Abstract

Soil and Water Bioengineering (SWB) is an environmentally friendly approach that integrates biological and engineering techniques applied to manage natural hazards and restore degraded ecosystems. Instead of using concrete or heavy machinery, SWB relies on vegetation, natural materials and ecological processes. This method helps to reduce disaster risks, improves soil quality, enhances water retention and increases biodiversity. SWB techniques, such as vegetative slope stabilization and erosion control, are cost-effective and adaptable to different landscapes. As a nature-based solution, SWB plays a key role in promoting sustainable land management, restoring ecological balance, and building resilience against climate-related hazards. This mini review aims to provide an overview of SWB as a sustainable, nature-based approach that integrates ecological and engineering principles to mitigate natural hazards and restore degraded ecosystems.

Keywords: Soil; Water bioengineering; Ecosystem; Biodiversity; Sustainable environmental management; Biodegradable materials; Wildlife habitats; Geotechnical engineering

Introduction

Knowledge Graphs (KGs) have emerged as powerful tools in environmental science, offering structured frameworks for integrating diverse datasets across ecological, climatic, and socio-environmental domains. While their utility in data organization and retrieval is well-recognized, this article critically examines the inherent limitations of KGs in capturing the dynamic complexities of environmental systems. Specifically, it explores how KGs may inadequately represent nonlinear interactions, such as feedback loops and tipping points, which are pivotal in ecological processes. Furthermore, the article highlights the epistemological and ethical concerns arising from biases embedded in source datasets, including geographic and taxonomic disparities, and the underrepresentation of Indigenous and local knowledge systems. The rigidity of predefined ontologies within KGs is also scrutinized, questioning their capacity to accommodate the fluid and evolving nature of environmental knowledge. Through this critical lens, the article advocates for a more nuanced application of KGs, emphasizing the need for integrating dynamic modeling approaches and inclusive data practices to more accurately reflect the intricacies of environmental phenomena.

Keywords: Knowledge graphs; Environmental science; Ecological; Climatic; Tropical biodiversity; Epistemological; Geographic regions

Methodology

Knowledge Graphs (KGs) have become increasingly prevalent in environmental science, promising structured, integrative approaches to managing diverse ecological, climate, and socio-environmental datasets [1,2]. While KGs offer compelling capabilities in terms of data integration, query efficiency, and decision support, their adoption raises profound philosophical and epistemological questions. This article critically examines inherent limitations, highlighting potential epistemological pitfalls, methodological biases, and ethical concerns regarding their implementation.

Reductionism and the Illusion of Completeness

The environmental domain encompasses complex systems characterized by non-linearity, emergence, and contextual variability. Knowledge Graphs typically simplify these complexities into discrete nodes and edges, potentially promoting an overly reductionist view. By structuring knowledge into fixed categories, KGs risk obscuring dynamic ecological interactions such as feedback loops, tipping points, or emergent behaviors [3,4]. Such simplifications might lead practitioners to underestimate the inherent uncertainty and complexity in environmental systems, creating an illusion of completeness.

Additionally, KGs often implicitly project certainty. Their structured format and semantic querying capabilities may inadvertently foster false confidence, masking underlying data limitations or gaps in environmental knowledge. This illusion of completeness is problematic, particularly when KGs inform critical environmental decisions such as biodiversity conservation or climate mitigation strategies [5]. Recognition of KG limitations and explicit acknowledgment of uncertainties are essential to prevent decision-making biases.

Data Biases and Epistemological Limitations

Environmental KGs inherit biases intrinsic to the datasets from which they are constructed. Many datasets reflect geographic, taxonomic, or thematic biases, disproportionately representing well-studied regions or taxa while neglecting less-documented contexts, such as tropical biodiversity or indigenous ecological knowledge [3,6,7]. Consequently, KGs can inadvertently reinforce existing knowledge inequities, privileging certain ecological viewpoints or geographic regions over others.

Moreover, knowledge graphs predominantly incorporate structured, quantitative, or formally documented data. This tendency marginalizes qualitative and experiential knowledge, including indigenous practices and local ecological understandings, crucial for nuanced environmental management. The exclusion of such tacit knowledge raises ethical and epistemological issues, underscoring a significant limitation: KGs may systematically overlook critical socio-cultural dimensions integral to holistic environmental science [8-11].

Ontological Rigidity vs. Environmental Fluidity

KG construction mandates defining ontologies—structured frameworks that determine how data is categorized and related. Ontologies, however, impose rigidity upon the representation of inherently fluid environmental phenomena [12]. For example, classifications of ecosystems, species interactions, or sustainability practices often vary contextually or evolve with new scientific discoveries [13]. Rigid ontologies constrain KGs’ flexibility, potentially excluding novel, hybrid, or intermediary ecological concepts and interactions.

Environmental knowledge evolves continually, challenging the static nature of predefined ontological structures. To counter this rigidity, KG designs must incorporate flexible ontological frameworks capable of adapting to scientific advances and incorporating diverse, context-specific knowledge [14,15]. Failing to address this rigidity may render KGs obsolete or exclusionary over time, limiting their relevance and applicability to dynamic ecological scenarios.

Ethical Considerations and Inclusivity

Deploying KGs in environmental contexts raises critical ethical considerations. Decisions embedded in KG construction— what data to include, what categories to use, whose knowledge to represent—are value-laden and have tangible consequences. For instance, predominantly Western scientific ontologies might marginalize indigenous or local ecological knowledge, perpetuating epistemic injustice by underrepresenting or misrepresenting these alternative knowledge systems [16].

Additionally, the portrayal of data as objective facts within KGs conceals subjective biases in data collection processes. Explicitly acknowledging these biases and engaging diverse stakeholders, particularly marginalized communities, in KG construction are ethical imperatives. Moreover, environmental data often intersects with sensitive human community information, raising privacy and data governance concerns. Ethical KG deployment thus necessitates careful consideration of inclusivity, representation, and data sensitivity issues to ensure fairness and transparency in knowledge representation.

Can Environmental Complexity Ever Be Fully Captured?

Fundamentally, environmental complexity may never be entirely captured within a KG or any singular modeling framework. Environmental systems are open, dynamic, and multi-layered, defying full encapsulation by structured relational models alone. While KGs effectively manage known, structured information, they inherently omit unknown interactions, emergent properties, and qualitative dimensions of ecological knowledge [17].

This limitation is not inherently problematic if recognized and explicitly accounted for in practice. Rather than viewing KGs as comprehensive or definitive repositories, environmental scientists and practitioners should treat them as complementary tools within broader analytical and interpretive frameworks. Embracing KGs as partial, evolving representations encourages humility, continuous learning, and integrative approaches that combine structured data with qualitative insights and uncertainty management strategies.

Conclusion

Knowledge graphs offer significant promise for integrating and managing environmental data but carry inherent philosophical, epistemological, and ethical limitations. Issues of reductionism, epistemological completeness, data bias, ontological rigidity, and inclusivity require thoughtful consideration. Ultimately, the efficacy and legitimacy of KGs depend upon recognizing their limitations and using them responsibly, with an awareness that complexity, uncertainty, and diverse ways of knowing must be continuously respected and explicitly integrated into environmental analysis and decision-making frameworks.

References

1. Ma X (2022) Knowledge graph construction and application in geosciences: A review. Computers & Geosciences 161: 105082.
2. Wu J, Orlandi F, O’Sullivan D, Dev S (2022) Link Climate: An interoperable knowledge graph platform for climate data. Computers & Geosciences 169: 105215.
3. Farahbakhsh I, Bauch CT, Anand M (2024) Tipping points in coupled human–environment system models: A review. Earth System Dynamics 15(4): 947-967.
4. Shortridge JE, Bukvic A, Mitchell M, Goldstein J, Allen T (2024) Characterizing Climatic Socio‐Environmental Tipping Points in Coastal Communities: A Conceptual Framework for Research and Practice. Earth’s Future 12(7): e2023EF004123.
5. Ge R, Xia Y, Ge L, Li F (2025) Knowledge Graph Analysis in Climate Action Research. Sustainability 17(1): 371.
6. Donaldson M, Burnett N, Braun D, Suski C, Hinch S, et al. (2016) Taxonomic bias and international biodiversity conservation research. FACETS 1: 105.
7. McElwee P, Fernández‐Llamazares Á, Aumeeruddy‐Thomas Y, Babai D, Bates P, et al. (2020) Working with Indigenous and local knowledge (ILK) in large‐scale ecological assessments: Reviewing the experience of the IPBES Global Assessment. Journal of Applied Ecology 57(9): 1666-1676.
8. Han D, Zhao Z, Liu C (2025) A LLM Enhanced Two-Stage Approach for Multi-hop Knowledge Graph Reasoning. In: H Sun, H Fan, Y Gao, X Wang, D Liu, et al. (Eds.), Computer Supported Cooperative Work and Social Computing. Springer Nature Singapore, Vol. 2344, pp. 309-323.
9. Han F, Deng Y, Liu Q, Zhou Y, Wang J, et al. (2022) Construction and application of the knowledge graph method in management of soil pollution in contaminated sites: A case study in South China. Journal of Environmental Management 319: 115685.
10. Wang X, Chen Z, Wang H, Hou UL, Li Z, et al. (2025) Large Language Model Enhanced Knowledge Representation Learning: A Survey. Data Science and Engineering.
11. Zhang M, Yang G, Liu Y, Shi J, Bai X (2024) Knowledge graph accuracy evaluation: An LLM-enhanced embedding approach. International Journal of Data Science and Analytics.
12. Sidelle A (1992) Rigidity, Ontology, and Semantic Structure. The Journal of Philosophy 89(8): 410-430.
13. Asokan VA, Yarime M, Onuki M (2020) A review of data-intensive approaches for sustainability: Methodology, epistemology, normativity, and ontology. Sustainability Science 15(3): 955-974.
14. Liu X, Xiao Y, Zhou S (2025) Knowledge-graph-driven environmental monitoring with cross-regions knowledge transfer. Knowledge and Information Systems 67(3): 2721-2744.
15. Wolffe TAM, Vidler J, Halsall C, Hunt N, Whaley P (2020) A Survey of Systematic Evidence Mapping Practice and the Case for Knowledge Graphs in Environmental Health and Toxicology. Toxicological Sciences 175(1): 35-49.
16. Holmgren SD, Boyles RR, Cronk RD, Duncan CG, Kwok RK, et al. (2021) Catalyzing Knowledge-Driven Discovery in Environmental Health Sciences through a Community-Driven Harmonized Language. International Journal of Environmental Research and Public Health 18(17): 8985.
17. Zhou L, Cai L, Janowicz K, Mai G, Stephen S, et al. (2021) Environmental Observations in Knowledge Graphs.