Constraining parameter uncertainty for predicting oxygen and hydrogen isotope values in fruit.

Understanding δ18O and δ2H values of agricultural products like fruit is of particular scientific interest in plant physiology, ecology, and forensic studies. Applications of mechanistic stable isotope models to predict δ18O and δ2H values of water and organic compounds in fruit, however, are hindered by a lack of empirical parameterizations and validations. We addressed this lack of data by experimentally evaluating model parameter values required to model δ18O and δ2H values of water and organic compounds in berries and leaves from strawberry and raspberry plants grown at different relative humidities. Our study revealed substantial differences between leaf and berry isotope values, consistent across the different relative humidity treatments. We demonstrated that existing isotope models can reproduce water and organic δ18O and δ2H values for leaves and berries. Yet, these simulations require organ-specific model parameterization to accurately predict δ18O and δ2H values of leaf and berry tissue and water pools. We quantified these organ-specific model parameters for both species and relative humidity conditions. Depending on the required model accuracy, species- and environment-specific model parameters may be justified. The parameter values determined in this study thus facilitate applications of stable isotope models where understanding δ18O and δ2H values of fruit is of scientific interest.

Effects of phenotypic variability on the oxygen and hydrogen isotope compositions of grains in different winter wheat varieties.

Stable isotope analyses are the leading method for geographic origin determination, especially of plant-based agricultural products. Origin analysis is typically done by comparing a suspicious sample to reference materials with known geographic origin. Reference materials are usually collected at the species level, assuming different varieties of a species to have comparable isotope compositions within a given location. We evaluated whether different phenotypes that are expressed in different varieties of winter wheat (Triticum aestivum L.) influence the oxygen (δ18O) and hydrogen (δ2H) isotope composition of plant tissue water and organic compounds. We found that mean δ18O and δ2H values among winter wheat varieties did not differ significantly in leaf water, however, differed significantly in bulk dried grain tissue. The differences in bulk dried grain δ18O and δ2H values among varieties can be related to differences in phenotypic trait expression among varieties. Despite this substantial phenotypic variability, the overall variability of bulk dried grain δ18O and δ2H values among varieties was small (SD 0.54 ‰ for oxygen, 3.60 ‰ for hydrogen). We thus conclude that reference materials collected at the species level should be sufficient for geographic origin analysis of winter wheat and possibly other cereals using δ18O and δ2H values.

Using plant physiological stable oxygen isotope models to counter food fraud.

Fraudulent food products, especially regarding false claims of geographic origin, impose economic damages of $30-$40 billion per year. Stable isotope methods, using oxygen isotopes (δ18O) in particular, are the leading forensic tools for identifying these crimes. Plant physiological stable oxygen isotope models simulate how precipitation δ18O values and climatic variables shape the δ18O values of water and organic compounds in plants. These models have the potential to simplify, speed up, and improve conventional stable isotope applications and produce temporally resolved, accurate, and precise region-of-origin assignments for agricultural food products. However, the validation of these models and thus the best choice of model parameters and input variables have limited the application of the models for the origin identification of food. In our study we test model predictions against a unique 11-year European strawberry δ18O reference dataset to evaluate how choices of input variable sources and model parameterization impact the prediction skill of the model. Our results show that modifying leaf-based model parameters specifically for fruit and with product-independent, but growth time specific environmental input data, plant physiological isotope models offer a new and dynamic method that can accurately predict the geographic origin of a plant product and can advance the field of stable isotope analysis to counter food fraud.