Introduction

Soil is a dynamic environment where biological, chemical, and physical components have complex interactions (Delgado and Gomez 2016) and are closely interrelated with the productivity and sustainability of the system. Most of its attributes directly or indirectly influence its quality (SQ) and biological production (De Deyn and Kooistra 2021; Ghorai et al. 2023). To effectively and sustainably improve soil quality and SOC sequestration, recommended practices should be region-specific and adapted to local pedoclimatic conditions (Amelung et al. 2020; Certini and Scalenghe 2023). Soil also provides and regulates ecosystem services that depend on formative factors, such as original parent material, relief, biota, weather, and climate, which affect it, as well as its management (Pereira et al. 2018; Vandermeer 2011; Certini and Scalenghe 2023). Drobnik et al. (2018), Diti et al. (2020), and Machado et al. (2019) propose to develop a soil quality index based on ecosystem services and their functions.

The conditions, distribution, and composition of plants respond to pH and nutrient availability (Bartelheimer and Poschlod 2016), and the integration of trees in productive systems improve SQ and nutrient availability; since increasing soil nutrient thresholds is a cost-effective and fast-efficient tool to assess SQ (Kisaka et al. 2023).

Soil services and their biota can sustain ecosystem biodiversity and support species heterogeneity and the presence of some unique species while providing genetic diversity several orders of magnitude greater than that found above ground (Parker 2010; Cooke et al. 2023). The abundance and diversity of the other organisms are generally positive for SQ and act as a buffer against climate change (Certini and Scalenghe 2023).

Soils are severely threatened, affecting yield and plant composition (Drobnik et al. 2018; Diti et al. 2020), and their rates of formation and recovery are far from balancing current rates of loss and degradation (Bindraban et al. 2012; Pulleman et al. 2012; Machado et al. 2019). Current practices of conservation and restoration of fragile ecosystems are an alternative to recover productivity through increased water and nutrient use efficiency, while the benefits that could be obtained from this restoration have rarely been evaluated (Zhao et al. 2005; Tittonell et al. 2012; Lewis et al. 2022). Such benefits are a renewed impetus to restore and return forests to their ecological function, protect species, sequester carbon, and secure livelihoods (Banin et al. 2022). The implementation of restoration programs in tropical environments restores soil biology and soil organic carbon and favors the early recovery of microbial communities (Bonner et al. 2020).

On the other hand, protected areas are a key conservation tool, although their effectiveness in maintaining biodiversity over time has yet to be quantified (Cooke et al. 2023). Global efforts are needed to define sustainable practices applicable in specific agroecological zones and food landscapes (Rumpel et al. 2023). Betancur-Corredor et al. (2023), Delgado et al. (2019), Lewis et al. (2022), Martinez and Zinck (2004) and Martinez et al. (2020), underline the difficulty and necessity of restoring degraded soils in the Andes and tropical savannas and the possibility of using microorganisms, reactivation of biochemical fluxes, transformation of degraded areas into technosoils or through other biotechnological alternatives Barrios et al. (2006) proposed technical indicators of agroecosystem change for farmers to assess unobservable changes in soil properties before they lead to more severe and visible forms of degradation. The use and evaluation of physical, chemical, and biological soil indicators based on land use dynamics in Colombia have been implemented in grazing systems by Teutscherová et al. (2021), degraded areas, agricultural and forestry systems (Velazquez et al. 2007; Machado et al. 2019), and forest restoration (Lozano-Baez et al. 2021). However, an integrated SQ indicator in different use systems is still lacking. The main objective of the present study was to evaluate SQ as an indicator of natural restoration and soil remediation. It established a baseline for physical, chemical, and biological indicators of Soil Quality under the five most common use scenarios in Andean ecosystems.

Scope

To evaluate soil quality as an indicator of recovery and natural restoration processes in the Civil Society Natural Reserve “La Montaña Mágica-El Poleo” (MMEP) in Zapatoca, Santander, Colombian Andes. Second, to analyze the soil's physical, chemical, and biological characteristics to establish its quality according to the land uses.

Methodology

Soil quality status was determined from three types of analysis: chemical, physical and biological. Five coverages were sampled: secondary forest (SF), natural regeneration (NR), active restoration (AR), cropland (CL), and degraded soil (DS) at the Civil Society Nature Reserve (CSNR) “La Montaña Mágica-El Poleo” at Zapatoca (Santander), based on Aguilar-Garavito and Ramírez (2015), Arshad and Martin (2002), Instituto Geografico Agustin Codazzi (IGAC) et al. (2006) and Pérez (2010) methodologies.

Study area

The CSNR is located in Zapatoca, Santander, Colombia (Fig. 1) in the buffer zone of the Serranía de los Yariguíes National Natural Park (06°50′05.4″ N latitude and 73°18′06.3″ W longitude). It is an 86.1 ha of premontane rainforest area between 1850 and 2300 m above sea level. Precipitation reaches 4000 mm y-1, with a bimodal regime (Espinal and Montenegro 1977). The mountains have steep to moderately steep slopes (slope > 50%) and moderate to severe erosion; soils derived from sandstones and shales are moderately acidic. The predominant soils have low pedogenetic development and, according to SSS (2022a) and WRB (2022), correspond to Typic Dystrudepts Sandy Loam Mixed Subactive Acid Hyperthermic (CAMBISOLS Endoleptic Dystric Arenic Loamic Ochric Raptic), Lithic Quartzipsamments Sandy Loam Mixed Subactive Acid Hyperthermic (REGOSOLS Epileptic Yermic Dystric Arenic Loamic Ochric) and Humic Psammentic Dystrudepts Sandy Loam Mixed Subactive Acid Hyperthermic (CAMBISOLS Dystric Arenic Loamic Ochric Raptic).

Fig. 1
figure 1

Map of the study area of the CSNR “La Montaña Mágica-El Poleo” in Santander, Colombia. It shows the land use coverages: cropland (CL), natural regeneration (NR), degraded soil (DS), active restoration (AR) and secondary forest (SF)

Full size image

Experiment design

Soils were studied in secondary forest (SF), natural regeneration (NR), active restoration (AR), cropland (CL), and degraded soils (DS) with 25, 11, 6, 25, and 25 years of age, respectively. Sampling was carried out from October 2021 to February 2022 in five coverages with four replicates per coverage in composite samples (10 subsamples) after removing the surface layer of vegetation and leaf litter (sensu IGAC et al. 2006). Each sample was homogenized and packed in a polyethylene bag for transport to the laboratory (Aguilar-Garavito and Ramírez 2015).

Chemical, physical and biological properties were analyzed at the Soil Science Laboratory—National University of Colombia at Medellin Campus. The status of each cover was estimated through 19 selected soil quality indicators.

Chemical, physical and biological characterization

Chemical analysis

We determined the following parameters: pH, organic carbon (OC), nitrogen (N), phosphorus (P), aluminum (Al), calcium (Ca), magnesium (Mg) and potassium (K) and cation exchange capacity in meq/100 g soil (CEC) using the methodologies of SSS (2022b); the main methods are summarized in Table 1.

Table 1 Chemical analysis of soils; methods and techniques applied at the Soil Science Laboratory—National University of Colombia at Medellin
Full size table

Physical análisis

Bulk density in (Bd), % porosity (Po), cm/hr infiltration (I), cm/hr percolation (Pe), penetration resistance (Pr), and texture (%) were determined (see Table 2).

Table 2 Soil physical analysis: methods and techniques applied according to IGAC et al. (2006) and Pérez (2010)
Full size table

Biological analysis

Macrofauna was collected in 1 m2 per sample, the litter layer was removed, and the surface layer (5 cm thick) was extracted, collected, and deposited on a 5 × 5 mm sieve for approximately three minutes (Krell et al. 2005). Each sample was mounted in a Winkler bag in the laboratory for 48 h. Macroinvertebrates were cleaned and stored in plastic containers with alcohol, individuals were morphologically described, and taxon assignment (family level) was based on morphological diagnosis (Krell et al. 2005) with a 30X Leica ES2 ® stereoscope. The result is expressed in the number of individuals per m2 (In) and families per m2 (Fn).

Statistical analysis

Statistical analyses were performed with R software. The Gaussian assumptions of normality and homoscedasticity were tested with the Shapiro–Wilk test (Shapiro.test) and the Levene test (Levene test) with a significance level of 0.01%. When these were not met, transformations were performed. A principal component analysis was performed to approximate the data's nature and a correlation matrix (corrplot) to know the relationships between parameters of different natures (physical, chemical, and biological). A one-way analysis of variance (ANOVA, aov) was applied to determine which hedges differed from Dunnet's test (DunnettTest). Both tests were analyzed at a significance level of 0.05% (null hypothesis in both cases: the means of the coverages are equal).

Results

The results of the nineteen variables assessed in the five analyzed coverages are presented in Table 3. The variation collected from PC1 and PC2 was 45.1% and 16.4%, respectively (Fig. 2). The grouping of the data in the principal component analysis showed two extreme land covers (SF and DS) that, given their characteristics, presented low standard variances. SF and DS coverages presented lower CL, AR, and NR variance. Bd, Al, and Pr had greater weight in PC1 positive towards DS, unlike pH, OC, N, Ca, In, and Fn, which had greater weight in PC1 negative towards SF. The variables with the greatest weight in PC1 were In, Fn, Ca, Mg, K, OC, N CEC, AL, and pH. In PC2, the variables with the greatest weight were K, texture, and Pr. The variance shows that CL, AR, and NR coverage is intermediate between DS and SF (Fig. 2).

Table 3 Results by cover
Full size table
Fig. 2
figure 2

Principal component analysis (PCA). Al: aluminum; Ca: calcium; K: potassium; Mg: magnesium; N: nitrogen; OC: organic carbon; P: phosphorus; pH; Clay; Sand; Silt; Bd: bulk density; Po: porosity; CEC: cation exchange capacity; Pr: penetration resistance; I: infiltration; Pe: percolation; Fn: families number; In: individuals number. Land use coverages were defined as cropland (CL), natural regeneration (NR), degraded soil (DS), active restoration (AR), and secondary forest (SF)

Full size image

In the matrix of Fig. 3, it is possible to distinguish a trend of negative correlation in red tones between texture, Bd, Al, and Pr, especially with Ca, Mg, K, and pH. A positive correlation was evidenced between N and OC with In and Fn.

Fig. 3
figure 3

Correlation matrix. Correlation between physical, chemical, and biological soil variables. The color bar indicates the direction and intensity of the correlation. Colors in the blue band signify a positive correlation, and those in the red band indicate negative correlations. Al: aluminum; Ca: calcium; K: potassium; Mg: magnesium; N: nitrogen; OC: organic carbon; P: phosphorus; pH; Clay; Sand; Silt; Bd: bulk density; Po: porosity; CEC: cation exchange capacity; Pr: penetration resistance; I: infiltration; Pe: percolation; Fn: families number; In: individuals number

Full size image

Analysis of variance tools and Dunnett's test

Dunnett's test showed no significant differences between active restoration (AR) and natural regeneration (NR) cover and secondary forest (SF). These treatments showed better soil conditions through the restoration processes. For AR, NR, and SF, Fn Bd and Pr had similar behavior. Active restoration (AR) did not differ significantly from the secondary forest (SF), while natural regeneration (NR) differed in pH, Al, and K concentration. AR showed better behavior than NR in these parameters. Crops (CL) did not show significant differences in pH, Al, or K concerning secondary forest (SF) associated with mechanical work or tillage on CL.

The pH (Fig. 4A) showed significant differences between SF, DS, and RN. Aluminum (Al) content (Fig. 4B) showed significant differences between SF, DS, and NR, and neither between AR and CL. An inverse relationship between pH and Al content was observed; cover crops with acid pH (< 5.0) also showed high Al levels (Fig. 4).

Fig. 4
figure 4

Analysis of variance between coverages, AR: active restoration, CL: cropland, DS: degraded soil, NR: natural regeneration, SF: secondary forest. A pH value (pH), p-value ANOVA: 0.000323 ***; B Aluminum (Al) content, p-value ANOVA: 7.84e-06 ***. The letters a and b in each analysis result from Dunnett's test, where all the land uses were compared with a control (SF). a: no significant difference; b: significant difference

Full size image

Regarding organic carbon (OC) and nitrogen (N) content (Fig. 5A and B), SF showed significant differences concerning NR, DS, CL, and AR. Regarding potassium (K) content (Fig. 5C), SF did not show significant differences to CL and AR but did show significant differences with NR and DS.

Fig. 5
figure 5

Analysis of variance between coverages, AR: active restoration, CL: cropland, DS: degraded soil, NR: natural regeneration, SF: secondary forest. A Organic carbon (OC) content, p-value ANOVA: 9.86e-06 ***; B Nitrogen (N) content, p-value ANOVA: 4.19e-05 ***; C Potassium (K) content, p-value ANOVA: 0.000236 ***. Letters a and b in each analysis result from Dunnett’s test, where all covers are compared with a control (SF). a: no significant difference; b: significant difference. The levels of N and K had the same behavior as CICE

Full size image

Bulk density (Bd) and penetration resistance (Pr) (Fig. 6A and B) showed significant differences between SF and DS and non-significant differences with NR, CL, and AR. Regarding the number of individuals (In) and families (Fn) of soil macroinvertebrates (Fig. 6C and D), SF presented significant differences to DS and CL and non-significant differences for NR and AR.

Fig. 6
figure 6

Analysis of variance between coverages, AR: active restoration, CL: cropland, DS: degraded soil, NR: natural regeneration, SF: secondary forest. A bulk density (Bd), p-value ANOVA: 0.00805 **; B penetration resistance (Pr), p-value ANOVA: 7.51e-05 ***; C number of individuals (In), p-value ANOVA: 0.0153 *; D number of families (Fn), p-value ANOVA: 0.011*. Letters a and b in each analysis result from Dunnett's test, where all covers are compared with a control (SF). a: no significant difference; b: significant difference

Full size image

The increase observed in Bd for all canopies was the response to a decrease in the number of individuals (In) and families (Fn) of soil macroinvertebrates. The highest values of Bd and Pr and the lowest values of individuals (In) and number of families (Fn) of soil macroinvertebrates were found in DS. The SF and NR coverages presented the highest values of In and Fn. The NR coverage approached the SF conditions for all parameters and moved away from DS (Fig. 6).

The infiltration analysis (I) (Fig. 7A) showed significant differences between SF and the other land uses (AR, CL, and DS). Percolation (Pe) (Fig. 7B) in SF showed significant differences with AR, DS, and NR and no significant differences for CL. The SF cover presented the highest infiltration (I) and percolation (Pe) values. The lowest values of infiltration (I) were found in NR and those of percolation (Pe) in AR (Fig. 7).

Fig. 7
figure 7

Analysis of variance between coverages, AR: active restoration, CL: cropland, DS: degraded soil, NR: natural regeneration, SF: secondary forest. A infiltration (I), p-value ANOVA: 0.0937; B percolation (Pe), p-value ANOVA: 0.0515. The letters a and b in each analysis result from Dunnett's test, where all the coverages are compared with a control (SF). a: no significant difference; b: significant difference

Full size image

Discussion

Bd, Al, and Pr are associated in the direction of SD. The increase in Al is associated with high acidity and limitation in developing vegetation and edaphic fauna (Forero 2011; Rahman and Upadhyaya 2021). In tropical soils, acidification is a natural process closely related to low base contents and soil impoverishment processes that affect soil fertility (Quinto-Mosquera et al. 2022). High Pr and Bd values are related to low soil quality and decreased macroinvertebrate populations (Capowiez et al. 2021; Lin et al. 2022), associated with a high degree of compaction that limits root growth and distribution and vegetation regeneration (Acton et al. 2011); at the same time affecting the porous soil system and air and water exchange. Basamba et al. (2006) stress the importance of biophysical measurements and bulk density and porosity as predictors of system performance and health. The pH, OC, N, Ca, In, and Fn parameters were associated in the direction of SF. In tropical forests, soil nutrient availability, organic matter content, and intense physicochemical changes in short periods condition vegetation development (Martínez et al. 2020; Quinto-Mosquera et al. 2022); at the same time, it has a direct effect on pH nutrient availability, SOM, CEC, and soil compaction (Betancur-Corredor et al. 2023). The variance in this analysis showed that the AR and NR covers are intermediate between DS and SF, suggesting the effectiveness of the restoration processes. The ecosystems were resilient and maintained fertility due to improved soil conditions (Chauveau 2015; Chapin et al. 2012).

The AR and CL covers did not show significant differences with SF in pH, K, and Al contents, contrary to the NR and DS covers, which showed significant differences. These relations suggest an improvement of Bd under AR and NR. This response may be associated with higher root activity, organic matter dynamics, and higher pedogenetic activity (Basamba et al. 2006; Betancur-Corredor et al. 2023) associated with native forets. According to Quinto-Mosquera et al. (2022), nutrient balances in tropical forests are not always conditioned by the limiting factor but by elements in optimal amounts. Tejnecký et al. (2020) suggest that forest composition and management practices influence element dynamics, rhizosphere content, and biogeochemical behavior of ecosystems.

CL did not present significant differences in pH with SF due to manual liming in cultivation, which can cause alkalinity while soil pH increases (Fageria and Baligar 2008). However, due to leaching in tropical conditions, Ca, Mg, and other nutrients, added via amendments, quickly leave the system. For Schlesinger (1997), leaching dynamics are more active than nutrient release from parent materials under high rainfall conditions. In and Fn of AR and NR did not show significant differences with SF, in contrast to CL and DS, probably due to changes in pH. Such changes can influence soil communities (Griffiths et al. 2011), whose macroinvertebrates can respond rapidly in the recovering ecosystem (Schloter et al. 2017). They were, therefore, a good indicator of SQ because changes in their population dynamics were associated with physicochemical in the soil and use dynamics (Velásquez and Lavelle 2019). pH and available nutrients may be important for detecting initial changes in the ecosystem in the short term (Tibbett et al. 2019).

Bd and Pr in NR, AR, and CL were not significantly different from SF; however, DS did show significant differences from SF. The results for AR and NR suggest a reduction of Bd in restoration, coinciding with Jiao et al. (2011) and Neris et al. (2012). Decreased Bd indicates improvements in structure and porosity (Li and Shao 2006) and soil quality at different successional stages (Martinez and Zinck 2004). The Bd values in CL were due to the chemical and manual treatment performed on the cover. Accumulation of leaf litter in coffee plantations increases physical quality, and vegetative cover improves soil conditions (Da Rocha Junior et al. 2020). Bd values in SD were associated with degradation, structure and porosity losses, water behavior, and soil aeration (Acevedo and Silva 2003; León et al. 2013). A direct effect was seen in the availability and efficiency of nutrients in the cycle with plants, animals, and the atmosphere, which made them more susceptible to erosion (Magdoff and Van Es 2021).

SF presented high values of N and OC, resulting in significant differences compared to NR, AR, CL, and DS, results that were due to the slow accumulation of N. Even in developed forests, the N cycle, characteristic of a mature tropical forest, may not be reestablished (Amazonas et al. 2011; Davidson et al. 2007; Vitousek et al. 1989). The relationship between OC and forest age is not always direct and may only present changes in 20 years of development (Marin-Spiotta et al. 2009; Neumann-Cosel et al. 2011). The results of León et al. (2013), Martínez et al. (2020), and Betancur-Corredor et al. (2023) for neotropical conditions in Colombia showed reactivation of soil nutrients and biogeochemical cycles associated with litterfall and high nutrient returns of N.

Infiltration (I) in AR, CL, and DS did not show significant differences to SF, while NR presented significant differences in I. This response is associated with the parent materials; the different coverages studied come from sandstone rocks (IGAC 2011). Heilweil et al. (2007) conducted hydrological studies in soils developed from sandstone, which showed a high net infiltration rate, mineralogy, and texture, influencing the soil infiltration rate (Wakindiki et al. 2002). In the study, site infiltration had high variability; which will be affected by soil heterogeneity and management, organic matter content, and soil porosity (Helalia 1994; Connolly et al. 1997; Heilweil et al. 2007; Leite et al. 2018). The importance is related to the direct relationship between parent material, texture, and porosity behavior (Gray et al. 2016; Araujo et al. 2017; Bonfatti et al. 2020) and responds to previous land use, soil type, and vegetation (Helalia 1994; Zimmermann et al. 2006). Soil infiltration (I) in SF was associated with an improvement in soil physical structure related to a higher organic matter content (Franzluebbers 2002; Teutscherová et al. 2021), a more developed root system (Basamba et al. 2006; Pohl et al. 2009) and higher macroinvertebrate activity (high pedoturbation) in the soil (Colloff et al. 2010). AR, DS, and NR presented significant differences for Pe in SF and did not present significant differences with CL. Plowing processes and improved mechanization in CL may explain the Pe conditions (Acevedo and Silva 2003; Basamba et al. 2006).

Conclusions

Significant differences in soil quality (SQ) were related to land uses for the study site. The soil's physical, chemical, and biological properties varied among the five evaluated land uses, indicating that the use significantly affects the properties and behavior of soils. Active restoration (AR) and natural regeneration (NR) effectively recover soil quality for Bd, Pr, In, and Fn. AR is more effective than NR in reducing acidity and aluminum concentration and increasing K in the soil. The number of individuals and families of macroinvertebrates showed significant differences according to land use in short periods, which suggests that these parameters will be a good indicator of soil quality. Infiltration was variable at the surface and showed significant differences at depth, influenced by parent material and affected by soil characteristics and land use. Behavior did not change significantly for AR, NR, OC, N, Ca, Mg, and CEC. It is important to highlight the importance of the species adaptability to the environment and not to confuse soil fertility with exuberance; under tropical conditions, plant species have a close relationship with SQ. For future studies, it is recommended to deepen the evaluation of soil biological diversity and its role in the behavior of SQ indicators.