Long term coverage across Spanish parks
The analysis follows five land-cover periods from 1990 to 2018 around 159 protected areas aggregated into 82 spatial units.
↳ Methods §2; Results §3.1
Abstract
Protected areas (PAs) are the most effective tools to protect biodiversity and ecosystem services. They have proven to be effective in stopping extensive land use conversion in well-conserved terrestrial ecosystems. However, land cover changes around PAs threaten biodiversity and ecosystem services within their limits and reduce ecological connectivity. In this study, we analysed the urban sprawls on the boundaries of 159 PAs (national, regional, and natural parks) in Spain, using 2.5 and 5 km non-protected buffer zones from 1990 to 2018. We clustered PAs based on biophysical and socio-economic characteristics and modelled urban sprawl in different buffers and periods. Hierarchical clustering revealed three groups of PAs: (a) proximate urban parks, (b) mountainous parks, and (c) parks in the Madrid autonomous region. We found that urbanisation in the surroundings of PAs in Spain has nearly doubled since 1990. General linear models explained a significant proportion of the urbanisation trends observed, with the number of municipalities in the boundary of the PA, the distance to a main road, and the distance to a big city acting as the most important drivers of urban sprawl. Our results also show that some PAs exert significant effects on urbanisation trends in their surroundings through the park-view effect. Finally, we highlight three coexisting phenomena that might explain the observed urban sprawl processes: (a) PAs attracting urbanisation in their surroundings due to the park-view effect, (b) PAs as a deterrent for urban sprawl within their limits, and (c) PAs occupying residual areas among previously urbanised lands.
Key Claims
Primary findingSupported
We found that urbanisation in the surroundings of PAs in Spain has nearly doubled since 1990.
Likely to hold — the claim's weakest requirement is that the effect exists at all. Experimental. Every point below is traceable to a quoted passage, but this forecast has not yet been validated against known replication outcomes.
descriptive land-cover trends establish substantial growth, although inconsistent area and percentage figures leave the precise magnitude unresolved
Abstract; Results §3.1; Conclusions
SecondaryTentative
distance to cities and roads, road density, and the number of municipalities are the factors that most influence urbanisation processes around PAs.
observational regressions identify associations rather than causal influence, with no significance marker for main-road distance
Conclusions; Abstract; Table 2
SecondaryTentative
Our results also show that some PAs exert significant effects on urbanisation trends in their surroundings through the park-view effect.
buffer contrasts lack a comparable unprotected counterfactual capable of identifying park attraction
Abstract; Results §3.3; Discussion §4.1
Derived from the full evaluation — not a separate score.
Strengths · Limitations
Strengths
Spatial workflow addresses known data discontinuities
The methods describe buffer construction and a correction process that checks selected CLC anomalies against high-resolution imagery, rather than treating the source series as directly comparable without adjustment.
↳ Methods §2; Figure 2
Discussion situates patterns within competing explanations
The paper engages contrasting literature and explicitly acknowledges that attraction, deterrence and residual-space designation cannot be isolated as single processes.
↳ Introduction; Discussion §4.1
Limitations
Park attraction lacks an identifying comparison
Differences between nearby rings and selected mapped examples do not distinguish park attraction from pre-existing development or park-location decisions. No comparable unprotected counterfactual establishes the claimed effect.
↳ Results §3.3; Discussion §§4.1–4.2
Driver models leave material uncertainty unresolved
Small stratified models use absolute urban-area increases without buffer-area normalisation or reported spatial diagnostics. Municipality counts may partly reflect park perimeter and available buffer area rather than a planning mechanism.
↳ Methods §2; Table 2; Discussion §4.2
Central wording exceeds the reported evidence
The Abstract presents significant park-view effects and names main-road distance among leading drivers, although Table 2 gives its coefficient no significance marker. The growth magnitude also needs reconciliation between the reported areas and percentages.
↳ Abstract; Results §§3.1–3.2; Table 2
Nabu Commentary
The main contribution is the differentiation of protected-area settings across a long land-cover series, with Figure 4 showing substantial contrasts in nearby urbanisation. The driver models and mapped examples do not separate park attraction from baseline development, location choices and planning conditions. Table 2 also provides less support for some named explanatory variables than the Abstract implies. The descriptive findings remain useful, but the proposed mechanisms require controlled comparisons and clearer treatment of measurement uncertainty.
Check another paper
Author Note / Field Feedback
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Dimension breakdown
Study Quality
Limited2.6
Confidence mediumContribution
3.4
The 1990–2018 description and differentiation of protected-area settings extend pooled accounts of nearby urban development. The empirical advance is primarily descriptive because the proposed attraction and planning mechanisms are not identified.
Results §3.1; Figure 4; Discussion §4.1
Methodological Rigour
2.0
The spatial workflow and partial imagery-based CLC correction support descriptive analysis. Unaddressed confounding and causal interpretation of uncontrolled buffer patterns impose the stacked methodological ceiling.
Methods §2; Results §3.3; Table 2; Discussion §4.2
Reporting
3.1
The sequence from clustering to models and mapped examples is followable. Central statements of significant effects and influential drivers materially overstate the identifying capacity of those analyses.
Abstract; Methods §2; Results §§3.1–3.3; Conclusions
Positioning
2.8
The discussion engages competing perspectives and acknowledges measurement and coverage limitations. It does not explain how missing counterfactual comparisons prevent attribution, or carry residual measurement uncertainty through to the headline magnitude.
Introduction; Discussion §§4.1–4.3
Reliability
Caveats4 of 4 checks
Internal CoherenceNoted
Internal numerical and methods–results inconsistencies require clarification, alongside uncertainty introduced by the land-cover correction. They qualify interpretation without establishing that the reported descriptive direction is invalid.
Research ConductClean
Unanimous clean assessment: this secondary geospatial analysis does not require human-participant ethics approval, and a competing-interest declaration is present. No research-conduct concern is identified in the supplied material.
Flags: 2 declared / 5 total
Reference IntegrityNoted
67 of 69 checkable references verified
73 references in manuscript 4 are books, websites or datasets — counted, but not index-checkable
Post-Publication RecordClean
No retraction notice found in Retraction Watch.
Last checked: 16 Oct 2026
Sources: Retraction Watch ✓
Readiness
Where this paper’s evidence sits on the path from initial observation to real-world use.
Mechanistic
The analysis supplies descriptive patterns and candidate correlates rather than an evaluated planning intervention. Recommendations therefore require further development before implementation.
Table 2; Discussion §4.2; Conclusions
AI-generated, human-governed. Something look off? Contact us to request a review.