Impact of climate change on microbial safety of leafy green vegetables

Summary Climate change is generally recognized as a major threat to humans and the environment. With respect to food production, climate change does not only affect crop production or food security, but possibly also effects on food safety by affecting the prevalence and levels of bacteria, fungi or other pests and pesticides. Fresh-cut or ready-to-eat leafy vegetables (e.g. lettuce and spinach) are increasingly consumed because they are promoted as part of a healthy diet. Such leafy green vegetables (LGVs) are identified as the fresh produce commodity group of highest concern from a microbiological safety perspective, because they are often grown in the open field and therefore vulnerable to contamination and contact with (faeces of) wildlife. Moreover, they are grown and consumed in large volumes and often consumed raw. Bacteria, such as Salmonella spp. and pathogenic Escherichia coli strains are the main pathogens causing foodborne disease through LGVs. A major knowledge gap is understanding how climate change may directly or indirectly affect the contamination of LGVs. This primarily relates to the current lack of methods and tools to link climate data and climate change scenarios to food safety. My thesis aims to quantify the impacts of climate change on microbial safety of pre-harvested LGVs. To achieve this, I reviewed the literature and synthesised major impacts of climate change on contamination sources and pathways of foodborne pathogens (focussing on Escherichia coli O157 and Salmonella spp.) on pre-harvested LGVs (Chapter 2). Subsequently, I developed a statistical model that identifies the weather and management variables that are associated with the LGVs contamination with generic E. coli using regression analysis (Chapter 3). To apply suitable climate data to this statistical model to assess future impacts, I have prepared a tool to downscale coarse climate and climate change data for local food safety scenario analysis (Chapter 4). Finally, I applied the downscaled data to the statistical model and used multi-criteria scenario analysis to explore future food safety (Chapter 5). E.coli is used as a hygienic indicator in this thesis to study microbial safety of LGVs. Its presence is indicative for an increased pathogen presence probability. E. coli and many foodborne bacteria share the same contamination pathways and climate change is expected to similarly impact on both bacteria. Hygienic status is therefore used in my thesis as a proxy for the microbial safety of LGVs. The major result of the literature review in Chapter 2 is that the impact of climate change on LGV contamination depends on the resulting local balance of the positive and negative impacts. The review shows that the interactions between climate change and contamination are real but poorly understood. Therefore, integrative quantitative modelling approaches with scenario analyses and additional laboratory experiments are needed. With this knowledge background, mixed effect logistic regression and linear regression models were developed to identify the climate and management variables that are associated with the presence and concentration of E. coli on LGVs (Chapter 3). These models used E. coli data of 562 lettuce and spinach samples taken between 2011 and 2013 from 23 open-field farms from Belgium, Brazil, Egypt, Norway and Spain. Weather and agriculture management practices together had a systematic influence on E.coli presence and concentration. Temperature explained most of the observed variation on E. coli prevalence and concentration on LGVs. Minimum temperature of the sampling day (odds ratio [OR] 1.47), region and application of inorganic fertilizer explained a significant amount of variation in E. coli prevalence. Maximum temperature on three days before sampling and region best explained the variation in E. coli concentration (R2= 0.75). Region is a variable masking many management variables including use of rain water, surface water, manure, inorganic fertilizer and spray irrigation. Climate variables and E. coli presence and concentration are positively related. The results indicate that climate change will have an impact on microbiological safety of LGVs. These impacts can be directly through an increasing temperature, but also indirectly through changes in irrigation water type, fertilizer type and irrigation method. Therefore, climate change and farm management should be considered more systematically in an integrated way in future studies on fresh produce safety. To prepare climate data for local food safety scenario analysis, a climate data downscaling tool was presented and demonstrated (Chapter 4). Coarse gridded data from two general circulation models, HadGEM2-ES and CCSM4, were selected and downscaled using the ‘Delta method’ with quantile-quantile correction for the Belgium meteorological station in Ukkel. Observational daily temperature and precipitation data from 1981 to 2000 were used as a reference period for this downscaling. Data were provided for four future representative concentration pathways (RCPs) for the periods 2031–2050 and 2081–2100. These RCPs are radiative forcing scenarios for which future climate conditions are projected. The climate projections for these RCPs show that both temperature and precipitation will increase towards the end of the century in Ukkel. The climate change data were subsequently used with Ratkowsky's bacterial growth model to illustrate how projected climate data can be used for projecting bacterial growth in the future. In this example, the future growth rate of Lactobacillus plantarum and the number of days that the bacteria are able to grow are both projected to increase in Ukkel. This example illustrates that this downscaling method can be applied to assess future food safety. This downscaling tool is relatively straightforward compared to other more complex downscaling tools, so the food safety researchers can easily understand and apply it to their impact studies. With the statistical model (Chapter 3) and downscaled climate data (Chapter 4), a multi-criteria scenario analysis tool was developed to explore future food safety using pre-harvest spinach in Spain as an example (Chapter 5). The future E. coli concentrations on spinach were projected to change in RCP 8.5 and RCP 2.6 by the end of the century in Spain. The E. coli concentration was projected to increase between 0.2 log10 CFU/g and 0.3 log10 CFU/g (depending on the climate scenarios and management options applied) due to higher temperature by the end of the century compared to the concentrations by the end of the last century. This comparison assumed no changes in agricultural management practices. This tool can be used to help selecting the best management practices considering climate change and other indicators. The pioneering research presented in my thesis brought new methods and tools, and another mind set to food safety research. The climate-change data downscaling tool provides detailed temporal and spatial climate data for climate scenario analysis in food safety assessment studies. The multi-criteria scenario analysis tool provides a platform to study changes in weather or climate, and management impacts on future food safety. This tool also allows for inclusion of different stakeholders’ perspectives or interests and supports their decision making processes. Moreover, the thesis presents a statistical model that can be used to study the relationship between climate and E. coli contamination. My thesis quantified the impacts of climate change on microbial safety of pre-harvested LGVs contaminated with generic E. coli for the first time. With one degree increase in minimum temperature of the sampling day, the odds of having E. coli presence on LGVs increase by a factor of 1.5. The mean E. coli concentrations are also expected to increase. Climate change should not be ignored in food safety management and research.