Operational comparability of ecological status assessment in Iberian transboundary rivers

Abstract

The European Union Water Framework Directive (WFD) requires member states to assess the ecological status of surface waters using biological quality elements (BQEs). Although the WFD includes intercalibration procedures to harmonise class boundaries, differences between national assessment systems may affect the comparability of ecological status classifications, particularly in transboundary rivers. This study evaluated the consistency of Portuguese and Spanish ecological status assessments in 32 water bodies across Iberian transboundary rivers in the Minho, Lima, Douro, Tagus and Guadiana basins, using data collected between 2019 and 2021. Comparisons included fish, macrophytes, benthic macroinvertebrates and diatoms. Biological assemblages, ecological quality ratios (EQRs) and ecological quality classifications were analysed. Results differed markedly among BQEs. Fish showed no significant differences in assemblage composition or EQR values between countries and a moderate positive correlation in ecological quality classifications. Macroinvertebrates displayed weak compositional separation, no differences in EQR values and a weak, nonsignificant positive correlation between countries. Macrophytes showed significant compositional differences despite similar EQR distributions and a weak, nonsignificant positive correlation in ecological quality classifications. Diatoms showed the strongest divergence, with clear assemblage separation, higher EQR values in Spain and a very weak, nonsignificant negative correlation between countries. Final ecological classifications also showed a weak, nonsignificant positive correlation between countries. These findings show that formal intercalibration alone does not guarantee operational comparability in shared river systems. Cross-border assessment consistency depends strongly on the BQEs considered and on methodological differences between national monitoring systems, highlighting the need for improved harmonisation of monitoring procedures and taxonomic practices.

Introduction

The assessment of ecological status is a central requirement of the European Union Water Framework Directive (European Commission, 2000), which establishes the objective of achieving at least “good ecological status” in all European surface waters. To operationalise this objective, member states (MS) are required to apply a multielement assessment framework that integrates biological quality elements (BQEs), including fishes, macrophytes, benthic macroinvertebrates, phytoplankton and diatoms, alongside physicochemical and hydromorphological indicators (European Commission, 2000). This multimetric structure reflects the WFD’s ecosystem-based paradigm, which recognises that ecological condition emerges from interactions among biological communities, environmental drivers and anthropogenic pressures, and cannot be reduced to any single indicator (Birk et al., 2012; Hering et al., 2010).

Although the Directive establishes a common conceptual framework, it also allows MS considerable, but constrained, latitude in implementing ecological status assessments. National authorities may develop assessment systems tailored to regional ecological characteristics, including defining water body (WB) types, selecting and aggregating metrics and establishing reference conditions and class boundaries, within the limits imposed by intercalibration requirements (European Commission, 2003; Poikane et al., 2014). This flexibility has enabled context-sensitive tools, but it also introduces structural heterogeneity in how ecological status is assessed across MS. Consequently, national assessment systems may differ in how ecological information is produced, integrated and translated into ecological status classes. Methodological variation can occur throughout the assessment process, from sampling design and temporal coverage to metric construction and index aggregation. These differences can propagate through the assessment chain, from sampling design to index aggregation, leading to contrasting ecological status classifications for the same WB, with implications for reporting, restoration prioritisation and river basin management (European Environment Agency., 2018; Poikane et al., 2015).

Such comparability challenges are not unique to the WFD, as ecological metrics are widely used as policy-relevant indicators despite their dependence on methodological design and context (Borja et al., 2008; Karr & Chu, 2000; Siddig et al., 2016). Assessment results are often treated as objective and comparable measures of ecosystem condition across space and time (European Environment Agency., 2018; Grizzetti et al., 2017). Yet they are inherently shaped by how biological data are collected, processed and aggregated, as well as by whether assessment approaches target specific pressures or integrate multiple stressors (Birk et al., 2012; Hering et al., 2006; Poikane et al., 2020).

To ensure comparability at the European level under the WFD, regional intercalibration exercises were conducted, in which the compliance of indices with the common conceptual framework was assessed, followed by cross-comparison of indices with common indices or metrics (De Hoyos et al., 2014; Poikane et al., 2014; Willby et al., 2014). The deviation of indices’ values was reduced through the harmonisation of class boundaries of moderate–good and good–high, when needed, in order to ensure operational comparability of ecological status across MS (European Commission, 2011; Willby et al., 2014). This procedure has been completed for Mediterranean biological quality elements such as phytoplankton, diatoms, macroinvertebrates, macrophytes and fish (Aguiar et al., 2014; Almeida et al., 2013; De Hoyos et al., 2014; Feio et al., 2013; Segurado et al., 2014). However, de facto confirmation of such comparability has not been tested further.

Transboundary river basins provide particularly valuable settings for examining these issues empirically. In such systems, neighbouring countries often apply distinct monitoring programmes and ecological assessment methods to the same or closely connected WB, creating opportunities to evaluate the field operational comparability of national assessment systems under shared environmental conditions. The Iberian Peninsula provides a relevant case study in this context. Portugal and Spain share several major international river basins, including the Minho, Lima, Douro, Tagus and Guadiana, which together encompass a large proportion of the territory of both countries. Cooperation in shared basins is agreed through the Albufeira Convention (Portugal & Spain, 1998), which promotes coordinated monitoring, data exchange and implementation of the WFD. Nevertheless, national monitoring programmes and assessment systems have developed largely independently, shaped by distinct administrative traditions, historical trajectories and technical choices (González-Paz et al., 2020; Poikane et al., 2015, 2020). As a result, transboundary river basins in the Iberian Peninsula provide a valuable setting to examine how differences in national assessment systems may influence ecological status classifications in shared WB, and to what extent national assessments can be considered operationally comparable.

Differences in typology assignment, index construction and sampling protocols may affect the degree of agreement between classifications, even where formal intercalibration procedures have been applied (González-Paz et al., 2020; Poikane et al., 2020), challenging the assumption that harmonised reporting structures necessarily translate into equivalent assessment outcomes and highlighting the need to analyse what is happening at the scale of sampling sites (Borja & Elliott, 2007; Solheim et al., 2020). Ensuring consistency in ecological status assessment is therefore both a regulatory requirement and a scientific challenge. Empirical evidence on how well such alignment holds when multiple assessment elements are combined into an overall ecological status remains limited (González-Paz et al., 2020). It also remains unclear how disagreement propagates when several elements jointly contribute to classification, as some elements are more prone to divergence than others, and how methodological differences nonetheless influence overall assessment consistency (Majaneva et al., 2024; Poikane et al., 2020).

This study addresses these gaps by analysing ecological status assessments produced by Portuguese and Spanish monitoring systems for shared river WB sampled over three consecutive years. By jointly evaluating multiple biological quality elements, the study examines the degree of agreement between national assessments and how this agreement varies across biological elements and assessment metrics. Beyond simple comparisons, the analysis evaluates differences in biological assemblage composition, ecological quality ratio (EQR) values and relationships among assessment metrics. In doing so, the study provides empirical insight into the operational comparability of WFD implementation in Iberian transboundary rivers and contributes to broader discussions on harmonisation and consistency in ecological status assessment under the WFD.

Materials and methods

Study area

The study area comprises 33 sampling sites located on the five international river basins shared by Portugal and Spain: the Minho, Lima, Douro, Tagus and Guadiana (Fig. 1), representing the transboundary hydrological systems of the Iberian Peninsula and extending along the Portuguese-Spanish border. These river basins encompass pronounced climatic and hydrological gradients. Northern basins are predominantly influenced by humid Atlantic conditions, whereas Southern basins experience semi-arid Mediterranean climates, reflecting the marked hydroclimatic gradient characteristic of the Iberian Peninsula (Gasith & Resh, 1999; Vicente-Serrano et al., 2014).

Fig. 1.

Location of the 33 sampling sites distributed across 32 Portuguese water bodies within the international river basins shared by Portugal and Spain (Minho, Lima, Douro, Tagus and Guadiana), representing the main transboundary river systems of the Iberian Peninsula

The 33 sampling sites included 21 boundary WB and 12 transboundary WB distributed across 16 sub-basins within five international river basins. One site was sampled in the Minho basin and one in the Lima basin. In the Douro basin, 13 sites were distributed across the Tâmega (n = 1), Tuela (n = 2), Rabaçal (n = 3), Sabor (n = 2), Maçãs (n = 2) and Águeda (n = 3) sub-basins. In the Tagus basin, five sites were sampled, including three in the Erges sub-basin and two in the Sever sub-basin. In the Guadiana basin, 13 sites were distributed across the Xévora (n = 5), Caia (n = 1), Alcarrache (n = 1), Ardila (n = 3), Murtega (n = 1) and Chança (n = 1) sub-basins. Reservoirs occupy the remaining border length and are considered highly modified WB and, as such, were not part of this study due to their assessment specificities.

Across the sampled sites, WB length ranged from 2.7 to 80.92 km, while drainage area ranged from 7.93 to 259.22 km². According to the criterion proposed by (Borgwardt et al., 2019), which defines very large rivers as systems with drainage areas exceeding 10,000 km², none of the sampled water bodies was classified as large river systems. Consequently, all sampled WB correspond to small- to medium-sized river types and consist of wadeable streams.

Sampling framework and biological quality assessment

Biological monitoring data were collected between 2019 and 2021 by Portugal and Spain separately, across the 33 sampling sites, and seven sites were sampled twice, resulting in a total of 40 records. In total, 23 fish records per country were available, 18 per country for macrophytes and 37 per country for macroinvertebrates and diatoms. Among the 40 sampling records, eight sampling sites included all four BQEs, 21 included three BQEs, nine included two BQEs and two included only one BQE.

All data used in this study were collected in accordance with the national protocols developed by Portugal and Spain under the WFD (European Commission, 2000). Field surveys and ecological status assessments were conducted independently by the two national teams, each applying its respective WFD-compliant monitoring programmes. Although both countries operate within the common normative framework established by the WFD, methodological differences exist between the national protocols. These differences include sampling season, spatial sampling design, field procedures, laboratory processing of samples and the indices used to assess ecological status. Table 1 summarises the main similarities and differences between the national protocols applied in Portugal and Spain for each BQE.

Table 1.

Summary of national sampling protocols, spatial coverage, field procedures and assessment indices applied for each biological quality element (BQE) in Portugal (PT) and Spain (ES) under the Water Framework Directive

Fishes		Macrophytes	Macroinvertebrates	Diatoms
Sampling season	PT: Spring (Mar–Jul, depending on region and flow conditions). SP: Spring–autumn (optimum summer–early autumn)	PT: Late spring–summer (vegetation development period). SP: Spring–early autumn during maximum vegetation development	PT: Spring sampling recommended to minimise seasonal variability SP: Annual sampling is typically in spring	PT: Preferably spring under stable flow conditions and good substrate visibility SP: Annual sampling spring–summer, depending on the monitoring programme
Sampling area/transect	PT: ≥ 20 × channel width (minimum 100 m); large rivers ≥ 10 × width SP: ≥ 10 × channel width (minimum 100 m) including all habitat units	PT: Representative reach (~ 100 m) surveyed along the entire channel SP: Standard reach (~ 100 m) covering macrophyte assemblage	PT: 50 m representative reach; 1.5 m2 SP: 100 m representative reach; 2.5 m2	PT: ~ 50 m reach including coarse substrate in turbulent flow; ~ 100 cm2 total area SP: ~ 10 m sampling section within representative river reach; ~ 100 cm2 total area
Sampling method	PT & SP: Electrofishing (DC/PDC), wading; zig-zag coverage of channel, typically single pass	PT: Visual survey of the macrophyte assemblage to estimate taxon cover SP: Visual survey with estimation of macrophyte cover classes	PT: 6 kick samples (1 m × 0.25 m) with a hand net (500 µm), habitat proportional SP: 20 kick samples (~ 0.5 m × 0.25 m) with a hand net (500 µm), habitat-proportional	PT: Scraping biofilm from ≥ 5 stones using a brush SP: Scraping biofilm from ≥ 5–10 stones
Sampling substrates/habitat coverage	PT & SP: All mesohabitats (riffles, runs, pools)	PT & SP: Whole channel vegetation including macrophytes, bryophytes and macroalgae	PT: 6 habitat categories (blocks, stones, gravel, sand/silt/clay, macrophytes/algae, coarse organic matter) SP: 5 habitat categories (hard substrates, detritus, vegetated banks, macrophytes, fine sediments)	PT & SP: Preferentially coarse substrates (stones, cobbles); artificial substrates or macrophytes if necessary
Sample treatment	PT & SP: Fish identified to species; individuals counted and measured; abundance estimated (density/CPUE)	PT: Macroalgae grossly identified; cover estimation used in index calculation. Indicator species list as in the original French index SP: Bryophytes grossly identified; cover classes converted to abundance scale. Indicator species lists are different from the original French index	PT & SP: Samples preserved in ethanol or formaldehyde; washed, sorted and identified mainly to family level	PT & SP: Samples preserved in Lugol or formaldehyde; permanent slides prepared for microscopy Higher taxonomic resolution in Spain (Varieties) Indicator values and calculation software differences;
Indices	PT: F-IBIP SP: EFI + integrado	PT & SP: IBMR	PT: IPtI SP: IBMWP	PT & SP: SPI

F-IBIP Fish-based Index of Biotic Integrity for Portuguese Wadeable Streams (Segurado et al., 2014), EFI+ European Fish Index “Integrado” (Confederación Hidrográfica del Duero, 2021), IBMR Macrophyte Biological Index for Rivers (Aguiar et al., 2014), IPtI Portuguese Invertebrate Index (INAG, 2009), IBMWP Iberian Bio Monitoring Working Party (Alba-Tercedor & Sánchez-Ortega, 1988), SPI Specific Pollution Sensitivity Index (Cemagref, 1982)

Despite these methodological differences, the biological assessment indices applied in both countries operate within the WFD intercalibration framework, which aims to ensure comparability of ecological status classifications among MS. Portugal and Spain use their own type-specific reference conditions, established within their respective national assessment systems, against which ecological quality ratios (EQRs) are calculated. European intercalibration aims to ensure that national ecological status class boundaries represent comparable levels of ecological alteration, while allowing MS to retain their own assessment methods and type-specific reference conditions (European Commission, 2011, 2015). Common Implementation Strategy (CIS) Guidance Document No. 30 specifically provides a procedure for fitting new or revised national classification methods to the results of a completed intercalibration exercise (European Commission, 2015). As the present study compares the operational outputs of the existing Portuguese and Spanish assessment systems, rather than developing or revising national classification methods, this fitting procedure was not applied. Consequently, methodological differences, including those associated with national reference conditions, may remain despite formal intercalibration. In the case of macrophytes and diatoms, the indices are even the same (Table 1). The overall ecological status for each sampling site followed the “one-out, all-out” principle, whereby the lowest ecological status class assigned among the assessed BQEs determined the final ecological classification.

To evaluate cross-border consistency between national assessment systems, the percentage of disagreement was calculated for each BQE as the proportion of WB for which Portugal and Spain assigned different ecological status classes. Disagreement was also calculated for the final ecological status classification obtained after applying the “one-out, all-out” principle.

Data analysis

To evaluate cross-border consistency between the Portuguese and Spanish biological assessment systems, three complementary analytical approaches were applied: i) nonmetric multidimensional scaling (nMDS) and analysis of similarities (ANOSIM) were used to compare the composition of biological assemblages between countries; ii) ecological quality ratio (EQR) values were visually explored using raincloud plots and compared between countries using Wilcoxon signed-rank tests across BQEs and iii) Kendall’s tau rank correlation analyses were conducted to evaluate the agreement between ecological status classes derived from national assessment systems.

Multivariate analysis of biological assemblage

Differences in biological assemblage composition between Portugal (PT) and Spain (SP) were examined using nonmetric multidimensional scaling (nMDS) for visual exploration of multivariate patterns and analysis of similarities (ANOSIM) for statistical testing of between-group differences.

Analyses were conducted separately for each BQE. Species-level data were used for fish and macrophytes, family-level data for macroinvertebrates and genus-level data for diatoms. Using genus-level resolution for diatoms ensured comparability between the Portuguese and Spanish datasets, which differed in taxonomic resolution. Before analysis, rare taxa were removed to reduce noise associated with sporadic occurrences. Taxa occurring in fewer than 5% of sampling sites were excluded. The remaining abundance data were square-root transformed to reduce the influence of dominant taxa (Clarke & Warwick, 2001). Bray–Curtis similarity matrices were constructed separately for each BQE. Two-dimensional nMDS ordinations were generated to visualise patterns of similarity among sampling sites. Ordination quality was assessed using stress values, following the interpretation guidelines proposed by (Clarke & Warwick, 2001), where stress values below 0.10 indicate a good representation of the multivariate relationships and values above 0.20 should be interpreted with caution.

Differences between countries were tested using a one-way ANOSIM, with country (Portugal vs. Spain) as the grouping factor. The ANOSIM R statistic was used to quantify the degree of separation between groups, with values approaching 1 indicating strong separation between groups and values near 0 indicating little or no difference. Statistical significance was assessed using 999 permutations. All multivariate analyses were conducted using PRIMER v7 (Clarke & Gorley, 2015).

Analysis of ecological quality ratios (EQRs)

EQR values were calculated for each BQE (fish, macrophytes, macroinvertebrates and diatoms) and were treated as continuous variables. EQRs are calculated as the ratio between observed values and type-specific reference conditions. The EQR range is conceptually bounded between 0 and 1 under the Water Framework Directive (European Commission, 2000) and related intercalibration procedures, to ensure comparability (Birk et al., 2012; Poikane et al., 2014). Each sampling site was treated as a paired observation, with corresponding ecological assessments from Portugal and Spain. Differences in EQR values between Portugal and Spain were assessed using Wilcoxon signed-rank tests for paired samples, conducted separately for each BQE. Statistical significance was assessed at α = 0.05. All statistical analyses were conducted using JASP (JASP Team, 2024).

Kendall’s tau correlation analysis

To enable comparison between ecological classifications, WFD ecological status classes were treated as ordinal variables and encoded on a numerical scale ranging from 1 (high status) to 5 (bad status), for Kendall’s tau correlation analyses. This transformation allowed the application of rank-based statistical methods to assess agreement between national assessment systems.

To assess the degree of association and potential redundancy among BQE-specific ecological quality classifications and final ecological classifications derived from the Portuguese and Spanish assessment systems, pairwise Kendall’s rank correlation coefficients (τ) were calculated. Kendall’s tau was selected because the variables were not assumed to be normally distributed, included ties and were expected to be monotonic rather than strictly linear. Kendall’s τ was calculated for each pair of variables using all sampling sites. Correlation coefficients range from −1 (perfect negative correlation) to +1 (perfect positive correlation), with values close to zero indicating weak or no correlation. Statistical significance was assessed using two-tailed tests with α = 0.05. The resulting coefficients were arranged into correlation matrices and visualised as correlograms to facilitate the interpretation of the strength and direction of relationships among variables.

Results

Multivariate analysis of biological assemblages

Nonmetric multidimensional scaling (nMDS), based on square-root–transformed data and Bray–Curtis similarity, was applied to fish fauna, macrophyte, macroinvertebrate and diatom assemblages to examine differences in community composition between countries (Fig. 2a–d). Two-dimensional ordinations produced stress values ranging from 0.14 to 0.24, indicating variable goodness of fit among ordinations, with the macrophyte ordination showing the highest stress value.

Fig. 2.

Nonmetric multidimensional scaling (nMDS) ordinations of biological assemblage composition for a fish, b macrophytes, c macroinvertebrates and d diatoms across Portuguese (PT) and Spanish (SP) sampling sites. Ordinations were based on square-root-transformed abundance data using Bray–Curtis similarity. Each point represents a sampling site, and colours indicate the country where samples were collected. Stress values indicate the goodness-of-fit of the two-dimensional ordinations

For fish assemblages (Fig. 2a; stress = 0.15), samples from PT and SP showed extensive overlap in ordination space. Consistent with this pattern, ANOSIM indicated no statistically significant difference between countries (R =  − 0.005, p = 0.466). For macrophyte assemblages (Fig. 2b; stress = 0.24), PT and SP samples were distributed in distinct regions of ordination space, with limited overlap. ANOSIM detected a statistically significant difference between countries (R = 0.409, p = 0.001), although the relatively high stress value suggests that the ordination should be interpreted with caution. For macroinvertebrate assemblages (Fig. 2c; stress = 0.14), PT and SP samples showed partial overlap in ordination space. Although ANOSIM indicated a significant difference between countries, the low R value suggests weak separation between groups (R = 0.109, p = 0.002). For diatom assemblages (Fig. 2d; stress = 0.17), PT and SP samples formed distinct clusters with minimal overlap. ANOSIM indicated a statistically significant difference between countries (R = 0.635, p = 0.001). Across the four biological elements, ANOSIM R values ranged from − 0.005 (fish) to 0.635 (diatoms), indicating differences in the degree of assemblage separation between countries.

Analysis of ecological quality ratios (EQRs)

The distribution of EQR values varied among BQEs and between countries (Fig. 3). Raincloud plots highlighted differences in EQR distributions and paired site-level comparisons between Portuguese (PT) and Spanish (SP) assessment systems.

Fig. 3.

Raincloud plots showing the distribution of ecological quality ratio (EQR) values for each biological quality element (fish, macrophytes, macroinvertebrates and diatoms) in Portugal (PT) and Spain (SP). Raincloud plots combine kernel density estimates, boxplots representing median values and interquartile ranges and individual paired observations to illustrate data distributions, variability and paired comparisons between countries

For fish, EQR values tended to be lower in Spain than in Portugal, although this difference was not statistically significant (Wilcoxon signed-rank test, p = 0.057). Macrophyte EQR values showed similar distributions between countries, with no significant differences detected (p = 1.000). For macroinvertebrates, EQR values tended to be higher in Spain, although differences were again not statistically significant (p = 0.055). In contrast, diatom EQR values were significantly higher in Spain than in Portugal (p < 0.001).

Overall, only diatom EQRs showed a significant difference between countries, whereas fish and macroinvertebrates exhibited nonsignificant trends, and macrophytes showed near-identical EQR distributions between countries.

Kendall’s tau correlation analysis

Kendall’s rank correlations revealed varying levels of agreement between Portuguese and Spanish ecological quality classifications across BQEs (Fig. 4). Significant correlations are indicated by asterisks in the correlation matrix.

Fig. 4.

Kendall’s tau correlation matrix among biological quality elements (BQE) classifications and final ecological classifications derived from the Portuguese (PT) and Spanish (SP) assessment systems. Correlation coefficients (τ) are shown in the upper triangle, while ellipses in the lower triangle indicate the direction and strength of correlations. Blue colours represent positive correlations, and red colours represent negative correlations. Asterisks denote statistically significant correlations (p < 0.05)

Between countries, a moderate positive and statistically significant correlation was observed for fish (τ = 0.40). Macrophytes showed a weak, nonsignificant positive correlation between countries (τ = 0.20), while macroinvertebrates showed a low, nonsignificant positive correlation (τ = 0.30). Diatom assessments showed a very weak, nonsignificant negative correlation between countries (τ = −0.10).

Within the Portuguese assessment system, a statistically significant negative correlation was observed between fish and diatoms (τ =  − 0.60). The Portuguese final classification showed statistically significant positive correlations with Portuguese fish (τ = 0.50) and with Portuguese macroinvertebrates (τ = 0.30).

Within the Spanish assessment system, statistically significant positive correlations were observed between fish and macroinvertebrates (τ = 0.50). The Spanish final ecological classification showed statistically significant positive correlations with fish (τ = 1.00), macroinvertebrates (τ = 0.40) and diatoms (τ = 0.40).

The final ecological classifications assigned by Portugal and Spain showed weak and nonsignificant agreement (τ = 0.30). Pairwise disagreement in BQE-specific between Portugal and Spain varied among BQEs, reaching 86.9% for fish, 44.4% for macrophytes, 59.5% for macroinvertebrates and 70.3% for diatoms. Overall disagreement in the final ecological status classification, obtained through the “one-out, all-out” principle, reached 80.0%.

Discussion

This study evaluated cross-border consistency in biological assemblage composition and ecological quality assessment under the WFD across five international Iberian river basins. Overall, the results indicate that cross-border comparability was strongly dependent on the biological quality elements (BQE) considered, particularly with respect to assemblage composition, EQR distributions and ecological quality classifications. Clear differences between Portugal and Spain were observed for some BQEs, whereas others showed substantial overlap in assemblage composition and only moderate agreement in assessment outcomes. These findings indicate that cross-border comparability under the WFD remains insufficiently resolved and depends on both the biological characteristics considered for each quality element and the translation of national assessment systems into ecological quality scores. Similar patterns have been reported in broader WFD literature, showing that biological assessment methods may differ substantially among MS even under a common normative framework and that intercalibration improves comparability of class boundaries but does not eliminate methodological differences among national systems (Birk et al., 2012; Hering et al., 2010; Poikane et al., 2014).

The multivariate analyses revealed contrasting patterns among BQEs’ taxa. Because these analyses considered Portugal and Spain as independent groups, the resulting patterns should be interpreted as broad comparisons of assemblage composition between national datasets rather than strict pairwise comparisons between corresponding sampling sites. Diatoms showed the strongest overall separation between Portuguese and Spanish assemblages, while macrophytes also displayed substantial separation between countries. In contrast, macroinvertebrates showed only weak separation, and fish assemblages did not differ significantly between countries. These differences may reflect methodological differences and differences in taxonomic expertise when applying national monitoring protocols, including sampling procedures and identification level.

The strong separation observed for diatoms is related to their bioassessment performance, which is sensitive to taxonomic resolution and sample processing. In this study, a genus-level harmonisation was necessary to ensure comparability between countries, which may have reduced biological detail while still retaining sensitivity to broad ecological patterns (Rimet & Bouchez, 2011; Taurozzi et al., 2024). However, the original datasets from each country showed substantial differences in the taxonomic resolution and taxa recorded by the respective identification teams. In addition, variability in diatom identification among analysts and taxonomic teams may introduce further inconsistencies between datasets and affect their comparability (Almeida et al., 2013; Riato et al., 2022; Tyree et al., 2020). Diatoms are highly responsive indicators of environmental gradients and water quality variation (Rimet & Bouchez, 2011), and differences in taxonomic resolution and identification may therefore contribute to the observed differences in assemblage composition and ecological assessment between countries. In contrast, the absence of significant compositional differences in fish suggests lower taxonomic uncertainty, lower species diversity and greater cross-border similarity in fish assemblages than in the other BQEs, albeit at different sampling times. Fish communities generally integrate ecological conditions over broader spatial and temporal scales and are influenced by connectivity and catchment-scale processes, which may reduce abrupt differences across political boundaries when hydrological connectivity is maintained (Pont et al., 2021; Roset et al., 2007). Macroinvertebrates showed only weak separation despite methodological differences between national protocols. Because WFD macroinvertebrate assessments are typically based on family-level identification, assemblage data reflect broader ecological patterns rather than fine-scale taxonomic differences, which may smooth cross-border contrasts (Birk & Hering, 2006; Jones, 2008). Macrophytes, in turn, showed clearer differentiation, which may reflect both their sensitivity to environmental gradients and potential differences in survey timing and field protocols between national monitoring programmes, which are known to influence macrophyte assemblage assessments (Aguiar et al., 2014; Kanninen et al., 2013). Differences in taxonomic expertise may also influence macrophyte assessments, particularly for bryophytes and macroalgae, which are often difficult to identify.

The EQR analysis further supports the conclusion that cross-border comparability was BQE-specific and broadly consistent with the patterns observed in biological assemblage composition. Only diatoms showed significant differences in EQR values between countries, whereas fish and macroinvertebrates showed nonsignificant differences with p-values close to the significance threshold, and macrophytes showed nearly identical EQR distributions, in spite of the taxonomic divergences.

MS apply a wide range of biological assessment methods that differ in metric structure, pressure–response relationships and calibration procedures, even when they are formally intercalibrated. Although intercalibration harmonises ecological status class boundaries, national differences in index formulation and sampling design may still influence EQR outcomes (Birk et al., 2012, 2013; Poikane et al., 2014). When interpreted together with the multivariate and correlation analyses, the EQR patterns show that differences in assemblage composition and assessment outcomes do not necessarily coincide across BQEs. The present study was not designed to disentangle the relative contribution of assemblage composition from that of the specific national assessment systems to cross-border disagreement. Rather, the complementary analyses allow divergences to be identified at different stages of the assessment process, from the biological assemblages recorded to the EQR values and ecological quality classifications.

Kendall’s tau analysis added an important complementary perspective by showing that agreement between Portuguese and Spanish ecological quality classifications also varied among BQEs. Significant positive correlations between countries were observed only for fish, whereas macrophytes and macroinvertebrates showed weak and nonsignificant positive correlations, and diatoms showed a very weak negative and nonsignificant correlation between countries. Final ecological classification also showed weak and nonsignificant positive correlation between countries.

Taken together with the multivariate and EQR analyses, these results suggest that agreement between national assessment systems broadly reflects the biological patterns observed in assemblages’ composition and their interpretation for the ecological assessment. Fish assemblages showed no significant compositional separation between countries, and fish-based ecological quality classifications were moderately and significantly correlated between national systems, indicating a similar ecological ranking of sites across the Portuguese and Spanish assessment frameworks. In contrast, diatom assemblages exhibited strong compositional separation, significant differences in EQR distributions and no cross-country agreement in ecological quality classifications, suggesting that cross-border divergence was evident at both the assemblage and assessment levels of this BQE. Within-country relationships also differed between national systems. In Portugal, fish and diatom quality classifications were negatively correlated, suggesting that these elements may respond differently to environmental pressures. In Spain, the perfect correlation between fish classifications and final ecological classification indicates that fish assessments strongly determined the final status in the analysed dataset. This highlights how individual biological elements may exert disproportionate influence on final classification outcomes, particularly under the one-out-all-out (OOAO) principle.

Although both countries operate under the common normative framework established by the WFD and its intercalibration procedure, the national systems still differ in sampling season, spatial coverage, sample treatment, taxonomic resolution and index construction. The WFD requires ecological assessment within a shared legal framework, but implementation remains method-specific and typology-specific at the national level. European intercalibration was designed precisely because independently developed methods were not automatically comparable (European Commission, 2003; Poikane et al., 2014; Willby et al., 2014). Previous studies have similarly highlighted that differences among biological indices and assessment approaches remain a challenge for consistent implementation of the WFD across Europe (Hering et al., 2010; Santos et al., 2021; Voulvoulis et al., 2017).

The OOAO principle plays a key role in shaping cross-border consistency. Under the WFD, final ecological status is determined by the lowest classified BQE, making the final assessment sensitive to whichever BQE is most restrictive (European Commission, 2003; Voulvoulis et al., 2017). In this study, the weak agreement in final ecological classification between Portugal and Spain suggests that the OOAO principle did not fully override methodological differences between national systems. For example, the perfect correlation between fish classifications and final ecological classification in Spain indicates that this element dominated the integrated assessment in the analysed dataset. In this context, OOAO may amplify differences among national assessment systems when a single BQE disproportionately influences the final ecological classification. This reinforces the need to improve comparability at the individual BQE level before their integration into the final ecological status. Greater coordination of monitoring and assessment procedures between neighbouring countries could reduce BQE-specific discrepancies that may subsequently be propagated through the OOAO principle. The present results therefore support strengthening cross-border harmonisation of the biological assessments on which the final classification is based.

The analysis focused on biological quality elements and did not explicitly incorporate supporting physicochemical or hydromorphological variables. Neither physicochemical variables nor hydromorphological quality assessments were directly comparable between countries and were not explicitly incorporated into the comparative analyses presented. In addition, because EQRs were derived using the type-specific reference conditions established within each national assessment system, the specific contribution of differences in reference conditions to cross-border disagreement could not be disentangled from other methodological differences between the Portuguese and Spanish assessment systems. Physicochemical and hydromorphological elements should receive further attention and harmonisation efforts across countries, while being given more relevance in intercalibration procedures. More broadly, the study design does not allow the relative contribution of assemblage composition and national assessment procedures to the observed cross-border disagreement to be quantified independently. Despite these limitations, the study provides a structured comparison of cross-border ecological assessment outcomes under the WFD and identifies comparability at both the biological assemblage and national assessment levels in shared river basins. These findings have direct implications for transboundary river management and highlight the need to harmonise procedures and taxonomic indicators. If neighbouring countries classify the same WB differently, management priorities, restoration measures and compliance reporting also diverge. Improved harmonisation of sampling procedures and taxonomic resolution could therefore enhance cross-border consistency, particularly for biological elements showing greater divergence. In fact, without such harmonisation, the restoration outcomes will remain uncertain. This broader reading is in line with recent calls to strengthen the effectiveness, transparency and ecological interpretability of WFD implementation across Europe (Birk et al., 2012; Santos et al., 2021; Voulvoulis et al., 2017). For the programme of measures (PoM), the best approach would be active collaboration between neighbouring countries and common decisions for boundary and transboundary waters.

Conclusion

This study evaluated the operational comparability of ecological status assessments conducted independently by Portugal and Spain in Iberian transboundary rivers under the Water Framework Directive. The results showed that cross-border consistency varied substantially among biological quality elements. Fish assessments showed the highest degree of comparability between countries, whereas diatoms presented the strongest divergence in assemblage composition, EQR values and ecological quality classifications. Macrophytes and macroinvertebrates showed intermediate patterns, with weak agreement between national assessment systems.

Overall, the results demonstrate that formal intercalibration of ecological assessment methods does not necessarily guarantee operational comparability at the scale of shared water bodies. Differences in sampling procedures, taxonomic resolution and index implementation can still produce divergent ecological classifications between neighbouring countries, even under a common regulatory framework.

These findings highlight the need for improved harmonisation of monitoring procedures, taxonomic practices and assessment protocols in transboundary river basins. Strengthening cross-border coordination is essential to ensure more consistent ecological assessments and to support coherent river basin management and restoration planning under the Water Framework Directive.

Author contribution

A.T.F.: conceptualisation, methodology, investigation and writing—original draft; J.O.: supervision, data curation and writing—original draft; G.D.: methodology; P.B.: supervision, methodology and writing—review and editing; P.S.: conceptualisation, methodology, investigation and writing—review and editing; T.F.: conceptualisation, supervision, methodology and writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

The Open Access was funded by FCT – Fundação para a Ciência e a Tecnologia, I.P., through the project UID/00239/2025 (https://doi.org/10.54499/UID/00239/2025) of the Forest Research Centre. This work was supported by the ALBUFEIRA project (INTERREG ALBUFEIRA–0489_Albufeira_6_E), co-financed by the European Regional Development Fund (ERDF) through the Interreg V-A Spain–Portugal Programme (POCTEP) 2014–2020. António Tovar Faro was additionally supported by a Ph.D. grant from the FLUVIO–River Restoration and Management programme funded by Fundação para a Ciência e a Tecnologia I. P. (FCT), Portugal (2021.06859.BD) and by Centro de Estudos Florestais (CEF), funded by FCT under UID/00239. Gonçalo Duarte has been financed by FCT through CEECINSTLA/0001/2022/TERRA–Laboratório Associado para a Sustentabilidade do Uso da Terra e dos Serviço de Ecossistemas. Paulo Branco was financed by national funds via FCT (LA/P/0092/2020). Forest Research Centre (CEF) is a research unit funded by FCT, through project reference UID/00239/2025 and UID/PRR/00239/2025. The Associate Laboratory for Sustainable Land Use and Ecosystem Services (TERRA) is also funded by FCT (LA/P/0092/2020–https://doi.org/10.54499/LA/P/0092/2020).

Data availability

The data are available from the corresponding author upon reasonable request. The data are not publicly available due to potential use in future works.

Declarations

Ethics approval

This manuscript is original, has not been published previously and is not under consideration for publication elsewhere. The work complies with the journal’s ethical standards, and all authors have approved the submitted version of the manuscript. To the best of our knowledge, the manuscript contains no fabricated, falsified or plagiarised material.

Competing interests

The authors declare no competing interests.