The exercise traces its origins to Pillar Five of the Global Soil Partnership, which calls for harmonized methods, measures, and indicators for sustainable soil management. In November 2017, FAO established the Global Soil Laboratory Network, GLOSOLAN, at its Rome headquarters to strengthen soil testing laboratories worldwide. In Spain, the Spanish Society of Soil Science, the University of Zaragoza, and the Agroenvironmental Laboratory of the Government of Aragon adopted the initiative, building an inventory of soil analysis laboratories and launching a series of intercomparison exercises. The first test, run in 2019 and presented in Madrid on 10 April 2019 by SECS president Jaime Porta, exposed significant discrepancies among laboratories and made clear that standardization was urgently needed.

Porta himself is a central figure in the history of Spanish soil analysis. Beginning as an engineer in the State Agricultural Laboratory Network, the so-called García Faure project, he took charge of the Galicia laboratory in 1972, then moved to the Ebro laboratory in Zaragoza, where he automated soil analysis after studying automated systems at the INRA laboratory in Arras, France, and a cooperative laboratory in the Netherlands. He installed Technicon devices, ancestors of modern segmented continuous flow systems, for serial determination of phosphorus and organic matter, and two Granulostas instruments that processed roughly one hundred soil textures per week. His most enduring contribution was in quality control: working in batches with multiple internal controls and establishing sample exchange networks, practices that anticipated formal laboratory quality protocols by two decades.

For the 2021 exercise, two agricultural soil samples were prepared by drying, disaggregation, and sieving at 2 mm, then homogenized in bulk. The samples, named BIPEA, prepared by the French intercomparison organizer, and HUERTO, prepared directly by the Spanish team, were taken from the superficial 30 cm horizon of cultivated soils. Aliquots of approximately 500 g were dispatched to 29 interested Spanish laboratories, of which 26 responded; 17 had participated in the first test, enabling direct comparison. Each laboratory received an Excel sheet specifying units and significant figures, and results were submitted in triplicate, ideally in different batches, under confidentiality codes ranging from 11 to 301 known only to the laboratory and the two organizers.

The statistical machinery followed the WEPAL methodology of the Wageningen evaluating programmes. Rather than trusting a plain mean, the organizers applied robust statistics: the median and deviation from the median were calculated, anomalous data were eliminated through iterations, and a consensus mean and standard deviation emerged from the accepted values. The acceptance interval was set symmetrically at 95% confidence. Each laboratory result was then converted into a Zscore, the distance from the consensus value expressed in consensus standard deviations. Scores with absolute values below 2 were satisfactory, between 2 and 3 questionable, and above 3 unsatisfactory. A combined Zscore, the root mean square across a laboratory’s results for a parameter or for all parameters, summarized overall performance. Because the combined Zscore distributions were not normal, a logarithmic transformation normalized them before one-factor analyses of variance tested for differences between years.

The 2021 results revealed a striking asymmetry between the two samples. HUERTO produced a higher percentage of satisfactory results than BIPEA across most parameters, mainly because its consensus standard deviations were larger, widening the range of acceptable values. For organic matter, 74.3% of HUERTO results were satisfactory against 45.9% for BIPEA; for assimilable potassium, 71.0% versus 49.3%; and for clay, 83.9% versus 77.8%. Assimilable phosphorus and magnesium were the stars of the exercise, with satisfactory rates reaching 94.0% and 90.0% for HUERTO. Sand content proved the most troublesome textural fraction, with only 64.3% and 67.9% satisfactory results, a consequence of very narrow acceptance ranges that punished even small deviations.

Organic matter emerged as the methodological fault line. Most laboratories measured oxidizable organic matter by the Walkley-Black method, which relies on partial carbon oxidation, but two laboratories used calcination and one used the DUMAS procedure, which determines total organic carbon released by combustion. These total-carbon methods systematically overestimate relative to Walkley-Black, and eleven laboratories posted unsatisfactory combined Zscores for organic matter. Four consistently overestimated across all samples, five underestimated only the BIPEA sample, and one laboratory’s pattern of underestimation in BIPEA and overestimation in HUERTO suggested a transcription error with the samples mixed up. Only three laboratories used the alternative methods, too few for separate statistical treatment, but the discrepancy mirrors problems documented in GLOSOLAN and other international exercises, where comparing organic matter values across methods has repeatedly proven difficult.

Assimilable phosphorus told a brighter story. All laboratories responded, and the Olsen method delivered the highest number of satisfactory results for both samples. The improvement was a direct payoff from the 2019 exercise, which had flagged phosphorus as a weak spot and prompted the triplicate design. In the June 2024 BIPEA test, 77% of Olsen phosphorus results were satisfactory within a narrow acceptance range of 0.047 to 0.075 g/kg; the Spanish exercise achieved 84% with slightly wider ranges. Magnesium showed similarly strong performance, with 22 of 25 responding laboratories rated satisfactory. Potassium, by contrast, remained stubbornly problematic, with 9 of 23 laboratories unsatisfactory, and the internationally recognized BIPEA sample again showed greater heterogeneity that the organizers attribute to less homogeneity in the prepared soil rather than to laboratory error alone.

The comparison between 2019 and 2021 added a longitudinal dimension. Of the 17 laboratories analyzed across both exercises, 10 improved their overall combined Zscore, 3 maintained it, and 4 deteriorated. Ratings shifted from 7 unsatisfactory and 4 questionable laboratories in 2019 to 5 unsatisfactory and 7 questionable in 2021. Phosphorus improved from a questionable to a satisfactory global rating, and magnesium moved from questionable to satisfactory, the only parameter showing a statistically significant decrease in Zscore in the ANOVA at P = 0.088. Clay and phosphorus remained satisfactory across both tests, but potassium and sand stayed unsatisfactory, and organic matter regressed from questionable to unsatisfactory. For laboratories 3 and 4, the ANOVA detected significant improvement at P < 0.05, both posting lower combined Zscores in 2021.

The stakes extend well beyond laboratory pride. The CAP Strategic Plan for Spain requires reliable data on the state of agricultural soils, and evaluating organic matter content in Spanish soils is an explicit objective of its environmental monitoring program, which will collect 128,000 samples. Variability among laboratories, the authors warn, makes analytical results a matter of chance, and applying such results can lead to radically wrong fertilizer and conservation decisions, an unacceptable outcome in the national and European context of soil interest. The findings have been communicated to the Ministry of Agriculture, Fisheries and Food, and on 1 October 2024, Order APA/1044/2024 designated the Agroenvironmental Laboratory of the Government of Aragon as the national reference laboratory for agricultural soil fertility analysis, reviving a harmonization effort that lapsed when the Official Commission on Analysis Methods ceased operations in the early 1990s after powers transferred to the Autonomous Regions.

The authors propose annual two-sample intercomparison exercises covering the usual parameters, sharing of analysis methods to standardize determination techniques even where extraction methods remain official, comparative method studies requiring more participating laboratories and more information with results, and expanded coverage of parameters offered by fewer laboratories, from saturated paste and heavy metals to biological parameters and phytosanitary residues. The two-sample format has proven convenient and efficient, and the triplicate requirement, while complicating exercise management, added limited information. The broader lesson resonates internationally: in the 2024 WEPAL exercise, 337 laboratories participated with soil samples and only 10% of values fell outside acceptance ranges, compared with 28% for BIPEA and 15% for HUERTO in the Spanish exercise, though WEPAL participants were pursuing certification. Repetition of proficiency tests, the evidence shows, is itself a driver of improvement, and regular maintenance of these exercises is now framed as an essential tool for the sustainable governance of soil resources.

Subject of Research: Proficiency testing of Spanish soil laboratories for soil fertility analysis quality and harmonization.

Article Title: Proficiency Testing for Soil Fertility Analysis

Article References: Usón Murillo, A., Betrán Aso, J., & Sampériz Sarvisé, M. (2025). Proficiency Testing for Soil Fertility Analysis. Spanish Journal of Soil Science, 15, Article 14838. https://doi.org/10.3389/sjss.2025.14838

Image Credits: AI Generated

DOI: 10.3389/sjss.2025.14838

Keywords: soil fertility, proficiency testing, Zscore, organic matter, Walkley-Black, Olsen phosphorus, GLOSOLAN, laboratory harmonization, soil texture, CAP Strategic Plan, robust statistics, Spain