Impact of Soil Management Practices on Olive Orchard Soil Health and Arthropod Diversity in Messenia, Greece

Abstract

Soil degradation driven by intensive management practices has become of increasing concern for olive cultivation, as trends for desertification and loss of arable land have emerged across the Mediterranean basin. Agroecological management practices, such as mulching made from olive tree pruning remains, have shown potential for improving soil structure, nutrient retention and biodiversity. This study aimed to enhance the understanding of how soil management influences soil properties and arthropod diversity in small-scale olive orchards in a heterogeneous landscape in south-west Greece. Soil was sampled from 11 orchards managed under one of two systems: conventional (herbicide use, tillage, mowing) and agroecological (cover cropping, mulching), encompassing a diversity of management practices. Physicochemical properties were measured alongside soil arthropod abundance and diversity, allowing for comparisons at two levels: between management systems and among practices nested within each system. When compared across broader systems, the agroecological orchards, compared to conventional orchards, had greater porosity (56.38% and 48.75%), and soil organic matter (8.99% and 6.87%), though differences in soil composition likely accounted for some of the variation. Additionally, metrics for arthropod diversity were improved under agroecological management, with 21% higher Shannon diversity and 16.8% greater evenness compared to conventional management. Ordination analysis and generalized linear models further supported these findings illustrating the relationship between agroecological management, soil health and arthropod diversity. These results support a growing body of research which illustrate the potential of agroecological management in enhancing soil health and biodiversity in olive orchards and contributing to the development of more resilient agroecosystems within the Mediterranean basin.

1. Introduction

Within the EU, 60–70% of soils are degraded as a direct impact of intensive agricultural management [1]. Intensive management, also known as conventional management, encompasses practices such as crop homogenization, extensive land use, tillage and chemical inputs [2,3]. Without intervention, soil degradation is projected to worsen, as 90% of the global growth in crop production is expected to come from higher yields and increased cropping intensity, with the remainder resulting from land expansion [4]. This degradation threatens the multifunctionality of agroecosystems—critically impairing ecosystem services, such as temperature regulation, carbon sequestration, nutrient cycling, and water infiltration [1,5]. Moreover, this has direct consequences to agro- and ecological biodiversity, manifesting as the loss of native plant species, microorganisms, and microhabitats [6,7], all of which are required for the persistence and resilience of global food systems [8].

Various policy and research platforms have begun to prioritize these challenges [2,9]. For example, the ‘EU Soil Strategy for 2030’, aims to restore degraded European soils by 2050 through various measures [10]. However, one key element to such a plan is accurate soil monitoring and assessment of soil health across a diversity of agroecosystems [11,12]. Soil health, characterized by its physical, chemical, and biological properties, can indicate a soil’s proneness to degradation [1,13]. Chemical and physical properties include pH, soil organic matter, and bulk density [7,14]. Biological properties include the activity of microorganisms, such as fungi and bacteria, and soil macrofauna such as soil arthropods [15]. Interestingly, attempts to use soil arthropod diversity and community composition as an indicator for soil health have been attempted [16,17]. However, more evidence is needed to support this relationship [15]. Another challenge is capturing the diversity of agricultural management as how this varies across crops, countries, and socioeconomic regions. While there is a growing body of research which helps explain the relationship between soil health and management, much of the existing data is based on experimental setups—testing an ideal that often does not exist in reality [13].

In Greece, olive farming exemplifies the aforementioned challenges. Olive (Olea europaea) cultivation covers an area of over 700,000 ha and is considered one of the most significant agricultural activities from a financial, social, and ecological perspective [18]. In recent decades, increased demand for olive oil and table olives has led to the industrialization and consequent expansion of olive farming into more accessible arable land [18,19]. This expansion has permitted the mechanization of olive production and harvest, as well as a need for more intensive use of irrigation and groundwater [20,21]. Farmers now prioritize management which requires minimal water usage and reduced physical maintenance such as low planting density, limiting canopy size, and the elimination of natural ground cover [22]. While many of these practices reduce the physical cost for the farmer, they simultaneously increase pests, fungal infections, and increase susceptibility to desertification [21,22]. Furthermore, the risk of drought, water insecurity and seasonal extremes has become increasingly prevalent in Greece due to climate change [1,20], all of which are actively feeding into a cycle of erosion, salinization, compaction, sealing, and contamination—ultimately degrading soil throughout the landscape [6,21]. The collective consequences of these factors have manifested as widespread ecological damage, declining olive yields, and increasing land abandonment [23,24,25]. Therefore, there is a pressing need for the adoption of management practices which support soil ecosystems and biodiversity [26].

Agroecological management is an increasingly popularized approach which supports soil function and biodiversity, through several principals including the limiting of mechanical disturbance to the soil, application of organic amendments, and the maintenance and diversification of ground cover [2,13,27]. These principals promote soil health, preventing further degradation [2,13]. Specific to olive orchards, promising results have been found in orchards applying cover crops and cover using mulching made of olive tree pruning remains. Studies have shown these practices increase soil organic matter (SOM), improving soil fertility and structure, preserve soil moisture, and prevent soil erosion [18]. Even when used in predominantly more intensive systems, these methods have been found to show generous benefits [18,28]. However, the use of these practices are still limited in Greece, and their effectiveness if implemented is not yet well understood due to the low availability of data in the country regarding soil health and management [29].

The aim of this study was to investigate the impact of soil management on soil health and biodiversity of olive orchards within the region of Messenia (Figure 1), one of the most productive olive growing regions in Greece [30]. To achieve this, the landscape was surveyed to identify the major practices used in the region. Once identified, representative orchards were selected, and soil samples were taken to both classify the soil texture and assess its physiochemical properties. Additionally, arthropod traps were set in several orchards within the landscape to measure several metrics of soil arthropod diversity. We hypothesized that orchards under agroecological management would have improved chemical and physical (structural) properties. We further expected that these improved physiochemical properties would correspond to higher arthropod diversity and community evenness.

2. Materials and Methods

2.1. Study Area and Orchard Selection

A total of twelve olive orchards were selected in the region of Messenia (36°59′33.8028″ N, 21°39′24.4476″ E), close to the Navarino Environmental Observatory. The landscape is characterized by a mosaic of coastal lowlands, mountainous terrain, and olive orchards resulting in a diverse range of microclimates within the region [30]. The climate is characteristic of a typical mediterranean climate, marked by hot, dry summers and mild, wet winters [8], with a mean annual air temperature (MAT) of 18 °C and mean annual precipitation (MAP) of 695 mm (mean between 1956 and 2011, from the Hellenic National Meteorological Service, Methoni) [31]. The local climatic conditions, combined with the underlying limestone geology, have resulted in the formation of a diverse range of soil types [32]. Furthermore, the interplay between climate, topography, and geology creates varied growing conditions (e.g., water availability and nutrient content) for olive trees [33].

The twelve orchards were managed under one of two systems: conventional (CON, hereafter) and agroecological (ECO, hereafter). Nested within each system were specific management practices (Figure 1). For ECO orchards, this included (i) mulching (MU) using olive tree pruning remains and (ii) cover cropping (CC). For CON orchards, this included (i) herbicide (H), (ii) tillage (T), and (iii) mowing (M) of the natural vegetation. For each management practice, two to three representative orchards were identified. Additional information on the specificities of each practice was collected through direct communication with the landowner (

Table 1. Summary of management practices found within the region of Messenia including a description of the practice as well as the general period in which each practice is executed.

System	Practice	Field ID	Description	Time Period
ECO	Cover Crops	CC1, CC2, CC3	Seeding of Vicia sativa and Lupinus albus, followed by shallow tillage (5–10 cm depth) to mix the seeds into the soil.	Every other year between October and November
	Mulching	MU1, MU2	Pruning remains finely ground using a woodchipper and applied as a cover layer.	Once a year between October and January
CON	Mowing	M1, M2	Mechanical mowing of the vegetation to prevent competition for water and nutrients.	1–2 times per year between April and June
	Herbicide	H1, H2, H3	Applied to ground vegetation to prevent growth. The dominant herbicide used is Flazasulfuron (Katana).	1–2 times per year between January and May
	Tillage	T1, T2	Mechanical tillage of the topsoil layer (20–30 cm).	1–2 times per year between April and May

).

All orchards were comparable in terms of agronomical features (planting density, variety and age) as well as the slope (approximately <20%). For further details of each practice, see Appendix A (

Table A1. Additional characteristic of orchards sampled (Field ID) including an estimation of vegetation cover (%), size (ha)—calculated in QGIS, planting density of the olive tree (trees/ha), fertilizer use, and if the orchards apply pesticides.

Management Practice	Field ID	Vegetation Cover (%)	Size (ha)	Density (Trees/ha)	Fertilizer Use	Pesticides
Cover Crops	CC1	75.00	1.29	112	Organic	no
	CC2	86.25	1.27	108	Organic	no
	CC3	84.50	1.41	98	Organic	no
Mulching	MU1	100	0.90	88	Mulching	no
	MU2	73.25	0.32	100	Mulching	no
Herbicide	H1	16.00	2.08	-	Inorganic	yes
	H2	50.00	0.42	100	Inorganic	yes
	H3	56.25	0.54	96	Inorganic	yes
Mowing	M1	95.00	0.60	192	Inorganic	-
	M2	97.50	0.42	80	Inorganic	yes
Tillage	T1	88.75	0.53	104	Inorganic	yes
	T2	99.50	0.70	144	Inorganic	yes

). All practices were established (>4 years) [34], with the exception of CC3, which had been applying cover crops for the last two years. Across agroecological orchards, no additional chemical amendments (fertilizers, pesticides, nor herbicides) were applied. The position of the orchards within the landscape varied due to the heterogeneity of the topography, with some orchards located in the hills and some within urban settlements.

2.2. Soil Sampling and Analysis

All soil samples were taken in November 2025 during the olive harvest season. Prior to sampling, a standard area of 0.25 ha was established for sampling, considering a ten-meter buffer zone for each orchard. The buffer zone was established to mitigate any edge effects from bordering orchards [35]. Within this plot, four transects of fifty meters in length were established. The transects were used as a reference for all soil and arthropod sampling points [35,36], all of which were pre-determined in QGIS (version 3.34.4, Zurich, Switzerland) using the research tool ‘random points on lines’ and exported to the app ‘SW Maps’ (version 1.2.4, New Delhi, India) for field reference.

The soil sampling procedure involved randomly selecting four points per transect (Figure 2). From the selected points, soil samples were taken using an Edelman augur, at 0–20 cm soil depth [37]. The samples were combined into one composite sample per transect, resulting in four composite samples per orchard [38]. These samples were used for both the classification of the soil (in terms of texture) as well as the analysis for soil organic matter (SOM) and pH. All samples were sieved (<2 mm), air-dried and stored at 20 °C until further analysis. All bulk density (BD) measurements were taken in January, in which one point was randomly selected per transect. All samples were taken using a cylindrical augur, at 0–5 cm soil depth [39]. The BD samples remained inside of the cylinder in order to maintain the original structure of the soil and the analysis was conducted on the same day. All sampling was conducted outside the tree canopy [34].

The texture of the orchard was quantified using the ‘Hydrometer Method’ [33,39,40,41]. Soil preparation required the creation of one representative sample (50 g), consisting of approximately 12.5 g of each composite sample per orchard. All samples were pulverized using a pestle and mortar. Laboratory preparation involved placing deionized water into a cylinder and leaving it to rest overnight, as well as a control sample consisting of 100 mL of NaPO3 and 900 mL of deionized water. Each soil sample was mixed with a 2:1 ratio of deionized water and a sodium hexametaphosphate (NaPO₃) dispersing solution and then placed in a sealable volumetric flask. Samples were placed in a shaker for 15 min. All solutions were left in the lab overnight to ensure they reached room temperature (20 °C).

The following day, the soil slurries were shaken for another 15 min and placed in 1000 mL volumetric flasks. The cylinder was then brought to its total volume using the deionized water that had been prepared. All samples were thoroughly mixed using a plunger before the first hydrometer reading. Following the mixing of the sample, readings were taken at 40 s and 2 h using a standard hydrometer (g L⁻¹). Temperature (°C) was also recorded, and this value was used to adjust the recorded buoyancy. For each degree above 20 °C, 0.36 was added to the buoyancy value, or subtracted if below. Once the corrections were made, Equation (1) was used to calculate first the collective fraction of clay and silt (reading after 40 s in g L⁻¹), followed by the clay fraction (reading after 2 h in g L⁻¹).

$${\displaystyle \frac{\mathrm{S}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\left(\mathrm{g} {\mathrm{L}}^{-1}\right)-\mathrm{R}\mathrm{e}\mathrm{a}\mathrm{d}\mathrm{i}\mathrm{n}\mathrm{g}(\mathrm{g} {\mathrm{L}}^{-1}))}{\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e} \mathrm{w}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}\left(\mathrm{g}\right)\ast 100}}$$

(1)

Sample (g L⁻¹) represents the initial concentration of total soil mass dispersed in the liquid suspension, while the reading (g L⁻¹) corresponds to the concentration of particles remaining in the suspension at a given time (40 s and 2 h, respectively). The fraction of silt was calculated according to Equation (2):

$$\mathrm{\%}\mathrm{S}\mathrm{i}\mathrm{l}\mathrm{t}=(\mathrm{\%}\mathrm{S}\mathrm{i}\mathrm{l}\mathrm{t}+\mathrm{\%}\mathrm{C}\mathrm{l}\mathrm{a}\mathrm{y})-\mathrm{\%}\mathrm{C}\mathrm{l}\mathrm{a}\mathrm{y}$$

(2)

Lastly, Equation (3) was used to calculate the fraction of sand. The final percentages were used to classify each of the samples according to the USDA soil texture triangle [42].

$$\mathrm{\%}\mathrm{S}\mathrm{a}\mathrm{n}\mathrm{d}=100-(\mathrm{\%}\mathrm{S}\mathrm{i}\mathrm{l}\mathrm{t}+\mathrm{\%}\mathrm{C}\mathrm{l}\mathrm{a}\mathrm{y})$$

(3)

The following soil properties were determined: soil organic matter (SOM), soil organic carbon (SOC), pH, bulk density (BD), porosity, and soil moisture. Organic matter was determined using a loss on ignition procedure (550 °C for 3 h) [43]. To estimate SOC, a conversion factor of 0.58 was multiplied by SOM [43]. Soil pH was measured in the lab [44] using a portable pH electrode (HI-99121; Hannah Instruments, Woonsocket, RI, USA). In order to measure pH, four replicates of 20 g of the air-dried soil samples were pulverized and mixed for one minute with 40 mL of deionized water in 100 mL cylinders [44]. Once all particles settled (taking approximately 30 min), the tip of the electrode was placed in the solution, and the pH was recorded [45].

In order to determine bulk density (BD; g/cm³), soil samples were heated at 105 °C for 24 h [46]. Prior to drying, the volume and mass (g) of the cylinder were recorded. After 24 h, the samples were cooled, and the dry weight was recorded. Porosity was derived from BD using an assumed particle density for limestone soils (s = 2.7 g) [47]. Equation (4) was used to calculate an approximation of soil porosity per BD sample where porosity is $\varphi$, ${\rho }_b$ is BD, and ${\rho }_{s }$ is the particle density [48]:

$$\varphi =1- {\displaystyle \frac{{\rho }_b}{{\rho }_s}}$$

(4)

Soil moisture was determined using a portable soil moisture sensor (TEROS 12, METER Group Inc., Pullman, WA, USA), measured as volumetric water content (m³/m³) [35]. Measurements required the creation of a borehole, allowing for the insertion of the sensor in the side wall of the cavity. Readings were taken every two to three days over the course of ten days, in which four readings were taken per field per sampling day [49]. The mean volumetric water content (m³/m³) per orchard was taken as the final metric for soil moisture.

2.3. Arthropod Sampling

The arthropod sampling procedure involved the installation of pitfall traps, which were implanted over the course of two days in November [50,51]. Six pitfall traps were implanted per orchard, with three traps spaced 15 m apart along the two centermost transects [51]. The traps were installed with an Edelman augur, using cups 10 cm in depth, and 8 cm in diameter—making sure that the top of cup was level with the surface of the ground. The cups were filled one-fourth with propylene glycol for the preservation of the arthropods [50,51,52]. In an attempt to prevent the traps from flooding due to rainfall, a cover was created using a plastic plate and wooden skewers. Additionally, the plates were covered with surrounding foliage to weigh down the materials and provide some camouflage (Figure 3).

After seven days, the samples were collected and transported back to the lab in sealed containers. Arthropods were identified predominantly to order level [50,53], with the exception of a few taxa which were identified to family level [54]. If the abundance was less than five, individuals across all samples’ orders were combined into ‘Other’ [55]. The following metrics were used to quantify arthropod abundance, diversity and community composition: (i) total arthropod abundance, (ii) relative abundance (RA), (iii) order richness, (iv) Shannon diversity (H), (v) Pielous evenness index (J), and (vi) inverse Simpson diversity (1/D) [17,55]. All metrics were calculated in R using the ‘vegan’ package [54].

2.4. Statistical Analysis

Statistical analysis was conducted in R studio (2024.12.0, Boston, MA, USA). Orchard H1 was excluded from the analysis, due to excessive flooding which took place over the duration of the sampling. This resulted in five ECO orchards (3 cover crops and 2 mulching), and six CON orchards (2 orchards per management practice). Statistical comparisons of soil properties were conducted at two levels: between management systems and among practices nested within each system.

In order to determine the appropriate statistical test, normality was assessed using the Shapiro–Wilk test and Levene’s test was used for measuring homogeneity of variances (p < 0.05) [56]. If assumptions were met, ANOVA was used followed by a Tukey test if post hoc analysis was necessary; Welch’s ANOVA was used in the case of normality, but unequal variances, and otherwise the Kruskal–Wallis test was used, followed by a post hoc Dunn test [56]. The same steps were followed to compare management systems, applying a two-sample t-test when assumptions were held, a Welch’s t-test for unequal variances, or a Wilcoxon rank-sum otherwise [57]. Using principal component analysis (PCA) it was possible to explore the dynamics between soil properties, soil composition, and management systems [58]. Prior to the PCA, the soil property data was standardized (mean-centered and scaled to unit variance) using the ‘prcomp’ R base function [58]. The PCA scores and variable loadings were illustrated as a biplot using the ‘factoextra’ package [58]. Permutational multivariate analysis (PERMANOVA) was used to test differences in the composition of soil properties between management systems based on Euclidean distances calculated from the standardized soil properties [58].

The same analytical approach was used for the statistical comparison of arthropod abundance, richness, evenness, Shannon diversity and inverse Simpson diversity. However, if significant differences were detected, p-values were adjusted using the Bonferroni correction method [58]. Additionally, relative abundances were used to conduct an indicator species analysis (ISA) in order to isolate taxa-specific differences between management types [59,60]. Furthermore, generalized linear models (GLMs) were used to assess the effects of management and PCA soil gradients on arthropod diversity metrics [58]. Shannon diversity, inverse Simpson diversity, and Pielous evenness were modeled using Gaussian error distributions, with management system and the first two PCA axes of soil properties included as predictors. Likelihood-ratio tests (LRTs) were used to assess the significance of these predictors used in GLMs by comparing nested models [58].

3. Results

3.1. Soil Properties

The most predominant soil types were sandy-silt-loam to sandy-loam with clay (%), sand (%), and silt (%) in the range of 3–17%, 20–66%, and 30–66%, respectively. On the field level, some notable differences in composition were found. For example, in orchard T2, there was a substantially higher concentration of sand (80.24%) compared to the other fields, as well as in M1 (65.68%), suggesting that conventional orchards had greater fractions of sand compared to agroecological orchards. A summary of these findings is presented in

Table A2. Soil composition for each orchard (Field ID) showing the fraction of clay, sand, and silt (%) and the corresponding soil type according to the USDA soil triangle.

System	Practice	Field ID	Clay (%)	Sand (%)	Silt (%)	Soil Type
ECO	Cover Crops	CC1	11.61	22.74	65.65	Sandy-Silt-Loam
		CC2	12.95	29.27	57.78	Sandy-Silt-Loam
		CC3	16.86	50.57	32.57	Sandy-Silt-Loam
	Mulching	MU1	10.89	27.03	62.08	Sandy-Silt-Loam
		MU2	8.89	37.09	54.02	Sandy-Silt-Loam
CON	Herbicide	H1	-	-	-	-
		H2	16.49	20.13	63.39	Sandy-Silt-Loam
		H3	7.90	39.79	52.31	Sandy-Silt-Loam
	Mowing	M1	3.95	65.68	30.37	Sandy-Loam
		M2	6.98	58.11	34.91	Sandy-Loam
	Tillage	T1	19.88	34.40	45.72	Clay-Loam
		T2	5.93	80.24	13.84	Loamy-Sand

, along with the corresponding soil type classification.

No significant differences were detected when comparing soil properties across individual management practices. However, when testing between broader systems, pH, porosity, and SOM (and SOC) were significantly higher in ECO orchards, while BD was significantly higher in CON orchards (

Table 2. Soil properties for each management system (mean ± SD) including bulk density (BD), porosity (%), pH, soil organic matter (SOM), soil organic carbon (SOC), and volumetric water content (VWC). Asterisks denote statistical significance, (* p < 0.05; ** p < 0.01).

Soil Property	Agroecological (ECO)	Conventional (CON)	p-Value
Bulk Density (g/cm3)	1.18 ± 0.11	1.38 ± 0.13	0.01 **
Porosity (%)	56.38 ± 3.94	48.75 ± 4.34	0.01 **
pH	8.30 ± 0.09	7.59 ± 0.65	0.04 *
SOM (%)	8.99 ± 0.82	6.87 ± 1.70	0.04 *
SOC (%)	5.21 ± 0.48	3.98 ± 0.98	0.04 *
VWC (m3/m3)	0.31 ± 0.01	0.30 ± 0.02	>0.10
Clay (%)	12.24 ± 2.97	11.58 ± 8.59	>0.10
Sand (%)	29.75 ± 5.45	51.12 ± 20.14	0.05 *

).

Comparing individual management practices using PCA was not conducted due to the low number of replicates (n = 2). However, a system-level comparison was sufficient. The variables used in the PCA included SOM, BD, porosity, pH, VWC, as well as clay (%) and sand (%). The results of the PCA (Figure 4) illustrated a moderate distinction between ECO and CON systems in terms of fundamental chemical and physical soil properties.

The PCA gave seven different principal components (

Table A7. Summary of eigenvalues, variance, and cumulative proportion of total variance captured in the principal component analysis (PCA).

Principal Component	Eigenvalue (λ)	Variance Explained (%)	Cumulative Proportion
PC1	4.004	57.20	0.572
PC2	1.953	27.90	0.851
PC3	0.544	7.80	0.929
PC4	0.298	4.30	0.971
PC5	0.156	2.20	0.994
PC6	0.044	0.60	1.000
PC7	0.000	0.00	1.000

). The first two principal components, PC1 (57.2%) and PC2 (27.9%), accounted for 85.1% of the total variance. The PCA ordination suggests that agroecological orchards were associated with nutrient-rich soils (higher SOM), wetter soils, and higher pH. This pattern was reflected by PC1, with negative loading for soil organic matter (−0.414), volumetric water content (−0.363) and pH (−0.465), and positive loading for sand (0.453), indicating that agroecological orchards tended to score negatively along PC1. Conventional orchards were associated with more compact soils (greater BD) and greater proportions of sand. This was reflected by the loadings of PC2, with negative loadings of BD (−0.484) and sand (−0.255). The independent loadings for each variable are summarized in Appendix C (

Table A8. Summary of variable loadings for principal components—PC1 and PC2 across variables including pH, soil organic matter (SOM; %), bulk density (g/cm³), porosity (%), sand (%), clay (%), and VWC (m³/m³).

Variable	PC1	PC2
pH	−0.465	−0.084
SOM (%)	−0.414	0.082
Bulk Density (g/cm3)	0.352	0.484
Porosity (%)	−0.349	−0.487
Sand (%)	0.453	−0.255
Clay (%)	−0.175	0.564
VWC (m3/m3)	−0.363	0.362

). The results of PERMANOVA further supported the results of the PCA, indicating that the overall configuration of soil properties was significantly different between agroecological and conventional orchards (F_1,9 = 4.45, R² = 0.33, p = 0.007).

3.2. Soil Arthropods

A total of 9138 arthropod specimens were collected over the course of the study. Cumulatively, orchards applying herbicides had the highest number of individuals (3100), followed by cover crops (2212), tillage (1640), mowing (1271) and mulching (915). When arthropod abundances were compared, using field-level replicates, no significant differences were found among management practices nor management systems (Figure A1). Among the arthropods collected, a total of 19 orders were identified. Among them included microarthropod orders, Collembola (Springtails) and Acari (Mites), as well as meso-arthropod orders including Isopoda (Woodlice), Coleoptera (Beetles), Julida (Millipedes) and more. Two orders were further identified to family level including Coleoptera: Carabidae, Scarabidae, and Tenebrionidae as well as Hymenoptera: Formicidae. Furthermore, nine taxa were merged into ‘Other’ due to low abundances across all traps: Geophilomorpha, Hemiptera, Hymenoptera, Isoptera, Lithobiomorpha, Orthoptera, Psocodea, Scolopendromorpha, and Staphylinidae (

Table 3. Arthropod taxa abundances (mean ± SD) across management systems; p-values indicate significant taxon-management system associations identified by the indicator species analysis (ISA), which combines taxa specificity and constancy. Mean values are presented for descriptive purposes only. Management system associations are denoted by asterisks (* p < 0.05; ** p < 0.01).

Arthropod Taxa	Agroecological	Conventional	p-Value
Acari	2.1 ± 3.1	1.3 ± 1.9	-
Araneae	1.4 ± 1.8	1.8 ± 1.5	-
Carabidae	1.7 ± 2.4 **	0.4 ± 0.7	0.005
Coleoptera	0.7 ± 1.0	0.9 ± 1.1	-
Collembola	21.9 ± 23.8	56.9 ± 51.4 **	0.005
Diptera	2.2 ± 2.2	8.5 ± 13.7 **	0.010
Embioptera	0.2 ± 0.6	0.3 ± 0.7	-
Formicidae	15.2 ± 22.4 **	4.9 ± 4.9	0.005
Isopoda	4.4 ± 8.2	4.9 ± 10.1	-
Julida	0.8 ± 1.0	2.6 ± 4.0 **	0.010
Opiliones	0.5 ± 0.7 *	0.1 ± 0.4	0.015
Pseudoscorpiones	0.1 ± 0.3	0.2 ± 0.7	-
Scarabidae	0.3 ± 0.6	0.2 ± 0.8	-
Tenebrionidae	0.3 ± 0.1	0.1 ± 0.5	-
Other	0.6 ± 1.1	0.3 ± 0.7	-

).

The results of the indicator species analysis (ISA) provided insight into the significant taxon-specific differences between management practices. Formicidae were significant to orchards applying cover crops (p = 0.015), while Opiliones were significant to mulching plots (p = 0.005). Collembola was explicitly associated with orchards that applied herbicides (p = 0.005), whereas Julida was associated with mowing (0.005). No orders were significant to orchards practicing tillage (Appendix B;

Table A4. Arthropod abundances of specific taxa (mean ± SD) across management practices; p-values indicate significant taxon-management system associations identified by the indicator species analysis (ISA), which combines taxa specificity and constancy. Mean values are presented for descriptive purposes only. Management practice associations are denoted by asterisks (** indicates p < 0.01).

Taxon	Mulching	Cover Crop	Mowing	Tillage	Herbicide	p-Value
Acari	2.5 ± 3.8	1.8 ± 2.5	0.5 ± 0.7	1.9 ± 2.3	1.6 ± 2.1	-
Aranae	2.0 ± 2.0	1.1 ± 1.6	2.0 ± 2.0	1.1 ± 1.0	2.2 ± 1.7	-
Carabidae	1.6 ± 1.3	1.7 ± 2.9	0.8 ± 1.0	0.2 ± 0.6	0.2 ± 0.5	-
Coleoptera	1.2 ± 1.2	0.2 ± 0.6	0.3 ± 0.5	1.2 ± 1.3	1.0 ± 1.2	-
Collembola	16.3 ± 16.6	25.6 ± 27.4	23.4 ± 13.5	55.9 ± 49.7	91.5 ± 56.5 **	0.005
Diptera	2.3 ± 2.5	2.1 ± 2.1	3.2 ± 2.1	1.7 ± 1.3	20.7 ± 18.6 **	0.005
Embioptera	0.3 ± 0.7	0.1 ± 0.5	0.1 ± 0.3	0	0.9 ± 0.9 **	0.005
Formicidae	8.3 ± 7.2	19.7 ± 27.7 **	6.9 ± 6.9	2.7 ± 2.3	5.2 ± 3.9	0.010
Isopoda	1.0 ± 1.0	6.7 ± 10.1	9.6 ± 15.5	0.9 ± 1.8	4.1 ± 6.3	-
Julida	0.8 ± 1.1	0.8 ± 0.9	4.8 ± 5.8 **	2.1 ± 2.7	0.8 ± 0.9	0.010
Opiliones	0.9 ± 0.9 **	0.2 ± 0.4	0.2 ± 0.5	0.1 ± 0.3	0.1 ± 0.3	0.005
Pseudoscorpiones	0.2 ± 0.4	0.1 ± 0.3	0.4 ± 1.2	0.1 ± 0.3	0.1 ± 0.3	-
Scarabidae	0.5 ± 0.9	0.1 ± 0.3	0.2 ± 0.6	0	0.4 ± 1.2	-
Tenebrionidae	0.1 ± 0.3	0.4 ± 1.2	0.2 ± 0.6	0	0.2 ± 0.6	-
Other	1.0 ± 1.5	0.3 ± 0.6	0.2 ± 0.6	0.5 ± 0.9	0.2 ± 0.4

). The same trends appeared across management systems, with the addition of Embioptera in ECO orchards (Table 3).

Management practices explained a significant proportion of variation in Shannon diversity (F = 13.58, p = 0.005), inverse Simpson diversity (F = 13.05, p = 0.005), and Pielous evenness (F = 42.95, p < 0.001). Post hoc analysis indicated that orchards managed with herbicides and tillage generally exhibited lower diversity and evenness compared to cover crops, mulching and mowing, with several significant associations across the management practices (

Table A5. Diversity metrics (mean ± SD) across management practices including Shannon diversity, inverse Simpson diversity, Pielous evenness, and taxon richness. Different lowercase letters indicate statistically significant differences among management practices based on post hoc comparisons following Kruskal–Wallis tests (p < 0.05). Metrics sharing the same letter in the same column are not significantly different.

Management Practice	Shannon Diversity	Inverse Simpson Diversity	Pielous Evenness	Taxon Richness
Cover Crop	1.43 ± 0.096 a	0.565 ± 0.013 a	0.565 ± 0.013 a	12.7 ± 2.520
Mulching	1.60 ± 0.001 a	0.592 ± 0.021 a	0.592 ± 0.021 a	15.0 ± 1.410
Mowing	1.52 ± 0.023 a	0.567 ± 0.001 a	0.567 ± 0.001 a	14.5 ± 0.707
Herbicide	1.20 ± 0.041 ab	0.459 ± 0.007 ab	0.459 ± 0.007 ab	13.5 ± 0.707
Tillage	1.11 ± 0.135 b	0.455 ± 0.021 b	0.455 ± 0.015 b	11.5 ± 2.120

). Across management systems, agroecological orchards exhibited higher Shannon diversity, inverse Simpson diversity, and evenness compared to conventional orchards (Figure 5); however, after Bonferroni correction, these differences were only marginally significant (p = 0.079 across all three). Unadjusted p-values indicated a marginally significant difference in Shannon diversity (t = 2.16, p = 0.058), and significant differences in inverse Simpson diversity (t = 2.29, p = 0.048), and evenness (W = 28, p = 0.022). No significant differences were found in taxonomic richness across management practices nor systems.

The results of the GLMs illustrate a relationship between management systems and arthropod diversity metrics (Appendix B;

Table A6. Results of generalized linear model (GLM) assessing the effects of management system and soil principal component axes on arthropod diversity metrics. Models were fitted at the field level using management system (conventional vs. agroecological), PC1, and PC2. Estimates (β) represent regression coefficients. Likelihood-ratio tests were used to evaluate statistical significance—significance denoted by asterisks (. indicates p < 0.1; * indicates p < 0.05; ** indicates p < 0.01).

Response Variable	Predictor	Estimate (β)	LR (X2)	p-Value
Shannon diversity	Management system (CON)	−0.129	0.259	0.015 *
	PC1	−0.013	0.258	0.781
	PC2	−0.077	0.160	0.039 *
Inverse Simpson diversity	Management system (CON)	−0.152	0.700	0.009 **
	PC1	−0.050	0.698	0.858
	PC2	−0.132	0.412	0.028 *
Pielous Evenness	Management system (CON)	−0.063	0.018	0.001 **
	PC1	−0.002	0.018	0.718
	PC2	−0.018	0.012	0.075 .

). The models indicate that CON management systems are related to lower diversity and evenness when compared to ECO orchards. The variation captured by PC2 significantly contributed to explaining differences in Shannon and inverse Simpson diversity, but not evenness; PC1 explained no variation in diversity metrics. These findings were supported by the results of the likelihood-ratio tests, reinforcing that management has an effect on arthropod diversity (Table A6). Soil variation captured by PC2 further explained differences in Shannon and inverse Simpson diversity, while PC1 explained no variation in diversity metrics.

4. Discussion

Linking soil management practices, orchard characteristics, and soil properties have revealed valuable insights into the effect of management on soil health and biodiversity. The findings of this study advocate for the adoption of agroecological practices and the reduction in tillage and herbicide use in olive orchards to preserve soil health, as also substantiated by previous studies [25,61,62]. When compared to conventionally managed orchards, the plots using mulching and cover crops were found to have higher greater soil organic matter. Generally, the range of pH across all orchards fell within the wide pH range for olive trees (5.5–8.0). However, the ECO plots generally had a higher pH (above 8.0), which could be attributed to the high levels of limestone in the soil in this region, a compound known to decrease soil acidity [63].

Mulching plots particularly stood out with the highest levels of organic matter (SOM). Several studies examining the effect of mulching using pruning remains support these findings [25,64]. Interestingly, the cover crop orchards were also applying pruning remains using an integrated approach, which has been suggested to be more effective [25,65]. However, the pruning remains were coarse and scarce compared to the more finely ground mulch used in the mulching plots, which is more effective than thicker mulch [64]. These findings are supported by a study conducted in Spain, which reported the most significant increase in OM from a finer mulch (<8 cm in diameter) at a higher density [64]. Nonetheless, the plots using cover crops presented similar findings to those of mulching, with a slightly reduced effect. The central construct of organic matter, SOC, is understood to improve soil structure as it acts as a binding agent for the aggregation process. This is reflected in the ECO orchards through findings of increased porosity and lower bulk density, which could suggest improved aggregate stability [20,29]. Studies investigating the impact of applying pruning remains have reported similar findings, as well as a correlated reduction in runoff and erosion [65]. While the trend for soil moisture was not as explicit in this study, there was generally higher volumetric water content in the ECO orchards, particularly in the plots using cover cropping. These findings are supported by a study which investigates the effect of cover cropping on soil moisture dynamics, which found cover crops to provide an immediate reduction in soil and/or water loss due to improved infiltration, particularly in clay-rich soils [66]. Moreover, a meta-analysis examined the relationship between ground cover management and soil moisture and identified a trend that both mulching and cover cropping enhance infiltration and retention capacities [61]. The link between these capacities and soil properties, including soil organic matter, porosity, and bulk density, is well documented and particularly relevant in drought-prone regions such as Messenia [16].

All conventional orchards were found to have relatively poor structure, with the highest bulk density and lowest porosity found in the tillage and mowing plots. While this might be attributable to the high percentage of sand in the soil, the use of heavy machinery for pruning and harvest in conventionally managed orchards would have contributed to the soil compaction [20,67]. Specifically, in plots applying herbicides, there was an additional effect of reduced or no ground vegetation (less than 60%), which likely contributed to the reduced organic matter (SOM), as well as the poor structure. There is strong evidence that maintaining a vegetative layer prevents compaction through the formation of aggregates and root channels in the soil [68]. Furthermore, conventionally managed orchards generally have lower volumetric water content (VWC). However, these orchards simultaneously had higher percentage of sand, which can directly impact water retention and infiltration [49]. This effect is visible in fields applying tillage which shows the lowest VWC across all orchards, as well as the highest fraction of sand (Table A2).

The principal component analysis (PCA) illustrated the dynamics between the selected soil properties, textural composition, and management. While many of the relationships observed in this model align with the findings of the univariate tests, this analysis provided a systems perspective, illustrating the interplay between the measured properties and how soil management practices influence them. More specifically, the model illustrated a dynamic in the ECO orchards, between soil organic matter (SOM), and VWC which was not visible from the univariate tests. A higher pH was also associated with a higher SOM content, which is not typically the case, as more acidic soils can inhibit the breakdown of organic matter [69]. Moreover, VWC followed a similar directionality to clay, whereas sand exhibited the opposite trend and was more associated with conventional orchards. There was also a clear structural difference between ECO and CON orchards, depicted by the directionality of bulk density and the clustering of conventional orchards in the biplot. This reinforces a consistent narrative that poor soil structure is linked to conventional management practices [65,67,70,71,72].

Overall, the results indicate that the composition of the soil was a strong driver of the variation in soil properties across orchards. This points to the complexity of measuring soil properties within such a heterogeneous region and isolating the effect of soil management. However, it could also be argued that these findings support the consideration of both soil composition and soil management practices in tandem. As the use of this ordination analysis was employed as an exploratory statistical tool rather than an inferential one, due to the limited size of the data set, the results should be interpreted with some caution. Regardless, the models suggest an interplay of properties and demonstrate how using multivariate projections has a greater capacity to capture more complex patterns than simply using univariate testing. Furthermore, it provides a means of quantifying soil health without the use of a soil quality index that requires a more extensive list of soil properties to be measured [56].

The results of the arthropod assemblages in winter were comparable to compositions found in a similar study conducted in olive orchards in Crete, Greece [15,73]. The five most dominant taxa in this composition included Collembola, Formicidae, Coleoptera, Diptera, and Isopoda which was consistent with other reports, making up more than 80% of the entire sample [52,60]. Interestingly, the plots applying herbicides had the greatest abundance of arthropods, while the lowest abundance was found in plots applying mulching, which does not align with other reports [74]. However, the main contributor to these higher abundances was Collembola, which corresponded with the findings of other studies [52]. The ISA allowed for investigation of taxa-specific variation between management types, revealing a trend where Coleoptera, and Formicidae were more strongly associated with ECO orchards. These are orders which have been reported as critical ecosystem engineers, contributing to detritus breakdown, and nutrient recycling [16,53]. In mulching plots, Opiliones were significant, alongside greater abundances of Aranae (non-significant) which are known to be important predators of common olive pests [53,75]. Collembola was significant to herbicide plots, and more broadly associated with CON orchards. However, this order is generally not found to be responsive to management intensity [74].

Diversity and evenness indices differed significantly between management systems. The highest values were found in ECO plots, possibly reflecting the improved soil health of these orchards. However, when compared across management practices, orchards implementing mowing had higher diversity and evenness than orchards applying cover crops. This was likely due to a greater volume of vegetation in the mowing plots compared to cover crops [60,75]. Furthermore, mowing is typically not as intensive, when compared to herbicide and tillage, making it somewhat of an intermediate practice. Despite the greater abundances, lower diversity was found in plots applying herbicides and practicing tillage. The use of GLMs to link observations made in the PCA to arthropod diversity metrics revealed PC2 as a potential driver of diversity variation. This principal component was defined by greater bulk density and greater concentrations of sand which aligns with the known impact of poor soil structure on soil arthropods [15]. However, PC1 which was defined by greater soil organic matter, pH, and improved soil moisture did not correspond to variation in diversity. This makes it difficult to say that arthropod diversity can act as an indicator for soil health alone, as there are likely several other drivers that were not accounted for in the models which could explain the presented differences in diversity between management systems.

The observed ambiguities in the results point to several limitations, which should be considered in the interpretation of these findings. Firstly, the number of replicates, as well as the range of soil properties and variables measured were limited. These constraints were primarily due to restrictions of time and resources. Furthermore, logistical constraints due to the region’s complex topography hindered accessibility to many orchards in the study area. Orchard accessibility was further diminished by the timing of the study, as it was conducted during both the rainy season and the harvest period (October to January). The timing of the study invoked additional uncertainty in the results due to weather phenomenon (e.g., flooding), which could have affected arthropod abundances. Simultaneously, the harvest period introduced human disturbances which could have influenced soil structure. Thus, future studies would benefit from increased orchard replication, which may require scaling up the study area. Furthermore, a broader set of soil indicators could offer a more holistic perspective. Multi-seasonal surveys of soil and biodiversity would capture temporal and seasonal variability. Lastly, farmer participation posed challenges, such as difficulties with communication, while variable willingness to participate reduced the number of orchards that could be sampled. Farmer reluctance also limited access to supplementary information such as data on the productivity of the orchards. Future research may benefit from compensation schemes, rewarding landowners for study participation. Feedback of the study results to the landowners may also encourage stronger participation willingness and cooperation.

Despite the presented limitations, the patterns in this study suggest that agroecological management is more beneficial than conventional management systems for physical, chemical, and biological soil properties. However, the regional bias towards conventional soil management is reflected by the poor soil structure, and low organic matter content of the orchards surveyed in this study. Previous studies have reported these conditions to enable desertification [1,21,76]. Reduced arthropod diversity was also found, which is supported by the findings of Seibold et al. [77] showing that the decline in arthropod populations, found in their study, occurred extensively across large spatial scales, and was intricately linked to conventional agricultural practices at the landscape scale. Collectively this can result in an irreversible loss of soil, inevitably impacting regional biodiversity, agricultural capacity, livelihood, as well as biogeochemical cycles and water balances [6,77]. This link between soil quality and poor soil structure, and consequently soil loss through erosion and runoff, has been captured by many studies investigating the effects of more intensive farming practices, particularly within the Mediterranean basin [40,78]. Moreover, the aforementioned consequences are further exacerbated by the impacts of climate change, which have already begun to impact olive growth and maturation in this region due to intense periods of drought. This was observed in the yellowing of the leaves due to water stress, despite the rather high tolerance of olive trees to drought [79]. Additionally, the soil becomes more susceptible to soil and water loss, intensifying the already high rates of erosion and runoff [62,67,78]. Therefore, applying practices which improve the fertility and overall structure of the soil serves as an important pillar of climate change adaptation in olive orchards and create resilience against soil loss on larger spatial scales [20,28,78,79].

5. Conclusions

The results of this study support the use of agroecological practices, including techniques such as soil mulching using olive tree pruning remains and cover cropping, all without the need for mechanical or chemical disturbance to the soil. These practices hold significant promise in enhancing soil health on both a field and landscape scale in terms of soil properties and arthropod diversity. On the contrary, conventional practices, particularly tillage and herbicide application, demonstrated a lesser capacity to support healthy soils and arthropod diversity.

A transition that supports the inclusion of agroecological practices could have a positive long-term impact on the ecology and resilience of the landscape and, thus, the livelihoods of stakeholders in this region. However, more efforts are needed to survey and collect data on the characteristics, soil properties, and biodiversity of these orchards, as the findings of this study were often predominated by the effect of soil composition. Therefore, similar studies should be conducted over long-term periods to provide a better understanding of how this agricultural landscape is affected by the various management practices implemented in this region. In addition, a study with a more integrated approach, examining ecological, social, and economic variables in relation to olive farming and soil management, could be of great value. Nevertheless, this study provides a strong foundation of data in an understudied region, supporting the development of resilient olive orchards and sustainable soil management, and lays the groundwork for future research in rural Greece.