Assessing Landscape-Level Biodiversity Under Policy Scenarios: Integrating Spatial and Land Use Data

Abstract

Spatially explicit tools are essential for assessing biodiversity and guiding land use decisions at broad scales. This study presents a national-level approach for evaluating habitat quality as a proxy indicator for biodiversity, using Latvia as a case study. The approach integrates land use data, landscape structure, and habitat characteristics to generate habitat quality indices for agricultural and forest land. It addresses a common limitation in biodiversity planning, namely, the lack of consistent species-level data, by providing a comparative and conceptually robust way to assess how different land use types support biodiversity potential. The methodology was applied to assess current habitat quality and to simulate changes under two policy-relevant land use scenarios: the expansion of protected areas and a shift to organic farming. Results showed that expanding protected areas increased the national habitat quality index by 8.47%, while conversion to organic farming produced a smaller but still positive effect of 0.40%. Expansion of protected areas, therefore, led to a greater improvement in habitat quality compared to converting farmland to organic systems. However, both strategies offer complementary benefits for biodiversity at the landscape scale. Although national-level changes appear moderate, their spatial distribution enhances connectivity, particularly near existing protected areas, and may facilitate species movement. This approach enables national-level modelling of biodiversity outcomes under different policy measures. While it does not replace detailed species assessments, it provides a practical and scalable method for identifying conservation priorities, particularly in regions with limited biodiversity monitoring capacity.

1. Introduction

Biodiversity, as defined by the Convention on Biological Diversity adopted at the United Nations Earth Summit in Rio de Janeiro in 1992, refers to the vast array of living organisms found across different environments, including land, marine, and freshwater ecosystems, along with their associated ecological systems [1]. This diversity manifests at multiple levels within species, among species, and across ecosystems. The primary categories of biodiversity, genetic, species, ecosystem, and landscape diversity, are interconnected, though they can also be studied independently.

Genetic diversity represents the variation in traits among individuals of the same species and their ability to adapt to environmental shifts [2]. This type of diversity plays a fundamental role in species survival and evolutionary processes, particularly in response to ecological changes and external pressures [3]. For example, maintaining genetic variation in agricultural crops is vital for breeding efforts aimed at enhancing disease resistance and resilience to climate fluctuations [4].

Species diversity, the most commonly referenced level, describes the number and significance of species (their abundance and richness) within a particular area [5,6,7]. At the ecosystem level, biodiversity characterises the interactions among different species living in the same environmental conditions. At the landscape level, understanding biodiversity is the most complex because it may encompass multiple ecosystems and their interactions, depending on the specific area. High species diversity often indicates a healthy and resilient ecosystem. For instance, tropical rainforests and coral reefs are known for their exceptional species diversity and are critical for global biodiversity [8,9]. These ecosystems support a vast array of species that perform essential ecological roles and provide valuable resources for humans.

At the ecosystem level, biodiversity characterises the interactions among different species living in the same environmental conditions. This involves understanding the roles species play within ecosystems, particularly keystone species, which are crucial for maintaining ecological balance [10]. Ecosystem diversity is vital for the provision of ecosystem services that humans rely on, such as food, feed, fibre, water purification, nutrient cycling, and climate regulation [11,12,13].

At the landscape level, understanding biodiversity is the most complex because it may encompass multiple ecosystems and their interactions, depending on the specific area [14]. Landscape diversity considers the arrangement and connectivity of different habitats and their influence on ecological processes and species distributions [15]. For instance, a landscape that includes forests, wetlands, and grasslands will support a wider variety of species and ecological functions than a homogenous landscape [16].

Understanding and preserving biodiversity at all these levels is essential for maintaining the health and stability of the biosphere. However, evaluating biodiversity across genetic, species, ecosystem, and landscape levels is challenging due to the complexity, time, and extensive data required. There is a growing need for efficient, spatially explicit methods that can integrate land use and ecological data to support biodiversity monitoring and planning.

In response to this challenge, numerous recent studies across Europe have employed landscape-scale or nation-wide modelling approaches to link biodiversity with land use and habitat structure. For instance, Dou et al. (2021) developed a continental-scale land systems typology for Europe that integrates landscape characteristics and land use intensity, offering valuable insights into spatial heterogeneity [17]. However, the system does not directly capture local species composition or habitat quality and may oversimplify ecological differences within classes [17]. Similarly, Virkkala et al. (2022) mapped forest biodiversity potential in Finland using indicator bird species, providing a strong example of species-based modelling at the national scale [18]. Still, the approach requires high-quality species data and may not be easily transferable to contexts with limited monitoring capacity [18]. Auffret et al. (2018) demonstrated that land use intensification has led to large-scale declines in grassland specialist flora across southern Sweden [19]. While this highlights the urgency of integrating land use change in biodiversity assessments, the study focuses on floristic data and does not address broader habitat quality indicators [19]. Andersson et al. (2022) further confirmed that mosaic forest-farmland landscapes with grazed semi-natural pastures support high levels of biodiversity in northern Europe, particularly for plants, bumblebees, and butterflies [20]. However, their study is based on intensive fieldwork in selected landscapes and may not scale easily to national-level planning frameworks [20].

Together, these studies highlight the critical role of land cover composition, landscape configuration, and land use intensity in shaping biodiversity outcomes at broad spatial scales. However, they also reveal key methodological challenges, such as limited taxonomic scope, the need for detailed and high-resolution species data, and difficulties in applying results across different geographic or ecological contexts. These constraints limit the scalability and operational use of such approaches in regions where monitoring data are sparse or fragmented.

To address the growing need for proxy-based frameworks that balance ecological relevance with practical applicability, this study adopts habitat quality as a proxy indicator. Habitat quality reflects the capacity of a landscape to support diverse and viable species populations and is increasingly used in spatial conservation planning [21,22]. Importantly, this method provides a relative indicator of biodiversity potential rather than a direct measure of species richness. This approach enables the evaluation of biodiversity potential using available land use data at broad spatial scales. Latvia was selected as a national case study to apply and demonstrate this approach. The country presents a relevant context due to its ecological heterogeneity, high proportion of natural and semi-natural habitats, and active engagement in EU biodiversity policy implementation. While species monitoring efforts are ongoing in Latvia, for taxa such as birds, butterflies, moths, dragonflies, amphibians, and mammals, they are taxon-specific, spatially limited, and not explicitly designed to link species occurrences with land use types at the polygon level. For example, bird monitoring is conducted along fixed 4 km transects selected to ensure representative coverage across the country, based on a 50 × 50 km national grid, within which observations are carried out in four selected 5 × 5 km cells. However, detailed information on land use or management practices is not systematically recorded alongside these observations, limiting their integration into spatial biodiversity modelling. Invertebrate monitoring includes basic land use classification within a 100 m radius around survey transects, but these surveys are only conducted in areas with suitable habitats, thereby introducing spatial bias. Furthermore, monitoring intensity and expert capacity vary over time, and improvements in detection methods or observer expertise can affect data comparability across years, potentially inflating perceived biodiversity trends [23,24]. To overcome the lack of comprehensive species occurrence data, particularly for large-scale applications, this study applies a spatially explicit habitat quality index. In addition to establishing a baseline of habitat quality, the approach supports scenario analysis. By applying it to alternative land use futures, including policy-driven changes, we assess how different targets may influence biodiversity. This scenario-based approach supports proactive decision-making by enabling stakeholders to explore potential outcomes and develop strategies tailored to specific ecological and geographical contexts. The Latvian case study not only demonstrates the practicality of the method but also highlights the importance of spatial prioritisation and context-specific conservation planning.

2. Materials and Methods

2.1. Case Study

Latvia is a Northern European country covering approximately 64,600 km², situated in the Boreal biogeographic region [25]. Its landscape is composed of forests (more than 50% of the land area), agricultural land, wetlands, grasslands, inland waters, and coastal habitats. Latvia is recognized for its relatively high proportion of natural and semi-natural habitats, including extensive peat bogs and species-rich grasslands. This ecological diversity, together with ongoing land use changes and policy commitments under the EU Biodiversity Strategy, makes Latvia a suitable case study for assessing habitat quality and modelling conservation scenarios at the national scale.

2.2. Agricultural Land

The evaluation of habitat quality in agricultural landscapes in this study is based on the ecosystem quality framework developed by Reidsma et al. (2006) [26], which defines ecosystem quality as the capacity of a system to sustain biodiversity under natural conditions. In this framework, values range from 0%, indicating a fully degraded ecosystem with no wild species, to 100% representing a natural, undisturbed state. This approach has been applied in previous landscape-scale assessments as a relative indicator of biodiversity, not an absolute measure, as it simplifies complex ecological processes into a conceptual scoring system suitable for spatial modelling [13].

Given the well-documented link between agricultural land use intensity and biodiversity loss [10,27,28,29], habitat quality values are assigned based on dominant land use types and typical management practices in Latvia, such as conventional versus organic farming, arable land versus grasslands, and areas under protection. These assignments draw on published empirical studies, including Katayama et al. (2019), which show that orchard systems under low-input or organic management support higher species richness across multiple taxa [30], and Horak et al. (2013), who found that traditional fruit orchards contribute significantly to structural habitat diversity and serve as biodiversity reservoirs in rural landscapes [31].

Meta-analyses consistently show that organic and low-intensity systems support higher biodiversity than conventional farming. Tuck et al. (2014) reported that organic management increases species richness by about 30% across multiple taxa, with the strongest effects in intensively managed arable landscapes [32], while Sanders et al. (2025) highlighted broader environmental and resource-conservation benefits across temperate climates [33]. Grassland systems in particular are recognised as biodiversity hotspots: extensively managed meadows and pastures host diverse plant and pollinator communities [34,35]. This evidence underpins the higher values assigned in Table 1 to semi-natural grasslands and other extensively managed farmland. Semi-natural grasslands maintain high biodiversity under both conventional and organic systems when grazing or mowing remains extensive, while perennial plantations can enhance heterogeneity in arable-dominated settings but may reduce diversity if they replace semi-natural habitats in mosaic landscapes [30]. Uncultivated patches can provide habitat opportunities in intensively farmed arable systems but are often degraded or hold limited ecological value in forest-dominated or mosaic landscapes [36]. Grassland overgrowth by shrubs typically reduces open-habitat specialists [37], peat extraction fields represent highly degraded conditions [38], and protected bogs achieve maximum values due to their status as internationally recognised biodiversity hotspots [39].

Table 1. Habitat quality index for different agricultural systems in relation to landscape diversity, farming systems, and land use in Latvia, where “1” reflects very intensive land use with almost no biodiversity potential and “8−10” reflects semi-natural or protected conditions with very high biodiversity potential.

Agricultural System in Latvia	Homogeneous (Arable) Landscape		Heterogeneous Landscape, Homogeneous (Forest) Landscape
	Conventional	Organic	Conventional	Organic
Grains, oilseeds, pulses	1	2	3	4
Vegetables and potatoes	1	2	3	4
Perennial plantations	3	4	1	2
Fallow	1	2	3	4
Other crops	1	2	3	4
Grasslands	1	2	3	4
Meadows and pastures	8		8
Managed agricultural land (not used for production)	10		10
Uncultivated	10		0	0
Overgrown	5		0	0
Peat extraction fields	1
Protected sites, bogs	10		10

At the same time, biodiversity responses are context-dependent and vary among taxonomic groups. Some studies find stronger benefits of organic farming in homogeneous arable landscapes compared with structurally complex mosaic ones [27,36]. To reflect this, landscape heterogeneity was incorporated as a modifying factor in the modelling framework. Consequently, the fixed values in Table 1 should be understood as relative proxies that capture the comparative ecological potential of land use types, rather than precise measures of species richness. Their limitations, including the lack of taxon-specific calibration, are addressed in Section 4.2.

To refine the spatial allocation of high-quality habitats, additional geospatial layers from the Nature Data Management System “Ozols” were integrated, including the boundaries of specially protected nature territories, their functional zones, and habitat types [40]. Polygons where no economic activity is permitted, such as strict nature reserves or core zones of protected areas, are assigned the maximum habitat quality index (10 points), reflecting near-natural conditions with minimal human disturbance. This modelling assumption is consistent with findings by Geldmann et al. (2013), who demonstrated that areas with stricter protection levels are generally more effective in maintaining biodiversity [41]. Conversely, for land polygons classified as ecologically valuable but lacking explicit information on restrictions related to economic activities, the habitat quality index is interpolated following the approach used by Valujeva et al. (2020) [13].

Another factor affecting habitat quality is landscape diversity or heterogeneity [36,42,43,44]. Accordingly, the landscape is classified into three groups: a heterogeneous (mosaic) landscape, a homogeneous landscape dominated by arable land, and a homogeneous landscape dominated by forests. To determine whether a landscape is homogeneous or heterogeneous, the entire territory of Latvia is divided into 100-hectare grids. If the area of arable land within a grid exceeds 70%, the territory is considered homogeneous with an arable land dominance. Conversely, if the forest area within a grid exceeds 70%, the territory is classified as homogeneous with a forest dominance. Areas that do not meet these criteria are considered as heterogeneous landscapes. This classification approach follows the methodology applied by Dou et al. (2021), who used a comparable 70% threshold to distinguish dominant land systems at the 1 km² scale across Europe [17]. Applying this threshold allows for a consistent and ecologically meaningful delineation of landscape types in biodiversity assessments.

The habitat quality index values assigned to different agricultural systems, based on land use type, farming system, and landscape diversity, are presented in Table 1.

2.3. Forest Land

The evaluation of biodiversity at the polygon level in forested areas is constrained by a lack of comprehensive biodiversity data. Bird habitat quality is, therefore, selected as a proxy indicator, as birds are widely recognised as reliable bioindicators of forest biodiversity due to their sensitivity to stand structure, their relatively well-studied ecology, and the availability of long-term monitoring data [18,45]. This quality is assessed based on key forest characteristics, including economic activities within the forest stand, stand age, and the dominant tree species (Table 2). Based on the dominant tree species, forests are grouped into seven categories: pine, spruce, birch, black alder, grey alder, aspen, and other species. Forest stands are further classified into six age categories: clear-cut area, young stand, middle-aged stand, maturing stand, mature stand, and over-mature stand. Each tree species has a specific age range corresponding to each age category. This classification system enables a more precise evaluation of bird habitat quality and its correlation with forest management practices, providing valuable insights into biodiversity conservation within different forest types.

Table 2. Age groups for different tree species, years.

Species	Clear-Cut Area	Young Stand	Middle-Aged Stand	Maturing Stand	Mature Stand	Over-Mature Stand
Pine	0–2	3–40	41–80	81–100	101–140	141+
Spruce	0–2	3–40	41–60	61–80	81–120	121+
Birch	0–2	3–20	21–60	61–70	71–90	91+
Aspen	0–2	3–10	11–30	31–40	41–60	61+
Black alder	0–2	3–20	21–60	61–70	71–90	91+
Grey alder	0–2	3–10	11–25	26–30	31–40	41+
Other species	0–2	3–40	41–60	61–80	81–120	121+

The assessment of habitat quality for forest birds in this study is carried out with input from the leading ornithologists representing the main institutions involved in bird studies in Latvia: the University of Latvia and JSC “Latvian State Forests”. Based on expert knowledge and field experience, correlations are established between bird species richness and forest stand age classes for each dominant tree species. These relationships reflect how attractive different forest stands are to bird communities.

In forests, habitat quality is assumed to increase progressively as the stand matures and transitions through successive age classes. This assumption reflects the general trend that older forests develop structural complexity, such as deadwood, cavities, and multi-layered canopies, that provides habitats for a wider range of bird species [46,47]. The rate of improvement varies among dominant tree species, depending on their ecological characteristics. For example, aspen and alder are assigned higher maximum scores because they typically support greater cavity-nesting and insectivorous bird diversity compared with spruce or birch [48]. Over-mature stands are generally assigned the highest habitat quality scores, although the maximum possible score differs by tree species to reflect their varying biodiversity contributions (Table 3). It is acknowledged, however, that some bird species benefit from young or early-successional stands, while others depend on mature or old-growth forests [49,50]. The progressive increase presented in Table 3, therefore, represents a simplified proxy for average biodiversity potential rather than a species-specific response.

Table 3. Assessment of habitat quality in forest stands, points.

Species	Clear-Cut Area	Young Stand	Middle-Aged Stand	Maturing Stand	Mature Stand	Over-Mature Stand
Pine	1	2	3	6	7	8
Spruce	1	2	3	5	7	8
Birch	1	2	3	5	6	7
Black alder	1	2	4	7	8	9
Grey alder	1	2	3	4	5	6
Aspen	1	2	5	7	9	10
Other species	1	2	5	7	9	10

The habitat quality scores presented in Table 3 were not developed within this study but are compiled based on pre-existing expert evaluations. These expert-derived values are used as input data for spatial modelling and reflect prior knowledge about the relationships between forest structure and bird diversity in Latvia. Their limitations, including the oversimplification of non-linear and taxon-specific responses, are further discussed in Section 4.2.

2.4. Data Sources

This study is based on spatially explicit land use data obtained from official national sources. Agricultural land use information is sourced from the Rural Support Service of Latvia, where most farmers submit spatial data on their cultivated crops through annual declarations required by the Common Agricultural Policy regulations [51]. This dataset provides detailed spatial information on cultivated crops and distinguishes between conventionally and organically managed fields, which is essential for scenario analysis.

Forestry data are sourced from the State Forest Service of Latvia, which manages the State Forest Register and ensures the quality and accuracy of forest inventory information [52]. All spatial datasets used in this study reflect land use conditions as of the year 2024 and include approximately 1.2 million agricultural land polygons and 4.0 million forest land polygons.

In addition, data from the national Nature Data Management System “Ozols” are integrated to identify whether specific land polygons fall within designated protected areas. This database includes the spatial boundaries of specially protected nature territories, their functional zones, and habitat classifications.

All datasets used in this study are publicly available through the Latvian Open Data Portal [53].

2.5. Scenarios Description

To evaluate the potential impact of future land use and management policies on biodiversity, two scenario-based simulations are developed: (1) Scenario 1: Expansion of protected areas; and (2) Scenario 2: Increase in organic farming. These scenarios are derived from the targets set by the EU Biodiversity Strategy for 2030, which outlines commitments for the expansion of protected areas and the promotion of sustainable agricultural practices, particularly organic farming. Both scenarios are implemented within the same spatial framework used in the baseline assessment.

Scenario 1 simulates an increase in the extent of protected areas in Latvia, aligning with the EU Biodiversity Strategy for 2030, which sets a target of protecting at least 30% of the total land area of the EU, including 10% under strict protection. In this scenario, additional protected territories are designated on agricultural land to reflect this objective. The selection of new areas for protection is based on current habitat quality assessments and focuses on regions already recognised as ecologically valuable or located near existing protected sites [40]. Specifically, areas with high habitat quality indices and regions surrounding current protected zones are prioritised for inclusion. In the context of this study, all newly designated protected areas under Scenario 1 are treated as strictly protected, where economic activities are either entirely excluded or significantly limited. Under such management, these areas are expected to maintain minimal human disturbance and high ecological integrity. For modelling purposes, they are, therefore, assigned the maximum habitat quality index value (10 points), reflecting conditions characteristic of undisturbed or near-natural habitats. This approach reflects the objective of maximizing biodiversity potential in areas selected for formal protection and is consistent with the ecological role such areas are intended to fulfil within conservation networks. Protected areas with higher levels of legal restriction and effective management have been shown to better maintain biodiversity and prevent habitat loss, reinforcing the rationale for modelling newly protected areas under strict conservation regimes [41].

Scenario 2 models the transition from conventional to organic farming, aiming to meet the EU Biodiversity Strategy target of having at least 25% of EU agricultural land under organic farming by 2030. In Latvia, the share of organic agricultural land was 15.97% in 2022 and decreased slightly to 15.65% in 2024 [54]. While various socio-economic and institutional drivers, such as market demand, policy incentives, advisory support, and farmer attitudes, can influence farmers’ decisions to adopt organic farming [55], the aim of this study is not to explore these drivers in detail or prescribe where such transitions should occur. Rather, the modelling approach focuses on evaluating potential biodiversity gains under a scenario where the national organic farming target is met.

To operationalise this, the scenario simulates organic conversion within cereal-dominated areas, as cereals represent a major share of conventional agricultural production in Latvia. This approach aligns with findings by Flick et al. (2012), who suggest that the positive impact of organic farming on biodiversity is strongest in landscapes where intensive agriculture is the main land use [27]. Spatial data from the Rural Support Service is used to identify conventional cereal fields. From this data, randomly selected cereal field polygons currently under conventional management are converted to organic systems until the target increase of +9.35% of agricultural land is reached. This approach reflects a realistic and unbiased distribution of potential conversions, without assuming specific policy mechanisms or location-based interventions. Converted fields are assigned a higher habitat quality index, as outlined in Table 1, according to their landscape context (homogeneous or heterogeneous), thereby enabling assessment of spatial impacts on biodiversity potential.

2.6. Technical Implementation

Habitat quality modelling and spatial data processing are conducted using a combination of R 4.5.1, Python 3.14, and a PostgreSQL 17 database with the PostGIS 3.5.3 extension. Initial data manipulation and spatial operations are carried out in R, within the RStudio environment, using packages such as dplyr (data processing), sf and lwgeom (spatial data handling), and data.table (efficient large-table operations).

For performance optimisation, especially with large datasets, data are imported into a PostgreSQL/PostGIS environment. Here, Python scripts, using the psycopg2 library, are used to run SQL queries, monitor progress, and manage memory more effectively. This setup allows for faster processing and parallel execution where applicable.

2.7. Modelling Approach

The modelling process starts with the baseline habitat quality assessment. Each land polygon in Latvia is assigned a habitat quality index based on its land use and management characteristics, using values from Table 1 for agricultural systems and Table 3 for forest stands. This index is then multiplied by the area of the respective polygon to calculate its total habitat quality score in points. These point values are spatially aggregated within a 100-hectare grid, primarily for visualisation purposes, by summing all scores of polygons located within each grid cell to create a national-scale baseline habitat quality map.

For each scenario, a set of polygons is then selected for modification. In Scenario 1, areas adjacent to existing protected sites and with high habitat quality were chosen for simulated designation as new strictly protected areas. In Scenario 2, randomly selected conventionally managed cereal fields are converted to organic management. For the selected polygons, updated habitat quality indices are applied, again following the classifications in Table 1 and Table 3. These new scores are multiplied by polygon area and aggregated within the same 100-hectare grid framework to generate habitat quality maps after scenario implementation.

The effect of each scenario is calculated by subtracting the post-scenario habitat quality value from the baseline value for each grid cell. The relative impact is then expressed as a percentage change.

A simple sensitivity test is performed for the baseline scenario by adjusting habitat quality indices by ±1 point for selected key forest classes. These included middle-aged birch stands, maturing pine stands, mature pine and birch stands, and over-mature aspen stands. The key forest classes are identified by multiplying the habitat quality index values from Table 3 with the corresponding forest area from Table A1 and selecting the five classes contributing the highest total habitat quality points. To assess spatial sensitivity in Scenario 2, the baseline random allocation of organic conversion is compared to an alternative clustered allocation, where conversions are concentrated in homogeneous arable landscapes. In addition, to evaluate uncertainty and test the robustness of both scenarios, a Monte Carlo simulation with 200 iterations is conducted. In each iteration, different combinations of land parcels, based on each scenario’s assumption, are randomly selected for conversion to either protected areas or organic farming. Habitat quality is recalculated accordingly. From these simulations, the mean, standard deviation, and 95% confidence intervals are computed to quantify the uncertainty and consistency of the modelled outcomes.

3. Results

3.1. Habitat Quality as an Indicator for Biodiversity

In this study, habitat quality is chosen as the primary indicator for biodiversity due to its integrative nature, combining ecological parameters such as species presence, ecosystem integrity, and land use impacts. According to the calculations, the actual habitat quality in Latvia for the year 2024 has been calculated at 32,219,607 points. Areas with the highest habitat quality scores are concentrated in regions containing protected sites, extensive forest landscapes, and wetland ecosystems (Figure 1). These include the northwest and southwest parts of the country, the surroundings of Kemeri National Park near Riga, the Gauja River valley in the northeast, and the Teicu Nature Reserve and Lubana wetland complex in the southeast (Figure 1). These spatial clusters represent regions with high biodiversity potential based on their current land cover and use characteristics. As such, these areas can serve as reference landscapes for guiding future conservation and land use planning efforts.

Figure 1. Habitat quality in Latvia.

A simple sensitivity test on the baseline habitat quality values, adjusting habitat quality indices by ±1 point for selected key forest classes, revealed that changes exceeded ±0.70% only for middle-aged birch stands (±0.93%) and maturing pine stands (±0.75%). For all other key classes, the effect remained below ±0.6%. These results highlight the disproportionate influence of certain widespread or high-scoring forest types on national habitat quality outcomes, underlining their importance in large-scale biodiversity assessments.

3.2. Scenario-Based Biodiversity Assessment

Implementation of Scenario 1, which simulates an expansion of protected areas in line with the EU Biodiversity Strategy, resulted in an increase in total habitat quality to 34,948,971 ± 1783 points (95% confidence interval: 34,948,722–34,949,219), representing an average gain of 8.47% compared to the baseline (Table A2). Figure 2 provides a spatial visualisation of the results. The results demonstrate that extending protection in areas that are already ecologically valuable can boost biodiversity potential. This supports the strategic logic of reinforcing existing conservation networks rather than creating isolated new ones, thereby enhancing overall ecosystem resilience and connectivity. In addition, spatial analysis revealed that 69.78% of high habitat quality cells, defined as those in the top 25% of all grid-based habitat quality scores, were located within 1 km of existing protected areas. This indicates a strong spatial relationship with protected zones and suggests that even moderate improvements in habitat quality may contribute meaningfully to functional connectivity across the landscape.

Figure 2. Habitat quality in Latvia after implementation of Scenario 1.

Scenario 2, which models a transition from conventional to organic farming to meet the 25% target set by the EU Biodiversity Strategy, resulted in an increase in total habitat quality to 32,347,071 ± 112 points (95% confidence interval: 32,346,870–32,347,320), representing an average gain of 0.40% compared to the baseline (Table A3). Figure 3a provides a spatial visualisation of the results. The impact was spatially dispersed and relatively limited in magnitude, as converted areas were distributed randomly across the landscape. Localised improvements were observed primarily in homogeneous arable landscapes, where the transition to organic practices provided a stronger increase in habitat quality than in more heterogeneous landscapes. To test spatial sensitivity, Scenario 2b applied organic conversion exclusively in homogeneous agricultural landscapes, resulting in a habitat quality of 32,347,140 ± 103 points (95% confidence interval: 32,347,125–32,347,154), again an average gain of 0.40% (Table A4). Although the numerical difference is negligible, the spatial outcome (Figure 3b) shows that targeting more intensively managed regions can lead to more coherent ecological improvements. These findings suggest that while organic farming expansion has a positive but modest impact at the national scale, its effectiveness can be significantly enhanced through spatial prioritisation and improved landscape connectivity.

Figure 3. Habitat quality in Latvia after implementation of Scenario 2: (a) randomly selected fields for conversion; (b) fields selected in homogeneous arable landscapes.

4. Discussion

4.1. Comparing Conservation Interventions

Long-term biodiversity conservation depends not only on the integrity of protected areas themselves but also on the ecological condition of their surrounding landscapes [56]. Scenario 1, simulating the expansion of protected areas, produced the largest improvement in national habitat quality, with a spatially uneven increase approaching 8.47%. Ensuring connectivity between protected areas is essential to achieving their conservation objectives, particularly by facilitating species movement and maintaining ecological processes [57]. In Scenario 1, we assumed that expanding protection to areas near existing protected sites would result in the strongest positive impact on biodiversity. Our results support this assumption, showing that targeted expansion in these areas can improve habitat quality. These findings highlight the importance of strengthening existing protected areas rather than creating new isolated protected areas that are not well linked to the surrounding landscape [58,59]. This interpretation is further supported by our finding that nearly 70% of high habitat quality areas are situated within close proximity (1 km) of existing protected areas, suggesting potential for strengthening ecological networks through targeted interventions.

In contrast, Scenario 2, representing a shift to organic farming on conventionally managed cereal land, led to a modest increase of approximately 0.4%. Although positive, this change was spatially dispersed and smaller in magnitude. The impact was more evident in homogeneous arable landscapes, confirming that landscape context and land use intensity critically shape biodiversity outcomes [32,33]. A sensitivity test (Scenario 2b), where organic conversion was applied only in homogeneous agricultural areas, yielded a similar overall gain but resulted in more spatially coherent improvements, underscoring the importance of spatial targeting.

These findings suggest that targeted expansion of protected areas can deliver greater benefits for biodiversity than widespread but less focused improvements in land management. Nevertheless, both approaches are complementary. Organic farming can improve biodiversity in the wider agricultural landscape and serve as a buffer around core conservation areas, while protected areas play a central role in safeguarding habitats for biodiversity.

Importantly, not all aspects of biodiversity require strict protection. Certain low-intensity land use systems, such as traditional orchards, extensively managed grasslands, and mixed-use mosaic landscapes, can support high levels of biodiversity, particularly for species adapted to low-intensity agricultural activities [34,35,60]. These land use types contribute meaningfully to biodiversity conservation outside strictly protected zones, especially in regions where formal designation may not be feasible. Therefore, in this study, biodiversity potential should be interpreted as a relative measure of how favourable different land use types and spatial patterns are for maintaining ecologically valuable habitats, whether through legal protection or sustainable land management practices. To enhance the precision and ecological relevance of habitat quality assessments in the future, it is essential to incorporate a broader spectrum of land use types and management regimes that contribute to biodiversity beyond formally protected areas.

4.2. Limitations and Future Research

The methodological approach developed in this study for assessing habitat quality potential offers a relative rather than an absolute measure of biodiversity within specific territorial units (grids). This approach allows for a comparative analysis of habitat quality potential at the ecosystem level across different regions and facilitates the evaluation of changes in habitat quality potential under two scenarios of land management change. Notably, Damiani et al. (2023) emphasise that current methodologies are inadequate for simultaneously addressing the wide range of pressures on biodiversity, including those across ecosystems, taxonomic groups, essential biodiversity variables, and key factors needed for comprehensive biodiversity impact assessments within a value-chain context [61]. Assessing biodiversity is complex because it varies globally in distribution, vulnerability, and irreplaceability, requiring regionalised analyses of land use impacts and consideration of how different land use types and management practices affect biodiversity and land use intensity [62].

Expert assessments of forest lands have identified several limitations in the approach. Relying solely on dominant tree species and stand age may not sufficiently represent ecosystem quality, as these variables vary by site conditions and management history. Additionally, key ecological factors such as soil type, hydrology, and past land use were not included in the current index. Additionally, the exclusion of indicator species, those reflecting naturalness and habitat specificity, limits the sensitivity of the assessment to key biodiversity patterns. Because each species responds to a unique set of habitat characteristics, it is difficult to generalise changes in habitat availability without species-specific data [63].

The indices used in this study, while informed by existing literature and expert knowledge, lack a quantitative calibration that captures the full ecological variability across habitat types. Although this study considers important factors like landscape structure, farming intensity, forest age, and dominant tree species, using just one habitat quality index per land use type still misses variation within these categories. Key aspects such as species richness, abundance, soil type, and past land use are not included, which may affect how well the index reflects real biodiversity patterns. Similarly, the use of single-point increments to represent forest age classes oversimplifies what are often non-linear and species-specific biodiversity responses to forest development, as highlighted by Spake et al. (2015) and Handegard et al. (2024) [46,47]. To partially address these uncertainties, we conducted a sensitivity analysis by adjusting habitat quality indices by ±1 point for key forest classes, selected based on their total contribution to the baseline habitat quality score. In addition, a Monte Carlo simulation with 200 iterations was performed for both scenarios to quantify stochastic uncertainty in habitat quality changes resulting from random or spatially targeted land conversions. These steps provide an initial test of the robustness of the model outputs. However, future development of the index should continue to incorporate empirical calibration using field-based biodiversity data and explore alternative scoring schemes to better reflect ecological complexity.

Furthermore, although the overall increase in national habitat quality under Scenario 1 is moderate (<10%), its spatial distribution may have an ecologically significant effect. Localised improvements, particularly near existing protected areas, can enhance landscape connectivity and facilitate species movement, even when total habitat gains are relatively limited. As emphasized by Brudvig et al. (2009) and Saura et al. (2018), connectivity plays a critical role in supporting species persistence and maintaining ecosystem functionality, beyond the total area of habitat [57,58]. To strengthen the ecological relevance of this framework, future development should incorporate explicit connectivity metrics and assess whether targeted improvements in habitat quality contribute to reducing fragmentation and species isolation.

This study presents a spatially explicit approach for assessing habitat quality as a proxy for biodiversity at the landscape level, offering a valuable tool for evaluating the effects of land use and management scenarios. While the approach provides relative estimates rather than absolute measures of biodiversity, it enables consistent comparisons across regions and scenarios. The integration of expert-based assessments and land cover data enhances its practical utility for conservation planning and policy development. Importantly, many of the methodological limitations identified in this study, such as the use of fixed habitat quality indices, exclusion of site-specific ecological factors (e.g., soil, hydrology, historical land use), and lack of species-level validation, are directly targeted for improvement within the framework of the newly launched national research project HiQBioDiv (VPP-VARAM-Daba-2024/1-0002). This programme aims to generate new knowledge and tools to support the EU Biodiversity Strategy goals through a more integrated and data-rich approach. The project will enhance ecological modelling by incorporating indicator species data, improving spatial coverage of field observations, and enabling empirical calibration of habitat-biodiversity relationships. It will also address gaps related to landscape connectivity, non-linear biodiversity responses, and land use intensity gradients, thereby building a more ecologically sensitive and policy-relevant foundation for biodiversity assessments. By combining ecological field research with geospatial modelling, data mobilisation, and socio-economic analyses, HiQBioDiv will deliver open-access tools and validated methodologies that extend and refine the approach developed in this study. As such, this research contributes to a growing knowledge base that supports data-driven, transparent, and scalable biodiversity assessment and planning in Latvia and beyond.

5. Conclusions

This study presents a spatially explicit approach for modelling habitat quality as a proxy for biodiversity at the national level. The results demonstrate that different policy measures have markedly different impacts on habitat quality. Scenario analysis revealed that targeted expansion of protected areas, especially near existing sites, provides stronger benefits for biodiversity than widespread but diffuse changes in land management.

Moreover, while habitat quality alone cannot capture all dimensions of biodiversity at the field or stand level, it offers a practical and comparative measure of how favourable various land use types are for supporting biodiversity across landscapes. Even modest national-level improvements can enhance ecological connectivity and landscape resilience, highlighting the importance of spatial configuration in addition to total area.

The inclusion of sensitivity testing and Monte Carlo simulations strengthens the robustness of the results and supports the credibility of the main findings. This approach enables the prediction of biodiversity outcomes under alternative land use scenarios and supports evidence-based policy planning. As such, it contributes to national and EU-level biodiversity goals by helping to identify priority areas for conservation and informing strategies that integrate ecological, spatial, and land use considerations. Future research should incorporate species-level data, connectivity metrics, and non-linear biodiversity responses to further enhance sensitivity and policy relevance.