An Integrated DPSIR-Based Assessment of Ecological Risks and Sustainable Management Options for the Caspian Sea

This study adopts a structured literature-synthesis design combined with a semi-quantitative risk-assessment and scenario-comparison methodology. No new field sampling was undertaken; instead, findings from peer-reviewed studies, institutional monitoring reports, and regional environmental datasets published between 2010 and 2026 were compiled, cross-checked, and integrated to characterize ecological trends across the Caspian basin. This was not a formal systematic review; sources were selected purposively, and one earlier basin-wide pollution-load assessment was retained.

2.1 Data sources and analytical framework

Evidence was drawn from foundational and regional studies on marine litter and microplastic dynamics [5-8], global policy assessments of plastic debris [9], groundwater-related contamination on the Absheron Peninsula [10], and coastal geomorphological change assessed through remote sensing and Geographic Information System (GIS) [11], a non-peer-reviewed report used as background only. To organize these heterogeneous sources into a coherent descriptive structure, the DPSIR framework was applied. This framework links the socio-economic activities that generate environmental pressure (Driving Forces), the resulting physical and chemical loading on the ecosystem (Pressures), the observed condition of environmental quality (State), the ecological and socio-economic consequences of that condition (Impacts), and the policy or technical measures adopted in response (Responses). DPSIR is used here as a descriptive organizing structure; it does not by itself establish causality between drivers and observed ecological change.

2.2 Semi-quantitative ecological risk assessment

Because the reviewed sources reported ecological outcomes using non-comparable methodologies and units, a semi-quantitative ERS was constructed to allow relative comparison among stressor categories. Each of six principal environmental stressors identified through the literature synthesis-oil and gas pollution, industrial and agricultural wastewater, plastic and microplastic pollution, biodiversity loss, hydrological change, and sea-level variability-was rated independently on two five-point ordinal scales (1 = very low, 5 = very high): a severity score, reflecting the intensity of documented ecological impact, and an occurrence score, reflecting the frequency with which the stressor was reported across the reviewed literature. Occurrence thus measures evidence frequency, not true frequency in the sea, and may favor well-studied stressors. Rubric for severity: 1 = no measurable effect; 2 = localized, reversible; 3 = localized, persistent; 4 = sub-basin scale or sensitive taxa affected; 5 = basin-wide or population-level effects. Rubric for occurrence: 1 = isolated reports; 2 = one locality; 3 = one sub-basin; 4 = two of the three sub-basins; 5 = all three sub-basins and national or regional assessments. The ERS was then calculated as the product of these two ratings:

ERS = Severity × Occurrence

A higher ERS value indicates greater cumulative ecological significance and, correspondingly, higher priority for management intervention. This scoring approach does not yield an absolute measure of ecological damage; rather, it provides a transparent, reproducible method for ranking stressors relative to one another based on the weight of current evidence. The multiplicative form follows the likelihood × consequence logic of risk matrices; additive and geometric-mean aggregation gave the same order (one tie under the additive form). Robustness was tested by changing each score by ±1 in turn (20 admissible perturbations) and comparing rankings with Spearman’s ρ.

2.3 Comparative evaluation of management strategies

Four candidate environmental management strategies were identified from the reviewed literature as having demonstrated applicability to semi-enclosed coastal and marine systems: smart sensor networks (continuous water-quality monitoring), UAV-based monitoring (coastal and surface surveillance), biomonitoring (using aquatic indicator species), and nature-based filtration (constructed wetlands and coastal vegetation buffers); these four labels are used throughout. Each strategy was scored on four evaluation criteria-pollution detection capability, spatial monitoring coverage, response speed, and long-term ecosystem-recovery potential-using a five-point scale (1 = very low, 5 = very high) informed by the performance characteristics reported for each technology in the reviewed monitoring literature. A score of 5 denotes a continuous, operational capability in comparable systems, 3 a partial or periodic one and 1 none; no expert panel was used, so scores are indicative of author judgments. An overall effectiveness score was calculated as the arithmetic mean of the four criterion scores for each strategy. Because equal weighting is an assumption, stability was tested with ±1 score changes and 10,000 random weight sets.

2.4 Scenario comparison

To illustrate the potential divergence in ecological trajectory associated with current versus improved management practice, two qualitative development scenarios were constructed for the period 2025–2035: a Business-as-Usual (BAU) scenario, in which existing fragmented and reactive management continues, and an Integrated Environmental Management scenario, in which the framework proposed in Section 2.3 is implemented. These scenarios were developed as a structured, evidence-informed comparison of plausible directions of ecological change rather than as a quantitative predictive model; the relative ecosystem-health trajectories are illustrative projections intended to visualize the qualitative direction and approximate magnitude of divergence between the two pathways, not statistically fitted forecasts. Decision rules: an index set to 100 in 2025 falls by 7–9 points per two-year step under BAU (pressures persist without new controls) and rises by 3–7 points per step under the integrated scenario (smaller early gains reflect deployment time); values are author-assigned and uncalibrated.

The findings of this synthesis reinforce the view that ecological degradation in the Caspian Sea is not attributable to a single pollution source; the reviewed evidence associates it with the cumulative and interacting effects of hydrocarbon extraction, industrial and agricultural discharge, plastic contamination, biodiversity loss, and climate-driven hydrological change. The closed nature of the basin substantially limits natural dilution and flushing of pollutants, so that contaminants entering the system through any one of these pathways tend to persist and accumulate rather than dissipate, a dynamic that helps explain why relatively moderate discharge volumes can produce disproportionately large and long-lasting ecological effects [25].

The ecological risk assessment indicates a clear management hierarchy, with oil and gas pollution and industrial-agricultural wastewater together accounting for the largest share of cumulative risk. This result is consistent with prior assessments identifying offshore drilling, pipeline operations, and river-borne nutrient and heavy-metal transport as the longest-standing and most severe pressures on the basin. It is also broadly consistent with earlier institutional river-load estimates (Section 3.3), which were not used in scoring. Because hydrological change and sea-level decline are documented mostly as drivers or projections, their lower scores partly reflect the distribution of research effort. At the same time, the comparatively high risk score attributed to plastic and microplastic pollution, despite its shorter documented history relative to oil contamination, signals an emerging threat whose persistence and capacity for bioaccumulation warrant closer monitoring attention in future assessments [26].

The comparative evaluation of management strategies suggests that no single technology is sufficient to address the Caspian Sea's environmental problems in isolation. Smart sensor networks and UAV-based monitoring offer strong detection and coverage capability but do not, by themselves, restore ecological function; biomonitoring and nature-based filtration provide longer-term ecological insight and habitat restoration but respond more slowly to acute pollution events. These findings support an integrated framework in which real-time technological monitoring is combined with ecosystem-based restoration measures, allowing early warning capability to be paired with the ecological recovery function that restoration provides.

The scenario comparison further indicates that the trajectory of ecological change in the Caspian Sea over the next decade will depend substantially on whether current, largely reactive management practices continue or are replaced by a coordinated, technologically supported framework. While these scenarios are illustrative rather than predictive, they are consistent with the broader literature suggesting that delayed or fragmented intervention in closed marine systems tends to produce compounding, and increasingly difficult to reverse, ecological decline. This underscores the importance of timely investment in monitoring infrastructure and of strengthening the regional cooperation mechanisms among the five Caspian littoral states that such investment requires.

This study has several limitations that should be considered when interpreting its findings. The analysis relies primarily on secondary data drawn from published literature and institutional monitoring reports rather than on original field sampling, which constrains the spatial and temporal resolution of the underlying evidence base. The severity and occurrence ratings used in the ERS, and the criterion scores used in the strategy comparison, are semi-quantitative expert judgments informed by the reviewed literature rather than measurements derived from a single standardized protocol, and should therefore be interpreted as indicative rankings rather than precise, statistically validated estimates. Similarly, the 2025-2035 scenario projections are qualitative, evidence-informed illustrations of plausible directions of ecological change rather than outputs of a calibrated predictive model. Future research incorporating long-term in-situ monitoring, remote-sensing time series, and hydrodynamic or machine-learning-based modeling would help to refine and validate the risk rankings and management priorities proposed here. Sources were selected purposively rather than by systematic search, and monitoring standards and sampling frequencies differ among the littoral states, limiting comparability. A few non-peer-reviewed and web-based sources [4, 11, 18, 22] were retained for background context only; no ERS score, strategy score, or ranking relies exclusively on them. Sensitivity tests show that the leading priorities are robust, whereas lower ranks are not.