Soil Enzymes and Stable Isotopes as Suitable Soil–Plant Indicators of Ecosystem Functionality in Mediterranean Forests

Abstract

Monitoring the soil–plant system in forest ecosystems is crucial for preserving their ecological functions and services. This study assessed carbon and nitrogen stable isotopes and ecoenzymatic stoichiometry as suitable indicators for characterizing the soil–plant system as a functional unit of ecological processes. To this end, in June 2021 six plots (1 m² each) were selected in two typical Mediterranean forest ecotypes: a coastal stone pine forest (Pinus pinea L., PF) and a meso-hygrophilous broadleaf forest (RV). Soil samples (0–15 and 15–30 cm depth) and litter samples (40 × 40 cm) were collected and characterized in terms of physical, chemical and biochemical properties. t-tests revealed significant differences between RV and PF, indicating distinct microbial nutrient acquisition strategies. The higher C:N ratio in PF suggested lower litter quality and greater recalcitrance to microbial decomposition. Consistently, RV showed a more pronounced ¹³C and ¹⁵N enrichment from litter to SOM down to a 30 cm depth, confirming faster organic matter decomposition and mineralization. Enzyme activity patterns supported these findings. The higher β-glucosidase and butyrate esterase activities in RV reflected its greater microbial potential to activate biogeochemical cycles. Both forests exhibited a higher microbial demand for C and P than for N to maintain ecological stoichiometric balance, with stronger C limitation at the surface and P limitation in the subsoil, particularly in RV soil. This integrated monitoring approach provides insights into nutrient cycling and ecosystem resilience and offers tools to evaluate ecosystem functionality under changing environmental conditions, supporting sustainable forest management.

1. Introduction

Mediterranean forests represent about 1.8% of worldwide forests, mainly located in the Mediterranean basin, and they host high levels of plant and animal diversity [1]. These ecosystems show vulnerability to climate change as well as to management policies that affect their biodiversity and the future provision of ecosystem services [2,3]. The Park of San Rossore extends over about 4800 ha on a sand dune system in the northern part of the Tuscany region (Italy) and hosts different plant communities of high ecological value, so the park is included in the world network of biosphere reserves (UNESCO) [4]. The main vegetation types include an old stone pine plantation on consolidated sand dunes and a meso-hygrophilous broadleaf forest in depressions. Stone pine forests have been historically planted for dune stabilization and sea–wind protection as well as for wood products and pine nuts [4,5]. The meso-hygrophilous broadleaf forest is situated in the sandy belt along the dunes (retrodunal area), which remains consistently humid throughout the summer.

At a small scale, there is a noticeable zonation pattern of plants in coastal dune ecosystems due to the strong ecological gradients caused by variations in abiotic conditions [6].

Although these environments have high biological and ecological values, they are threatened by human disturbance and climate change. The evaluation of ecosystem functionality through the application of suitable ecological indicators can enable the selection of appropriate conservation and management strategies. These ecological indicators should provide an indirect measure of specific ecosystem processes and functions, being easy to measure, sensitive to environmental stress and disturbances, and suitable for the prediction of changes [7,8,9]. Ecological indicators of soil–plant systems can be useful for evaluating the carbon and nutrient cycles in the environment, including soil enzymes and stable isotopes. Soil microorganisms can be effective indicators, as they quickly respond to environmental changes by altering community composition or activities (e.g., soil enzymes), often preceding changes in soil physical and chemical properties [10]. In particular, microorganisms are involved in the production of extracellular enzymes as a foraging strategy, including β-glucosidase (BG), acid phosphatase (AP) and N-acetyl-β-D-glucosaminidase (NAG), which are the enzymes most studied as proxies for soil C, P and N cycles, respectively. BG catalyzes the hydrolysis of cellulose residues in plant debris, producing glucose that is an energy-C source for microbial growth and activity. AP catalyzes the hydrolysis of esters and anhydrides of phosphoric acid, while NAG is a N-acquiring enzyme from chitin and peptidoglycan [11,12]. The application of ecoenzymatic stoichiometry theory (EST) links enzyme activities to microbial nutrient and energy requirements, as well as to resource availability and quality, to assess carbon and nutrient limitations in the environment [13,14,15,16]. According to the EST approach, microbial nutrient limitation can be estimated by deviations from the global mean ratio of C-, N- and P-acquiring enzyme activities (1:1:1) [14,15]. Changes in microbial nutrient limitation control the rate and efficiency of organic matter decomposition and immobilization, thus influencing the balance between the C pool stored and CO₂ emissions. The quantification of microbial limitation under specific environmental conditions and spatial scales contributes to the evaluation and prediction of C and nutrient cycles driven by microorganisms [17]. The EST approach has been successfully applied as an indicator of microbial nutrient limitations in sediments [18], natural soils [19,20,21] and agricultural soils [22,23,24].

In addition, carbon and nitrogen stable isotopes in soil components are valuable tools for studying C and N biogeochemical cycles, providing insights into the sources, transformations and pathways of these elements across multiple ecosystem types, as well as into their linkage with climate change and anthropogenic impacts [25,26].

The aim of this study is to evaluate the suitability of stable C and N isotopes in plant litter and soil, together with soil microbial carbon and nutrient limitations expressed through ecoenzymatic stoichiometry, as indicators for characterising the soil–plant functional unit in Mediterranean forests. To this end, we tested the sensitivity of these indicators in two forest ecotypes of high biological and ecological value that coexist within the same stand: an old coastal pine forest and a meso-hygrophilous broadleaf forest.

2. Materials and Methods

2.1. Study Area

The study was carried out in the Regional Park of Migliarino, San Rossore, Massaciuccoli, central Italy, (43.732022 N, 10.290910 E), 4 m above sea level. The park is included in the Natura 2000 site Selva Pisana (IT5170002), and it extends over a coastal alluvial plain, characterized by a coastal dune system, i.e., foredune, interdune, and backdune habitats [27].

The southern part of the park, delimited by the Arno River, is characterized by wetlands, a network of ponds directly connected to the dune system, originating from the deposition of sediments at the river mouth. In this scenario, superficial aquifers provide a constant water supply, especially in depressions where wetlands occur. The park hosts a forest mainly composed of even-aged pine stands historically managed through cuttings with a rotation age of 90 years, plantations, or seeding [5].

In the study area, two typical Mediterranean forest ecotypes within the Regional Park were selected: a coastal stone pine forest (PF), dominated by Pinus pinea L., and a meso-hygrophilous broadleaf forest located in humid retrodunal depressions, mainly composed of retrodunal depressed plain vegetation (RV), dominated by Alnus glutinosa (L.) Gaertn and Fraxinus angustifolia Vahl subsp. oxycarpa (M. Bieb. ex Willd.) Franco & Rocha Afonso (Figure 1). The observed variations in plant species richness between the two study areas (dunal and retrodunal) can be attributed to the different small-scale abiotic conditions, which determine a strong ecological gradient and marked vegetation zonation.

In the pine forest area, plant species experience more severe abiotic conditions that limit the establishment of highly specialized herbaceous species. However, in the plain of the retrodunal depression, the severity of these conditions decreases, allowing meso-hygrophilous plant communities to reach greater biodiversity.

The following climatic traits characterize the location: mean annual temperature: 15.3 °C, mean annual precipitation: 950 mm, mean annual radiation: 175 W m⁻².

2.2. Soil and Plant Litter Sampling

In a homogeneous area of each ecotype, litter (size 40 × 40 cm) was collected in June 2021 in six random sampling plots (1 m²) for each selected ecotype (PF and RV). The distance between plots ranged between 100 and 300 m. After the litter removal, three soil subsamples were collected in each plot at 0–15 cm and 15–30 cm depths and mixed to obtain one composite soil sample for each plot (6 soil samples for each ecotype). Before physical, chemical and biochemical analysis, litter samples were oven-dried at 50 °C and ground to a fine powder, while soil samples were air-dried and sieved at 2 mm.

2.3. Soil Physical Characterization

In dried soil samples the electrical conductivity (EC) and pH were determined on water extract, 1:5 and 1:2.5 (w:v), respectively, using specific electrodes (ASA-SSSA, 1996). The size distribution of aggregates (0.02–2000 μm) was measured using a laser granulometer, the Mastersizer 2000 (Malvern Panalytical Ltd., Malvern, UK), equipped with a wet-sample dispersion unit. Afterwards, soil samples were sonicated in the granulometer recirculation system and measured again. Ultrasound was applied until the size distribution of dispersed particles was constant. The aggregate granulometric curve was compared to the curve obtained at the end of ultrasound application. Following Rawlins et al. [28], we refer to the difference in MWD (µm) of these two size distributions as disaggregation reduction, DR, a quantitative estimate of aggregate stability.

2.4. N and C Concentrations and Stable Isotopes

Carbon and nitrogen concentrations (%C and %N) and isotope compositions (δ¹³C and δ¹⁵N) were determined for the plant litter, soil samples and soil organic matter fractions. All samples were dried and subsequently ground to a fine powder with a mortar. An aliquot of dry powder (0.5–2 mg) of litter and soil samples was then directly used for %N and δ¹⁵N analyses, while soil subsamples of about 100 mg were treated with hydrochloric acid to remove carbonates before %C and δ¹³C determinations. Analyses of ¹³C/¹²C and ¹⁵N/¹⁴N isotope ratios were performed using an isotope ratio mass spectrometer (Isoprime, GV, Cheadle, UK) connected to an elemental analyzer (NA1500; Carlo Erba, Milan, Italy). The isotopic compositions were expressed as δ notation vs. VPDB for δ¹³C and vs. atmospheric N₂ for δ¹⁵N, according to the expression: δ = (R_s − R_std)/R_std × 1000, where R_s is the isotope ratio of the sample and R_std is the isotope ratio of the international standard. In particular, IAEA-CH₆ sucrose and IAEA-600 caffeine were used for scale normalization of measured δ¹³C values to the VPDB scale; IAEA-600 caffeine and USGS40 glutamic acid were used for scale normalization of measured δ¹⁵N to the atmospheric N₂ scale. The precision values of measurements, expressed as standard deviations, were determined against replicate measurements (n = 10) for IAEA standards for each considered element and were better than ±0.1‰ for both δ¹³C and δ¹⁵N.

2.5. Enzyme Activities and Stoichiometry

β-glucosidase (BG; EC 3.2.1.21), acid phosphatase (AP; EC 3.1.3.2), and N-acetyl-β-D-glucosaminidase (NAG, EC 3.2.1.14) hydrolytic enzyme activities and butyrate esterase enzyme activity were analyzed at a 0–15 cm soil depth, according to the fluorometric method proposed by Marx et al. and Vepsäläinen et al. [29,30]. Briefly, a moist sample was treated with 25 mL Na-acetate buffer (pH 5.5). A suspension was obtained by treating a sample with an UltraTurrax homogenizer (IKA, Staufen, Germany) for 1 min at 9600 g. Aliquots of 100 μL were withdrawn and dispensed into a 96-well microplate. Finally, 100 μL of 1 mM substrate solution was added, giving a final substrate concentration of 500 μM. Fluorescence (excitation: 360 nm; emission: 450 nm) was measured after 0, 30, 60, 120, and 180 min of incubation, with an automated fluorimetric plate-reader (Infinite F200 pro TECAN, Männedorf, Switzerland). The enzyme activities (BG, AP and NAG) were normalized to TOC and log-transformed.

Vector analysis was used to indicate microbial resource limitations [11]. A longer vector represents a greater carbon (C) limitation, while a vector angle greater than 45° indicates phosphorus (P) limitation. Vector length and angle were calculated according to the method described by Wang et al. [31] as follows:

$$VL=\sqrt{({lnBG/lnBG+lnNAG)}^2+{\left(lnBG/\mathit{ln}BG+lnAP\right)}^2}$$

$$VD=Degrees(ATAN2\left(lnBG/lnBG+lnAP\right), \left(lnBG/lnBG+lnNAG\right))$$

Greater values for VL indicate greater C-limitation vs. N/P limitation; a VD < 45° indicates N-limitation, while a VD > 45° indicates P limitation.

2.6. Statistical Analysis

The statistical analysis was performed using R software (4.5.2 version). After confirmation of the normality (Shapiro–Wilk normality test) and homoscedasticity (Bartlett test) of the data, differences in soil and litter properties were subjected to a one-tail t-test. Data not normally distributed and with unequal variances were log-transformed prior to statistical analysis.

3. Results

3.1. Litter and Soil Properties

The particle size distributions of the two forest soils at depths of 0–15 and 15–30 cm are shown in Figure 2. Based on the graphs of sonicated samples, the PF soil exhibited a sandy texture, whereas the RV soil showed a sandy loam texture. The comparison of the size distributions of sonicated and non-sonicated samples revealed that the RV soil had a more aggregated structure compared to PF. Moreover, both forest soils showed greater stability in their surface layer (Figure 2).

The chemical properties of soil and litter samples collected from the two forest sites (PF and RV) are summarized in

Table 1. Leaf litter and soil (at 0–15 cm and 15–30 cm) chemical properties measured in the coastal stone pine forest (PF) and in the meso-hygrophilous broadleaf forest (RV). Asterisks represent a significant difference of p < 0.05. ns = not significant.

	Litter			0–15 cm			15–30 cm
	PF	RV		PF	RV		PF	RV
pH				6.0 ± 1.23	7.6 ± 0.24	ns	6.9 ± 1.28	8.2 ± 0.19	ns
EC				0.7 ± 0.49	3.1 ± 0.62	*	1.0 ± 0.31	1.6 ± 0.20	*
TOC	49 ± 0.8	46 ± 0.7	*	2.7 ± 1.69	2.7 ± 1.80	ns	0.4 ± 0.14	0.4 ± 0.17	ns
TN	0.7 ± 0.05	0.9 ± 0.16	*	0.13 ± 0.04	0.33 ± 0.19	*	0.03 ± 0.01	0.09 ± 0.06	*
TOC/TN	70 ± 5.4	52 ± 9.2	*	21.3 ± 13.86	8.7 ± 2.77	*	12.9 ± 3.66	4.9 ± 1.48	*

EC = electrical conductivity (dS m⁻¹); TOC = total organic carbon (%); TN = total nitrogen (%).

. Litter from PF exhibited higher total organic carbon (TOC) and lower total nitrogen (TN) compared to RV. The lower TN content in PF was also observed in the soil samples at both depths (0–15 cm and 15–30 cm). Similarly, electrical conductivity (EC) was lower in PF than in RV at both soil depths. However, no significant differences in soil TOC content were detected between the two forests at either depth. The TOC/TN ratio in both litter and soil samples was higher in PF than in RV (Table 1). The vertical distribution of TN and TOC in the soil profiles displayed a decrease from the topsoil (0–15 cm) to the subsoil (15–30 cm) in both forest ecosystems.

3.2. Carbon and Nitrogen Stable Isotopes

The isotope compositions of C and N in litter and soil samples are reported in Figure 3. The δ¹³C values of the litter did not differ significantly between PF and RV, whereas the δ¹⁵N values were significantly higher in RV compared to PF. An increasing trend in both δ¹³C and δ¹⁵N was observed from the litter to the 15–30 cm depth soil samples in both forest ecosystems; however, RV showed significantly enriched values at both the 0–15 cm and 15–30 cm depths compared to PF.

3.3. Enzyme Activities and Stoichiometry

In the litter, the BG and AP activities were higher in RV than in PF. Regarding soil enzyme activities, the main differences among ecotypes were observed at the subsurface layer (15–30 cm). Specifically, RV soil showed higher BG and BE activities and lower NAG and AP activities than PF. At the surface layer (0–15 cm), only NAG was statistically higher in RV compared to PF, and no significant differences were observed between the two ecotypes at the surface layer (0–15 cm) for BG, AP, and BE activities (Figure 4).

Lower values of the ratio ln(BG)/ln(AP) were observed in the surface layer, while higher values of the ratio ln(BG)/ln(NAG) were observed at the subsurface layer, particularly in RV (Figure 5). In both forests, litter exhibited stoichiometric ratio values greater than 1.

The vector angle (VA) of both forest soils was greater than 45°, with higher values in the surface soil layer than in the subsoil (Figure 5B). VA was below 45° in litter samples, with lower values observed in the PF forest. The RV forest exhibited greater differences between surface and subsurface soils compared to PF for ecoenzyme stoichiometry.

4. Discussion

Soil aggregation and water-stable aggregates are usually related to C content [32]. However, given that the soil carbon content is similar in the two ecotypes, the greater aggregate stability observed in RV is likely attributable to its higher proportion of finer mineral particles [33].

The stable isotopes and enzyme activities were sensitive indicators of the functionality of the two forest ecotypes. In particular, these indicators can reflect soil–plant multifunctionality and ecosystem processes, thus providing a better understanding of drivers and interactions affecting the soil–plant system. The uniformity of δ¹³C in the plant litter layer between the two forest ecotypes suggested that plants did not differ in carbon isotope discrimination and relied on a similar isotopic CO₂ source during photosynthetic assimilation [34]. Conversely, the lower δ¹⁵N of plant litter in PF than RV indicates possible differences in availability, sources and forms of soil nitrogen between the two forest ecotypes. Several factors can explain this result, including differences in inorganic or organic nitrogen sources, root depth, and mycorrhizal-derived nitrogen [35,36]. Furthermore, the lower soil pH in PF compared to RV could reduce nitrification, leading to changes in the ratio between nitrate and ammonium N forms [37].

The increase in δ¹³C and δ¹⁵N from litter to soil indicated an enrichment in the heavier isotopes ¹³C and ¹⁵N during plant litter decomposition due to the loss of C- and N-containing compounds. A gradual ¹³C and ¹⁵N enrichment is usually observed with soil depth [38,39,40]. However, it is worth noting that RV showed a more pronounced ¹³C and ¹⁵N enrichment from litter to SOM down to the 30 cm depth, suggesting a faster SOM decomposition and mineralization rate compared to PF [41]. This may be partly attributed to differences in the physical form and biochemical quality of the litter inputs, which influence microbial activity and isotope fractionation in the two ecotypes. In particular, PF exhibited poorer litter quality, as indicated by its higher C:N ratio. Such lower quality could be related to the high content of recalcitrant compounds in the P. pinea litter, which are more resistant to decomposition than those in the RV litter [42,43]. In addition, the lower values of δ¹⁵N in PF confirmed lower decomposition, mineralization, nitrification, and denitrification rates than in RV, as these soil processes typically discriminate against the heavier ¹⁵N isotope [44]. The higher C:N ratio in PF litter reflected microbial nitrogen limitation, as highlighted by a lower VD, limiting nitrogen mineralization and increasing nitrogen immobilization in microbial biomass [45]. In addition to litter quality, other factors may contribute to the contrasting isotopic profiles between RV and PF, including differences in microbial community composition, dissolved organic carbon (DOC) transport, and soil physical properties [46]. Soil characteristics such as clay content and aggregate stability strongly affect SOM δ¹³C and δ¹⁵N signatures. Fine-textured and aggregate-rich soils—such as those in RV—typically exhibit higher δ¹³C and δ¹⁵N values and greater enrichment with depth because soil aggregates preferentially stabilize the ¹³C- and ¹⁵N-enriched products of microbial decomposition [47]. In contrast, coarse-textured soils like PF offer weaker physical protection, resulting in smaller isotopic variations in soil depth [46]. Furthermore, fine-textured soils provide a greater specific surface area and higher cation exchange capacity, enhancing the sorption and stabilization of ¹³C-enriched DOC within soil aggregates. Finally, soil texture also determines nitrogen loss pathways. The high permeability and low chemical reactivity of sandy soils promote a rapid leaching of DOC and mineral N (often relatively ¹⁴N-rich), which induces an isotopic enrichment of the remaining nitrogen pool.

Vegetation, through litter inputs and decomposition, can also affect soil enzyme activities, leading to changes in soil microorganisms and their metabolic patterns.

The higher stimulation of soil organic matter decomposition in RV relative to PF is supported by the enzymatic activity patterns. Butyrate esterase activity (BE) has been used as an indicator of overall soil microbial activity [48], while β-glucosidase enzyme activity reflects the ability of soil to hydrolyze low-molecular-weight carbohydrates [49]. The higher activity of these enzymes at the 15–30 cm layer of RV compared to PF suggests a greater microbial potential to activate biogeochemical cycles and decompose organic matter.

Additionally, RV showed higher surface (0–15 cm) activity of N-acetylglucosaminidase, an enzyme involved in the nitrogen cycle, indicating enhanced microbial utilization and assimilation of nitrogen through enzyme production.

Enzyme stoichiometry revealed a higher microbial demand for C and P relative to N to maintain ecological stoichiometric balance [14]. Microbial P limitation may be linked to microbial requirements, since P is an essential element for microbial growth [44]. In particular, P limitation was higher in the 0–15 cm soil layer, whereas C limitation was more pronounced in the 15–30 cm layer, especially in RV soil.

Although TOC levels in the soils of both ecotypes were not significantly different, the vector length (VL) was higher in RV at the subsurface layer (15–30 cm), confirming that the relative degree of microbial C limitation was higher in RV than in PF [50].

However, the lower AP activity in RV compared to PF at the 15–30 cm soil layer may indicate a higher availability of inorganic P, which is the product of AP-mediated hydrolysis [51]. At the same time, the RV forest exhibited higher AP activity in the litter layer, suggesting that microbial communities invest more in phosphorus acquisition. In both forests, litter stoichiometric ratios greater than 1 confirm that litter, especially in RV, may have alleviated soil microbial P limitation, consistent with previous findings [52].

The vector angle, an indicator used to assess whether soil microorganisms are limited by N or P availability [53], exceeded 45° in both forest soils, with higher values in the 0–15 cm soil layer compared to the subsoil. This pattern reflects a decline in P limitation with increasing depth [54]. Subsoils typically exhibit low P availability due to reduced SOM content and strong P binding to clay particles [31]. At the 15–30 cm layer, the decline in microbial biomass and activity may further slow P cycling. Overall, microbial carbon and nutrient limitations reflect the long-term adaptation of microbial communities to the site-specific soil and environmental conditions [55].

5. Conclusions

This study highlighted the effectiveness of ecoenzymatic stoichiometry and stable isotope analysis as suitable soil–plant indicators for assessing ecosystem functionality, in terms of carbon and nutrient cycles, in coastal Mediterranean forests. The observed differences between the pine (PF) and broadleaf (RV) forests highlighted the influence of vegetation and environmental conditions on soil–plant functionality and microbial nutrient dynamics. The δ¹³C and δ¹⁵N signatures reflected the rate of decomposition and mineralization of organic matter, while enzyme activities indicated the microbial capacity to activate biogeochemical cycles and maintain stoichiometric balance. Future research should integrate these indicators with broader biogeochemical assessments to improve predictive models of ecosystem functionality related to management strategies and climate change, including cross-scale measurements and season dynamics.