Soil ecosystem changes by vegetation on old-field sites over five decades in the Brazilian Atlantic forest

Danielle Cristina Ortiz·Tancredo Augusto Feitosa de Souza·Tatiani Maria Pech·

Marie Luise Carolina Bartz3·Dilmar Baretta4·Alexandre Siminski2·Júlia Carina Niemeyer2

Abstract Vegetation types alter soil ecosystems by changing soil fauna community activities and soil physical-chemical properties. However, it is unclear how tree species (natural forest, native and exotic tree plantations) promote changes in the soil ecosystem, and if these changes alter functional groups of soil fauna and ecosystem services. To determine the effects of five decades of old-field vegetation on soil ecosystems in the Brazilian Atlantic Forest, field sampling of three ecosystems (exotic tree species Pinus elliottii Engelm. plantation, endangered tree species Araucaria angustifolia (Bertol.) Kuntze plantation, and a natural ecosystem) were carried out, as well using bait-lamina tests and bioassays with collembolans, earthworms and seeds of Lactuca sativa L. Field sampling evaluated the soil fauna community and soil physical-chemical properties. The bait-lamina test in situ was carried out for 14-days to determine fauna feeding activity, and the bioassays evaluated the reproduction of Folsomia candida, the avoidance of Eisenia andrei, and germination of L. sativa in the soil from each ecosystem. The results are: (1) vegetation type altered the soil fauna community composition; (2) soil fauna feeding was reduced in the plantations compared to the natural ecosystem; (3) a physical barrier was created by recalcitrant litter that compromised fauna community structure and seed bank germination in situ; and, (4) changes in soil physicalchemical properties promoted decomposers.

Keywords Endangered tree species·Exotic tree species·Forest ecosystem·Soil fauna functional groups·Soil invertebrates

Introduction

Understanding changes in soil ecosystems of native (Araucaria angustifolia(Bertol.) Kuntze) and exotic (Pinus elliottiiEngelm.) plantations in the Brazilian Atlantic Forest that influence soil biota composition, fauna feeding activity, and soil physical-chemical properties is fundamental to determine the development of both plantation species. Abandoned old-field areas changed to tree plantations are recognized as a southern Brazilian feature to purposeful changing tree community composition to meet fiber and timber needs (Ortiz et al. 2019). Their impact on soil fauna composition and activity remains unclear. According to Choo et al. (2020) and Jo et al. (2020), soil fauna contributes to organic matter decomposition, nutrient cycling, and soil ecosystem sustainability.

Soil ecosystems also depend on tree community composition that, in turn, create habitat and food resources for the soil fauna community, (ecosystem engineers, litter transformers, herbivores, predators, and decomposers), as described by Souza and Freitas (2018). These fauna organisms also directly affect soil physical-chemical properties through their activities (litter decomposition, nutrient cycling, bioturbation, soil excavation, and seed dispersion), and the loss of a diverse fauna community creates negative changes in the soil ecosystem over decades (Yang et al. 2018; Anderson and Ingram 1989; Myer and Forschler 2019).

When tree species successfully establish in a new area, they start changing soil fauna groups and consequently ecosystem services (Duval et al. 2020). It is nearly universally accepted among ecologists and researchers in southern Brazil thatP. elliottiiplantations are problematic due to their growth characteristics (Tesfaye et al. 2020), litter composition and physical barriers to seed bank germination (Gioria and Pysec 2016; Chen et al. 2019; Jo et al. 2020), inhibitory compounds and allelopathic effects (Gioria et al. 2014; Santonja et al. 2015), native species regeneration (Gioria et al. 2012; Bechara et al. 2014; Kimura et al. 2015), shifts in soil properties (e.g., pH and nutrient contents) (Bechara et al. 2013; McTavish et al. 2020), and environmental degradation (Ziller and Galvão 2003). For example, in a 40-year field study analyzing the seed bank dynamics of an Atlantic Coast restinga, a distinct type of coastal broadleaf forest, under invasion byP. elliottii, Bechara et al. (2013; 2014) concluded that the pine litter acts as a physical barrier to the germination of seed bank seeds and prevents the arrival of the seed rains directly into the soil, hindering the establishment of native plants and, consequently, the natural regeneration of the undergrowth (Fockink et al. 2019).

We examined the soil fauna community structure (richness, abundance and diversity), fauna feeding activity (baitlamina test-performed to evaluate the feeding activity of the soil fauna), ecotoxicity and allelopathic effects (avoidance, germination and reproduction tests), and soil changes (chemical and physical properties) ofA. angustifoliaandP. elliottiiplantations and compared them with the natural forest. Owing to the impact of the type of vegetation on the soil ecosystem, it was hypothesized that: (1) tree plantations can establish monodominance for long periods of time that decreases soil macroarthropod abundance and diversity by decreasing habitat structure and food sources. Based on the resource concentration hypotheses of Melo et al. (2019), we expected to find low macroarthropod numbers and diversity in plots of monospecific plantations due to their high needle litter production, acting as a physical barrier (Santonja et al. 2015); (2) fauna feeding activity is low in plots dominated by monospecific species by harming multiple ecosystem functions promoted by soil fauna (decomposition of soil organic matter and nutrient cycling) (Crotty et al. 2015). Based on the studies by Trentini et al. (2018) and Chae et al. (2019), we expected to find slow decomposition rates of the needle litter due to its chemical and physical nature and a physical barrier to areas of monospecific plantations (Bezkorovaynaya et al. 2017); (3) in soil ecosystems dominated by monospecific plantations would have chemically-mediated plant to plant interference (i.e., rhizodeposition). Tree species contain bioactive compounds with recognized activity into the rhizosphere that promote negative effects for the germination of other species (Sartor et al. 2009; Tigre et al. 2012; Leandro et al. 2014); and, (4) tree species can change the physical and chemical properties of soil by root development that establish and spread. Based on the work by Souza and Freitas (2017) and Souza et al. (2019), we expected to find changes in soil properties (pH, nutrient content, organic matter content and moisture) in the monospecific plantations that reflect the services provided by soil fauna after five decades ofP. elliottii(exotic) andA. angustifolia(native) monodominance.

The aim was to determine the effects after five decades of plantations on old-field soil ecosystems in the Brazilian Atlantic Forest. We hypothesized that: (1) tree plantations can establish monodominance for long periods of time that decreases soil macroarthropod abundance and diversity by decreasing habitat structure and food resources; (2) fauna feeding activity would be low in plots dominated by monospecific stands because multiple ecosystem functions promoted by soil fauna (decomposition of the organic matter and nutrient cycling) would be affected; (3) soil ecosystems dominated by monospecific stands would have chemicallymediated plant-plant interference (i.e., rhizodeposition); and, (4) tree species would change physical and chemical properties of the soil by the establishment and spread of their roots. To this end, we combined field samples from three vegetation types (P. elliottii,A. angustifoliaand a natural ecosystem) to characterize soil biota composition and soil physical-chemical properties, with in situ tests using bait-lamina and bioassays with soil organisms and seeds of annual plant species.

Materials and methods

Vegetation types, climatic conditions, and soil type

Our study was carried out in a Federal Conservation Unit (FCU), the National Forest of Três Barras (TBNF), located in Três Barras, Santa Catarina, Brazil. This conservation unit was created in 1968 by the Brazilian Institute of Environment and Renewable Natural Resources. The TBNF (4,459 ha), is a conservation area designed to combine nature conservation with the direct use of natural resources. It is a mosaic of areas and uses, includingA. angustifoliaplantations (native species),P. elliottiiplantations (exotic species), and natural forest (Atlantic Forest biome composed by mixed ombrophilous or moist forest) (Ferreira et al. 2011). This site is located in the northern plateau of Santa Catarina, with an altitude of 802 m. Three vegetation types were selected, considering adjacent areas (i.e., adjacent to each other within 500 m) of independent longterm field experiments which were begun before the FCU had been created. ThePinusplantation was established in 1963 in a randomized block design, comprising an area of 2 ha with seedlings spaced at 2.5 m × 2.0 m. There were two thinning in 1973 and 1980. TheAraucariaplantation was established in 1953 following a block design of 8 ha with seedlings spaced at 1.0 × 1.0 m. The plantation was thinned in 1977 and in 1984. The natural forest is a 24-ha area which has been used as reference area since 1950 for floristic and phytosociological studies. It is in an advanced stage of succession and some trees were extracted before the creation of the FCU in 1968.

The climate is humid subtropical without a dry season and with a temperate summer (Cfb-type according to Köppen-Geiger climate classification), with average annual precipitation of 1790 mm and temperatures + 17 °C. Climate data, monthly rainfall and mean temperature for Três Barras (November 2014 to December 2015; Fig. 1) were obtained online at: https:// ciran. epagri. sc. gov. br. The soil type is classified as ferralsols, deep, intensely weathered soils, often high in iron or aluminum oxides (WRB 2007).

Fig. 1 Monthly rainfall and mean air temperature data from the experimental area in Três Barras (November 2014 to December 2015)

Experimental design

Since all study areas were independent projects with different experimental designs, and to avoid possible random effects through dissimilarities of each, a rigid and pre-established design was used, selecting plots in all ecosystems with trees covering more than 95% of the plot. Homogeneous plantations provide true replicates of each plot and avoidance of random effects. After previous systematization, three plots in each ecosystem were identified, and sampling started December 2014 for: theP. elliottiiplantation (exotic species),A. angustifoliaplantation (dominant native species) and a natural mixed ombrophilous forest (Table 1).

Table 1 Geographical coordinates and altitude of each study ecosystem

Table 2 Means of relative abundance of macroarthropod taxonomic groups, ecological indexes, feeding activity (%), avoidance of earthworms (%), reproduction (number of juveniles), and seed germination (%)

Three 10 m × 10 m plots were delimitated for each ecosystem, maintaining 30 m between plots and 20 m from the border of the fragment. In each plot, pre-determined sampling included three pitfall traps for collecting active surface organisms, three monoliths for collecting earthworms, four groups of five bait-lamina sticks for determination of soil fauna feeding activity, and five subsamples of surface soil for physical and chemical analysis and ecotoxicity tests. Sampling of soil macroarthropods occurred in the spring (December 2014) and the fall (April 2015). Sampling for physical and chemical analysis and ecotoxicity tests, as well as the bait- lamina test, occurred in April 2015.

Soil macroarthropod sampling

Two methods as described in the Tropical Soil Biology and Fertility (TSBF) protocol were combined (Anderson and Ingram 1989). Three pitfall traps were installed and three soil monoliths (0.30 × 0.30 × 0.20 m) per plot collected to extract and identify soil macroarthropods. In the pitfall traps, 200 mL of a detergent solution at a concentration of 10% were added. The traps remained in the field for 48 h. For soil monoliths, the macroarthropods were manually extracted and preserved in pots containing 10 mL of 70% alcohol. Sampling was performed in each plot of the experimental area, totaling 54 sampling points (N= 54). Only individuals longer than 2 mm were considered in our analyses. They were counted and identified under a stereoscopic microscope at the level of order. The communities were characterized based on the following: (a) relative abundance expressed by (N/n) × 100, where N = abundance of an order; n = abundance in the area; (b) richness data expressed in numbers of the observed taxonomic groups; (c) Shannon diversity index (Shannon and Weaver 1949); and, (d) Pielou′s evenness index (Pielou 1969).

Soil chemical and physical properties

Soil from each ecosystem was analyzed for: pH, exchangeable cations (Al3+, Ca2+, Mg2+, H++ Al3+, and K+), available phosphorus, active hydrogen (H+), sulphate (S-SO4-2), micronutrients (B, Cu2+, Zn2+, and Mn2+), a saturation of bases, organic matter, moisture and total porosity (five subsamples by plot). The pH was measured in a suspension of soil and distilled water (1:2.5 v: v) (Black 1965). All exchangeable cations (Ca2+, Mg2+and K+) were determined by the extraction method using an atomic absorption spectrophotometer for Ca2+and Mg2+and a flame photometer for K+(IITA 1979). The potassium chloride extraction method was used to determine exchangeable Al3+. Available phosphorus (Olsen′s P) was determined colorimetrically using a spectrophotometer at 882 nm by extraction with sodium bicarbonate for 30 min (Olsen et al. 1954). Sulphate and micronutrient contents were determined following protocols described by Black (1965). Saturation of bases was measured using the equation: (sum of bases/CEC) × 100 (%) (Black 1965). Organic matter was estimated according to Okalebo et al. (1993). Soil moisture and total porosity were determined according to IITA 1979.

Soil fauna feeding activity

This was measured using lamina sticks as described by von Torne (1990) and standardized by ISO 18,311 (ISO 2016). The test consists of using PVC lamina (120 × 6 × 1 mm) with 16 holes 2 mm diameter 5 mm apart. These holes are filled with the substrate to be consumed by the organisms. The substrate was a homogeneous mixture of powdered cellulose (70%), wheat flour (27%), activated charcoal (3%), and distilled water. A 32 bait-lamina was inserted per plot, totaling 288 sampling points (N=288). The lamina sticks remained in the field for 14 days, then carefully removed and visually analyzed against the light. The feeding activity of the soil fauna was quantified following the method used by Podgaiski et al. (2011) by considering the percentage of empty holes at the end of the exposure time.

Ecotoxicity and allelopathic tests

Avoidance tests using earthworms were carried out as recommended by ISO 17,512 - 1 (ISO 2008) in two-section plastic boxes, filled on one side with test soil and the other with control soil, and adding ten clitellate earthworms to the centre of each box. There were five replicates per treatment and combinations ofP. elliottii,A. angustifoliaand natural ecosystem soils were evaluated, and a comparison made betweenP. elliottiiand tropical artificial soil (TAS) as described by Garcia (2004). The tests were incubated at 20 °C ± 1 °C for 48 h. The number of earthworms in the test and control soils in each replicate were recorded. A control or dual test was also carried out using control soil on both sides of the box to verify the random distribution of the earthworms with the same soil in the two halves of the box.Reproduction tests with juvenile Collembola (springtails) were carried out following ISO 11,267 (ISO 2014). In the tests, plastic vessels containing 30 g of the study soils and TAS as a control, ten collembolans were added in each replicate (five replicates per treatment). The tests lasted 28 days, and the vessels were then filled with water and drops of stamp ink, stirred, and floating juveniles on the surface photographed. Counting was carried out using the images on the ImageJ software (Schneider et al. 2012). Our reproduction test with Collembola fulfilled the validation criteria of minimal reproduction in control (TAS), showing 275.8 ± 81.9 juveniles, not exceeding 30% of the coefficient of variation.For the germination test, we obtained leachates of the soils from theP. elliottiiandA. angustifoliaplantations and the natural ecosystem. Each leachate consisted of 1000 mL of distilled water in a column with 300 g of fresh soil with a screen to prevent the passage of soil. The leachate was collected after 30 min of percolation. For the germination test, 10 seedsLactuca sativaL. were placed on Petri dishes lined with filter paper with leachate or distilled water as a control. The tests were in triplicate with exposure time of 72 h at 20 °C. After this period, the germination percentage was evaluated using the following: G (%) = N/A × 100, where G is the germination percentage, N the number of germinated seeds, and A the total number of seeds.

Statistical analyses

All data were tested with Shapiro-Wilk′s test for normality and log transformation applied when necessary. To avoid and detect spatial autocorrelation, the Moran’s I function as described by Gittleman and Kot (1990) was used. All variables were analyzed with a one-way ANOVA with the vegetation type and plot number as random factors. Bonferroni′s test was used as the post hoc test (p＜ 0.05). A non-metric multidimensional scaling (NMDS) was carried out to analyze differences between the studied ecosystems in terms of soil biota composition using the “metaMDS” function with Euclidean dissimilarities (Schmitz et al. 2020a). The “adonis” function was used to run a PERMANOVA with 999 permutations. The functions for the NMDS and PERMANOVA were available within the vegan package. Soil physical-chemical properties were summarized using a principal component analysis (PCA) to identify possible vegetation dissimilarities and to reduce the n-dimensional nature of variables, (some redundant soil properties), to two linear axes explaining all the variance and to explore the influence of vegetation types. Pearson correlation was used between the PCA axes and soil physical-chemical properties and carried out using the “vegan” package and following procedures described by Schmitz et al. (2020b). All statistical analyses were performed in R 3.4.0 (R Core Team 2018).

Results

Macroarthropod community structure and ecotoxicity tests

Fourteen taxonomical orders of the macroarthropod community were identified (Table 2). Their abundance varied significantly among ecosystems (p＜ 0.001). The most frequent taxonomic group was Hymenoptera, followed by Coleoptera and Araneae. The ANOVA results showed significant differences among the ecosystems on Shannon′s diversity index, Pielou′s evenness index, % avoidance of earthworms, and the germination (%) ofL. sativaseeds. There was no significant different among the ecosystems for species richness, soil fauna feeding activity, and reproduction of Collembola. ThePinusplantation supported the presence of Araneae, Haplotaxida, and Hymenoptera, while theAraucariaplantation supported Coleoptera, Dermaptera, Ixodida, and Pseudoscorpiones. It also showed the highest values of earthworm avoidance and germination ofL. sativa. Chilopoda and Diplopoda were exclusive in the natural ecosystem, and had the highest Shannon′s diversity indexes and Pielou′s evenness indexes (Table 2).

The non-metric multidimensional scaling (NMDS) revealed that the soil biota composition, feeding activity, avoidance, reproduction, and seed germination varied significantly between vegetation types (PERMANOVA:F= 42.69,p＜ 0.01; Fig. 2). The ordination had a good fit (stress value = 0.11). Biota composition was highly correlated with vegetation types. The variation in biota composition in each ecosystem was explained by Acarina (96.3%), Araneae (90.8%), avoidance of earthworms (94.6%), Blattodea (55.8%), Chilopoda (98.6%), Dermaptera (84.3%), Diplopoda (95.3%), feeding activity (94.0%), germination ofL. sativa(52.4%), Hymenoptera (70.0%), Orthoptera (94.9%), richness (99.8%), reproduction of collembolans (99.3%), and Shannon′s index (72.3%) (Fig. 2).

Fig. 2 Non-metric multidimensional scaling (NMDS) based on soil biota composition, feeding activity, avoidance, reproduction, and seed germination from the study ecosystems

Changes in soil physical–chemical properties by ecosystems

Soil physical-chemical properties were significantly affected by the vegetation, with some exceptions (pH:F2,78= 0.87,p= 96.987, and available P:F2,78= 1.21,p= 90.764) (Table 3). The one-way ANOVA showed significant differences among the ecosystems on Al3+(F2,78= 10.23,p＜ 0.01), Ca2+(F2,78= 13.33,p＜ 0.01), Mg2+(F2,78= 8.98,p＜ 0.05), H++ Al3+(F2,78= 7.17,p＜ 0.05), K+(F2,78= 8.12,p＜ 0.05), S-SO4-2(F2,78= 15.34,p＜ 0.01), B (F2,78= 10.86,p＜ 0.01), Cu2+(F2,78= 12.34,p＜ 0.01), Zn2+(F2,78= 10.27,p＜ 0.01), Mn2+(F2,78= 13.45,p＜ 0.01), saturation of bases (F2,78= 9.81,p＜ 0.01), organic matter (F2,78= 13.24,p＜ 0.01), soil moisture (F2,78= 12.56,p＜ 0.01), and total porosity (F2,78= 12.91,p＜ 0.01). Exchangeable Al, and H++ Al3+showed the lowest values from the natural ecosystem. In theP. elliottiiplantation, the lowest values were found for Ca2+, available K+, Zn2+, sum of bases, organic matter, moisture, and total porosity. In theP. elliottiiandA. angustifoliaplantations, of B and Mn2+levels were lowest (Table 3).

Table 3 Soil physical-chemical properties among the three ecosystems

Pearson′s correlation showed that most of the soil physical-chemical properties were correlated with each other (Table 4). The pH was correlated with Al3+, H++ Al3+, S-SO42-, B, Zn2+, sum of bases and Mn2+. Exchangeable Al was correlated with Ca2+, H++ Al3+, K+, B, Zn2+, sum of bases, Mn2+, and total porosity. Exchangeable Ca2+was correlated with all variables, except Mg2+, available P, and S-SO42-. Exchangeable Mg2+was correlated with K+, Zn2+, and sum of bases. Exchangeable K+was correlated with all variables except available P, S-SO42-, Cu2+, and Mn2+. Sulfate was only correlated with Cu2+, and Mn2+. Boron was correlated with Cu2+, Zn2+and sum of bases, while Cu2+was correlated with organic matter, and soil moisture. Organic matter was only correlated with soil moisture, and total porosity (Table 4).

According to the PCA analysis, the three ecosystems were dissimilar. The first two axes of the PCA explained 87.7% of the variation in physical-chemical properties (Fig. 3). The first axis explained 62.8% of variance and was positively correlated with Ca2+(R = 0.78,p＜ 0.01), B (R = 0.80,p＜ 0.01), K+(R = 0.75,p＜ 0.01), and the sum of bases (R = 0.83,p＜ 0.001) and negatively correlated with Al3+(R = - 0.80,p＜ 0.01), and H++ Al3+(R = - 0.75,p＜ 0.01). The second axis explained 24.8% of the variation in physical-chemical properties and was positively correlated with S-SO42-(R = 0.77,p＜ 0.01) and negatively with Cu2+(R = - 0.80,p＜ 0.01), organic matter (R = - 0.61,p＜ 0.01) and moisture (R = - 0.58,p＜ 0.01) (Fig. 3).

Fig. 3 Principal component analysis (PCA) for soil physical-chemical properties of the three different ecosystems

PT 62.0*37.0 - 68.0 55.0*46.0 - *68.0 73.0 - 51.0 - 46.0 55.0 46.0 47.0 25.0*98.0*58.0 tsioM 30.0 - 35.0 - *77.0 95.0*53.0 - *78.0 52.0 - 73.0 - 65.0*18.0 85.0 36.0 62.0*69.01 MO 80.0 - 16.0 - *37.0 85.0 43.0 - *28.0 23.0 - 54.0 - 84.0*38.0 05.0 95.0 25.01+2nM*88.0*29.0 - *47.0 25.0*89.0 - 16.0 85.0 - *57.0*09.0 10.0 - *38.0*09.01 BS *87.0*79.0 - *09.0*86.0*29.0 - *78.0 96.0 - 54.0*79.0 21.0*39.01 4 elbaT smetsysoce eerht fo seitreporp lios gnoma noitalerroC+2+2-2++3++2+2+3 nZuCBOS-SPKl A + HgMaClAHpselbairaV 4 *47.083.0 - *97.0*57.035.0 - 34.0*48.0 - 25.016.0*97.0 - 1Hp+3*19.0 - 30.0 - *69.0 - 94.0 - 45.0*28.0 - *69.016.0 - *78.0 - 1lA+2*97.094.0*78.090.0 - 45.0 - *58.0*08.0 - 93.01 aC+2*07.025.046.040.0 - 73.0 - *27.065.0 - 1 gM+3+*58.0 - 30.0*09.0 - 36.0 - 53.0*96.0 - 1 l A + H+*98.025.0*38.030.0 - 73.0 - 1 K 92.0 - 30.0 - 65.0 - 62.0 - 1 P-2 44.008.0 - *23.01 OS-S 4*39.020.01 B+2 30.01 uC+2 1 nZBS+2nMMO tsioM 3 elbaT ni nevig stinu ;50.0 ＜ p ta tnaicfingis* ;ytisorop latot = PT ;erutsiom lios = tsioM ;rettam cinagro lios = MO ;sesab fo muS = BS

Discussion

Our results emphasize the influence of vegetation types in the humid subtropics on soil macroarthropod community composition (i.e., abundance, Shannon′s diversity index and Pielou′s evenness index), soil fauna behavior (i.e., feeding activity, avoidance, and reproduction), seed germination, and soil physical-chemical properties. We wanted to understand how various ecosystems change the macroarthropod communities when acting as a fauna habitat inhibitor (i.e., when litter could act as a physical barrier) and feeding inhibitor (i.e., when litter does not serve as a food resource). The results reveal that there were significant differences on soil biota composition, feeding activity, avoidance of earthworms, and reproduction of collembolans and germination ofL. sativaamong the three ecosystems. According to Roy et al. (2018), Prayogo et al. (2019) and Pompermaier et al. (2020), soil fauna communities in ecosystems dominated by monospecific stands (such asP. elliottiiandA. angustifoliain this study) are less rich and diverse and with structures that differed from natural ecosystems dominated by a wide range of plant species. Consistent with previous studies (Roy et al. 2018; Chae et al. 2019; Prayogo et al. 2019; Jo et al. 2020), macroarthropod community diversity was strongly associated with the following characteristics: (1) natural ecosystems with intermediate and rapid litter decomposition rates; (2) high soil root development in the rhizosphere (the extrusion of H+and organic compounds) of native plant species; (3) high litter N content; (4) low acid-insoluble residue concentrations; and, (5) leaf structure (e.g., hairs) related to litter palatability. Our hypothesis that vegetation types change soil macroarthropod abundance and diversity by altering habitat structure and food source.

For the fauna feeding activity, there were no statistical differences among the ecosystems. However, the results demonstrate that the natural ecosystem has greater soil fauna feeding activity. In both plantation ecosystems, when a natural ecosystem is converted to a monodominance of tree species, there were losses in abundance (by 56.4%) and diversity (by 82.2%) of soil fauna (Prayogo et al. 2019). In another study, Bezkorovaynaya et al. (2017) found both abundance and biomass of soil fauna (microarthropods) positively correlated with the feeding activity. Our second hypothesis that fauna feeding activity would be low in the plantation plots was supported. The concept of decreasing multiple ecosystem functions promoted by soil fauna in a monodominance ofA. angustifoliawas expanded. TheP. elliottiiplantation exhibited low soil fauna diversity due to its poor quality fauna habitat structure and food resources with a low degree of biological breakdown by litter transformers (e.g., Coleoptera, Diplopoda and Isopoda).

The soil ecosystem of a particular vegetation type can affect the rate of soil fauna feeding activity through the chemical and physical properties of the litter on the soil surface (Rożen et al. 2010). According to Jo et al. (2020), the accumulation of a recalcitrant type of litter promotes both physical and chemical barriers to soil fauna activity. First, this type of litter acts as a fauna habitat inhibitor acting as a physical barrier to ecosystem engineers (e.g., Blattodea and Hymenoptera). Second, it has negative effects on litter transformers (e.g., Coleoptera, and Diplopoda) by reducing their feeding activity that in turn negatively affects both predators (Araneae, and Chilopoda) and herbivore (Orthoptera) abundance and function. The physical barrier promoted by recalcitrant litter is difficult to breakdown due to its chemical composition (low N, high acid-insoluble and phenolic compounds, and low palatability). Chen et al. (2019) observed that both the physical barrier and its chemical composition of the litter ofPinusspecies are determining factors for the increase of decomposer (fungal and bacterial) activities. However, as described by Santonja et al. (2015), long-term deposition ofPinuslitter with high amounts of inhibitory compounds (phenolic compounds) would have negative effects on decomposer activity. Finally, with a recalcitrant litter, there is a decrease in soil fauna communities when measured by richness, abundance, and activity.

Our results for the avoidance test using earthworms also recognized the negative effects of pine on soil habitat quality. This shows that earthworms are highly sensitive to changes in the soil ecosystem (Cardinael et al. 2018; Kamau et al. 2020). These invertebrates avoid areas of disturbance such as (1) monodominance of plant species; (2) litter with potential inhibitory or toxic effects to feeding activity; and (3) litter with low palatability (Martins et al. 2013; Chomel et al. 2014; Anwar et al. 2018; Fockink et al. 2019; Santonja et al. 2019). In general, leaves or needles of gymnosperms contain a series of organic compounds with low palatability to soil fauna (Barreta et al. 2008; Kimura et al. 2015; Santonja et al. 2019). However, our results on the reproduction of collembolans were not compromised under theAraucariaplantation, suggesting that these organisms group can reproduce without difficulties under this condition. Collembolans are abundant in plantations with high inputs of recalcitrant litter and high abundance of decomposers (fungi) as described by Ribeiro-Troian et al. (2009), and they are also widely used as bioindicators of soil quality (Ribeiro-Troian et al. 2009; Santos et al. 2018; Ortiz et al. 2019). According to Ponge (1991) and Aupic-Samain et al. (2019), recalcitrant litter promotes microorganisms that degrade lignin (e.g., white-rot fungi) that in turn, provide food resources for collembolans.

Recalcitrant litter (as found with bothAraucariaandPinusplantations) has the following characteristics on the soil surface: (1) low rates of decomposition, and (2) high rates of accumulation/deposition (Vestgarden et al. 2004). These features modify detrimentally the diversity of ecosystem engineers, litter transformers, predators, herbivores, and decomposers in the soil ecosystem. The possible causes are related to the physical barrier presented by litter accumulation and its chemical composition inhibiting fauna feeding. The hypothesis that recalcitrant litter can promote allelopathic effects on natural regeneration as described by Kimura et al. (2015) by decreasing seed germination must also be considered. However, our hypothesis thatP. elliottiiwould have allelopathic effects on seed germination was not supported. These results corroborate Fockink et al. (2019), who demonstrated thatPinuslitter acts as a physical barrier to seed germination and emergence, but the seed bank was activated when the litter was removed. Thus, one of the problems ofP. elliottiiplantations is litter accumulation, causing a physical barrier to soil fauna and to seed germination.

The changes observed in the soil chemical and physical properties (increasing H+and Al3+contents and decreasing saturation of bases (sum of bases, K+, Cu2+, Ca2+and S-SO42-), increasing organic matter content and soil moisture) only reflect the unsustainable ecosystem services by a disturbed soil fauna community in monospecific plantations. Monospecific plantations can change soil chemical and physical properties that may contribute to constraining the occurrence of native plant species, reducing seed bank germination, and decreasing native species diversity (Souza et al. 2016, 2017; Shiferaw et al. 2019). According to research elsewhere, (Basirat et al. 2019; Chen et al. 2019; Ge et al. 2019; Liu et al. 2019), it was hypothesized that three different mechanisms may be involved in the disturbance of soil fauna diversity and activity. First, monospecific plantations create a large physical barrier (recalcitrant litter), thus compromising the abundance of ecosystem engineers (Hymenoptera, Blattodea, and Diplopoda). Consequently, this changes soil porosity due to the lack of fauna galleries and nests (Ashwood et al. 2019) and soil moisture by increasing water runoff (Guareschi et al. 2020). Second, the chemical quality of recalcitrant litter negatively affects the abundance of litter transformers (Coleoptera and Diplopoda) associated with bothA. angustifoliaandP. elliottiiplantations. Thus, it promotes decreasing soil organic matter content, nutrient cycling (K+, Ca2+and S-SO42-) (Goss-Souza et al. 2019; Liao et al. 2020; McTavish et al. 2020), affecting litter palatability and feeding activity of soil organisms. Consequently, recalcitrant litter favored decomposers and associated soil fauna (collembolans). Finally, monospecific plantations might present high root development in their rhizosphere that increases the H+and Al3+contents. By altering these two cations,AraucariaandPinusplantations may alter the nutrient cycle (Souza et al. 2017) and decomposer composition under their canopies (Aupic-Samain et al. 2019).

Conclusions

The findings of this study are: (1) monospecific plantations such asA. angustifoliaandP. elliottiialter soil fauna community composition; (2) there is evidence of reduced soil fauna feeding activity in monodominant ecosystems (P. elliottiiandA. angustifolia); (3) there is a physical barrier created by recalcitrant litter that compromises soil fauna community structure and bank seed germination; and, (4) there are changes in soil physical-chemical properties that promote decomposers. Our findings suggest vegetation changes to the soil ecosystem through three mechanisms: creating a physical barrier that compromises the ecosystem engineers; the chemical compounds of litter reduce activity and abundance of litter transformers, herbivores, and predators; and finally, vegetation changes nutrient cycling and soil ecosystem. Our results are an important contribution to our understanding of the importance of the role of soil fauna underlying soil ecosystem sustainability. Thus, future studies should include decomposers identification to fully understand the effects of monospecific plantations and recalcitrant litter on fauna feeding activity.