Introduction

The Araucaria Forest is a unique ecosystem with a mix of broad-leaf and coniferous species that is restricted to Southern Brazil and northeastern Argentina in whose canopy is dominated by the presence of Araucaria angustifolia (Bert.) Kuntze (Mallmann et al. 2018). As result of the history of continuous anthropogenic intervention in the AF, including extensive exploitation of timber species with high economic value and agricultural and urban expansion, primary forests have been reduced dramatically, while secondary forests currently cover around 20–25% of the original area (Vibrans et al. 2011; Lacerda 2016).

Some authors have argued that the maintenance of forest fragments on private properties (i.e., outside protected areas) is related to traditional agroforestry systems established throughout the Araucaria Forest, particularly caívas in Northern Santa Catarina State and faxinais in Southern Paraná State (Mello and Peroni 2015; Marques et al. 2019; Lacerda et al. 2020). Currently, caíva systems cover approximately 140,000 hectares across the region, with a heterogeneous density and composition of the tree, shrub, and herbaceous strata, allowing for different modes of production (Hanisch et al. 2010). Araucaria Forest biodiversity is also a source of a range of non-timber forest products (e.g., pinhão,Footnote 1 and erva-mateFootnote 2), and its understory is used for animal husbandry (forage and shade), all of which support household food security and income for small-scale farmers (Mello and Peroni 2015).

Although livestock are an important element of caíva systems, there is evidence that their presence does not have the negative impacts on tree diversity that is generally expected in sustainability and conservation analyses (Souza et al. 2010). Previous studies have shown that there are high levels of tree species richness (average of 40 species) with a density ranging from 220 to 1300 trees per hectare with DBH > 0.05 m (Hanisch et al. 2010; Mello 2013; Pinotti et al. 2018), which confirms the importance of this traditional silvopastoral system for forest conservation. Marques et al (2019) classified caívas into three types based on tree density: closed caívas, open caívas, and very open caívas.

In these systems, cattle generally graze on natural pastures that consist of the herbaceous stratum, with no grazing control or soil fertilization. Consequently, pasture yields are low and cease altogether during the autumn and winter months, which results in a low stocking rate (0.35 cow/ha) (Hanisch et al. 2014a). Such a situation is not economically attractive for small-scale farmers, resulting in pressure on caíva farmers to substitute this system with other more profitable alternatives, such as reforestation with exotic species or annual commodity crops (Lacerda et al. 2020).

In this context, research is needed to develop appropriate technologies with farmers who have conserved AF through historical land use such as caivas. Combining high-levels of biodiversity conservation with increased yield in traditional silvopastoral systems can be feasible through moderate use of inputs and the adoption of technologies based on the maximization of biological factors (e.g. photosynthesis and nutrient cycling), selection of pasture species, and grazing control (Feldhake et al. 2010; Clough et al. 2011; Calle et al. 2012; Dhanya et al. 2014; Lessard-Therrien et al. 2017; Keeley et al. 2019). These different technologies can be adopted without compromising natural tree species regeneration or soil integrity, particularly in terms of carbon cycling and other soil indicators (Castañeda et al., 2014; Feliciano et al. 2018; Sá et al. 2017).

Undoubtedly, the allocation of areas solely for ecosystem preservation is a valid strategy for biodiversity conservation worldwide (Alvarado et al., 2018) at least in the short term (Hill et al., 2015). However, the role of traditional systems in landscape conservation cannot be minimized as this may penalize farmers who have preserved forests on their land through use rather than converting them into less biodiverse land uses such as conventional agriculture (Michon et al. 2007; Steenbock et al. 2011; Fantini et al. 2017; Marques et al. 2019).

Proposals to improve caívas in terms of pasture yield must be based on a wider understanding of how natural systems that have been subjected to continuous modification through use as productive systems are maintained over time. For example, would practices such as pasture fertilization, controlled grazing with paddocks, and introduction of more productive pasture species negatively affect tree regeneration? Or would such strategies improve regeneration? Could an increase in pasture yield through fertilization reduce the soil carbon stock of these forest areas? Or could greater biomass and root mass of the pasture contribute to increases in the carbon stock? This interaction of factors can be better understood by evaluating the dynamics of the system’s components over time.

From this perspective, the use of two different strategies to intensify pasture use in caíva agroforestry systems were analyzed in terms of pasture yield and quality, basic soil indicators, and forest regeneration. The first strategy included winter overseeding of cultivated pastures on natural pasture in caívas, while the second was replacing natural pasture with giant missionary perennial pasture, a native species from Southern Brazil that has shown adequate productivity and persistence in silvopastoral systems (Baldissera et al. 2016). The results are discussed in comparison to traditional caiva management, to provide a basis for understanding how to support the conservation of AF remnants as well as achieve higher productivity in caivas or similar agroforestry systems.

Materials and methods

The study area is located in the Northern Plateau of Santa Catarina State, Southern Brazil (26o16’–26o33’ S; 50o32’–50o91’ W) (Fig. 1). The experiment was undertaken in eight 0.5 ha plots on six farms in which caíva systems are used. On two properties with larger caívas, we installed two plots in each caíva. On the other four farms, only one plot per caíva was established (mean size of caívas = 3.3 ± 1.2 ha). All eight caíva plots occurred at a similar range in elevation (770–815 m above sea level) and within the subtropical humid climate zone, which is characterized by frequent frosts in the winter, mean annual temperature of 17.6 °C, ranging from 14 °C in July to 21 °C in January, and 1400 mm average annual rainfall (Vibrans et al. 2013). Climate monitoring showed that, during the period of the experiment, the average temperature and rainfall were similar between these sites and within the expected for the region. The eight study sites were located on flat to slightly undulating terrain, deep and well-drained soil (classified as Oxisol; Embrapa 2013) and occurs within the range of the Araucaria Forest, Atlantic Forest biome. Soil analyzes carried out before the beginning of the experiment showed that had very similar soil composition, with a clay loam texture (32 to 41% clay content),

Fig. 1
figure 1

Location and aerial view of the studied caívas (n = 8) and the respective pasture management strategies tested to evaluate pasture yield (Santa Catarina State, Southern Brazil; 26o16’–26o33’ S; 50o32’–50o91’ W). TC = traditional management, WOC = winter overseeding of cultivated pastures, GMC = giant missionary perennial pasture

Full size image

These caívas had similar tree canopy cover (76 ± 8% = average ± SD measured using Spherical Crown Densiometer) and have been used for the last 50 years for animal husbandry (dairy cattle), with selective tree harvesting for domestic fuelwood and erva-mate production, and no history of forest clear cutting.

Among the eight 0.5 ha plots established in eight caivas, three were maintained with traditional management practices (TC), three with winter overseeding of cultivated pastures on natural pasture (WOC), and two with natural pasture replaced by giant missionary evergreen pasture (GMC)Footnote 3 (Fig. 2). The treatments were characterized by the use of different agronomic practices such as fertilization, overseeding, mowing, and others.

Fig. 2
figure 2

Overview of a traditional caíva (TC) in Summer (a) and in Winter (b), and caívas with the adoption of two pasture management strategies: GMC (c) and WOC (d), with dairy cattle grazing. Santa Catarina state, Brazil

Full size image

The giant missionary perennial pasture (Axonopus catharinensis Valls) cv. SCS315 Catarina-gigante was planted manually in October 2014 after natural pasture has been eliminated using a desiccant herbicide. No soil tillage was used and seedlings were spaced 0.50 × 0.50 m. After planting, the area was fenced for eight months.

Prior to establishing the experiment, soil samples (0–10 cm deep) were taken for nutrient and pH assessment. In WOC and GMC plots, dolomitic lime was applied directly onto the soil to achieve a target pH of 5.5; fertilizer was applied twice a year (Abril and October) to increase soil nutrients to an average fertility level based on the recommendations for Warm Season Grasses (CQFS RS/SC, 2004). Gafsa's natural phosphate was used as a source of phosphorus and urea as a source of nitrogen (100 kg N.ha.year−1). The lime was divided in two applications (April and October 2015) to avoid potential damage to the roots of native trees, as they are naturally adapted to the low soil fertility of caívas. In 2015 and 2016, Lolium multiflorum (30 kg.ha−1) and Vicia sativa (10 kg.ha−1) were overseeded manually in April and May for WOC and GMC treatments, respectively.

In all caívas, the herbaceous layer was periodically trimmed by farmers using a motorized brush cutter or a scythe, at an average height of 0.10 m. This is common practice in caívas and is associated with reducing invasive species and to facilitate the harvesting of erva-mate (Mattos 2015).

Grazing by animals in the WOC and GMC plots was managed rotationally. Animals entered the paddocks when the pasture reached 0.30 m in height and remained until the forage in each paddock was reduced to about a 0.10 m in height within one day. Thus, the time intervals between grazing for each strategy was variable. Animal response data was not analyzed, thus, for this analysis animals are considered as forage trimmers (Feldhake et al. 2010).

Photosynthetically active radiation (PAR) was measured in September 2016 in each caíva, between 11:00 and 13:00, with clear skies, using a portable digital PAR meter and LightScout® photosynthetic photon flux (PPF; μmol.m−2.s−1) at 1 m above ground level.

Pasture yield and quality

The pasture yield was assessed continuously from July 2015 to November 2016. Herbage mass (HM, kg.ha−1 of dry matter, DM) in WOC and GMC plots was estimated from random 0.25 m2 samples (n = 10) cut with pruning shear and collected before animals entered the plots (when pasture reached 0.30 m in height). Prior to each cutting, ten pasture height measurements were taken using a sward stick.

In traditional caíva plots without control of animal grazing, HM was determined based on herbage material collected from three grazing exclosures in each 0.5 ha plot, whenever the pasture reached 0.15 m in height. The grazing exclosures were reallocated at the end of each grazing period to avoid the long-term effects of maintaining the exclosure.

All pasture samples were dried at 60 °C for 72 h and then weighed. The HM data of each cutting was grouped together for total HM.

Samples harvested during the Spring–Summer 2015 were ground to pass through a 1 mm screen and analyzed for crude protein (CP), dry matter digestibility (DMD), neutral detergent fiber (NDF), and acid detergent fiber (ADF) using near-infrared reflectance spectroscopy (NIRS) and a spectral range of 1100–2500 nm (FOSSNIR Systems 5000; CEPA Laboratory, Passo Fundo/RS, Brazil).

Soil indicators

Twenty soil sub-samples were randomly collected, in March 2015 and February 2017, from each caíva plot at a depth of 0.00 to 0.10 m. For each collection, the soil sub-samples were grouped to form three composite soil samples per caíva plot. After collection, each composite sample was sent to the laboratory for the following analyses: total organic soil carbon (wet oxidation with potassium dichromate and sulfuric acid); clay content (Densitometry); pH (Potentiometry); soluble phosphorus (P) content (Mehlich-1/Calorimetry); exchangeable ion (Al3+); base saturation (V%); and soil organic matter (Atomic Absorption Spectrophotometry).

Forest regeneration

After the establishment of the 0.5 ha plots, a floristic survey of trees (height > 1.30 m and diameter at breast height (DBH) ≥ 0.05 m) was carried out (see Hanisch et al. 2016; Pinotti et al 2018).

Tree species regeneration data was obtained from 18 plots of 4 m2, that were distributed evenly throughout each of the eight 0.5 ha plots. The objective of these plots was to evaluate the effects of cattle grazing on the dynamics of forest regeneration in the caívas. The plots were not fenced, thus allowing free access for cattle grazing, but were marked with wooden stakes (0.10 m high) painted red to ensure visibility and to avoid mowing within these areas.

Forest regeneration was assessed in each plot at 0, 7, 13, and 20 months after experiment implementation. For each evaluation, all tree seedlings (from 0.05 m up to 1.30 m high) were tagged and identified (Flora do Brasil 2016) to evaluate richness (total number of species) and density (number of individuals/ha). In each survey, the botanical identification was carried out in loco. When this was not possible, samples were collected for later confirmation and curation at the Escola de Florestas de Curitiba herbarium. The regeneration data were used to quantify the structural parameters of density and frequency (Mueller-Dombois and Ellenberg, 1974), which were then used to calculate Shannon diversity index (H') and Pielou evenness index (J’). The rate of natural regeneration (Jardim, 1986) was also calculated to verify if there was densification, mortality, or stability in the number of regenerating individuals over time.

Data analysis

Prior to final site selection, a covariance analysis was performed with their location (latitude and longitude) as covariates to rule out spatial effects. As no effects were observed for either latitude (p = 0.840) or longitude (p = 0.0921), the eight sites were accepted for installation of the experiment.

Data from the WOC (n = 3) and GMC (n = 2) treatments were compared with data from traditional pasture management (TC; n = 3) using ANOVA with repeated measures over time. All results were tested for normality (Shapiro–Wilk) and homogeneity of variance (Levene’s test). When necessary, the data were log-transformed to normalize the residuals (pasture yield and density of regeneration). Means were compared using Fisher's test at 5% significance. The analyses were performed with the support of R, version 3.3.2 (R Core Team, 2016).

Results

Pasture yield and quality

The total pasture yield (from July 2015 to November 2016) was significantly higher in WOC and GMC management strategies compared to the caívas with traditional pasture management (TC) (Table 1). The pasture quality did not differ between management strategies for crude protein (CP), digestibility (DMD), and acid detergent fiber (ADF), although higher values for neutral detergent fiber (NDF) were found in TC.

Table 1 Pasture yield and chemical composition (mean ± standard deviation) in caívas submitted to three pasture management strategies: GMC = giant missionary perennial pasture; WOC = winter overseeding of cultivated pastures; TC = traditional management
Full size table

The pasture yield of natural pasture in TC was almost nil for nearly seven consecutive months (April to October). Meanwhile, during its growth period (November to March), total production was around 1000 kg.DM.ha−1 (Fig. 3), a value considered low and below the minimum recommended for animal production, which is 1200 kg.DM.ha−1 (Hodgson, 1990).

Fig. 3
figure 3

Distribution of pasture yield (kg DM.ha−1) throughout the evaluation period in caívas submitted to three pasture management strategies: GMC = giant missionary perennial pasture; WOC = winter overseeding of cultivated pastures; TC = traditional management (Santa Catarina State, Southern Brazil; 26o16’–26o33’ S; 50o32’–50o91’ W). HM = herbage mass. Bars followed by the same letters, on the same cut-off date, do not differ from each other by Tukey's test (5%)

Full size image

The most significant increase in pasture yield occurred in the GMC plots, indicating that replacing natural pasture can be an effective strategy to increase animal production in caívas. However, during the second winter period (June to September/2016) there was a significant reduction in pasture yield in this treatment because of continued vegetative growth of the giant missionary perennial pasture which delayed the winter overseeding (Fig. 3).

Soil indicators

The studied soil variables in WOC and GMC were affected by the use of corrective treatments and fertilizers, as indicated by an increase in soil pH, soluble phosphorus content, and base saturation (V%), and a reduction in exchangeable aluminum compared to TC in the first year (Table 2). These results confirm the efficiency of top-dressing fertilizer application in perennial pastures and the partitioning of the total dose for soil correction. However, we found a significant reduction in soil pH and V% over the two-year study period in WOC and GMC, which clearly indicates that annual fertilization for both strategies is necessary.

Table 2 Soil variables in caívas subjected to three pasture management strategies at the beginning (2015) and at the end (2017) of the experiment: GMC = giant missionary perennial pasture; WOC = winter overseeding of cultivated pastures; TC = traditional management (Santa Catarina State, Southern Brazil; 26o16’–26o33’ S; 50o32’–50o91’ W)
Full size table

Soil organic matter (SOM) content did not differ between treatments and remained above 4% until the end of the experiment. On one hand, this confirms the importance of traditional caíva systems for soil carbon conservation, since they present high carbon stock levels; while on the other, the results indicate that the two proposed strategies do not negatively affect carbon stock in the medium term. The TOC results followed the same trend as SOM, between treatments and over time. TOC did not differ between GMC and TC in the two evaluation periods, although the forage yield was six times greater in GMC, indicating that pasture management by grazing height contributed to the maintenance of C in the soil.

Forest regeneration

Pasture management strategies show no significant differences in relation to species diversity (S) or forest regeneration density (D) (Table 3). The mean density of trees in TC, WOC, and GMC treatments was 575, 447, and 400 trees per hectare, respectively. Although not statistically significant, the values were dissimilar with high standard deviations. This result is expected as the caívas have different management histories in terms of forest species and pressure from land use in the surrounding region. Finally, significant differences in relation to rate of natural regeneration (NRR) were found, suggesting an impact of the alternative treatments (GMC and WOC) on forest regeneration.

Table 3 Means and standard deviation of species diversity (richness, S) and density (D), Shannon diversity index (H’), Pielou's evenness index (J’), natural regeneration rate (NRR) of individuals, and photosynthetically active radiation (PAR) in caívas with three pasture management strategies: GMC = giant missionary perennial pasture; WOC = winter overseeding of cultivated pastures; TC = traditional management
Full size table

Discussion

Pasture

Shifts from low input agriculture to intensified systems tend to reduce functional and species diversity (Cotter et al. 2017). However, in the caívas studied herein, the adoption of techniques that are universally accepted as sustainable for livestock production enabled increases in pasture yield and maintained active tree regeneration in the areas, an important indicator for sustainability. These techniques include: (1) introduction of a productive and regionally adapted pasture to replace natural pasture; (2) increasing pasture botanical composition with winter overseeding of cultivated pastures; (3) adequate soil management through fertilization and soil correction; and (4) grazing control.

Increases in pasture yield occurred despite the inherent restrictions of traditional silvopastoral systems, particularly low initial soil fertility (see TC soil characteristics in Table 2) and excessive shade due to the occurrence, on average, of more than 400 native trees ha−1. Similar results have been observed in other studies on natural pastures and silvopastoral systems with native trees (Calle et al. 2012; Vallejo et al. 2012; Marty 2015; Lessar-Therrien et al. 2017) where the intensification of the system occurs with moderate use of inputs.

The WOC total pasture yield was more than three times greater than in TC although most of the increase took place in late winter and spring which corresponds to the annual growth period of overseeded forage. This demonstrates that the performance of natural grasses typical of TC did not improve despite the application of fertilizer and lime to increase soil pH and nutrients; in fact, natural grass development during the warmer months of the year was low and inconsistent (Fig. 3).

On the other hand, total pasture yield in GMC was consistently much higher throughout the first year, confirming the potential use of giant missionary grass in shaded systems. Its lowest levels of production occurred during winter in the second year of the study (July and August 2016) which is likely related to the delay in overseeding. However, this continued growth is a characteristic of giant missionary perennial pasture that has also been observed in areas under full sun; in the autumn months during years with mild temperatures, this forage continues to grow, thus delaying the practice of overseeding and, consequently, the use of the area for winter grazing. Subsequent research in caívas with giant missionary perennial pasture should include intercropping with perennial legumes such as Trifolium sp, which in studies under full sun conditions has helped to increase productivity in periods of transition (Hanisch et al., 2014b).

Furthermore, it is possible that the potential to increase forage is even greater in this strategy, since the yield in GMC was 40% lower than that observed in full sun conditions for this forage species (Miranda et al. 2012; Baldissera et al. 2016). In the experimental plots there was a clear limitation in terms of light availability, as the tree canopy cover was greater than 70%. Therefore, in selecting caívas to plant giant missionary, the ideal is to choose “very open” caívas (Marques et al., 2019) with lower tree density, which are common in the region.

Our results indicate that the two proposed strategies were effective in increasing pasture yield in caívas, confirming the hypothesis that the top-dressing application of inputs and grazing control are efficient for use in these traditional agroforestry systems. Furthermore, the results suggest that the adoption of the two strategies together could minimize the negative effects observed in both. That is, the use of overseeding in March in the WOC would permit the availability of winter forage beginning the month of June and avoid the reduction observed in this period with the delay of overseeding in the GMC. However, such a strategy requires adequate planning and selection of the areas appropriate for each strategy.

Soil indicators

The SOM levels observed in the TC were higher than those found in other agrosilvopastoral systems, in the faxinais (another traditional agroforestry system in Southern Brazil), as well as in long-term no-tillage agricultural systems in similar climates (Frazão et al. 2010; Cunha et al. 2011; Vicente and Araújo 2013; Sá et al. 2017). This result confirms the importance of agroforestry systems for the sequestration and storage of carbon in the soil (Feliciano et al., 2018) and demonstrates the important ecosystem services traditional caívas provide in the region.

Soil organic matter plays a key role in the sustainability of agricultural systems, influencing physical, chemical, and biological attributes of soil with implications for stability and productivity of agroecosystems. One of the challenges in intensifying the use of caívas is to avoid the mineralization of SOM, which may be intensified due to productive activities such as grazing and intense pasture management. This fraction of the soil is a fundamental component of carbon, nitrogen, and nutrient stocks in the system (Gruba and Mulder, 2015). Therefore, practices that accelerate the decomposition of organic matter through oxidation, such as soil tillage, should be strictly avoided in caívas, since revolving the soil disrupts soil structure and aggregation, decreasing biological activity and depleting the organic C in the soil.

To avoid the oxidation of SOM and the consequent reduction in TOC during the implementation of the giant missionary grass in the GMC, which in full sun conditions is carried out through soil preparation, the use of desiccant herbicide was tested. We found no visual evidence that the use of herbicide in the GMC treatments caused damage to the environment. The high SOM content may have contributed to this result, since the rate at which glyphosate is mineralized is directly related to the activity of microorganisms in the soil (Gomez et al., 2009). The fact that the application occurred almost 12 months before the first soil sample collection may have contributed to the absence of identifiable effects. Despite the environmental risk, the use of herbicide in the early phase of the GMC implementation was the most efficient option, given the negative effects that conventional soil preparation brings to the system, particularly the loss of SOM. The decomposition of the roots of natural pasture after applying herbicide apparently provides a large amount of plant biomass to the soil. This was verified by the reduction of compaction in the soil surface layer, which has even facilitated the planting of grasses.

In general, the significant increase in pasture yield in the WOC and GMC treatments did not affect the maintenance of high levels of SOM and TOC. This confirms the role of perennial pastures in the maintenance of C in the soil (Castañeda et al. 2014), particularly when appropriate management strategies are used, such as controlled grazing height and pasture fertilization. The grazing impacts in TOC storage are climate-dependent and the optimal use of grazing intensity and the grass species has the potential to increase soil organic carbon sequestration (Abdalla et al. 2018). Although further research is required that considers the effects of grazing on carbon stock in complex systems, in the short term our results are promising, particularly because they were obtained under conditions of increased forage production.

During the first year of evaluation there were no limitations on soil fertility for forage yield, either in GMC or WOC, for the main nutrients currently recommended (CQFS RS/SC, 2004). However, in complex systems such as caívas where the interaction between forage, trees, and animals is highly dynamic, the process of nutrient exportation in GMC and WOC was more intense than expected and the amounts initially recommended were not sufficient to maintain soil fertility over two years, with reduction in soil pH and V% values in 2017 (Table 2). The reduction of these attributes indicates that it is necessary to adjust pasture fertilization for a period greater than that tested in this study to avoid pasture degradation. On the other hand, and surpassing our expectations, a significant increase in the phosphorus content in the soil in GMC was observed in a short period of time (Table 2), confirming the efficiency of the top-dressing application of natural phosphate for the giant missionary grass pasture.

Forest regeneration

Overall, the analyzed indicators confirmed that forest regeneration is active and dynamic in caívas, despite the constant occurrence of animal grazing for several generations. The Araucaria Forest stands out among the forest types included in the Atlantic Forest biome by greater resilience to anthropogenic disturbances (Orihuela et al., 2015), what seems to be confirmed in this research. The results of absolute density and species diversity as an average of the four surveyed areas for all evaluated caívas (Table 3) were higher than those obtained in the Forest Floristic Inventory for Santa Catarina State (Vibrans et al., 2013), where an average of 14 regenerating species were recorded in forest fragments.

Some of the evaluated caívas showed species diversity values that reached almost 70% of the diversity observed in protected areas of Araucaria Forest in the region (Lacerda, 2016). These results confirm the importance of this agroforestry system in providing another important ecosystem service for the region: the maintenance of biodiversity. Clearly, caívas are important corridors of both floral and faunal biodiversity across the landscape. In addition, landscape heterogeneity itself can contribute to increased species diversity, and the presence of forest fragments in the landscape, despite being managed, tends to positively influence biodiversity (Reis et al., 2018).

Nevertheless, attention must be given to the lower density in the GMC, which may be related to the use of herbicide before planting the giant missionary grass. These effects must be evaluated over a longer period. An encouraging result is that the increase in animal load promoted by the greater availability of forage in WOC and GMC did not seem to have a negative impact on forest regeneration, as some TCs had lower density values than those observed in WOC (Table 3), which was also observed by Pinotti et al. (2020). In general, increasing animal load or grazing intensity tends to reduce the density of individuals rather than reduce the number of species (Dumont et al., 2009; Herrero-Jáuregui and Oesterheld, 2018; Sampaio and Guerino, 2007). However, there are no studies in which an increase in animal load is associated with the use of strategies to increase pasture yield, such as those assessed in this research. In addition, the vast majority of studies on the effects of animal grazing in forest areas are short-term, with no study to our knowledge that has evaluated the effect of cattle grazing and pasture intensification over a period of almost two years, without interruption.

Conclusions

The high percentage of forest cover, active forest regeneration, and high levels of organic matter and carbon in the soil observed in the studied caívas clearly indicate the important role these agroforestry systems play in providing ecosystem services and maintaining the AF. This study also shows that the impact of animal grazing is much lower than assumed, since forest regeneration occurs dynamically and actively over time.

The two proposed pasture management strategies (WOC and GMC) were effective in increasing pasture yield in the caívas over the study period, and the use of techniques such as soil correction and top-dressing were effective in quickly improving the soil conditions.

There is a need for medium and long-term monitoring of the studied management strategies to better understand the continued environmental impacts and the productive evolution of the system. In this perspective, a more detailed tracking of nutrient cycling is essential.

Finally, we must consider that the value of caívas as areas of use and environmental conservation does not substitute the need to maintain permanent preservation areas and create areas exclusively aimed at protecting Araucaria Forest landscapes. Nevertheless, strategies for productive improvement as well as programs aimed at payment for ecosystem services, particularly for more closed caívas, must be developed in order to ensure the continuation of these historic agroforestry systems in the long-term. The development of safe and economically viable technologies will specially benefit landowners who have suffered negative consequences because they have conserved the forest.