Stable isotopes as a predictor for organic or conventional classification of berries and vegetables

Xia Zhu-Barker; Michael Liou; Diana Zapata; Jingyi Huang; William R. Horwath

doi:10.1371/journal.pone.0318179

Abstract

Organic agriculture is expanding worldwide, driven by expectations of improving food quality and soil health. However, while organic certification by regulatory bodies such as the United States Department of Agriculture and the European Union confirms compliance with organic standards that prohibit synthetic chemical inputs, there is limited oversight to verify that organic practices, such as the use of authentic organic fertilizer sources, are consistently applied at the field level. This study investigated the elemental content of carbon (C) and nitrogen (N) and their stable isotopes (δ¹³C and δ¹⁵N) in seven different crops grown under organic or conventional practices to assess their applicability as a screening tool to verify the authenticity of organic labeled produce. Holm corrected Welch t-tests and a generalized linear mixed model (GLMM) were used to assess the potential of stable isotope or crop elemental content to differentiate organic vs. conventional production systems. Total C and N content or C/N ratio was not significantly different between production systems or among geographic origins for most crops. However, the average N stable isotope (δ¹⁵N) content differed, with conventional crops at 1.8 ± 2.2‰ and organic at 6.0 ± 3.4‰. A mixed model incorporating elemental contents and stable isotopes identified δ¹⁵N as the primary predictor in discriminating organic and conventional production systems. A δ¹⁵N threshold is suggested to differentiate conventional from organic grown raspberries (δ¹⁵N < 2.17‰) and strawberries (δ¹⁵N < 3.22‰), for an estimated false negative rate of 1%. Although further evaluation is needed, our extensive dataset (n = 791) captures key predictors of agricultural production systems and holds potential as a benchmark for future organic production verification.

Citation: Zhu-Barker X, Liou M, Zapata D, Huang J, Horwath WR (2025) Stable isotopes as a predictor for organic or conventional classification of berries and vegetables. PLoS ONE 20(3): e0318179. https://doi.org/10.1371/journal.pone.0318179

Editor: Taimoor Hassan Farooq, Central South University of Forestry and Technology, CHINA

Received: August 1, 2024; Accepted: January 10, 2025; Published: March 12, 2025

Copyright: © 2025 Zhu-Barker et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The datasets used in this study and all the scripts for data analysis were deposited in the public data Repository Figshare (DOI: 10.6084/m9.figshare.24844026 or https://figshare.com/articles/software/Stable_Isotopes_as_a_Predictor_for_Organic_or_Conventional_Classification_of_Berries_and_Vegetables_Code_Supplement_/24844026).

Funding: The study was supported by the Office of the Vice Chancellor for Research and Graduate Education at the University of Wisconsin-Madison with funding from the Wisconsin Alumni Research Foundation and the J.G. Boswell Endowed Chair in Soil Science. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Abbreviations: C, carbon; N, nitrogen; C/N, carbon to nitrogen ratio; δ¹⁵N, the N stable isotopic signature at natural abundance; δ¹³C, the C stable isotopic signature at natural abundance; GLMM, generalized linear mixed model; AICc, Akaike Information Criteria, corrected; SS, subject-specific; PA, population-averaged; AUROC, areas under the receiver operating characteristic curve; FNR, false negative rate.

Introduction

Organic agriculture has a reputation for having fewer negative consequences to the environment, and the consumption of organic produce is considered to promote human health [1,2]. However, while organic farming has advantages, it also presents unique challenges. For produce to be certified organic, such as in the U.S., it must be grown without synthetic fertilizer and pesticide inputs, adhere to defined standards of production, and undergo an annual certification audit [3]. The exclusion of conventional agrochemicals in organic farming often leads to higher production costs compared to conventional farming, primarily due to lower yields, increased labor and input costs, and greater crop losses from pests and diseases [4,5]. Among these challenges, organic farming systems navigate the complexities introduced by the prohibition of synthetic inputs, which adds layers of management difficulty and economical constraints that are not typically encountered in conventional production systems [6,7]. This prohibition poses a particular challenge in aligning nitrogen (N) availability from organic sources with the timing of crop demand. The difficulty in synchronizing N availability with crop demand in organic farming systems has led to one of the most commonly reported organic violations: the use of synthetic N fertilizers [8]. Consequently, the development of tools or protocols to verify the exclusive use of organic fertilizer material in organic farming has become a priority to help maintain integrity of organic production [9].

Organic systems typically rely on organic sources of N, derived from animal or plant matter, such as by-products of fish and food processing, animal manures, biological N fixation by legumes, and other organic amendments like compost [10]. This contrasts with conventional systems, which utilize synthetic fertilizers to supply N in available inorganic forms. Organic N sources undergo microbial mineralization and nitrification to be converted into plant-available N forms such as ammonium (NH₄⁺) and nitrate (NO₃⁻) [11–14]. Proteins and their constituents, peptides and amino acids, are the main N source in organic fertilizers. Due to enzymatic reactions that discriminate against the heavier isotope during biochemical transformations, organic N sources often exhibit enrichment in ¹⁵N (δ¹⁵N > 0) [15]. Conversely, conventional agriculture primarily depends on synthetically produced N fertilizers, which are manufactured with the Haber-Bosch process to convert atmospheric dinitrogen into ammonia. The δ¹⁵N values of synthetic fertilizer N sources closely mirror that of atmospheric N (δ¹⁵N = 0).

Often described as “nature’s ecological recorder”, stable isotopes are widely used to provide insights into plant resource use, trophic food web interactions, and as biochemical and environmental probes to disentangle various temporal and spatial scales during plant and animal growth [16–19]. The stable isotope signature (e.g., δ¹³C or δ¹⁵N), defined as the ratio of an element’s stable isotope in relation to an international reference standard, can vary based on factors such as the element’s initial concentration, the biochemical discrimination processes it undergoes, and environmental conditions [20–24]. This variability in the stable isotope signature is particularly noticeable among different crop production inputs, especially in fertilizer materials containing N, a critical plant macronutrient that significantly regulates plant productivity [25,26]. This makes it possible to use the δ¹⁵N value of produce to indicate the specific N-based fertilizing material used in a cropping system.

Given the distinct differences in the N isotopic signature (δ¹⁵N) between synthetic and organic fertilizer N sources, this metric has been used to discriminate between organic and synthetic fertilizers [15] and, therefore, has potential to distinguish the sources of applied N fertilizer in crop from the different production systems [27–29]. Bueno et al. [30] demonstrated that the δ¹⁵N value of organically grown tomatoes was higher than conventional production systems. A study investigating isotopic compositions of the soil inorganic N pool after the application of compost or fertilizer found that its δ¹⁵N value ranged from −3 to + 2‰ in soils that received fertilizer, while compost-treated soils exhibited a value greater than + 8‰ [31]. This suggests that the δ¹⁵N value of organically grown plants (i.e., those received compost) would be higher than that of conventionally grown plants (i.e., those treated with synthetic fertilizer), given that inorganic N is the most abundant form of N available for plant uptake. Although the δ¹⁵N value of produce may serve as an indicator of organic or synthetic fertilizer sources in crop production, broader investigations encompassing various production regions and crop species are necessary. Further considerations should include other potential indicators closely related to N cycling in cropping systems, such as carbon (C) content and its isotopic signature (δ¹³C). These investigations are necessary to evaluate the potential of these stable isotopic indicators as screening tools to verify the authenticity of organic labeled produce [32].

This study aims to evaluate the use of C and N contents and their stable isotope signatures as screening tools to discriminate between organic and conventional crops using a mixed model approach. We examined δ¹⁵N, δ¹³C, total N, total C, and C:N ratio of imported and locally produced fresh produce samples found across California markets to evaluate the authenticity of organic produce. The collection of a large number of samples is crucial for the development of robust predictive and explanatory models.

Materials and methods

Sample collection

Fresh samples of organically or conventionally grown produce were purchased from January through December of 2018 from food markets located across Fresno, Monterey, Salinas, and Sacramento Counties in California (n = 791). All samples were kept on ice while transported to the laboratory and stored in a freezer until preparation and analyses. Fresh produce geographical region of origin included the U.S., Mexico, Canada, Chile, China, Guatemala, Ecuador, Serbia, Peru, and Poland. Nearly 97% of the food samples analyzed were grown in Mexico and the U.S. (Fig S1A in S1 File). The remaining countries were grouped as “others” for the analysis. A similar number of organic and conventional fresh produce were collected (Fig S1B in S1 File, Table 1). The fresh produce samples represented 99 commercial food producers (companies). Seven major organically grown crops were selected for the study and consisted of berries (strawberry (Fragaria × ananassa), blueberry (Vaccinium sect. Cyanococus), blackberry (Rubus subg. Rubus), and raspberry (Rubus idaeus)) and vegetables (broccoli (Brassica oleracea var. italica), celery (Apium graveolens), and lettuce (Lactuca sativa var. longifolia)) (Fig S1C in S1 File).

Download:

Table 1. Number of the organic and conventional fresh produce samples collected by crops.

https://doi.org/10.1371/journal.pone.0318179.t001

Elemental composition and stable isotope analysis

Samples were blended in a food processor, freeze-dried, and ground into powder with a mortar and pestle for elemental analysis of total C and N content and relative abundance of δ¹³C and δ¹⁵N using a combustion analyzer (Elementar Analysensysteme GmbH, Hanau, Germany) interfaced to an isotope ratio mass spectrometer (PDZ Europa 20-20 IRMS, Sercon Ltd., Cheshire, U.K.) at the Stable Isotope Facility, University of California, Davis. Isotope abundances are expressed in δ notation as parts per mil (‰) and calculated as δ¹³C = [((¹³C/¹²C)_sample/(¹³C/¹²C)_standard) − 1) × 1000] and δ¹⁵N = [((¹⁵N/¹⁴N)_sample/(¹⁵N/¹⁴N)_standard − 1) × 1000].

Statistical analysis

To evaluate the potential of using either elemental contents (e.g., C, N, C/N ratio) and or stable isotopes (e.g., δ¹³C, δ¹⁵N) for differentiating conventional and organic production, Holm corrected Welch t-tests and Tukey mean difference tests were applied to compare the mean values of these predictors between different crops and from different regions. Visualizations and data manipulation throughout use the “tidyverse” packages [33] in the R statistical computing language [34] (R Core Team, 2022).

A generalized linear mixed model (GLMM) with binary responses (logit link) of conventional vs. organic produce was used as the fundamental regression model. The classification model thresholds fitted probabilities from the regression model for prediction. This method was chosen for the ideal balance of classification flexibility, interpretability, and power. Predictor variables were selected by a combination of graphical examination, and automated model selection with Akaike Information Criteria, corrected (AICc) as the objective to minimize. AICc was chosen to discourage overfitting of the model and optimize for interpretability alongside performance. Model selection up to the second-order interaction across nine variables relating to elemental contents, stable isotopes, and source of crop were considered. GLMM fitting was achieved by adaptive Gaussian Quadrature for maximum likelihood, with the package “GLMMadaptive” [35]. The top 5 models were manually reviewed with a battery of graphical diagnostics (residual and influence plots), hypothesis testing (Osius-Rojek, Hosmer-Lemeshow, Stukel’s goodness-of-fit), numerical stability (optimization parameters), and basic classification metrics (ROC curves, sensitivity, specificity, and accuracy).

The final chosen model is a function of C/N ratio, C percentage, δ¹⁵N, Crop type, and Company:

where E[Y] is the expected probability of a data point being observed as organic, and parentheses denote a random effect on the intercept. “Company” is modeled as a random effect to reflect the assumption that each producer company can have specific effects on the crop properties but belong to a more general population of companies that may not be represented in the data. This has advantages especially when some companies have a limited number of samples by “borrowing information” from companies with more data.

Diagnostics of the fitted model show several influential points around the extremes of the fitted distribution, likely due to the sampling structure of the dataset. There is poor fit around the logistic function in the 3^rd and 7^th deciles of fitted probability ranges, ( and respectively. The main effect coefficients of celery and blackberry appear the most sensitive to optimization parameters (~1 se), all other coefficients are sufficiently stabilized with 11 quadrature points. The observed tails of the company random effect are slightly larger than expected from the normal distribution, but not large enough for the normality assumption to be inappropriate (Shapiro-Wilk 0.354). Variable importance in the model is quantified by (hierarchical) term deletion (Rao) score test statistics in the fully fixed model, i.e., to evaluate the importance of C/N ratio, C/N ratio is removed from the model as well as any other higher order interactions involving C/N ratio.

The non-linearity of the link function and random effect of “Company” produce two sets of estimate targets, either coefficients conditional on the value of “Company” or with random effects marginalized out. These targets are commonly described as “subject-specific” (SS) models or “population-averaged” (PA) models in biostatistical literature [36]. We opt for the PA interpretation because the aggregate response of the entire population is our focus. Marginalized coefficients are estimated by Monte Carlo integration, using robust standard errors to protect against possible likelihood misspecification.

Overall quality of classification model is evaluated with areas under the receiver operating characteristic curve (AUROC). Searching for appropriate cutoff values for classification should be “Company” agnostic with the primary goal of < 1% false negative rate (FNR) of misclassification with “organic” as the positive case. FNR control is chosen to avoid penalizing honest organic farmers, and flag cases that are most likely conventionally grown but are labeled as organic. Prediction accuracy is used as a secondary characteristic for evaluating model usefulness. Separate effect profiles for each crop are created at the dataset average levels of C/N ratio, C percentage, and δ¹⁵N by crop, and 95% level confidence intervals for estimated probability are generated by normal approximation. For practicality of summarizing results, we consider single dimensional cutoffs with other variables held at crop specific averages but note that presence of interaction effects imply relative values of cutoffs may differ across different profiles of predictive variables. Classification criteria is developed for each crop, and different threshold values for the regression model are decided by taking the highest predicted probability that will produce a classification with FNR < 1% with 95% confidence. The confidence intervals around the FNR were generated by nonparametrically bootstrapping the dataset with 2000 replicates.

Results

Total C and N content, and C/N ratio

Carbon content varied among crops, with no differences among production systems or geographic origin (Table S1 in S1 File). Overall, carbon content ranged from 31.7 to 48.2%. Among the crops, celery contained the lowest average C (35.4%), and raspberry had the highest C content (43.8%). As indicated by the standard deviation, the highest C content variation was found in raspberry and blackberry (1.7% and 1.8%, respectively) (Fig 1b).

Download:

Fig 1. Boxplots of nitrogen (a) and carbon (b) content (%), and C/N ratio (c) of conventional and organic crops.

Bars above boxplots denote a Welch two-sample t-test between conventional and organic crop, with adjusted (***), (**), (*), and non-significant otherwise (ns). p-values are Holm corrected by each family defined by tests within the same subfigure.

https://doi.org/10.1371/journal.pone.0318179.g001

Nitrogen content was significantly different among crops and geographic regions (p < 0.0001); however, there was no evidence of difference between production systems, except for celery for which higher N content was found in conventional systems than in organic systems (Fig 1a). More variability in mean N content was found among crops, ranging from 0.2 to 6.0%, and followed the order: blueberries (0.5%) < blackberries (1.1%) < strawberries (1.2%) < raspberries (1.3%) < celery (1.7%) < lettuce (3.4%) < broccoli (4.6%). Among geographic origin, N content was highest in the U.S. (2.27%), followed by Mexico (1.79%) and other countries (0.90%). Within geographic origin, only the U.S. showed significant differences in N content between conventional and organic systems (p < 0.0001) (Table 2).

Download:

Table 2. Isotopic composition of C (δ¹³C, ‰) and N (δ¹⁵N, ‰) by production system and geographic origin. Entries are in the format “mean (sd) letter”. Letter displays show Tukey range test significance for each isotope, among countries (columns). Different letters indicate the group difference of means is significantly different than 0 at the 0.05 level. The p-value column shows a Welch t-test between production systems within each country for each measured variable (rows).

https://doi.org/10.1371/journal.pone.0318179.t002

The C/N ratio was significantly different between crops and geographic origin (p < 0.0001), with no differences between production systems (Table S1 in S1 File). Overall, vegetables had a lower C/N ratio compared to berries (Fig 1c): broccoli (10.4) < lettuce (12.1) < celery (23.2) < strawberries (36.1) < raspberries (37.2) < blackberries (40.5) < blueberries (81.8). Within geographic origin, the U.S. had significant differences in C/N ratio between organic and conventional systems in celery (p < 0.0001).

Natural δ¹³C abundance

Significant differences in δ¹³C value were found among geographic origin, production systems and crops (p < 0.001). Overall, δ¹³C values ranged from −30.95‰ to −21.28‰. Generally, crops grown in Mexico exhibited higher natural δ¹³C abundance than that grown in the U.S. and the rest of the countries (Table S1 in S1 File). Overall, organic produce revealed a slightly, but significantly lower δ¹³C than conventional produce (Fig 2a, Table 2). Within a geographic origin, only Mexico had significant differences in δ¹³C values between organic and conventional crops (p = 7.41e-07) (Table 2). Besides, the analyses of intrinsic differences among the crops showed that vegetables (broccoli, celery, and lettuce) had lower δ¹³C values than berries (blackberry, blueberries, raspberry, and strawberry) (Fig 2b, Table S1 in S1 File).

Download:

Fig 2. Stable isotope ratio of δ¹³C (a and b) and δ¹⁵N (c and d) of organic and conventional crops.

Bars above boxplots denote a Welch two-sample t-test between conventional and organic crop, with (***), (**), (*), and non-significant otherwise (ns). p-values are Holm corrected by each family defined by subfigure.

https://doi.org/10.1371/journal.pone.0318179.g002

Natural δ¹⁵N abundance

The δ¹⁵N values ranged from −4.2 to + 17.5‰, with an overall average and standard deviation of 4.04 ± 3.54‰. The δ¹⁵N values also varied among crops as: blueberries < broccoli < strawberry < raspberry < blackberry < lettuce < celery (Fig 2d). The geographical origin and production system significantly affected the δ¹⁵N values of the crops (p < 0.001). The crops grown in the U.S. and Mexico generally exhibited significantly higher average δ¹⁵N, with 3.9 ± 3.9‰ and 4.8 ± 3.2‰, respectively, than crops grown in other countries (Canada, Chile, China, Guatemala, Ecuador, Serbia, Peru, and Poland) with 1.2 ± 1.6‰ (Tables S1 and S2 in S1 File).

A strong separation in δ¹⁵N values was observed between conventional and organic systems for most crops (Fig 2c). Regardless of the differences in δ¹⁵N among the geographical origins, there was a three-fold increase in mean δ¹⁵N values of organic crops compared to conventionally produced crops (Fig 2c). Average δ¹⁵N value was 1.8 ± 2.2‰ for conventional and 5.9 ± 3.4‰ for organic crops. Celery, lettuce, raspberry, and strawberry showed the largest difference in δ¹⁵N values between conventional and organic systems.

Predicting crop production system

In the fully fixed version of our final model, the ranking of highest variable importance (respecting the hierarchical principle by removing dependent higher order terms) is: company (score = 294.64, p < 1.0e⁻¹⁶), δ¹⁵N (score = 168.45, p < 1.0e⁻¹⁶), crop (score = 61.28, p = 1.3e⁻⁰⁶), C/N ratio (score = 35.88, p = 7.7e⁻⁰⁶), C %(score = 8.91, p = 2.8e⁻⁰³) (Table S2 in S1 File). The presence of crop × C/N ratio and crop × δ¹⁵N interactions imply each crop has a different relationship between C/N Ratio, δ¹⁵N and organic classification. For δ¹⁵N, the strongest slopes (log-odds scale) are for raspberry (slope = 3.36, p = 6.95e⁻⁰⁴), strawberry (slope = 0.80, p = 9.66e⁻⁰⁵), lettuce (slope = 0.35, p = 0.242) and celery (slope = 0.30, p = 3.89e⁻⁰⁵). For C/N Ratio, the strongest slopes (log-odds scale) are for raspberry (slope = −0.30, p = 1.87e⁻⁰⁴), broccoli (slope = 0.28, p = 0.16), strawberry (slope = 0.08, p = .068). C percentage across all crops has an estimated slope (log-odds scale) of 0.09 (p = 0.11) (Table S3 in S1 File).

In classification, the SS model of fitted probabilities give an AUROC of 0.983, while the PA model has an AUROC of 0.863, showing the drop in accuracy with a company agnostic classification mode (Fig 3). Since each crop’s characterization is different for classification (Fig S2 in S1 File), controlling the FNR < 1%, we get different probability thresholds and performance metrics for each crop (Table S4 in S1 File). Holding C % and C/N ratio at their crop specific sample means, the cutoff values were δ¹⁵N < 2.17 ‰ for raspberries and δ¹⁵N < 3.22 ‰ for strawberry. Similarly, holding C % and δ¹⁵N at crop specific averages, raspberries with C/N ratio > 69.2 should be flagged as suspiciously non-organic. Other crops and covariates do not reach the < 1% FNR threshold standard for univariate cutoffs for the average crop (Fig 4).

Download:

Fig 3. ROC curves based on predicted probabilities of organic classification from GLMM model across dataset.

Population averaged and subject specific models for “company” are both shown. The drop in AUROC between the models show the loss of company information in classification. The tradeoff between true positive rate (TPR) and false positive rate (FPR) is relatively symmetric in the tails. We focus our attention on regions of high TPR for developing crop specific cutoffs for covariates.

https://doi.org/10.1371/journal.pone.0318179.g003

Download:

Fig 4. Effect profile plots of (a) δ15N, (b) C/N Ratio, (c) C percentage with crop specific sample averages for other covariates.

Fitted probabilities are population-averaged across company/regions. 95% confidence intervals (normal approximation) are shown for fitted probability of organic classification in dotted lines. Marginal sample distributions of covariates for each crop are shown as ticks across the bottom axis.

https://doi.org/10.1371/journal.pone.0318179.g004

Discussion

The impact of production systems on crop elemental compositions

We found consistently greater δ¹⁵N values in organic (5.98 ± 3.38 ‰) than in conventional (1.84 ± 2.19 ‰) systems (Table 2). Classification analysis indicated that δ¹⁵N natural abundance was the biological variable with the highest predictive capability for distinguishing between organic and conventional food production systems. This consistent difference aligns with previous studies [27,30], which report higher δ¹⁵N values in organic crops compared to conventional crops. The elevated δ¹⁵N values in organic farming systems are primarily attributed to the exclusive use of organic fertilizers and amendments such as manure or compost, which inherently possess higher δ¹⁵N signatures [15]. In contrast, conventional farming systems predominantly rely extensively on synthetic fertilizers, which typically have lower δ¹⁵N values [15]. While the occasional use of organic amendments such as manure or compost in conventional systems may lead to higher δ¹⁵N values compared to exclusive synthetic fertilizer use, the δ¹⁵N values of certified organic produce are generally expected to remain above those of conventional counterparts due to the prohibition of synthetic fertilizers in organic farming. Contamination from adjacent conventional fields may also contribute to δ¹⁵N variability in organic produce, potentially narrowing the differences between organic and conventional systems. However, such contamination is unlikely to cause δ¹⁵N signatures of certified organic produce to fall below those commonly observed in conventional systems. Therefore, although the use of a single δ¹⁵N threshold may seem arbitrary and does not fully capture variations in climate regions, soil conditions, land management practices, or organic fertilizer sources, δ¹⁵N remains a promising preliminary assessment tool. Specifically, it can help identify food products with exceptionally low δ¹⁵N values that warrant further investigation to verify the authenticity of organic labels.

Overall, differences in δ¹⁵N between production systems were more prominent in broccoli, celery, lettuce, raspberries, and strawberries (Fig 2). Blackberries showed little differences in δ¹⁵N abundance between production systems, being relatively high in both types of farming systems, which is possibly due to the heavy use of natural N sources (bloodmeal, feather meal, composted manure) in both organic and conventional blackberries [37]. A study in a selected group of vegetables (potato, leek, cauliflower, tomato) in Slovenia found that the N stable isotope ratio was a successful technique to identify production practices [38]. Vegetables like garlic, onion, parsley, sweet pepper, and carrots showed δ¹⁵N values overlapped between conventional and organic production practices. In the Slovenian study, the differences in δ¹⁵N values found between organically and conventionally grown products ranged from 3 to 6‰; nevertheless, it is essential to mention that the analyzed sample size was relatively small (n = 107, approx. 3–6 samples per vegetable). A larger sample size, as conducted in the current study (n = 791), could reduce the dispersion in δ¹⁵N values, which are known to vary widely by species, soil management (e.g., N input, crop rotation), and environmental conditions [39]. For example, in an extensive study analyzing over 100 soils from 20 U.S. states, the δ¹⁵N of soil was found to vary from 5 to 12‰, influenced by factors such as climate, depth, soil pH, and land use [40].

The C and N content of berries and vegetables were not significantly different between the two production systems (organic and conventional), except conventionally grown celery showed a higher average N content than organic (2.17% vs. 1.45%). Similar findings were reported in celery, in which the high N content was found in conventional grown celery after supplying a high amount of inorganic N, mainly as ammonium (NH₄⁺) [41]. After further analyzing the N forms in the previous study, it was found that higher NH₄⁺ content occurred in the conventional grown celery than the organic grown celery, while nitrate (NO₃⁻) content was not significantly different between these two production systems.

Predicting cropping systems and perspectives

The elemental composition analysis of food has been demonstrated in several studies to differentiate between crop production systems from different geographical origins [9,28,30]. However, to the best of our knowledge, no study has attempted to determine a threshold for specific crops, possibly due to the limitations in the sample size used in these studies. In the current study, an extensive crop sampling (n = 791) was conducted to select variables that could help to verify the authenticity of organic production of organic berries and vegetables.

The covariates used in the final fitted model were evaluated by associative strength and do not imply a causal relationship. Nor does it exclude other variables from potentially having strong relationship with organic and conventional classification. “Company” emerges as the strongest predictor in these models, surpassing any biological measurement. This is understandable considering that some companies exclusively produce organic crops, while others focus solely on conventional crops. Moreover, some companies produce specific subsets of crops, and estimating the company effects for crops they do not produce requires imputation. Knowing the crop company would thus significantly bias the prediction one way or another. Our cross-sectional summaries displayed relatively good coverage of both crops and organic systems. However, developing screening criteria should be company agnostic; thus, these considerations point to modeling the company as a more general population. The variance structure was not deeply explored in this application and would require more structured sampling and larger datasets for numerical feasibility. We did observe slightly larger variability in the actual dataset than was assumed during modeling. A deeper exploration of the source of this variability could yield more accurate models.

Although model misfit was observed in regions of extreme probability, the increased variability in these regions probably won’t affect the classification problem application. The increased variability in these extremes is largely due to data points from sparsely sampled companies or extreme values of biological measurements. All other regions of the model fit well and have been shown to reliably differentiate between organic and conventional crops. These biases and the dependence on dataset sampling should be considered when interpreting model results.

From this dataset, we have found compelling evidence for δ¹⁵N to be used as screening criteria for classifying organic and conventional systems in raspberries and strawberries (Fig 4). The cutoffs developed are in tail regions of our data as expected, but this makes precision of lower FNR control harder to scrutinize. Increased systematic sampling, i.e., examples of truly, fraudulently grown crops, would further ensure that the classification criteria suggested is robust to any unmeasured confounders. There is insufficient evidence for many of the other crops to achieve low levels of false negatives due to their weaker distinguishing features. We note that the relationship of lettuce with δ¹⁵N appears to be robust and would benefit from a further investigation with an increased sample size. C/N ratio also has a strong association with broccoli, raspberries, and strawberries, indicating that these covariates have potential to serve as a useful criterion for classification. Interestingly, raspberries display a trend where higher C/N ratios reduce the chance of being classified as organic – a pattern only evident when conditioned on C % and δ¹⁵N and absent in marginal distributions, meriting further investigation.

Our findings suggest that by implementing crop elemental composition analysis – particularly N stable isotopic analysis – it is feasible to predict the food production system at the market level. The classification tool we employed can pinpoint produce potentially fertilized with synthetic N sources, guiding further authenticity investigations. Although distinct differences in δ¹⁵N among produces were observed and preliminary crop-specific thresholds set, the N stable isotope ratio’s variance associated with soil organic matter and climate conditions [22,42] might constrain its broader application. Future studies should focus on structured sampling with a larger dataset to investigate relationships among soil properties, climatic conditions, land management, and crop elemental compositions to better develop and apply such screening tools.

Conclusions

Our study showed that the total C and N content, as well as the C/N ratio, showed little variation between production systems and geographical origins for most crops. It’s particularly significant to note that δ¹⁵N, with observed values averaging 1.8 ± 2.2‰ for conventional crops and 6.0 ± 3.4‰ for organic, emerged as the most consequential predictor in distinguishing production systems. Indeed, organic produce consistently exhibited higher δ¹⁵N values. Models were developed for seven crops with isotope data for which strawberries, raspberries, lettuce, and celery were identified as the most promising candidates for classification of production methods based on these models. A δ¹⁵N threshold is suggested to differentiate conventional from organic grown raspberries (δ¹⁵N < 2.17‰) and strawberries (δ¹⁵N < 3.22‰), for an estimated false negative rate of 1%. While further investigations are necessary to solidify the δ¹⁵N thresholds between organic and conventional production methods, our extensive dataset of 791 samples provides invaluable insights. It not only sheds light on the characteristics of different agricultural production systems but also lays the groundwork for the enhancement of organic verification processes in the future.

Supporting information

S1 File. Supplementary for “Stable isotopes as a predictor for organic or conventional classification of berries and vegetables”.

https://doi.org/10.1371/journal.pone.0318179.s001

(PDF)

Acknowledgments

We thank Eshan R. Toosi and Martin Burger for their comments on earlier versions of this manuscript.

References

1. Skinner C, Gattinger A, Krauss M, Krause H-M, Mayer J, van der Heijden MGA, et al. The impact of long-term organic farming on soil-derived greenhouse gas emissions. Sci Rep. 2019;9(1):1702. pmid:30737429
- View Article
- PubMed/NCBI
- Google Scholar
2. Baudry J, Assmann KE, Touvier M, Allès B, Seconda L, Latino-Martel P, et al. Association of frequency of organic food consumption with cancer risk: findings from the NutriNet-Santé Prospective Cohort Study. JAMA Intern Med. 2018;178(12):1597–606. pmid:30422212
- View Article
- PubMed/NCBI
- Google Scholar
3. Willer H, Schlatter B, Trávnícek J, Kemper L, Julia L. The World Of Organic Agriculture. Statistics & emerging trends 2020. Research Institute of Organic Agriculture (FiBL) & IFOAM‑Organic International. 2020. 2020.
4. Meemken E-M, Qaim M. Organic agriculture, food security, and the environment. Annu Rev Resour Econ. 2018;10(1):39–63.
- View Article
- Google Scholar
5. Knapp S, van der Heijden MGA. A global meta-analysis of yield stability in organic and conservation agriculture. Nat Commun. 2018;9(1):3632. pmid:30194344
- View Article
- PubMed/NCBI
- Google Scholar
6. Seufert V, Ramankutty N, Foley JA. Comparing the yields of organic and conventional agriculture. Nature. 2012;485(7397):229–32. pmid:22535250
- View Article
- PubMed/NCBI
- Google Scholar
7. Kirchmann H, Thorvaldsson G. Challenging targets for future agriculture. Eur J Agron. 2000;12(3–4):145–61.
- View Article
- Google Scholar
8. Downing J. Organic farms unknowingly used a synthetic fertilizer. Sacramento Bee, Sacramento. 2008.
9. Novak V, Adler J, Husted S, Fromberg A, Laursen KH. Authenticity testing of organically grown vegetables by stable isotope ratio analysis of oxygen in plant-derived sulphate. Food Chem. 2019;29159–67. pmid:31006471
- View Article
- PubMed/NCBI
- Google Scholar
10. Alneyadi KSS, Almheiri MSB, Tzortzakis N, Di Gioia F, Ahmed ZFR. Organic-based nutrient solutions for sustainable vegetable production in a zero-runoff soilless growing system. J Agric Food Res. 2024;15101035.
- View Article
- Google Scholar
11. Parton WJ, Schimel DS, Cole CV, Ojima DS. Analysis of factors controlling soil organic matter levels in great plains grasslands. Soil Science Soc of Amer J. 1987;51(5):1173–9.
- View Article
- Google Scholar
12. Cabrera F, Martín-Olmedo P, Lopez R, Murillo J. Nitrogen mineralization in soils amended with composted olive mill sludge. Nutr Cycl Agroecosys. 2005;71(3):249–58.
- View Article
- Google Scholar
13. Melillo JM, Aber JD, Linkins AE, Ricca A, Fry B, Nadelhoffer KJ. Carbon and nitrogen dynamics along the decay continuum: Plant litter to soil organic matter. Plant Soil. 1989;115(2):189–98.
- View Article
- Google Scholar
14. Muller A, Schader C, El-Hage Scialabba N, Brüggemann J, Isensee A, Erb K-H, et al. Strategies for feeding the world more sustainably with organic agriculture. Nat Commun. 2017;8(1):1290. pmid:29138387
- View Article
- PubMed/NCBI
- Google Scholar
15. Mukome FND, Doane TA, Silva LCR, Parikh SJ, Horwath WR. Testing protocol ensures the authenticity of organic fertilizers. Cal Ag. 2013;67(4):210–6.
- View Article
- Google Scholar
16. Blackmer AM, Bremner JM. Nitrogen isotope discrimination in denitrification of nitrate in soils. Soil Biol Biochem. 1977;9(2):73–7.
- View Article
- Google Scholar
17. Boutton TW, Archer SR, Midwood AJ. Stable isotopes in ecosystem science: structure, function and dynamics of a subtropical Savanna. Rapid Commun Mass Spectrom. 1999;13(13):1263–77. pmid:10407309
- View Article
- PubMed/NCBI
- Google Scholar
18. West JB, Bowen GJ, Cerling TE, Ehleringer JR. Stable isotopes as one of nature’s ecological recorders. Trends Ecol Evol. 2006;21(7):408–14. pmid:16753238
- View Article
- PubMed/NCBI
- Google Scholar
19. Barbour MM. Stable oxygen isotope composition of plant tissue: a review. Funct Plant Biol. 2007;34(2):83–94. pmid:32689335
- View Article
- PubMed/NCBI
- Google Scholar
20. Robinson D. delta(15)N as an integrator of the nitrogen cycle. Trends Ecol Evol. 2001;16(3):153–62. pmid:11179580
- View Article
- PubMed/NCBI
- Google Scholar
21. Ma J-Y, Sun W, Liu X-N, Chen F-H. Variation in the stable carbon and nitrogen isotope composition of plants and soil along a precipitation gradient in northern China. PLoS One. 2012;7(12):e51894. pmid:23272186
- View Article
- PubMed/NCBI
- Google Scholar
22. Craine JM, Elmore AJ, Wang L, Augusto L, Baisden WT, Brookshire ENJ, et al. Convergence of soil nitrogen isotopes across global climate gradients. Sci Rep. 2015;58280. pmid:25655192
- View Article
- PubMed/NCBI
- Google Scholar
23. Craine JM, Brookshire ENJ, Cramer MD, Hasselquist NJ, Koba K, Marin-Spiotta E, et al. Ecological interpretations of nitrogen isotope ratios of terrestrial plants and soils. Plant Soil. 2015;396(1–2):1–26.
- View Article
- Google Scholar
24. Chung I-M, Lee T-J, Oh Y-T, Ghimire BK, Jang I-B, Kim S-H. Ginseng authenticity testing by measuring carbon, nitrogen, and sulfur stable isotope compositions that differ based on cultivation land and organic fertilizer type. J Ginseng Res. 2017;41(2):195–200. pmid:28413324
- View Article
- PubMed/NCBI
- Google Scholar
25. LeBauer DS, Treseder KK. Nitrogen limitation of net primary productivity in terrestrial ecosystems is globally distributed. Ecology. 2008;89(2):371–9. pmid:18409427
- View Article
- PubMed/NCBI
- Google Scholar
26. Fowler D, Coyle M, Skiba U, Sutton MA, Cape JN, Reis S, et al. The global nitrogen cycle in the twenty-first century. Philos Trans R Soc Lond B Biol Sci. 2013;368(1621):20130164. pmid:23713126
- View Article
- PubMed/NCBI
- Google Scholar
27. Trandel MA, Vigardt A, Walters SA, Lefticariu M, Kinsel M. Nitrogen isotope composition, nitrogen amount, and fruit yield of tomato plants affected by the soil-fertilizer types. ACS Omega. 2018;3(6):6419–26. pmid:30023947
- View Article
- PubMed/NCBI
- Google Scholar
28. Rogers KM. Nitrogen isotopes as a screening tool to determine the growing regimen of some organic and nonorganic supermarket produce from New Zealand. J Agric Food Chem. 2008;56(11):4078–83. pmid:18489112
- View Article
- PubMed/NCBI
- Google Scholar
29. Inácio CT, Chalk PM, Magalhães AMT. Principles and limitations of stable isotopes in differentiating organic and conventional foodstuffs: 1. Plant products. Crit Rev Food Sci Nutr. 2015;55(9):1206–18. pmid:24915332
- View Article
- PubMed/NCBI
- Google Scholar
30. Martínez Bueno MJ, Díaz-Galiano FJ, Rajski Ł, Cutillas V, Fernández-Alba AR. A non-targeted metabolomic approach to identify food markers to support discrimination between organic and conventional tomato crops. J Chromatogr A. 2018;154666–76. pmid:29526497
- View Article
- PubMed/NCBI
- Google Scholar
31. Choi W-J, Ro H-M, Lee S-M. Natural 15N abundances of inorganic nitrogen in soil treated with fertilizer and compost under changing soil moisture regimes. Soil Biol Biochem. 2003;35(10):1289–98.
- View Article
- Google Scholar
32. Georgi M, Voerkelius S, Rossmann A, Graßmann J, Schnitzler WH. Multielement isotope ratios of vegetables from integrated and organic production. Plant Soil. 2005;275(1–2):93–100.
- View Article
- Google Scholar
33. Wickham H, Averick M, Bryan J, Chang W, McGowan L, François R, et al. Welcome to the Tidyverse. JOSS. 2019;4(43):1686.
- View Article
- Google Scholar
34. Team RC. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2022. p. https://www.R-project.org/.
35. Rizopoulos D. GLMMadaptive: Generalized Linear Mixed Models using Adaptive Gaussian Quadrature. 2023;https://github.com/drizopoulos/GLMMadaptive.
- View Article
- Google Scholar
36. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics. 1988;44(4):1049–60. pmid:3233245
- View Article
- PubMed/NCBI
- Google Scholar
37. Strik BC, Bryla DR. Uptake and partitioning of nutrients in blackberry and raspberry and evaluating plant nutrient status for accurate assessment of fertilizer requirements. hortte. 2015;25(4):452–9.
- View Article
- Google Scholar
38. Sturm M, Lojen S. Nitrogen isotopic signature of vegetables from the Slovenian market and its suitability as an indicator of organic production. Isotopes Environ Health Stud. 2011;47(2):214–20. pmid:21644134
- View Article
- PubMed/NCBI
- Google Scholar
39. Bateman AS, Kelly SD, Woolfe M. Nitrogen isotope composition of organically and conventionally grown crops. J Agric Food Chem. 2007;55(7):2664–70. pmid:17341092
- View Article
- PubMed/NCBI
- Google Scholar
40. Shearer G, Kohl DH, Chien S. The nitrogen‐15 abundance in a wide variety of soils. Soil Science Soc of Amer J. 1978;42(6):899–902.
- View Article
- Google Scholar
41. Gąstoł M, Domagała-Świątkiewicz I, Krośniak M. Organic versus conventional–a comparative study on quality and nutritional value of fruit and vegetable juices. Biol Agric Hortic. 2011;27(3–4):310–9.
- View Article
- Google Scholar
42. Choi W-J, Kwak J-H, Park H-J, Yang H, Park S-I, Xu Z, et al. Land-use type, and land management and disturbance affect soil δ15N: a review. J Soil Sediment. 2020;20(9):3283–99.
- View Article
- Google Scholar

[ref1] 1. Skinner C, Gattinger A, Krauss M, Krause H-M, Mayer J, van der Heijden MGA, et al. The impact of long-term organic farming on soil-derived greenhouse gas emissions. Sci Rep. 2019;9(1):1702. pmid:30737429
View Article
PubMed/NCBI
Google Scholar

[2] View Article

[3] PubMed/NCBI

[4] Google Scholar

[ref2] 2. Baudry J, Assmann KE, Touvier M, Allès B, Seconda L, Latino-Martel P, et al. Association of frequency of organic food consumption with cancer risk: findings from the NutriNet-Santé Prospective Cohort Study. JAMA Intern Med. 2018;178(12):1597–606. pmid:30422212
View Article
PubMed/NCBI
Google Scholar

[6] View Article

[7] PubMed/NCBI

[8] Google Scholar

[ref3] 3. Willer H, Schlatter B, Trávnícek J, Kemper L, Julia L. The World Of Organic Agriculture. Statistics & emerging trends 2020. Research Institute of Organic Agriculture (FiBL) & IFOAM‑Organic International. 2020. 2020.

[ref4] 4. Meemken E-M, Qaim M. Organic agriculture, food security, and the environment. Annu Rev Resour Econ. 2018;10(1):39–63.
View Article
Google Scholar

[11] View Article

[12] Google Scholar

[ref5] 5. Knapp S, van der Heijden MGA. A global meta-analysis of yield stability in organic and conservation agriculture. Nat Commun. 2018;9(1):3632. pmid:30194344
View Article
PubMed/NCBI
Google Scholar

[14] View Article

[15] PubMed/NCBI

[16] Google Scholar

[ref6] 6. Seufert V, Ramankutty N, Foley JA. Comparing the yields of organic and conventional agriculture. Nature. 2012;485(7397):229–32. pmid:22535250
View Article
PubMed/NCBI
Google Scholar

[18] View Article

[19] PubMed/NCBI

[20] Google Scholar

[ref7] 7. Kirchmann H, Thorvaldsson G. Challenging targets for future agriculture. Eur J Agron. 2000;12(3–4):145–61.
View Article
Google Scholar

[22] View Article

[23] Google Scholar

[ref8] 8. Downing J. Organic farms unknowingly used a synthetic fertilizer. Sacramento Bee, Sacramento. 2008.

[ref9] 9. Novak V, Adler J, Husted S, Fromberg A, Laursen KH. Authenticity testing of organically grown vegetables by stable isotope ratio analysis of oxygen in plant-derived sulphate. Food Chem. 2019;29159–67. pmid:31006471
View Article
PubMed/NCBI
Google Scholar

[26] View Article

[27] PubMed/NCBI

[28] Google Scholar

[ref10] 10. Alneyadi KSS, Almheiri MSB, Tzortzakis N, Di Gioia F, Ahmed ZFR. Organic-based nutrient solutions for sustainable vegetable production in a zero-runoff soilless growing system. J Agric Food Res. 2024;15101035.
View Article
Google Scholar

[30] View Article

[31] Google Scholar

[ref11] 11. Parton WJ, Schimel DS, Cole CV, Ojima DS. Analysis of factors controlling soil organic matter levels in great plains grasslands. Soil Science Soc of Amer J. 1987;51(5):1173–9.
View Article
Google Scholar

[33] View Article

[34] Google Scholar

[ref12] 12. Cabrera F, Martín-Olmedo P, Lopez R, Murillo J. Nitrogen mineralization in soils amended with composted olive mill sludge. Nutr Cycl Agroecosys. 2005;71(3):249–58.
View Article
Google Scholar

[36] View Article

[37] Google Scholar

[ref13] 13. Melillo JM, Aber JD, Linkins AE, Ricca A, Fry B, Nadelhoffer KJ. Carbon and nitrogen dynamics along the decay continuum: Plant litter to soil organic matter. Plant Soil. 1989;115(2):189–98.
View Article
Google Scholar

[39] View Article

[40] Google Scholar

[ref14] 14. Muller A, Schader C, El-Hage Scialabba N, Brüggemann J, Isensee A, Erb K-H, et al. Strategies for feeding the world more sustainably with organic agriculture. Nat Commun. 2017;8(1):1290. pmid:29138387
View Article
PubMed/NCBI
Google Scholar

[42] View Article

[43] PubMed/NCBI

[44] Google Scholar

[ref15] 15. Mukome FND, Doane TA, Silva LCR, Parikh SJ, Horwath WR. Testing protocol ensures the authenticity of organic fertilizers. Cal Ag. 2013;67(4):210–6.
View Article
Google Scholar

[46] View Article

[47] Google Scholar

[ref16] 16. Blackmer AM, Bremner JM. Nitrogen isotope discrimination in denitrification of nitrate in soils. Soil Biol Biochem. 1977;9(2):73–7.
View Article
Google Scholar

[49] View Article

[50] Google Scholar

[ref17] 17. Boutton TW, Archer SR, Midwood AJ. Stable isotopes in ecosystem science: structure, function and dynamics of a subtropical Savanna. Rapid Commun Mass Spectrom. 1999;13(13):1263–77. pmid:10407309
View Article
PubMed/NCBI
Google Scholar

[52] View Article

[53] PubMed/NCBI

[54] Google Scholar

[ref18] 18. West JB, Bowen GJ, Cerling TE, Ehleringer JR. Stable isotopes as one of nature’s ecological recorders. Trends Ecol Evol. 2006;21(7):408–14. pmid:16753238
View Article
PubMed/NCBI
Google Scholar

[56] View Article

[57] PubMed/NCBI

[58] Google Scholar

[ref19] 19. Barbour MM. Stable oxygen isotope composition of plant tissue: a review. Funct Plant Biol. 2007;34(2):83–94. pmid:32689335
View Article
PubMed/NCBI
Google Scholar

[60] View Article

[61] PubMed/NCBI

[62] Google Scholar

[ref20] 20. Robinson D. delta(15)N as an integrator of the nitrogen cycle. Trends Ecol Evol. 2001;16(3):153–62. pmid:11179580
View Article
PubMed/NCBI
Google Scholar

[64] View Article

[65] PubMed/NCBI

[66] Google Scholar

[ref21] 21. Ma J-Y, Sun W, Liu X-N, Chen F-H. Variation in the stable carbon and nitrogen isotope composition of plants and soil along a precipitation gradient in northern China. PLoS One. 2012;7(12):e51894. pmid:23272186
View Article
PubMed/NCBI
Google Scholar

[68] View Article

[69] PubMed/NCBI

[70] Google Scholar

[ref22] 22. Craine JM, Elmore AJ, Wang L, Augusto L, Baisden WT, Brookshire ENJ, et al. Convergence of soil nitrogen isotopes across global climate gradients. Sci Rep. 2015;58280. pmid:25655192
View Article
PubMed/NCBI
Google Scholar

[72] View Article

[73] PubMed/NCBI

[74] Google Scholar

[ref23] 23. Craine JM, Brookshire ENJ, Cramer MD, Hasselquist NJ, Koba K, Marin-Spiotta E, et al. Ecological interpretations of nitrogen isotope ratios of terrestrial plants and soils. Plant Soil. 2015;396(1–2):1–26.
View Article
Google Scholar

[76] View Article

[77] Google Scholar

[ref24] 24. Chung I-M, Lee T-J, Oh Y-T, Ghimire BK, Jang I-B, Kim S-H. Ginseng authenticity testing by measuring carbon, nitrogen, and sulfur stable isotope compositions that differ based on cultivation land and organic fertilizer type. J Ginseng Res. 2017;41(2):195–200. pmid:28413324
View Article
PubMed/NCBI
Google Scholar

[79] View Article

[80] PubMed/NCBI

[81] Google Scholar

[ref25] 25. LeBauer DS, Treseder KK. Nitrogen limitation of net primary productivity in terrestrial ecosystems is globally distributed. Ecology. 2008;89(2):371–9. pmid:18409427
View Article
PubMed/NCBI
Google Scholar

[83] View Article

[84] PubMed/NCBI

[85] Google Scholar

[ref26] 26. Fowler D, Coyle M, Skiba U, Sutton MA, Cape JN, Reis S, et al. The global nitrogen cycle in the twenty-first century. Philos Trans R Soc Lond B Biol Sci. 2013;368(1621):20130164. pmid:23713126
View Article
PubMed/NCBI
Google Scholar

[87] View Article

[88] PubMed/NCBI

[89] Google Scholar

[ref27] 27. Trandel MA, Vigardt A, Walters SA, Lefticariu M, Kinsel M. Nitrogen isotope composition, nitrogen amount, and fruit yield of tomato plants affected by the soil-fertilizer types. ACS Omega. 2018;3(6):6419–26. pmid:30023947
View Article
PubMed/NCBI
Google Scholar

[91] View Article

[92] PubMed/NCBI

[93] Google Scholar

[ref28] 28. Rogers KM. Nitrogen isotopes as a screening tool to determine the growing regimen of some organic and nonorganic supermarket produce from New Zealand. J Agric Food Chem. 2008;56(11):4078–83. pmid:18489112
View Article
PubMed/NCBI
Google Scholar

[95] View Article

[96] PubMed/NCBI

[97] Google Scholar

[ref29] 29. Inácio CT, Chalk PM, Magalhães AMT. Principles and limitations of stable isotopes in differentiating organic and conventional foodstuffs: 1. Plant products. Crit Rev Food Sci Nutr. 2015;55(9):1206–18. pmid:24915332
View Article
PubMed/NCBI
Google Scholar

[99] View Article

[100] PubMed/NCBI

[101] Google Scholar

[ref30] 30. Martínez Bueno MJ, Díaz-Galiano FJ, Rajski Ł, Cutillas V, Fernández-Alba AR. A non-targeted metabolomic approach to identify food markers to support discrimination between organic and conventional tomato crops. J Chromatogr A. 2018;154666–76. pmid:29526497
View Article
PubMed/NCBI
Google Scholar

[103] View Article

[104] PubMed/NCBI

[105] Google Scholar

[ref31] 31. Choi W-J, Ro H-M, Lee S-M. Natural 15N abundances of inorganic nitrogen in soil treated with fertilizer and compost under changing soil moisture regimes. Soil Biol Biochem. 2003;35(10):1289–98.
View Article
Google Scholar

[107] View Article

[108] Google Scholar

[ref32] 32. Georgi M, Voerkelius S, Rossmann A, Graßmann J, Schnitzler WH. Multielement isotope ratios of vegetables from integrated and organic production. Plant Soil. 2005;275(1–2):93–100.
View Article
Google Scholar

[110] View Article

[111] Google Scholar

[ref33] 33. Wickham H, Averick M, Bryan J, Chang W, McGowan L, François R, et al. Welcome to the Tidyverse. JOSS. 2019;4(43):1686.
View Article
Google Scholar

[113] View Article

[114] Google Scholar

[ref34] 34. Team RC. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2022. p. https://www.R-project.org/.

[ref35] 35. Rizopoulos D. GLMMadaptive: Generalized Linear Mixed Models using Adaptive Gaussian Quadrature. 2023;https://github.com/drizopoulos/GLMMadaptive.
View Article
Google Scholar

[117] View Article

[118] Google Scholar

[ref36] 36. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics. 1988;44(4):1049–60. pmid:3233245
View Article
PubMed/NCBI
Google Scholar

[120] View Article

[121] PubMed/NCBI

[122] Google Scholar

[ref37] 37. Strik BC, Bryla DR. Uptake and partitioning of nutrients in blackberry and raspberry and evaluating plant nutrient status for accurate assessment of fertilizer requirements. hortte. 2015;25(4):452–9.
View Article
Google Scholar

[124] View Article

[125] Google Scholar

[ref38] 38. Sturm M, Lojen S. Nitrogen isotopic signature of vegetables from the Slovenian market and its suitability as an indicator of organic production. Isotopes Environ Health Stud. 2011;47(2):214–20. pmid:21644134
View Article
PubMed/NCBI
Google Scholar

[127] View Article

[128] PubMed/NCBI

[129] Google Scholar

[ref39] 39. Bateman AS, Kelly SD, Woolfe M. Nitrogen isotope composition of organically and conventionally grown crops. J Agric Food Chem. 2007;55(7):2664–70. pmid:17341092
View Article
PubMed/NCBI
Google Scholar

[131] View Article

[132] PubMed/NCBI

[133] Google Scholar

[ref40] 40. Shearer G, Kohl DH, Chien S. The nitrogen‐15 abundance in a wide variety of soils. Soil Science Soc of Amer J. 1978;42(6):899–902.
View Article
Google Scholar

[135] View Article

[136] Google Scholar

[ref41] 41. Gąstoł M, Domagała-Świątkiewicz I, Krośniak M. Organic versus conventional–a comparative study on quality and nutritional value of fruit and vegetable juices. Biol Agric Hortic. 2011;27(3–4):310–9.
View Article
Google Scholar

[138] View Article

[139] Google Scholar

[ref42] 42. Choi W-J, Kwak J-H, Park H-J, Yang H, Park S-I, Xu Z, et al. Land-use type, and land management and disturbance affect soil δ15N: a review. J Soil Sediment. 2020;20(9):3283–99.
View Article
Google Scholar

[141] View Article

[142] Google Scholar

Figures

Abstract

Introduction

Materials and methods

Sample collection

Elemental composition and stable isotope analysis

Statistical analysis

Results

Total C and N content, and C/N ratio

Natural δ13C abundance

Natural δ15N abundance

Predicting crop production system

Discussion

The impact of production systems on crop elemental compositions

Predicting cropping systems and perspectives

Conclusions

Supporting information

S1 File. Supplementary for “Stable isotopes as a predictor for organic or conventional classification of berries and vegetables”.

Acknowledgments

References

Natural δ¹³C abundance

Natural δ¹⁵N abundance