A Quantitative Investigation of the Effects of Landscape Composition and Spatial Configuration on Epigaeic Arthropods

Abstract

In recent years, the homogenization and fragmentation of agricultural landscapes have intensified, leading to a decline in epigaeic arthropods. Landscape heterogeneity is a core factor regulating biodiversity, encompassing two key dimensions: composition heterogeneity and spatial configuration heterogeneity. Both landscape composition and spatial configuration heterogeneity influence the distribution of epigaeic arthropods through independent and joint effects. However, quantitative evidence addressing their relative and combined influences remains limited. This study was conducted across 30 independent landscape units (1 km × 1 km) in Changtu County. Pitfall traps were deployed across different habitat types, with three traps per habitat. The proportion of semi-natural habitats was used as an indicator of landscape compositional heterogeneity, while multiple landscape metrics were used to characterize spatial configuration heterogeneity. The effects of landscape heterogeneity on epigaeic arthropods were evaluated using two response variables: activity density (mean number of individuals captured per trap) and diversity (effective number of species). Variance partitioning analysis (VPA) and Bioenv analysis were applied to explore their individual and joint effects on epigaeic arthropods. The results showed that higher landscape composition heterogeneity was associated with greater activity density of epigaeic arthropods, but no significant correlation was found with arthropod diversity. In terms of landscape spatial configuration, patch density (PD) and landscape division index (DIVISION) constituted the optimal model explaining the activity density of epigaeic arthropods, highlighting the importance of patch structure within landscapes. Furthermore, spatial configurational heterogeneity showed a stronger independent contribution than compositional heterogeneity, although their joint effect accounted for the largest proportion of explained variation. These findings provide a theoretical basis for landscape optimization and biodiversity conservation in intensive agricultural regions of Northeast China.

1. Introduction

Biodiversity loss in agricultural landscapes is a critical global issue [1]. Agricultural intensification and landscape simplification have been identified as key drivers of biodiversity decline, threatening essential ecosystem services such as pollination and biological pest control [2,3]. Balancing agricultural productivity with biodiversity conservation has therefore become a major challenge for sustainable land management [4,5]. This challenge has shifted research attention toward landscape-level approaches, particularly the role of landscape heterogeneity in reconciling agricultural production with biodiversity conservation.

The core goal of sustainable intensification is to significantly increase agricultural output while protecting, and even enhancing, biodiversity within agricultural landscapes. Achieving this relies on the fine-scale management of landscape patterns to optimize the long-term supply of multiple ecosystem services. Recent research further confirms that landscape heterogeneity, specifically landscape composition (the richness and proportion of different land use types) and spatial configuration (the size, shape, and connectivity of habitat patches), is a dominant factor regulating agricultural biodiversity and driving key ecosystem services like pollination and pest control [6]. Recent global meta-analyses further confirm that both crop diversity and landscape heterogeneity significantly enhance biodiversity across agricultural systems, highlighting the importance of integrating compositional and spatial configuration heterogeneity in landscape management [7]. On one hand, landscape composition heterogeneity and landscape spatial configuration heterogeneity can independently affect agricultural biodiversity. Expanding the variety of land cover types can fulfill the ecological requirements of different species, ultimately contributing to increased species richness [8]. Similarly, a high level of land cover diversity can benefit species that rely on multiple habitats at different stages of their life cycle or across seasons [9]. Increasing spatial configuration heterogeneity can also be important, as it increases the length of edge zones and the fragmentation or juxtaposition of habitats, which benefits many species [10]. On the other hand, composition and spatial configuration can also interact to jointly influence animal species in farmland, as they are interrelated rather than completely independent. Spatial configuration heterogeneity, reflected by patch number, may influence the variety of land cover types [11]. Examining how different aspects of landscape heterogeneity influence biodiversity is essential for revealing the processes that drive ecological patterns [12,13]. In particular, both habitat composition and spatial arrangement may play distinct as well as interactive roles in shaping farmland biodiversity, which still require further clarification. However, the relative importance of these independent and joint effects remains unclear.

Epigaeic arthropods are important biological resources in agricultural landscapes [14]. Due to their strong dispersal ability, sensitivity to environmental changes, and capacity to move freely between farmland and surrounding semi-natural habitats, they are often used as indicator organisms in studies of agricultural landscape and biodiversity. epigaeic arthropods are crucial for ensuring agricultural production and maintaining ecosystem stability. They contribute to the formation of cultivated land productivity. Their movement activities help loosen soil and improve soil conditions, while their feeding and metabolic activities promote nutrient cycling, increase soil fertility, benefit crop nutrient uptake, and enhance grain production capacity. Epigaeic arthropods contribute to the stability of farmland ecology. They can decompose pollutants in the soil environment and help control crop diseases and pests through predation, reducing pesticide use [15]. High activity density of epigaeic arthropods in an agricultural landscape indicates higher productivity, buffering capacity, self-purification, and resilience [16,17]. However, intensive production methods have led to a sharp decline in epigaeic arthropods in agricultural landscapes, which is consistent with recent global meta-analyses showing that land-use intensification significantly reduces soil fauna abundance and diversity [18].

This study selected 30 independent 1 km × 1 km landscape units within a major grain-producing county in Northeast China, with landscape composition ranging from simple to complex. Landscape composition heterogeneity was represented by the relative coverage of semi-natural habitats within each grid. Landscape metrics were used to represent landscape spatial configuration heterogeneity. This study examined how varying levels of semi-natural habitat (as an indicator of landscape composition) influence the activity density and diversity of epigaeic arthropods. In addition, we identified which landscape metrics describing spatial configuration exert the strongest influence at the landscape scale. Finally, we assessed both the separate and combined influences of landscape composition and spatial configuration on epigaeic arthropod communities. Specifically, we hypothesized the following: (1) The activity density and diversity of epigaeic arthropods would respond to changes in the proportion of semi-natural habitats, with higher landscape composition heterogeneity (i.e., a greater proportion of semi-natural habitats) promoting increased activity density and diversity. (2) Specific combinations of landscape spatial configuration metrics jointly influence epigaeic arthropods at the landscape scale. (3) The relative contributions of landscape composition and spatial configuration to epigaeic arthropod communities would differ, with each component explaining distinct aspects of community variation.

2. Materials and Methods

2.1. Study Area

Located at 42°33′–43°29′ N and 123°32′–124°26′ E, Changtu County is situated in the northernmost part of Liaoning Province. The climate is a mid-temperate continental monsoon climate. Annual precipitation averages approximately 600 mm, while the mean annual temperature is around 7 °C. Rainfall and heat occur concurrently, and the annual temperature range is large. The total area is 4317 km², of which cultivated land area is 2667 km². It is known for grain production, and agricultural intensification is increasing. Thirty 1 km × 1 km grids were selected as sampling units using the proportion of semi-natural habitats as a basis (Figure 1). The percentage of non-crop habitats within the 30 grids ranged from 0% to 30%, divided into intervals of 5%: 0–5%, 5–10%, 10–15%, 15–20%, 20–25%, and 25–30%. Five grids were selected as replicates for each proportion interval, and the habitat types within each grid were recorded. Since the spatial distribution of epigaeic arthropods varies among different habitat types, pitfall traps were established in each habitat type within each grid to accurately capture their spatial distribution. Three traps were installed per habitat type, with a minimum distance of 10 m between traps to reduce spatial autocorrelation. A schematic representation of the sampling design and trap structure is provided in Figure 2 to enhance methodological clarity and reproducibility. To avoid pseudoreplication, data from the three traps within each habitat were averaged. Subsequently, data were further aggregated at the landscape unit level, and the grid (1 km × 1 km) was used as the unit of analysis in all statistical analyses.

2.2. Epigaeic Arthropod Collection

Epigaeic arthropods were sampled using pitfall traps. Each trap consisted of a Polypropylene (PP) cup (500 mL capacity; 6 cm bottom diameter, 10 cm top diameter, and 11 cm height) inserted into the soil with the rim flush with the ground surface. A solution of 20% ethylene glycol (150–200 mL; Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) with a small amount of detergent (Nice Group Co., Ltd., Ningbo, China) was added to each cup to reduce surface tension. To prevent rainwater from entering, a cover was installed above each trap using three metal wires anchored into the surrounding soil. Traps were deployed for six consecutive days. After retrieval, specimens were preserved in Pre-labeled Polyethylene (PE) bottles containing 75% ethanol (Aladdin, Shanghai, China) and transported to the laboratory for identification and classification. Sampling was conducted under relatively stable and dry weather conditions to minimize short-term climatic effects on arthropod activity. Identification was performed using a ZEISS Stemi2000-c stereo microscope (Carl Zeiss Microscopy GmbH, Jena, Germany), with reference to identification guides such as “Pictorial Keys to Soil Animals of China” [19], “Identification of Agricultural Insects” [20], “The Color Iconography of Insects in Shenyang” [21], “Spiders from Agricultural Regions of China” [22], and “The Insect Family Trees” [23]. All species were identified to the family level. The epigaeic arthropod community was characterized using two primary indices: mean activity density (hereafter referred to as activity density) and a diversity index. Activity density was defined as the mean number of individuals captured per trap. Epigaeic arthropod diversity was expressed as the effective number of species (true diversity) derived from the Simpson’s dominance index conversion. Common index conversions to effective numbers of species are shown in

Table 1. Transformation between common indices and effective number of species (true diversity).

Index x:	Diversity Transformed by x	Diversity Transformed by pi
Species richness x ≡ ${\sum }_{\mathrm{i}=1}^{\mathrm{S}}$${\mathrm{p}}_{\mathrm{i}}^0$	x	${\sum }_{\mathrm{i}=1}^{\mathrm{S}}$${\mathrm{p}}_{\mathrm{i}}^0$
Shannon entropy x ≡ ${-\sum }_{\mathrm{i}=1}^{\mathrm{S}}$pilnpi	exp(x)	exp(${-\sum }_{\mathrm{i}=1}^{\mathrm{S}}$pilnpi)
Simpson index x ≡ ${\sum }_{\mathrm{i}=1}^{\mathrm{S}}$${\mathrm{p}}_{\mathrm{i}}^2$	1/x	1/${\sum }_{\mathrm{i}=1}^{\mathrm{S}}$${\mathrm{p}}_{\mathrm{i}}^2$
Gini-Simpson index x ≡ 1 − ${\sum }_{\mathrm{i}=1}^{\mathrm{S}}$${\mathrm{p}}_{\mathrm{i}}^2$	1/(1 − x)	1/${\sum }_{\mathrm{i}=1}^{\mathrm{S}}$${\mathrm{p}}_i^2$

Where S represents the total number of species, and p denotes the proportion of individuals of a given species (n) relative to the total number of individuals (N)(n/N).

[24].

2.3. Landscape Composition Heterogeneity and Spatial Configuration Heterogeneity

Composition heterogeneity was represented by the proportion of semi-natural habitat. Spatial configuration heterogeneity was represented by landscape metrics. To comprehensively understand the landscape characteristics of each sampling grid, metrics were selected at both the type level and landscape level for landscape pattern analysis: Patch Density (PD), Patch Richness (PR), Contagion Index (CONTAG), Landscape Division Index (DIVISION), Shannon’s Diversity Index (SHDI), and Shannon’s Evenness Index (SHEI). The formulas and ecological meanings of these landscape metrics are listed in

Table 2. Landscape metric formulas and ecological implications.

Landscape Index	Abbreviation	Formula	Ecological Implication
Patch Density	PD	$\mathrm{P}\mathrm{D}={\displaystyle \frac{N}{A}}$	Patch Density (PD) represents the density of a given patch type within the landscape and reflects landscape heterogeneity and fragmentation, allowing comparisons among landscapes of different sizes.
Patch Richness	PR	$\mathrm{P}\mathrm{R}=m$	Patch Richness (PR) represents the number of different patch types in the landscape and reflects landscape composition and heterogeneity, often showing a positive relationship with species richness.
Contagion Index	CONTAG	$\mathrm{C}\mathrm{O}\mathrm{N}\mathrm{T}\mathrm{A}\mathrm{G}=\left[1+{\displaystyle \frac{{\sum }_{i=1}^m{\sum }_{j=1}^m(p_{ij}\ln (p_{ij}))}{2\ln (m)}}\right]\times 100$	The Contagion Index (CONTAG) reflects the degree of aggregation and connectivity among patch types, with higher values indicating greater connectivity and lower values indicating increased landscape fragmentation.
Landscape Division Index	DIVISION	$\mathrm{D}\mathrm{I}\mathrm{V}\mathrm{I}\mathrm{S}\mathrm{I}\mathrm{O}\mathrm{N}=1-{\sum }_{i=1}^m{\sum }_{j=1}^n{\left({\displaystyle \frac{a_{ij}}{A}}\right)}^2$	The Landscape Division Index (DIVISION) reflects the degree of patch dispersion; values close to 0 indicate a single patch, whereas values approaching 1 indicate a highly fragmented landscape with many small patches.
Shannon Diversity Index	SHDI	$\mathrm{S}\mathrm{H}\mathrm{D}\mathrm{I}=-{\sum }_{i=1}^m[P_i\ln (P_i\left)\right]$	The Shannon Diversity Index (SHDI) reflects landscape heterogeneity and the diversity of patch types, with higher values indicating greater diversity and more even distribution.
Shannon Evenness Index	SHEI	$\mathrm{S}\mathrm{H}\mathrm{E}\mathrm{I}={\displaystyle \frac{-{\sum }_{i=1}^m{p}_i\ln (p_i)}{\ln (m)}}$	The Shannon Evenness Index (SHEI) reflects the evenness of patch type distribution, with values close to 1 indicating a more even distribution and low dominance, and lower values indicating dominance by one or a few patch types.

Note: PD = Patch Density; PR = Patch Richness; CONTAG = Contagion Index; DIVISION = Landscape Division Index; SHDI = Shannon Diversity Index; SHEI = Shannon Evenness Index.

[25]. FRAGSTATS 4.2 was used to calculate landscape metrics for each 1 km × 1 km grid.

2.4. Statistical Analysis

Non-metric Multidimensional Scaling (NMDS) was used to examine differences in the community composition of epigaeic arthropods at the family level, based on relative abundance data, among landscape units with varying proportions of semi-natural habitats. The effects of different levels of composition heterogeneity on the activity density and diversity of epigaeic arthropods were analyzed using one-way ANOVA. The Bioenv function was used to calculate the correlation between different semi-natural habitat structural characteristics (i.e., landscape metrics) and epigaeic arthropod communities. The results were tested using the Mantel function to identify the combination of landscape metrics that best explained the variation in epigaeic arthropods. VPA was applied to quantify the independent and shared contributions of composition and spatial configuration heterogeneity to epigaeic arthropod communities. VPA is an analysis that combines RDA and partial RDA to partition the variation of a response variable among two, three, or four explanatory variables. With two explanatory variables, VPA can effectively determine their common and unique contribution rates to the explained variable. Bioenv analysis and VPA were conducted in R (version 4.1.2) using the vegan package.

3. Results

3.1. Characteristics and Summary Statistics of Epigaeic Arthropods

In total, 4795 epigaeic arthropods were collected, representing 5 classes, 12 orders, and 51 families. The captured arthropods were categorized into dominant, common, and rare groups based on the percentage of the total catch. Dominant groups comprised ≥10% of the total catch; common groups comprised between 0.5% and 10%; rare groups comprised ≤0.5%. The dominant epigaeic arthropod groups in this study were Formicidae (ants), Carabidae (ground beetles), and Gryllidae (crickets), totaling 2734 individuals, accounting for 57% of the total catch. Common groups included 16 families such as Aphodiidae, Tenebrionidae, Hydrophilidae, Cicadellidae, and Chrysomelidae, totaling 1889 individuals, representing 40% of the total catch. The remaining 32 families were relatively rare, collectively accounting for 3% of the total catch. The types and numbers of epigaeic arthropods are listed in

Table A1. Types and quantities of epigaeic arthropods.

Epigaeic Arthropods			Taxa Codes	Numbers	Percentage	Dominance
Classes	Orders	Families
Insecta	Coleoptera	Carabidae	SP4	891	18.58	+++
		Aphodiidae	SP7	34	0.71	++
		Crioceridae	SP8	2	0.04	+
		Cleridae	SP9	1	0.02	+
		Elateridae	SP13	2	0.04	+
		Dytiscidae	SP15	1	0.02	+
		Nitidulidae	SP17	6	0.13	+
		Tenebrionidae	SP21	298	6.21	++
		Dermestidae	SP22	4	0.08	+
		Coccinellidae	SP23	7	0.15	+
		Geotrupidae	SP24	8	0.17	+
		Lucanidae	SP25	1	0.02	+
		Melolonthidae	SP28	2	0.04	+
		Mycetophagidae	SP31	5	0.10	+
		Meloidae	SP34	1	0.02	+
		Curculionidae	SP36	12	0.25	+
		Hydrophilidae	SP39	35	0.73	++
		Histeridae	SP40	2	0.04	+
		Chrysomelidae	SP42	66	1.38	++
		Staphylinidae	SP44	18	0.38	+
	Orthoptera	Oedipodidae	SP1	74	1.54	++
		Catantopidae	SP2	71	1.48	++
		Acrididae	SP12	18	0.38	+
		Gryllotalpidae	SP16	12	0.25	+
		Gryllidae	SP35	587	12.24	+++
		Tettigoniidae	SP49	1	0.02	+
		Pyrgomorphidae	SP51	110	2.29	++
	Hemiptera	Aradidae	SP3	159	3.32	++
		Pyrrhocoridae	SP10	4	0.08	+
		Nabidae	SP11	1	0.02	+
		Reduviidae	SP14	8	0.17	+
		Miridae	SP18	1	0.02	+
		Cydnidae	SP33	1	0.02	+
		Nepidae	SP37	1	0.02	+
		Coreidae	SP47	4	0.08	+
		Lygaeidae	SP48	1	0.02	+
	Homoptera	Cicadellidae	SP41	42	0.88	++
	Hymenoptera	Formicidae	SP43	1256	26.19	+++
	Dermaptera	Labiduridae	SP27	66	1.38	++
Arachnida	Araneae	Linyphiidae	SP20	1	0.02	+
		Theridiidae	SP26	155	3.23	++
		Salticidae	SP32	178	3.71	++
		Thomisidae	SP38	13	0.27	+
		Araneidae	SP46	7	0.15	+
		Trochanteriidae	SP50	1	0.02	+
	Opilliones	Protolophidae	SP19	190	3.96	++
Malacostraca	Isopoda	Oniscidae	SP5	158	3.30	++
Diplopoda	Polydesmoidae	Paradoxosomatidae	SP29	195	4.07	++
Chilopoda	Scutigeromorpha	Scutigeridae	SP45	18	0.38	+
	Lithobiomorpha	Geophilomorpha	SP6	8	0.17	+
		Lithobiidae	SP30	58	1.21	++
		total		4795	100

Note: +++ indicates that the number of individuals captured accounts for more than 10% of the total catch; ++ indicates 0.5–10% of the total catch; + indicates less than 0.5% of the total catch.

.

3.2. Effect of Composition Heterogeneity on Epigaeic Arthropods

The NMDS results showed a stress value of 0.18 (<0.2), indicating a good model fit suitable for data interpretation. In the NMDS ordination plot, smaller distances between sample points indicate more similar community structures of epigaeic arthropods within those samples. The distribution of epigaeic arthropod communities differed among gradients of semi-natural habitat proportion, i.e., different levels of composition heterogeneity. Grids with semi-natural habitat proportions of 0–5%, 5–10%, and 10–15% were primarily located on the right side of the first axis. Grids with semi-natural habitat proportions of 25–30% were primarily located on the left side of the first axis. Figure 3 shows that species composition was relatively similar within the same proportion range, but differed between different proportion ranges. However, the species similarity was high among the 0–5%, 5–10%, and 10–15% proportion ranges. The epigaeic arthropod community in the 25–30% semi-natural habitat proportion range had the lowest similarity with communities in other proportion ranges. The results indicate that community characteristics of epigaeic arthropods differ between low and high composition heterogeneity landscapes.

Significant differences in the activity density of epigaeic arthropods were detected among different proportions of semi-natural habitats based on one-way ANOVA. The activity density in grids with 25–30% semi-natural habitat was significantly higher than that in the three proportion intervals of 0–15%. There was no significant difference between these groups and the grids with 15–20% and 20–25% semi-natural habitat (Figure 4a). This suggests significant differences in activity density between high and low landscape heterogeneity. However, no significant effect of composition heterogeneity on the diversity of epigaeic arthropods was detected. Diversity did not show significant differences across the different semi-natural habitat proportions (Figure 4b).

3.3. Effect of Spatial Configuration Heterogeneity on Epigaeic Arthropods

The Bioenv function was used to analyze the correlation between different combinations of landscape metrics and epigaeic arthropod activity density and diversity, and the Mantel test was used to verify the results.

Table 3. Relationship between activity density of epigaeic arthropod and landscape spatial configuration heterogeneity based on Bioenv function.

Landscape Spatial Configuration Combination	Correlation Coefficient	Mantel Test (P)
PD	0.6482	0.02
PD DIVISION	0.7857	0.01
PD DIVISION SHDI	0.6857	0.03
PD CONTAG DIVISION SHDI	0.5821	0.04
PD CONTAG DIVISION SHDI SHEI	0.5607	0.04
PD PR CONTAG DIVISION SHDI SHEI	0.5107	0.07

shows the Bioenv results for different landscape metric combinations versus activity density and diversity. The results indicated that the combination of PD and DIVISION was the best model, with a correlation of 0.7857 with activity density. This was followed by the combination of PD, DIVISION, and SHDI, with a correlation coefficient of 0.6857 with activity density. The model with the lowest correlation coefficient with activity density was the one containing all variables. This suggests that patch number is a key factor influencing activity density at the landscape scale, as both PD and DIVISION reflect the quantitative characteristics of patches within a landscape unit. At the landscape scale, having multiple small patches in a unit of equal area can maintain higher activity density than having a single large patch. The Mantel test results in Table 3 showed that the effect of the PD and DIVISION combination on activity density was highly significant (p < 0.01), the effect of the full variable set was not significant (p > 0.05), and the other combination models showed significant correlations with activity density.

3.4. Quantifying the Effects of Landscape Composition and Spatial Configuration on Epigaeic Arthropod Communities

VPA was used to assess the independent and joint effects of landscape composition and spatial configuration on epigaeic arthropods. Using epigaeic arthropod activity density as the response variable, and the semi-natural habitat proportion (representing composition heterogeneity) and the optimal spatial configuration model (PD and DIVISION) as explanatory variables, the independent and common contribution rates of composition and spatial configuration heterogeneity to activity density were assessed. The results of the VPA are presented in Figure 5 and

Table 4. Variance variation decomposition of landscape composition and landscape spatial configuration.

Variable	df	Explanatory Rate
Landscape Composition	2	0.03
Landscape spatial configuration	1	0.07
Landscape Composition + Landscape spatial configuration	3	0.61

.The results indicate that epigaeic arthropod activity density at the landscape scale largely depends on both landscape composition and spatial configuration, and their joint effect best explains the variation in activity density. However, the explanatory power of composition and spatial configuration also differed, with spatial configuration having a higher explanatory rate. In our results, the independent explanatory rate of landscape spatial configuration was 7%, while that of landscape composition was 3%. Their common explanatory rate was 61%. This indicates that the influence of landscape spatial configuration on epigaeic arthropods is greater than that of composition. The role of landscape spatial configuration in maintaining activity density is more pronounced. Furthermore, their combined effect cannot be ignored. Landscape heterogeneity is mainly reflected in composition and spatial configuration. Our results show that their common explanatory rate accounts for the majority of the impact on epigaeic arthropods.

4. Discussion

Differences in semi-natural habitat proportion, as an indicator of landscape composition, were reflected in the activity density and diversity patterns of epigaeic arthropods, the impact of different combinations of landscape spatial configuration metrics on epigaeic arthropods at the landscape scale, along with their individual and combined contributions to shaping epigaeic arthropod communities in agricultural landscapes. The results show that epigaeic arthropod community characteristics differ among landscape units with different levels of semi-natural habitat proportion, which is consistent with recent studies showing that both compositional and spatial configuration heterogeneity significantly shape arthropod community structure in agricultural landscapes [26]. Epigaeic arthropod activity density in landscape units with high composition heterogeneity was significantly higher than in those with low composition heterogeneity. However, arthropod diversity did not exhibit a similar pattern. Bioenv analysis revealed that, in terms of landscape spatial configuration heterogeneity, the combination of the patch density (PD) and landscape division index (DIVISION) constituted the best explanatory model for epigaeic arthropod activity density. Epigaeic arthropod activity density at the landscape scale in agricultural landscapes largely depends on both landscape composition and spatial configuration heterogeneity, with their joint effect providing a better explanation. However, the explanatory power of these two factors for activity density varied, with landscape spatial configuration explaining a greater proportion of the variance.

A higher proportion of semi-natural habitats contributes to greater biodiversity and improved ecosystem functioning in agricultural landscapes [9,27], thereby contributing to agricultural sustainability. This pattern is further supported by recent global meta-analyses demonstrating that complex agricultural landscapes consistently sustain higher levels of biodiversity than simplified ones [28]. Our findings demonstrate that landscapes with higher compositional heterogeneity tend to support greater activity density of epigaeic arthropods, although no significant response was observed in diversity. Increasing semi-natural habitat proportion within the landscape to 15–20% leads to an increase in arthropod activity density, with this effect becoming significant at 25–30%. Some studies suggest that a semi-natural habitat proportion of approximately 5% appears to represent a threshold at which landscape heterogeneity in agricultural systems becomes relatively high [29]. However, a semi-natural habitat proportion of 5% is insufficient to support biodiversity conservation in agricultural landscapes. Many “greening” measures have been implemented in Europe to preserve landscape features or ensure a minimum area of Ecological Focus Areas (EFAs) at the farm level. In recent years, agricultural policy in Europe has increasingly emphasized environmental sustainability. Reforms of the Common Agricultural Policy (CAP) have introduced a framework that integrates ecological considerations into agricultural management, particularly through the incorporation of enhanced baseline requirements and voluntary eco-schemes. These measures encourage farmers to allocate a portion of their land to non-productive elements, such as semi-natural habitats, with the aim of improving landscape heterogeneity and supporting biodiversity within agricultural systems. Under the post-2020 CAP proposal (2021–2027), EFAs in their existing form were to be replaced by a requirement for Member States to designate a minimum proportion of agricultural land as non-productive features or areas. Continuing with the 5% minimum threshold likely offers little additional benefit for semi-natural habitat conservation on most Irish farms or across wider European farmland. Even with self-determined minimum shares, the proportion is unlikely to exceed 10%. Recent research also found that similar schemes (e.g., ESA, covering 5% of arable area) did not yield significant benefits for biodiversity [30]. This may be explained by the fact that a semi-natural habitat proportion of 5% at the landscape scale is insufficient to sustain multi-trophic epigaeic arthropod communities, and partly because landscape composition is only one factor influencing epigaeic arthropods in agricultural landscapes; their activity density and diversity also depend on habitat quality, landscape spatial configuration, etc. Recent studies further confirm that in intensive agricultural landscapes, when the proportion of semi-natural habitats falls below 10%, their capacity to maintain biodiversity significantly diminishes, landscapes tend toward biotic homogenization, and gains in community compositional diversity are limited [31]. Specifically within farmland ecosystems, research indicates that if the proportion of semi-natural habitats is below 15–20%, their function as biological habitats and ecological corridors is substantially weakened, making it difficult to effectively support long-term stability of natural enemy populations (e.g., predatory arthropods) [32]. Recent EU policies also emphasize the importance of substantially increasing habitat proportions. The EU Nature Restoration Law intends to restore 20% of land and sea ecosystems by 2030 and is committed to rehabilitating degraded ecosystems to mitigate biodiversity loss, rather than merely meeting low-proportion habitat conservation targets.

Landscape spatial configuration heterogeneity explained a greater proportion of the variance in epigaeic arthropod activity density than composition heterogeneity. Increased spatial configuration heterogeneity promotes the presence of individual arthropods. This supports the landscape-moderated trait selection hypothesis [33].The more heterogeneous the spatial configuration of a landscape unit, the greater the number of coexisting species or functional groups expected. The impact of landscape composition heterogeneity on the diversity of biological taxa has been widely demonstrated. For instance, composition heterogeneity has been shown to strongly drive bird population declines in France. However, landscape spatial configuration heterogeneity has an equally significant effect on biodiversity in agricultural landscapes. Studies show that spatial configuration heterogeneity shapes the functional composition of grassland butterflies [34]. The combined model of patch density (PD) and landscape division index (DIVISION) is used to characterize landscape spatial configuration in this study, best explained the variation in epigaeic arthropod activity density (R = 0.7857; p = 0.01). This aligns with previous research suggesting that landscapes with smaller average patches and larger boundary areas are more conducive to the survival of biological communities. Patch size and geometric complexity affect relative edge length, which is hypothesized to enhance diversity, as boundary areas offer unique resources for species at various life stages. Edge-biased distribution of epigaeic arthropods has also been confirmed. Relevant studies found that patch density (PD) has a positive correlation with the richness and abundance of ground beetles. The mechanism lies in that combinations of small patches significantly increase the total length of suitable edge habitats, providing diverse microhabitats for arthropods with different ecological niches [35]. This indicates that, per unit area, multiple small patches can provide richer edge resources than a few large patches, better supporting the survival and development of epigaeic arthropod populations. Furthermore, ensuring landscape connectivity is a key element in designing and planning agricultural landscapes. Research suggests that creating a cohesive ecological network of semi-natural landscape elements enhances predation and parasitism, potentially reducing pest damage to crops [36]. Constructing corridors between different semi-natural habitats to form an interconnected ecological network within landscape units is an important approach to enhancing biodiversity in agricultural landscapes. However, recent studies indicate that excessive connectivity may lead to high synchrony in pest or natural enemy populations at the landscape scale, increasing the risk of regional collapse. Conversely, moderate spatial isolation may cause arthropod populations in different patches to fluctuate asynchronously due to local environmental disturbances [37]. Related research suggests that the dispersed distribution of semi-natural habitats helps maintain spatiotemporal heterogeneity in natural enemy populations, thereby providing more stable biological pest control services at the landscape scale [38]. This indicates that, within a landscape unit of the same area, a spatial configuration of multiple small cropland patches interspersed with semi-natural habitats supports the migration, foraging, and habitation of epigaeic arthropods, thereby supporting higher activity densities.

Landscape composition and spatial configuration heterogeneity influence epigaeic arthropods through both independent effects and joint interactions. VPA results show that composition and spatial configuration heterogeneity jointly explained 61% of the variance in epigaeic arthropod activity density, with spatial configuration heterogeneity independently explaining 7% and composition heterogeneity explaining 3%. Landscape composition and spatial configuration heterogeneity are considered strong filters for epigaeic arthropods [39]. Studies have shown that composition heterogeneity primarily acts as an environmental filter shaping arthropod community structure within a landscape unit; reducing composition heterogeneity at the landscape scale naturally selects for more specialized and homogeneous species. Another study found that landscape composition heterogeneity (measured by the proportion of semi-natural habitats) had a strong influence on the composite traits of carabids [40]. For example, smaller carabids decreased with reduced composition heterogeneity, while larger carabids tended to be associated with increased composition heterogeneity. The impact of landscape spatial configuration heterogeneity on epigaeic arthropod communities mainly exhibits scale dependency. At different scales, the same spatial configuration characteristic can have opposing effects on arthropods. Nevertheless, the ecological role of spatial configuration heterogeneity in agricultural landscapes is a recognized fact, as it can provide more opportunities for animal spillover [41]. Larger patches with fewer surrounding land-use types provide fewer spillover opportunities. Composition and spatial configuration heterogeneity have distinct roles and implications in terms of ecological benefits and production system management. However, their combined promoting effect on epigaeic arthropods is certain. Therefore, it is crucial to consider both landscape composition and spatial configuration or the conservation of epigaeic arthropods at the agricultural landscape scale. Considering only one aspect does not adequately reflect its impact on arthropod activity density.

5. Conclusions

At the landscape scale, landscape heterogeneity typically includes two aspects: composition heterogeneity and spatial configuration heterogeneity. Both have separate and joint effects on epigaeic arthropods. In this study, to understand how landscape heterogeneity affects epigaeic arthropods at the landscape scale, the proportion of semi-natural habitat within a landscape unit was used as the measure of composition heterogeneity, and landscape metrics were used as the measure of spatial configuration heterogeneity. One-way ANOVA together with the Bioenv procedure were used to examine how compositional and spatial configurational heterogeneity relate to the activity density and diversity of epigaeic arthropods. VPA was applied to quantify the independent and common explanatory rates of composition and spatial configuration heterogeneity on epigaeic arthropods. The results suggest the following: (1) Significant differences exist in the activity density of epigaeic arthropods among landscape units with different composition heterogeneity, but no such significant difference was observed for diversity. Higher composition heterogeneity tended to be associated with higher activity density, while low composition heterogeneity tends to show lower activity density. (2) For spatial configuration heterogeneity, an optimal explanatory model exists for its effect on activity density. PD and DIVISION formed the model with the highest explanatory power for activity density. This indicates that regarding landscape spatial configuration, the number of patches within a landscape unit is crucial for activity density. (3) The explanatory power of landscape composition and spatial configuration on epigaeic arthropods differs. The independent explanatory rate of composition was 3%, while that of spatial configuration was slightly higher at 7%. Their common explanatory rate reached 61%. This indicates that both composition and spatial configuration heterogeneity jointly determine epigaeic arthropod activity density. This study demonstrates that it is necessary to consider both landscape composition and spatial configuration when optimizing agricultural landscape planning. Based on maintaining a certain level of composition heterogeneity within landscape units, ensuring an appropriate number of patches is a key focus of landscape management. This study has several limitations that should be acknowledged. First, arthropods were identified at the family level, which may not fully capture species-level biodiversity patterns, particularly species turnover within families. This limitation may lead to an underestimation of ecological responses to landscape heterogeneity. However, higher taxonomic levels have been shown to effectively reflect community responses to environmental gradients, especially in large-scale ecological studies where species-level identification is constrained [42]. Second, the sampling period was relatively short and may not fully capture seasonal dynamics of arthropod communities. Future studies should incorporate longer-term monitoring and species-level identification to better understand ecological responses to landscape heterogeneity. In addition, further research should explore multi-scale effects and integrate functional traits to improve the mechanistic understanding of biodiversity patterns in agricultural landscapes.