resistance to penetration(SRP, MPa) were collected with Digital Falker PLG1020 Penetrograph equipment, with three repetitions per plot, totaling 9 measurements per fragment. The SRP data were separated into 6 classes of depth: 1 − 10, 11 − 20, 21 − 30, 31 − 40, 41 − 50,and 51 − 60 cm.

Table 1 Mean (SE) features of the fragments

Inf iltration rate

For understanding the variability in inf iltration characteristics in each passive restoration site, the mini disk inf iltrometer (MDI), (tension disk inf iltrometer) was used to measure the accumulated inf iltration. The inf iltration samples were collected monthly with 3 sample repetitions in each plot,totaling 9 independent samples at random for each fragment and month. The MDI consists of a tube with an upper water chamber and an infer water chamber or a connection between them with a suction tube, which must be used with the recommended measure for each type of soil. In this case, considering that the soil in the research area is sandy, the suction rate 2 was used. At time zero, the inf iltrometer has been placed on the soil surface, assuring that it makes solid contact with the soil surface. The volume of water that inf iltrates into the soil has been recorded as a function of time in the record sheet.The procedure has been repeated for a suction head of 0.5 cm set in the MDI. Cumulative inf iltration (I, cm) is obtained by dividing the volume of inf iltrating water by the area through which water is inf iltrating. A graph is plotted between cumulative inf iltration on Y-axis and square root of time on X-axis.The data obtained are then f itted to the two-term equation for describing inf iltration under disk inf iltrometers and the values of constantsC1 andC2 were obtained by Zhang ( 1997):where,Iis the cumulative inf iltration (cm), andt(min) is the time.

Rainfall, throughfall and stemf low f ield sampling and storage

Rainfall,throughfall, and stemf low were measured from May/2018 to April/2019 within the three fragments. The rainfall was measured with three rain gauges installed near the stand, with a maximum distance of 30 m. The rainfall(P, mm) was calculated by the ratio between the rain gauge volume (V, L) and rain gauge opening (A, m 2 ). The throughfall was obtained from 24 rain gauges collectors located 1.20 m above the forest f loor and distributed throughout the plots. As rainfall, the throughfall was calculated by the ratio between the rain gauge volume (V, L) and rain gauge opening (A, m 2 ).

Stemf low was measured on trees with DBH (diameter breast height at 1.3 m) > 5 cm. The stemf low was measurement using a polyurethane gutter system based on the methodology (Likens and Eaton 1970) and f ixed at 1.30 m above the ground among the trees. The water that drained through the stem surface was directed by a 5/8 inch hose to 20 L collectors. The stemf low yield (SF, mm) was calculated considering the ratio between the volume accumulated in the collectors (V, L) and canopy area (CA, m 2 ).

Data analyses

Kolmogorov–Smirnov test—Lilliefors test (“lillie.test” function, “nortest” package) and the Bartlett test (“bartlett.test”function, “stats” package) were applied to verify the homoscedasticity in R software (R Development Core Team 2018)for statistical analysis. Analysis of variance (“aov” function, “stats” package) was applied to normal data through Tukey test (“TukeyHSD” function, “stats” package) at 5%probability level to analyze the means of the soil physical properties between the dif ferent passive restoration stages.Data that did not meet ANOVA assumptions were subjected to non-parametric Kruskal–Wallis test (“kruskal.test” function, “stats” package). The Spearman test (“cor” function,“spearman” method, “stats” package) and regression analyses was adopted to determine whether there were signif icant correlations between soil physical properties and net precipitation in the study.

Results

Soil properties

The soil density ordering in the current study has decreased as the passive restoration age increased (FC > FB > FA).Mean SD in the 0 − 20 cm soil layer was 7% and 10% lower in FA than in FB and FC, respectively (Table 2). SD values ranged from 1.31 to 1.46 g cm −3 in FA, from 1.33 to 1.55 g cm −3 in FB and from 1.34 to 2.13 g cm −3 in FC. Only FA (p-value = 0.31) and FB (p-value = 0.52) recorded normal SD distribution and homogeneous variance, although there was statistical dif ference between them (p-value < 0.05).The Kruskal–Wallis test also pointed out dif ferences in SD among FA, FB and FC (p-value = 0.0015). The investigated fragments have shown statistically signif icant dif ferences in SD median over the months (FA:p-value = 0.0181, FB:p-value = 0.0012, FC:p-value = 0.0044).

The annual mean soil moisture observed for FA was 14.9% (± 1.3%), whereas 11.5% (± 1.8%) for FB and 10.0%(± 1.1%) for FC (Table 2). Soil moisture ranged from 11.8%to 19.8% in FA, from 7.8% to 15.3% in FB and from 8.1%to 13.8% in FC. Maximum soil moisture in all fragments was recorded in October/2019. FA (p-value = 0.089) and FC(p-value = 0.1011) recorded normal distribution, but none of them showed homoscedasticity. Based on the Kruskal–Wallis analysis, soil moisture recorded statistically signif icant difference among fragments (p-value = 0.0010). Unlike FA and FB, FC has shown signif icant dif ferences in soil moisture over the months (p-value = 0.0177), which indicated that FA and FB tended to conserve soil moisture.

Soil resistance penetration (SRP) at 0 − 20 cm layer was signif icantly higher in FA than in FB and FC; however, FA presented overall trend to reduce SRP from the 20-cm soil layer downwards. FC showed the greatest resistance to penetration from the 20-cm soil layer downwards, whereas SRP in FB only started to decrease from the 40-cm soil layer downwards. The comparison of resistance to penetration means, FA presented the lowest SPR (0.87 MPa); it was followed by FB (0.91 MPa) and FC (1.10 MPa) (Table 3 and Fig. 1).

The highest soil resistance to penetration was observed from May to July in FA (1.32, 1.67 and 1.25 MPa, respectively) and FB (1.45, 1.64 and 1.14 MPa, respectively),whereas FC recorded the highest values in May and June(1.57 and 1.33 MPa, respectively). FA recorded the lowest soil resistance to penetration in November (0.36 MPa),whereas FB (0.35 MPa) and FC (0.45 MPa) recorded the lowest values in October (Fig. 2). Based on ANOVA, the passive restoration dif fered from each other in SRP at 95%conf idence level and months dif fered from each other at 95%conf idence level. FA and FB presented similar SRP between depth classes (p-value > 0.05), although this parameter presented statistically signif icant dif ference between months at 95% conf idence level (p-value < 0.05). The greatest dif ferences were recorded between June and September, October and November (p-value < 0.05). FC did not present statistically signif icant dif ference in SRP between months and between depth classes.

Table 3 Mean recorded for soil resistance to penetration (MPa),based on depth class in passive restoration at Cerrado for 46 (FA), 11(FB) and 8 years (FC). Águas Perenes Forest, Brotas County, Brazil

Table 2 Soil density (SD, g cm −3 ) and moisture (SM, %)over the months in passive restoration at Cerrado for 46(FA), 11 (FB) and 8 years (FC).Águas Perenes Forest, Brotas County, Brazil

Fig. 1 Mean SRP (MPa) prof ile in sites undergoing passive restoration at Cerrado for 46 (FA), 11 (FB) and 8 years (FC). Águas Perenes Forest, Brotas County, Brazil

Inf iltration rate

The average infiltration rate was similar in FA and FB(FA = 76.43 mm h −1 ; FB = 77.48 mm h −1 ) and approximately 39% higher than FC (55.79 mm h −1 ) (Table 4), and both were dif ferent than FC. The cumulative inf iltration had a polynomial trend and was similar in FA and FB and lower in FC (Fig. 3 a). The average of cumulative inf iltration was 2.2 cm for FA and FB and, 1.7 cm for FC. FA also showed the highest inf iltration velocity (132.21 mm h −1 ) and FC, the lowest amplitude (42.31 mm h −1 ) (Fig. 3 b).

Fig. 2 Boxplot of soil resistance to penetration (SRP)among months in sites undergoing passive restoration for ( a)46 years (FA), ( b) 11 years (FB)and ( c) 8 years (FC). Águas Perenes Forest, Brotas County,Brazil

Table 4 Inf iltration rate (IR, mm h −1 ) at dif ferent ages of passive restoration (FA = 46 years; FB = 11 years; FC = 8 years)

Net precipitation

The accumulated rainfall rate recorded during the survey period was 936.7 mm; FA and FB recorded net precipitation(NP) of 894.6 and 824.1 mm a −1 , respectively, i.e., values were 4.5% and 12% lower than that of rainfall. For FC, the NP was similar to rainfall. The highest ef fective precipitations were recorded from January to March in FA, as well as from February to March in FB and FC. Net precipitation was not recorded in June or July; March and December recorded the lowest NP values in the three sites (Fig. 4).

Correlation between soil physical properties and net precipitation

Correlation analysis has indicated that net precipitation was negatively correlated to soil resistance to penetration(r2 = − 0.65) and inf iltration rate (r2 = − 0.80) in FA, but it was weakly correlated to the other soil parameters (Table 5).Correlation analysis in FB has indicated that soil density was positively correlated to net precipitation (r2 = 0.70)and weakly correlated to the inf iltration rate (r2 = − 0.33),as also observed to FC (r2 = 0.45). The regression analysis showed that all sites have a practically linear relationship with ef fective precipitation (Fig. 5 a). The same behavior was observed between soil moisture and net precipitation in FA:as precipitation increase, soil moisture remains practically constant. In FC, in turn, it appears that the increase in soil moisture is caused by the increase in precipitation (Fig. 5 b).The mechanical resistance of the soil showed an inverse behavior with the net precipitation, indicating that with the increase of the rain, the mechanical resistance to the penetration decreases, especially in FA and FB (Fig. 5 c). The relation between net precipitation and inf iltration rate were similar and more pronounced in FA and FB (Fig. 5 d), but for all passive restoration sites, the inf iltration rate increased with the net precipitation.

Fig. 4 Net precipitation (mm)from May/2018 to April/2019.Passive restoration for 46 (FA),11 (FB) and 8 years (FC). Águas Perenes Forest, Brotas County,Brazil

Table 5 Regression models for Soil density (A), moisture (B), resistance penetration (C) and inf iltration (D) in function of net precipitation in passive restoration for 46 (FA), 11 (FB) and 8 years (FC).Águas Perenes Forest, Brotas County, Brazil

Discussion

Knowledge about the physical properties of soils plays a key role in studies about the ef ficiency of soil edaphic recovery practices, since soil recovery is much slower than plant recovery (Sparling et al. 2003; Li et al. 2007; Huang et al.2015). Soil physical properties analyzed in the current study have signif icantly changed afterEucalyptustrees were cut and passive restoration was implemented. This process led to dif ferences in soil density, as well as in soil moisture,resistance to penetration and inf iltration rate.

Soil density has decreased along passive restoration stages and decreased from 1.53 in the young to 1.38 in the old passive restoration. Similar results were reported by Huang et al. ( 2015), who observed that mean soil bulk density decreased from 1.56 in 8-year-old vegetation on the abounded land to 1.24 in 18-year-old vegetation at the refuse dumps. The highest soil density observed in FC may be associated with intense trampling caused by overgrazing or by machine traf fic for extended periods-of-time (shorter than 14 years). Based on the history of theEucalyptusharvest area, the soil was subjected to forest management,which may have contributed to the higher soil density in the subsurface layer in FC. The intense traf fic of agricultural machines and tools used in conventional management systems can lead to soil compaction over time (Bono et al.2013), since it increases soil density and decreases total soil porosity in deeper layers due to compacted subsurface layer formation—often up to 20 cm (Pessoa et al. 2018). Although soil density was lower than the limit suggested for sandy soils (< 1.6 g.cm −3 ), it was not a limiting factor for optimum plant growth (Reichert et al. 2003). On the other hand, FA showed even lower values than the ones observed in mature forest areas (Lima et al. 2009; Ramos et al. 2013; Ortiz et al.2017), probably due to the higher organic waste, litter and microorganism deposition in these areas, which indicated the ef ficiency of passive restoration in improving this soil feature (Albuquerque et al. 2005; Secco et al. 2005; Pessoa et al. 2018). Soil density in all sites was higher than the one observed in the regeneration of coarse-textured soils inEucalyptussp (Jankovský et al. 2019; Page-Dumroese et al. 2006; Pessoa et al. 2018) and in savannah (Pessoa et al.2018). On the other hand, the highest soil density in FC was similar to that ofEucalyptussp. monoculture and pasture(Sattler 2006; Prevedello et al. 2007), as well as lower than that inCerrado(Ramos et al. 2013).

Fig. 5 Soil density ( a), moisture ( b), resistance penetration ( c) and inf iltration rate ( d) in function of net precipitation in passive restoration for 46 (FA), 11 (FB) and 8 years (FC). Águas Perenes Forest, Brotas County, Brazil

In addition, passive restoration has signif icantly af fected soil moisture variability and levels over time. The lowest mean soil moisture and variability over the months in young passive restoration may be due to combined action of soil water retention capacity, vegetation transpiration, soil evaporation and meteorological elements. Because it is mostly featured as an open f ield, this site is mainly covered by herbaceous species, which are subjected to direct incidence of radiation and constant wind, fact that increases the loss of water from this environment through evapotranspiration processes (Tonello and Teixeira Filho 2011; Liuzzo et al. 2016;Anderson et al. 2017; Gaertner et al. 2019). Forest coverage in FA and FB can reduce wind speed near the ground and make soil surface more stable. According to Meerveld and McDonnell ( 2006) and Seneviratne et al. ( 2010), moisture in the upper soil layers is strongly inf luenced by vegetation transpiration and soil evaporation; the combination of these processes could consume approximately 60% of the total precipitation input (Oki and Shinjiro 2006; Zhang and Shangguan 2016; Zhang et al. 2016). Topographic features such as slope gradient (which can af fect water runof f and inf iltration) and aspect (which can af fect local climate parameters and plants’ water demand) can af fect evapotranspiration, but these features were closer among the three investigated sites (the same watershed hillshade). Thus,strong evaporation capacity in the young passive restoration could lead to continuously low soil moisture contents(Yang and Tian 2004).

The highest mean soil resistance to penetration in the current study was observed in young passive restoration,although the old passive restoration recorded the greatest SRP in the f irst soil layer (0–10 cm). Several studies conducted with visually healthy plants have shown soil resistance to penetration values signif icantly higher than the ones recorded in the current study (Busscher et al. 1997; To and Kay 2005; Whalley et al. 2007; Carvalho et al. 2012; Bolat et al. 2016; Pessoa et al. 2018), even than the ones recorded for young passive restoration, which showed the highest mean SRP. Sousa et al. ( 2015) also conducted a study inCerradowith Quartzarenic Neosol (the same biome and soil of the present research) and found SRP higher to the herein observed ones: from 0.75 to 1.75 MPa (Quartzarenic Neosol − AnthropizedCerrado), whereas values recorded in the current study ranged from 0.42 to 1.26 MPa in the young passive restoration, from 0.22 to 1.35 MPa in the medium passive restoration and from 0.24 to 1.66 MPa in the old passive restoration.

Jankovský et al. ( 2019) have compared SRP after forest cutting, and again, 5 years after the cutting – there was no intervention in the area during this period. Results reported by the authors ranged from 5.15 to 6.95 MPa; these values were much higher than that observed for young passive restoration, where machine traf fic and the cutting ofEucalyptussp. were more recent. SRP values higher than 2.5 MPa indicate compaction in agricultural soils (Silva and Castro 2015)—which overall results from the use of heavy machinery, mainly during harvesting—and is often classif ied as signif icantly restricting (Whalley et al. 2007; Ortiz et al.2017). This was not the case of this study, even in the young passive restoration, showing that after 8 years ofEucalyptussp harvesting, the soil structure is better than other similar studies.

Based on the USDA (1993), both old (0.87 MPa) and medium (0.92 MPa) passive restoration were classif ied as having low soil resistance to penetration, and the young passive restoration was classif ied as having moderate SRP(> 1.0 MPa). As observed for density and moisture, this outcome has indicated that the soil resistance to penetration improved from the young to old passive restoration and even the forest at the youngest passive restoration recorded values lower than the one classif ied as restrictive to root development.

Soil water inf iltration is controlled by the rate and duration of precipitation, soil physical properties, slope, vegetation, and surface roughness. Although there were improvements in density, moisture and resistance to soil penetration between the sites, the inf iltration rate was the only one that showed similarity between the sites, specif ically between medium and old passive restoration. It means that after 11 years, the passive restoration has the similar inf iltration rate properties than the old passive restoration in Cerrado.In spite of the fact that the young passive restoration showed the lowest inf iltration rate, the speeds for all passive restoration sites are considered to be “very high” (Bernardo et al.2008). The behavior observed in this study was dif ferent from that of Vilarinho et al. ( 2019) − that obtained a higher inf iltration rate in pasture than in the Cerrado area − but indicate that even with a history of using the soil withEucalyptussp. and heavy machinery for its harvest, after 8 years of conducting passive restoration, it improved the condition of water inf iltration in the soil. This information is extremely important for eco-hydrology, and must be associated with the forest restoration processes, in order to subsidize the management of water resources, since the water that inf iltrates the soil, will be associated with the maintenance of the soil water and its percolation in depth, that is, the maintenance of groundwater.

Ebeling et al. ( 2016) have found that the structural disturbance in sandy soil persisted for few more decades, which suggested that this soil type has exceptionally low ability to recover. According to Rab ( 2004), it takes more than 20 years to regenerate the top 10-cm soil layer. These results are in compliance with Page-Dumroese et al. ( 2006), who reported that coarse-textured soils presented clear recovery in the surface (0–10 cm) soil layer after 5 regeneration years, although such recovery was lesser evident in the subsoil (10–30 cm soil layer). The recovery of topsoil features through the establishment of sand-binding vegetation in sand-burial environment in Tengger Desert was estimated to take from 23 to 245 years. It is worth emphasizing that,besides using and covering the current soil, it is fundamental recording its history, and the management carried out in it,in order to assess soil recovery behavior based on specif ic forest restoration techniques. Many studies have focused on the ef fects of compaction on SPR; however, comparing these results is not an easy task due to strong variation in experimental conditions such as soil physical features (Mohieddinne et al. 2019) and land-use history. The recovery process depends on the initial compaction severity; therefore, used equipment, harvesting (Picchio et al. 2018; Mohieddinne et al. 2019) and management systems may help controlling recovery dynamics.

Conclusion

The herein investigated sites have in common the soil use system −Eucalyptussp. planted for cellulose production purposes − and their ages of passive restoration, correspond to the end of the soil management for this purpose. The passive restoration under these conditions, has favored soil recovery due to increased soil moisture content, as well as lower soil density and resistance to penetration. The measured soil parameters were strongly related to site’s age,which suggested that the passive restoration approach is ef fective in enhancing ecosystem’s recovery afterEucalyptussp. harvesting. These features allowed seeing improvements in hydrological maintenance in the soil, showing that the inf iltration rate in medium passive restoration (11 years)was almost similar to the old (46 years) passive restoration.This behavior proves the importance of forest cover to the maintenance of the physical properties of soil subjected to the investigated conditions, since they directly inf luence the quality of the soil, hydrological process and the success of forest restoration processes.

AcknowledgementsWe appreciate the cooperation of the International Paper from Brazil LTDA (Process 23112.000670/2015-59),Federal University of São Carlos and the Brazilian National Council for Scientif ic and Technological Development (CNPq).