Quantifying current and future raw milk losses due to bovine mastitis on European dairy farms under climate change scenarios

Quantifying current and future raw milk losses due to bovine mastitis on European dairy farms under climate change scenarios Paola Guzmán-Luna a, ⁎, Rajat Nag b , Ismael Martínez c,d , Miguel Mauricio-Iglesias a , Almudena Hospido a , Enda Cummins b a CRETUS (Cross-disciplinary Research Center in Environmental Technologies), Department of Chemical Engineering, Universidade de Santiago de Compostela, 15782 Santiago de Compostela, Spain b UCD School of Biosystems and Food Engineering, University College Dublin, Belfield, Dublin 4, Ireland c CLUN (Cooperativa Lácteas UNidas), Department of R&D, Ponte Maceira, 15864 Ames, Spain d Department of Analytical Chemistry, Nutrition and Food Science, Universidade de Santiago de Compostela, 15782 Santiago de Compostela, Spain HIGHLIGHTS •Current and future raw milk losses due to bovine mastitis were quantified. •An exposure assessment and stepwise probabilistic model were implemented. •Bedding material is found to be the main source of S. aureus exposure. •Higher raw milk losses are estimated for the Mediterranean regions of Europe. •Annual milk losses ranged from 0.37% (RCP2.6) to 3.21% (RCP8.5). GRAPHICAL ABSTRACT ABSTRACTARTICLE INFO Editor: Huu Hao Ngo Bovine mastitis is an infectious disease that causes udder inflammation and is responsible for raw milk losses across European dairy farms. It is associated with reduced cow milk yield and contributes to elevated Somatic Cell Count (SCC) in raw milk. Staphylococcus aureus is one of the most prevalent mastitis pathogens that cause subclinical and clinical mastitis and can be present as a coloniser bacterium in cows. Climate change and geographical variability may influence the prevalence of this pathogen. Thus, this research aimed to predict the raw milk losses in three major dairyproducing regions across Europe (i.e. Mediterranean, Atlantic and Continental) under climate change scenarios. An exposure assessment model and a stepwise probabilistic model were developed to predict potential cow milk yield reduction, S. aureus and SCC concentrations in the bulk tank milk at dairy farms. Baseline (i.e. present) and future climate change scenarios were defined, and the resultant concentrations of SCC and S. aureus were compared to the actual European regulatory limits. Across the three regions, raw milk losses ranged from 1.06% to 2.15% in the baseline. However, they increased up to 3.21% in the climate change scenarios when no on-farm improvements were considered. Regarding geographical variation, the highest potential milk losses were reported for the Mediterranean and the lowest for the Continental region. Concerning the fulfilment of the regulatory limits, the mean of S. aureus and SCC levels in milk did not exceed them either in any region or scenario. Nevertheless, when looking at percentiles, the 10th percentile remained above the limits of S. aureus in Atlantic and Mediterranean, but not in the Continental region. The findings provide a snapshot of climate change impacts on raw milk losses due to mastitis. They will allow farmers to detect weaknesses and prepare them to develop adaptation plans to climate change. Keywords: Food losses Risk assessment Pathogen infection Predictive modelling Stepwise probabilistic model Staphylococcus aureus Science of the Total Environment 833 (2022) 155149 ⁎Corresponding author. E-mail addresses: [email protected] (P. Guzmán-Luna), [email protected] (R. Nag), [email protected] (I. Martínez), [email protected] (M. Mauricio-Iglesias), [email protected] (A. Hospido), [email protected] (E. Cummins). http://dx.doi.org/10.1016/j.scitotenv.2022.155149 Received 4 February 2022; Received in revised form 18 March 2022; Accepted 6 April 2022 Available online 11 April 2022 0048-9697/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Contents lists available at ScienceDirect Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv 1. Introduction The dairy sector is a leader in the agricultural economy of the European Union (EU). In 2020, milk farmers across the EU produced 154 million tonnes of cow's milk (Eurostat, 2021), which is expected to increase (Bórawski et al., 2020). Even though milk is produced across all the EU member states, approximately 77% (119 million tonnes per year) of the total raw milk produced in the EU is located in seven countries (Eurostat, 2021). Bovine mastitis causes bacterial intramammary infections, and it is one of the most widespread global diseases that impact European dairy farms (Nalon and Stevenson, 2019). On dairy farms, the total raw milk losses due to mastitis can occur in two ways. On the one hand, this infection leads to a significant decrease in the cows' milk yield, being more pronounced in cows with clinical mastitis (where milk is typically discarded) than in cows with subclinical infection (Ingalls, 2001); on the other, this infection alters the raw milk quality, becoming unsuitable for human consumption and further processing (Gonçalves et al., 2018). The somatic cell count (SCC) is an indicator used in the dairy industry to determine the raw milk quality (Pantoja et al., 2009). Another relevant indicator is the total bacteria count (TBC), which determines hygienic on-farm conditions during milk production (Robles et al., 2020). Many pathogens can induce bovine mastitis such as Escherichia coli, Streptococcus spp.,and Staphylococcus spp. Across European dairy farms, Staphylococcus aureus has been the most frequent isolated mastitis pathogen from raw milk samples (Mekonnen et al., 2018;More et al., 2013;Tegegne and Tesfaye, 2017). It can appear as a coloniser on cows (coM) and remain unnoticed without inflammatory symptoms. However, it can also transmit from cow to cow, turning into subclinical bovine mastitis (sM) or evolving into clinical mastitis (cM), where symptoms are visible in the latter (Wald et al., 2019; Wellnitz and Bruckmaier, 2012). Besides being a bacteria found in the cows' udders, S. aureus has also been found at dairy facilities such as on milking systems, bedding material, faeces and feed (Zadoks and Fitzpatrick, 2009). The proliferation of pathogens is expected to benefit from the changes in climatic conditions (European Food Safety Authority, 2020)and,therefore, increase the microbial load in raw milk (Misiou and Koutsoumanis, 2021). In particular, the rise of average temperatures as a consequence of climate change around the globe (IPCC, 2014) is expected to increase the prevalence of bovine mastitis among dairy herds and, consequently, contribute to higher raw milk losses (Jingar et al., 2014). Projected climate change effects are not expected to occur uniformly worldwide. Across Europe, six biogeographical regions have been defined by the European Environment Agency (2017) depending on the climate change impacts and vulnerabilities, being the Mediterranean (i.e. southern Europe) one of the most vulnerable regions. This region is experiencing and is projected to continue experiencing temperature rises and a reduction of rainfall, mainly in summer, leading to droughts (European Environment Agency, 2017, 2019;Giorgi and Lionello, 2008). Similarly, the Continental region (i.e. central and eastern Europe) will suffer from an increment of heat extremes and rainfall reduction during the summer, increasing the risk of drought. On the contrary, an increment of extreme precipitation events is projected to occur in the Atlantic regions (i.e. north-western Europe) (European Environment Agency, 2017). Quality assurance of dairy products commences on the dairy farms, and thus, regulatory limits for SCC and S. aureus in raw milk have been set in Europe (European Commission, 2003, 2004). Both SCC and S. aureus load in raw milk have been widely studied previously (Jayarao et al., 2004; Kateřina et al., 2016;Malek dos Reis et al., 2013;Mekonnen et al., 2018; Naito et al., 2013;Wang et al., 2018), but none of these studies analysed the contamination pathway of S. aureus from the dairy farm to the farm bulk tank milk (BTM), which includes the intramammary and environmental S. aureus. Also, to the best of the authors' knowledge, even though the prevalence of bovine mastitis is affected by climatic conditions, all the available studies only estimate raw milk losses from bovine mastitis under current conditions without considering the future effects of climate change, which are expected to modify climatic variables. Quantitative Microbial Risk Assessment (QMRA) is a tool that aims to develop knowledge of the spread and control of infectious diseases (Mitchel et al., 2021). QMRA is particularly useful in developing control strategies and evidence-based policy decisions through the lens of health risk. QMRA is not limited to human health risks; it can also be developed for animal and plant pathogens (Mitchel et al., 2021). The overall hypothesis of this research is “Bovine mastitis may pose a threat to European dairy farms causing milk losses, which may change under projected climate change scenarios”. Hence, under climate change scenarios, this study aimed to predict the annual raw milk losses due to bovine mastitis across the main European milk producer regions. Climate change scenarios were considered to investigate the influence of temperature increment on the annual raw milk losses at dairy farms located in the Mediterranean, Atlantic and Continental. Herds were assumed to be comprised of 100 lactating cows in an indoor dairy housing system. Moreover, given the lack of studies that includes the complete contamination pathway of S. aureus,this research includes both the environmental and intramammary contamination of this pathogen from the dairy cow to the BTM. 2. Material and methods A model framework was developed, composed of an exposure assessment and a stepwise probabilistic model, to predict the annual raw milk losses based on the SCC and S. aureus concentrations at BTM under climate change scenarios (Fig. 1). The total raw milk losses include losses due to (i) exceeding the S. aureus and SCC limits in BTM, (ii) losses due to a cow's milk yield reduction, and (iii) raw milk discarded due to cow with clinical mastitis. The model inputs, computations and simulated outputs used in the model framework are detailed in Table 1.Specific parameters affected by temperature increase due to climate change are marked with an asterisk in the same table. The framework begins by defining the baseline scenario and climate change scenarios by 2050 (step 1). Then, an exposure assessment adapted from Vissers et al. (2006) was performed to estimate the amount of S. aureus in the environment to which cows are exposed. In the exposure assessment section, the different production stages 1 involved at a dairy farm are included (steps 2 to 8). Later, a stepwise probabilistic model was necessary to estimate the annual mastitis prevalence (step 9) and predict the cows' milk yield reduction and SCC and S. aureus concentrations at BTM (step 10). Finally, to estimate the raw milk losses associated with not meeting the standards, the S. aureus and SCC concentrations in the different scenarios were compared to the regulatory limits set in the EU (step 11). A sensitivity analysis was performed to identify the most influential model's inputs and observe its influence on the predicted raw milk losses. 2.1. Definition of baseline and climate change scenarios (step 1) The baseline scenario was built on current average annual temperatures (T) representative of the regions (i.e. Atlantic, Mediterranean, and Continental) that deviated from the World Bank Group (2021).Theclimate change scenarios were constructed based on the Representative Concentration Pathways (RCP) developed by the Intergovernmental Panel on Climate Change (i.e. RCP 2.6, 4.5 and 8.5) (IPCC, 2014). Unlike other emission scenarios, the RCPs project totals radiative forcing, considering the effect of efforts (e.g. international policies and agreements) to reduce greenhouse gas (GHG) emissions and mitigate climate change. Therefore, the RCP2.6 is considered a scenario with great mitigation efforts. RCP4.5 refers to an intermediate stabilised scenario, and RCP8.5 refers to a scenario where no efforts to reduce GHG emissions are made (IPCC, 2014). Each of the mentioned RCP scenarios was split into two sub-scenarios (Table 2), assuming ‘on-farm improvements’and ‘no on-farm improvements’. ‘On-farm improvements’scenarios consider an increase in the annual average milk yield (AAMY) and a decrease in annual mastitis prevalence. 1 Production stages refer to the sources and route of transmission of S. aureus at dairy farms. Six production stages are identified throughout the exposure assessment model. P. Guzmán-Luna et al. Science of the Total Environment 833 (2022) 155149 2 The former results from the continuous efforts to increase the AAMY through feeding strategies and genetic enhancement, which are expected to keep increasing in the near future (Bórawski et al., 2020). According to Reijs et al. (2013), there is a constant increment in the milk yield per dairy cow in different countries. By applying a linear model on the data on milk yield over the years, an increment of 50% by 2050 was projected compared to the AAMY in 2019, obtaining an R 2 of 0.9. Thus, a projected AAMY of 18,800 (SD 1979) L year −1 by 2050 was calculated. The latter results from farm enhancements such as improvement in udder health and novel mastitis treatments (IDF, 2018). These improvements reduced the rate of mastitis prevalence over the past years, and it is expected to continuously improve in the near term. However, data on the potential prevalence of mastitis by 2050 and historical records to predict it were not found. Thus, this study assumes a 10% reduction of annual prevalence by 2050, compared to the present prevalence. Conversely, in the ‘no on-farm improvements’scenarios, upgrades are not considered. The AAMY remains stable in 12,500 year −1 ,and the annual mastitis prevalence increases by 10% due to a rise in average global temperatures, leading to an increase of bovine mastitis (Jingar et al., 2014). The Tchange by 2050 is expected to vary among RCPs, so, Table 2 shows the corresponding increments for each scenario. Stepwise probabilistic model Data output Data output Data input Data input Step 1: Definition of baseline and climate change scenarios •Baseline temperature (T) in European regions (World Bank Group, 2021) •Global temperature change by 2050 for RCP2.6, 4.5 and 8.5 (IPCC, 2014) •Improvements by 2050: oIncrease annual average milk yield oDecrease mastitis prevalence •Baseline scenario •Scenario 1 = No improvement + RCP2.6 •Scenario 2 = Improvement + RCP2.6 •Scenario 3 = No improvement + RCP4.5 •Scenario 4 = Improvement + RCP4.5 •Scenario 5 = No improvement + RCP8.5 • Scenario 6 = Im p rovement + RCP8.5 Exposure assessment Step 2: Definition of input parameters for the exposure assessment •Data from the literature (Table 3) •Feed •Faeces •Bedding material Step 3: Quantification of S. aureus in mixed feed ration •Fraction silage in ration (F silage ) •Mean contamination level in silage (C silage ) •Mean contamination level in other feeds (C otherfeed ) •Concentration of S. aureus in the feed ration (C ration(0) ) Step 4: Calculation of S. aureus after growth due to feed storage •C ration (0) •Time between two feed ration refreshments •Lag time and growth rate of S. aureus •Factors to estimate growth rate: Maximum, minimum and optimum pH of the ration •Factors to estimate growth rate: maximum, minimum and optimum temperature of the ration •Concentration of S. aureus in the ration after storage (Ln[C ration (t)]) Step 5: Quantification of S. aureus concentration in faeces •Fraction of the feed ration digested (F digested ) •C ration (t) •Concentration of S. aureus in the faeces (C faeces ) Step 6: Quantification of S. aureus concentration in mixed faeces with contaminated dirt soil •C faeces •S. aureus concentration in the soil including bedding material (data from literature) •Fraction of soil in dirt (Table 1) •Concentration of S. aureus in the dirt (C dirt ) Step 7: Estimation of S. aureus concentration in the environment (cross-contamination) •C dirt •Mass of dirt attached to the udder teats •Number of microbial cells per cow before treatment (N before_treatment ) Step 8: Estimation of S. aureus concentration in the environment after udder teat treatment •N before_treatment •Efficiency of the treatment equal to the percentage of microbial cells removed by the treatment •Number of microbial cells per cow after treatment (N after_treatment ) Step 9: Calculation of the annual mastitis prevalence •Mastitis classification from Wald et al. (2019) •Data provided by Yang et al. (2012) •Data provided by Jingar et al. (2014) •Annual average temperature in the three regions •Clinical (cM prev ) and subclinical (sM prev ) mastitis prevalence •Number of cows with cM, sM and coM Step 10: Performance of Binomial flag Step 11: Quantification of total milk losses in current and climate change scenarios •Output step 8 •Output step 9 •S. aureus and SCC concentration for each type of mastitis according to Wald et al. (2019) •Milk yield reduction for each type of mastitis according to Sharma et al. (2011) •S. aureus (S BTM ) and SCC (SCC BTM ) concentration in the BTM considering cow milk yield reduction •Output step 10 •Limits of S. aureus and SCC in milk set by the EU regulation 853/2004 •Milk production if all cows are healthy vs milk production after cows’ milk yield reduction (output 9) •Raw milk losses due to exceeding the regulatory limits •Raw milk losses due to a reduction of cow’s milk yield Fig. 1. Schematic of the model framework for estimating bovine milk losses. The exposure assessment model (i.e., steps 2 to 8) was adapted from Vissers et al. (2006). P. Guzmán-Luna et al. Science of the Total Environment 833 (2022) 155149 3 2.2. Input parameters definition for the exposure assessment model (step 2) This step includes collating information for the input parameters in the exposure assessment model for S. aureus concentration in the dairy facilities environment. Three sources of contamination of S. aureus were identified (bedding material, silage, and the concentrated feed), and the number of S. aureus on each of these sources was retrieved from the literature (Table 3). Silage includes grass and maize ensilage. Also, the material used for the cows' bed is assumed to be straw and sawdust since both are the materials most used by dairy farmers (Leso et al., 2020). Table 1 Description and distributions of the inputs for the stepwise probabilistic model to predict the concentration of S. aureus inthefarmbulktankmilk(S BTM ), adapted from Vissers et al. (2006). Inputs and computations affected by the climate change scenarios are pointed out with a asterisk (*). Symbols Description and references Model/distribution baseline Units Model inputs F silageb Fraction silage in the ration (Driehuis et al., 2008; based upon expert opinion, CLUN) Uniform (min 0.6, max 0.84) % C silage Mean contamination level in silage Table 3 Log 10 CFU g −1 C concentrated Mean contamination in concentrated feed Table 3 Log 10 CFU g −1 C ∝ Maximum attainable contamination level in the feed ration Vissers et al. (2006). 8 Log 10 CFU g −1 t b The time between two feed ration refreshments (based upon expert opinion, CLUN) Uniform (min 12, max 24) Hours λLag time (Zwietering et al., 1996) 1 Hours μ opt Growth rate under optimal conditions (Vissers et al., 2006) 0.12 Hours T ration * The temperature of the ration, which was assumed to be the same as the current shade temperature and varies depending on the region under study (World Bank Group, 2021) Atlantic: Uniform (min 3.3, max 18.3) Mediterranean: Uniform (min 4.2, max 21.9) Continental: Uniform (min 0.5, max 17.8) °C T min Minimum growth temperature (Medveov and Valk, 2012)7 °C T opt Optimal growth temperature (Medveov and Valk, 2012) Uniform (min 37, max 40) °C pH ration The pH of the feed ration (Borreani and Tabacco, 2010) 4.1 pH min Minimum growth pH (Medveov and Valk, 2012)4 pH max Maximum growth pH (Medveov and Valk, 2012) 9.8 pH opt Optimal growth pH (Medveov and Valk, 2012) Uniform (min 6, max 7) F digested Fraction of the feed ration digested (Lassey, 2007)75% F beddinga Fraction of bedding material in the dirt (Vissers et al., 2006) Uniform (min 0, max 20) % C bedding Bedding material contamination level Table 3 Log 10 CFU g −1 M dirt Mass of dirt attached to the udder teats (Vissers et al., 2006)1 G PT efficiencya % of spores removed without pre-treatment (Vissers et al., 2006) Pert (30, 75, 90) % AAMY b * Data on annual average milk yield from a healthy cow (records provided by CLUN) Lognormal (mean 12,500, SD 1318.99) L year −1 SCC Wald Data on Somatic Cell Count from Wald et al. (2019) required in the binomial flag Table 4 Cells mL −1 S Wald Data on S. aureus concentration from Wald et al. (2019) required in the binomial flag Table 4 CFU mL −1 MP aft_red Annual raw milk production by the herd considering milk yield reduction. Data on milk yield reduction per type of mastitis was retrieved from Sharma et al. (2011) and required in the binomial flag Table 5 L year −1 Model outputs C ration(0) The concentration of S. aureus in the ration after mixing feed components (adapted from Vissers et al., 2006) Eq. (1) CFU g −1 C ration (t) The concentration of S. aureus in the ration considering the effect of feed storage (adapted from Vissers et al., 2006) Eq. (2) CFU g −1 C faeces The concentration of S. aureus in the faeces (adapted from Vissers et al., 2006) Eq. (8) CFU g −1 C dirt The concentration of S. aureus in the dirt because of a mix of faeces and bedding material (adapted from Vissers et al., 2006) Eq. (9) CFU g −1 N before_treatment Most probable number of S. aureus in cow's udders due to cross-contamination from dirt before treatment (adapted from Vissers et al., 2006) Eq. (10) CFU N after_treatment Most probable number of S. aureus in cow's udders after treatment (adapted from Vissers et al., 2006) Eq. (11) CFU sM prev * Prevalence of subclinical mastitis based on the temperature from each region Eq. (12) % cM prev * Prevalence of clinical mastitis based on the temperature from each region Eq. (13) % SCC BTM Somatic Cell Count concentration in the bulk tank milk Eq. (14) Cells mL −1 M yield_after Actual annual milk yield after considering cow milk yield reduction. Data from Sharma et al. (2011) was required Eq. (15) L year −1 S BTM S. aureus concentration in the bulk tank milk, considering the load in the environment and the raw milk due to the intramammary infection Eq. (16) CFU mL −1 a In the sensitivity analysis, F bedding and PT efficiency were modified by the lowest and highest value presented in this table. b Data related to on-dairy farms was retrieved from expert opinion (i.e. dairy farmers and veterinaries) (CLUN, 2021). Table 2 Description of the six climate change scenarios applied in this research. Temperature change by 2050 (°C) RCP2.6 (+0.3 to 1.7) a RCP4.5 (+1.1 to 2.6) a RCP8.5 (+2.6 to 4.8) a No Improvement Scenario 1 Scenario 3 Scenario 5 Improvement Scenario 2 Scenario 4 Scenario 6 a Data retrieved from IPCC (2014). Table 3 Data collation on the number of S. aureus for each input parameter of the model. Input Symbol Raw data (CFU g −1 ) Reference Bedding material C bedding 3.1 × 10 8 (Rendos et al., 1975) Bedding material C bedding 4.9 × 10 7 (Rendos et al., 1975) Bedding material C bedding 2.2 × 10 9 (Rendos et al., 1975) Bedding material C bedding 10–8×10 6 (Bradley et al., 2018) Bedding material C bedding 1.4 (± SD 2.5) 10 8 (Black et al., 2014) Bedding material C bedding 6.9 log 10 (Hogan et al., 1990) Bedding material C bedding 5.5 log 10 (Hogan et al., 1990) Bedding material C bedding 7.7 log 10 (Hogan et al., 1990) Silage C silage <10 (CESFAC, 2007;CLUN, 2021) Concentrated feed C concentrated <10 (CESFAC, 2007;CLUN, 2021) Note: For bedding material, a distribution is fitted as ‘Expon (345172279)’CFU g −1 . P. Guzmán-Luna et al. Science of the Total Environment 833 (2022) 155149 4 2.3. Quantification of S. aureus when feed components are mixed (step 3) Step 2 addresses the first production stage at dairy farms, consisting of the potential contamination of S. aureus in the cows' feed ration. The feed ration is a mix between concentrated feed and silage. However, the process of mixing potentially results in a contaminated feed ration with pathogens (Hope et al., 2009).Eq. (1) was applied to quantify the initial S. aureus concentration in the feed ration (C ration(0) ) after mixing the silage and concentrated feed, in which the obtained unit is CFU g −1 . Cration 0ðÞ¼Fsilage Csilage þ1−Fsilage  Cconcentrated (1) where F silage is translated as the fraction of silage in the cows' diet ration, which varies depending on dairy farmers' practices and dairy production systems (i.e. confined indoor and grazing outdoor). The value for the F silage for an indoor system was obtained from the farmers' opinion (Table 1). Regarding C silage and C concentrated , they refer to the most probable number of S. aureus in silage and concentrated feed, respectively. Respective data was deviated from the allowable limits of S. aureus in ruminant feed production (Table 3) according to the Guide of Feed Sanitization Standards (CESFAC, 2007). This guide is commonly used among farmers from the cooperative to reference quality controls at dairy farms. 2.4. Calculate the effect of pathogen growth due to feed storage (step 4) During feed storage, the growth of S. aureus can occur, and it can be influenced by environmental conditions such as temperature, and consequently, affected by the climate change scenarios. Thus, to calculate the pathogen concentration in the ration after storage (ln[C ration (t)]), Eq. (2) was retrieved from Vissers et al., 2006. The obtained unit is ln CFU g −1 . ln Cration tðÞ½¼ln Cration 0ðÞ½þμAntðÞ−ln 1þeμ∙AntðÞt eln C∝=Cration 0ðÞðÞ  (2) where C ration(0) is the initial concentration previously calculated in Eq. (1);μ is the microbial growth in the feed ration, and it is calculated based on the Baranyi and Roberts model (Baranyi and Roberts, 1994); A n is equal to the time between two feed ration refreshments (t), which are obtained in hours; eis a mathematical constant equal to 2718; C ∝ refers to the maximum level of contamination achievable in the feed ration, which according to Vissers et al. (2006), it is assumed equal to 8 log 10 pathogen g −1 . Also, Eq. (3) was needed to solve μin Eq. (2). It followed a gamma concept model since temperature (Eq. (4))andpH(Eq.(5)) effects are considered separately. S. aureus prefers an optimum temperature (T opt )thatranges from 37 °C to 40 °C and a minimum temperature (T min )of7°C.TheT ration was considered equal to the environment temperature. Thus, the average temperature was used from each biogeographical region for the baseline and different climate change scenarios (Step 1). Concerning the pH (γ(pH)), this mastitis pathogen prefers an optimum pH (pH opt ) that ranges from 6 to 7. It tolerates a minimum and maximum pH of 4 and 9.8, respectively (Medveov and Valk, 2012). The pH from the ration (pH ration ) was 4.1 (Borreani and Tabacco, 2010). μ¼γTðÞγpHðÞμopt (3) γTðÞ¼ Tration−Tmin Topt −Tmin  2 (4) γpHðÞ¼pHration−pHmin ðÞ pHmax−pHration ðÞ pHopt−pHmin  pHmax−pHopt  "# (5) Lastly, Eq. (6) (Vissers et al. (2006))solvesA n (t)inEq.(2), in which the obtained unit is hours. λrefers to the lag time, which is equal to 1, and t(in h) refers to the time between one feed ration and the next refreshment, which depends on dairy farmer's practices, so data from CLUN was used (Table 1). AntðÞ¼tþ1 μ ln e−μ∙tþq0 1þq0  (6) q0¼1 eλ∙μ−1(7) 2.5. Effect of animal digestion in the S. aureus concentration of faeces (step 5) The next production stage is associated with the S. aureus concentration in cows' faeces. Microorganisms can survive the ruminant digestion process and return to the environment throughout the animals' excretion. Eq. (8) quantifies the S. aureus concentration in the faeces (C faeces ) in CFU g −1 . To do so, the fraction of the feed ration that is digested (F digested ) is retrieved from Lassey (2007), which is equal to 75%. Cfaeces tðÞ ¼1 1−Fdigested  Cration tðÞ (8) 2.6. Effect of mixing the faeces with soil resulting in contaminated dirt (step 6) Another production stage is linked to a mix between faeces and bedding material, which are commonly mixed due to the natural movement and displacement of the cows resulting in contaminated dirt. Following Vissers et al. (2006),Eq.(9) estimates the S. aureus concentration in the contaminated dirt (C dirt )inCFUg −1 . Cdirt ¼Fbedding Cbedding þ1−Fbedding  Cfaeces (9) where F bedding refers to the fraction of bedding material in the dirt, as shown in Table 1. The pathogen concentration in bedding material (C bedding )was calculated based on the values extracted from the literature, as shown in Table 3,andC faeces was obtained previously from Eq. (8). 2.7. Evaluation of the cross-contamination from dirt to udder teats (step 7) This production stage refers to the S. aureus contamination resulting in cross-contamination from the dirt to the dairy cows' teats. Thus, Eq. (10) quantifies the number of microbial cells per cow (stated in CFU) before a common treatment that dairy farmers carry out to rinse off as much as possible the dirt on teats: Nbefore_treatment ¼Mdirt Cdirt (10) where M dirt is the mass of dirt stuck to the cows' teats (in g) and obtained from Vissers et al. (2006),asshowninTable 1, while C dirt was previously calculated in Eq. (9). 2.8. Effect of the treatment process of udder teats (step 8) In this step, the final concentration of S. aureus to which cows are exposed environmentally was calculated. During the production stage, teats are rinsed off, and microorganisms are removed on the cows' teats. Eq. (11) estimate the number of microbial cells per cow (N after_treatment ) after the treatment process, obtaining a value in CFU. Nafter_treatment ¼PTefficiency Nbefore (11) where PT efficiency refers to the treatment efficiency based on the percentage of microorganisms removed. The PT efficiency was assumed 75%, as reported by Vissers et al. (2006) (Table 1). P. Guzmán-Luna et al. Science of the Total Environment 833 (2022) 155149 5 2.9. Annual mastitis prevalence (step 9) An estimation of the annual mastitis prevalence was required to calculate the concentration of intramammary S. aureus. Since the study looks at the effect of climate change on raw milk losses due to mastitis, the prevalence of each mastitis was calculated in function of temperature. To do so, the sM prevalence (sM prev ) was estimated with a linear model, using the reported data on temperature and subclinical prevalence by Yang et al. (2012) and obtaining the result in percentage. The following equation (Eq. (12)) was obtained with an R 2 equal to 0.6: sMprev ¼7:734 TðÞ 0:4262 (12) Similarly, the cM prevalence (cM prev ) was calculated with a linear model, implementing data on temperature and clinical prevalence reported by Jingar et al. (2014) and obtaining the result in percentage. The following equation (Eq. (13)) was obtained with an R 2 equal to 0.9: cMprev ¼0:0026 TðÞ−0:0263 (13) Later, T(Eqs. (12) and (13)) was substituted by the average annual temperature representative of each region (Atlantic, Mediterranean, and Continental) in the baseline and climate change scenarios. Lastly, due to a lack of data on the prevalence of mastitis provoked by coloniser S. aureus, the prevalence was estimated with two conversion factors. These factors were obtained from the subclinical-to-coloniser (70:12) and clinical-to-coloniser (18:12) ratios provided by Wald et al. (2019). Then, a uniform distribution was applied by using both conversion factors. 2.10. Binomial flag (step 10) A binomial flagging method (Vose, 2000) was used to estimate the S. aureus and SCC concentration in the BTM. This approach is built on the binomial distribution, and it can be useful to resolve problems and estimate the probability of success of a trial, such as done by Coffey et al. (2009). They implemented the binomial flagging approach to estimate the presence of Aflatoxin B1 in maize, finding 51 samples positive out of 139. Translating it to the present research, the mastitis prevalence previously calculated together with the binomial flag allowed the calculation of the number of infected cows out of a herd of 100. A second binomial flag was applied only for the infected cows to determine the number of cows with cM (labelled as category 3), sM (labelled as category 2) and/or coM (labelled as category 1) (Table 4), while healthy cows were labelled as category 0. Once the number of cows within each category was estimated, data on the concentration of SCC and S. aureus per cow was derived using data from Wald et al. (2019) (Table 4). It allowed the calculation of the SCC (SCC BTM )andS. aureus (S BTM ) concentrations per mL in the BTM. Also, the cows' milk yield reduction was estimated since these concentrations depend on how much each cow contributes to the BTM (Table 5). The SCC BTM was estimated by Eq. (14) and stated in cells mL −1 .Cows with clinical mastitis were excluded since they are assumed to be separated from the rest of the herd, and their milk does not enter the BTM. SCCBTM ¼CowcoM SCCWald ðÞMyield after  þCowsM SCCWald ðÞMyield after  1000 MPaft red 1000 ð14Þ where the total number of cows with coloniser (Cow coM ) and subclinical mastitis (Cow sM ) in the herd were multiplied by the respective SCC concentration stated by Wald et al. (2019) (SCC wald )inTable 4.Later,thiscontribution was multiplied by the actual annual milk yield after considering cow milk yield reduction (M yield_after ) (L year −1 ) using Eq. (15): Myield_after ¼AAMY−MYRmastitis (15) Where AAMY refers to the annual average milk yield (L year −1 ) from healthy cows considering data provided by CLUN (mean 12,545, SD 1319). MYR mastitis refers to the milk yield reduction according to the type of bovine mastitis (sM or coM), calculated from Sharma et al. (2011) (Table 5), who provided the respective milk yield reductions based on SCC concentration in the raw milk per cow. Finally, the total concentration of SCC (cells year −1 ) from sM and coM was divided by the annual raw milk production in L year −1 (MP aft_red ), in which a reduction of the cows' milk yield due to bovine mastitis in the herd was also considered. In order to cover the whole contamination pathway, Eq. (16) estimates the S BTM (stated in CFU mL −1 ) by adding up the S. aureus load to which cows are exposed in the facilities environment (S env ) and the S. aureus concentration in the milk (S sa ) coming from the intramammary infection. Cows with clinical signs were also excluded here. SBTM ¼Senv þSSa (16) The first summand is given by Eq. (17) andstatedinCFUmL −1 ,where Cow coM and Cow sM refer to the number of cows with either coloniser or subclinical mastitis in the herd, N after_treatment to the number of microbial cells a cow is exposed to after treatment and is retrieved from step 7 (Eq. (11) in CFU) and MP aft_red to the total milk produced in l day −1 after milk yield reduction, as mentioned in Table 5. Senv ¼CowcoM þCowsM ðÞNafter_treatment MPaft_red (17) The second summand is given by Eq. (18) andisalsostatedinCFU mL −1 .S Sa the calculation is similar to SCC BTM , but considering data on S. aureus from Table 4, where the total number of cows eitherwith coloniser (Cow coM ) or subclinical mastitis (Cow sM ) were multiplied by the respective S. aureus concentration estimated by Wald et al. (2019) (CFU mL −1 ). The contribution of S. aureus per mL of raw milk was multiplied by the actual Table 4 S. aureus and SCC concentration in the milk corresponding to each type of bovine mastitis. Adapted from Wald et al. (2019). Data on the column of SCC concentration (SCC Wald )isusedinEq.(14) and data on the column of S. aureus concentration (S Wald )isusedinEq.(18). Flag label SCC concentration (cells mL −1 ) S. aureus concentration (CFU mL −1 ) Median Mean SD Distribution Min Max Distribution 3cM >10 6 2.18 × 10 6 1.58 × 10 6 Lognormal(2.18 × 10 6 , 1.58 × 10 6 )10 5 10 6 Uniform(10 5 ,10 6 ) 2 sM 200,000–500,000 1.00 × 10 6 1.21 × 10 6 Lognormal(1.00 × 10 6 , 1.21 × 10 6 )10 3 10 4 Uniform(10 3 ,10 4 ) 1 coM ≤100,000 4.50 × 10 4 3.30 × 10 4 Lognormal(4.50 × 10 4 , 3.30 × 10 4 )10 3 10 4 Uniform(10 3 ,10 4 ) P. Guzmán-Luna et al. Science of the Total Environment 833 (2022) 155149 6 annual milk yield once cow milk yield reduction was considered (M yield_after ). SSa ¼CowcoM SWald ðÞMyield_after  þCowsM SWald ðÞMyield_after  1000 MPaft_red 1000 (18) 2.11. Total raw milk losses under current and climate change scenarios (step 11) To obtain the raw milk losses related to not meeting the quality standards, the S BTM and the SCC BTM were compared to the regulatory limits for raw milk intended for processing in the EU: i.e. 400,000 cells mL −1 of SCC (European Commission, 2004) and 2000 CFU mL −1 of S. aureus (European Commission, 2003).The raw milk losses due to milk yield decrease were calculated from the difference between the total milk production assuming all cows are healthy and the total milk production with reduced milk yield due to bovine mastitis, divided by the total milk production to obtain a percentual value. The milk yield reduction from clinical cows was excluded since their milk production is directly considered a loss. 3. Results and discussion 3.1. Current raw milk losses due to bovine mastitis As a result of step 9, the highest annual prevalence of S. aureus mastitis was estimated for the Mediterranean. For this region, the proportions of each mastitis category were 22.7% for sM, 0.6% for cM, and 2.5% for coM (detail values per biogeographical region and mastitis category in Supplementary Fig. S1). The prevalence of mastitis varies across Europe and precise percentages on the bovine mastitis prevalence are not available since national recording systems on cows' welfare and mastitis rates are unavailable in most European countries. However, according to the literature, the annual mastitis prevalence in Europe ranges from 8 to 48% per herd (European Food Safety Authority, 2009;IDF, 2018), matching the annual prevalence calculated in the present research. Similarly, step 10 showed the highest annual concentration of S. aureus and SCC throughout the whole contamination pathway for the Mediterranean region, 1.5 × 10 3 CFU mL −1 and 2.0 × 10 5 cells mL −1 , respectively (Fig. S2 and Fig. S3). The results of the annual cumulated raw milk losses are shown in detail in Fig. 2. In a baseline scenario and across the three regions, the average annual raw milk loss due to bovine mastitis caused by S. aureus was 1.44%. In order to validate the calculated raw milk losses, the results of this research were compared to reported values. Houben et al. (1993) estimated the cumulated milk yield losses due to bovine mastitis, ranging from 0.5 to 2.0% when the cows were during their first lactation. The prevalence of mastitis used in that study was 13.7% (SD 10.3%). Another study found a similar percentage (i.e. 0.5%) when the prevalence of mastitis ranged from 18.1 to 27.4% Myllys and Rautala (1995). The percentages of mastitis prevalence calculated in the present study align with the values used in the present research, which vary from 21.7 to 30.5%. In a more recent study, Heikkilä et al. (2018) quantified the raw milk losses due to bovine mastitis caused by S. aureus. They estimated a value of approximately 4.3% across Finnish dairy farms. The slight difference between the outcome found in that research and this research is attributed to the different annual mastitis prevalence selected by the authors, ranging from 72% for subclinical mastitis to 28% for clinical mastitis. They assumed higher values than the used in the present research. In addition, considering that S. aureus is not the only pathogen that induces bovine mastitis, the obtained results are compared to reported values from different microorganisms. In the same study, Heikkilä et al. (2018) reported an annual raw milk loss of roughly 5% from Streptococcus spp. and up to 10.6% from Escherichia coli. Also, Feliciano et al. (2021) estimated the raw milk losses from E. coli with a probabilistic exposure assessment model, reporting raw milk losses of 10% in dairy farms located in hot weather. Even though E. coli presents the largest percentage, it is important to mention that this microorganism Table 5 Milk yield reduction per cow depending on its SCC concentration in the raw milk per cow. Adapted from Sharma et al. (2011). SCC (cells mL −1 ) Flag label of the type of mastitis Milk yield reduction Minimum Mid-point Maximum (l 305-days −1 ) a 0 1250 17,000 0 0 18,000 25,000 34,000 0 0 25,000 50,000 70,000 1 0 566,000 800,000 1,130,000 2 726 1,131,000 1,600,000 2,263,000 3 907 Note: a) Lactation period in a year = 305 days. Fig. 2. Predicted percentage of raw milk losses from the cumulated annual milk yield reduction due to bovine mastitis. Scenarios 1, 3 and 5 refers to no improvement scenarios, while scenario 2, 4, 6 refers to improvement scenarios, respectively RCP2.6, RCP4.5 and RCP8.5. P. Guzmán-Luna et al. Science of the Total Environment 833 (2022) 155149 7 is not considered a major issue in dairy farms since its visual symptoms make its detection unchallenging. Finally, the obtained results are compared with other contagious pathogens similar to S. aureus; so, Gonçalves et al. (2018) reported 3% of raw milk losses from bovine mastitis caused by Streptococcus agalactiae in warm weather conditions, which is similar to the obtained here for the Mediterranean region in a baseline scenario (i.e. 2.15%). 3.2. Future raw milk losses due to bovine mastitis under climate change scenarios Mastitis prevalence was one of the computations influenced by the modification of temperatures according to each RCP (i.e. 2.6, 4.5 and 8.5). As a result of step 9, the highest annual prevalence of mastitis was also observed for the Mediterranean in all the climate change scenarios since temperatures in this region are higher than in the Atlantic and Continental regions. As a result, the proportions for each mastitis category increased for this region and different scenarios. In a pessimistic climate change scenario without farm improvements, the proportions increased up to 25.4% for sM, 1.6% for cM and 3.5% for coM due to a rise of temperature by 2050 (detail values per biogeographical region and mastitis category in Supplementary Fig. S1). On the contrary, they dropped to 15.4%, 1.6% and 2.5%, respectively, as an effect of the on-farm improvements. Similarly, the highest annual S. aureus and SCC concentrations were also presented in the Mediterranean region, increasing up to 1.8 × 10 3 CFU mL −1 and 2.3 × 10 5 cells mL −1 , respectively, in scenario 5. On-farm improvements have a significant positive effect on the S. aureus and SCC concentration, dropping to 1.0 × 10 3 CFU mL −1 and 1.4 × 10 5 cells mL −1 , respectively, when scenario 6 is considered (Figs. S2 and S3). As a consequence, in the same scenario, annual raw milk losses reached only 2.45% in dairy farms located in the Mediterranean. However, when no on-farm improvements are considered, raw milk losses increased by 3.21% (Fig. 2). Due to the lack of similar studies, the comparison of the predicted raw milk losses for future scenarios obtained in this section was not possible. 3.3. Sensitivity analysis A sensitivity analysis was conducted to investigate which input parameters had the largest influence on the predictive model, which quantified the S. aureus from the environment and cross-contamination. According to Fig. 3,C bedding resulted as the most influential parameter on the concentration of S. aureus in the BTM, showing a positive correlation coefficient of 0.96. The value used for the S. aureus concentration in the bedding material contributes significantly to the pathogen concentration due to environment and cross-contamination. It explains that the concentration of S. aureus from the environment obtained in all the scenarios remains similar. Another input parameter of importance is F bedding , reaching a positive correlation coefficient of 0.20. Besides, temperature and pH were also part of the sensitivity analysis; however, both showed an insignificant influence, 0.01 and 0, respectively. In other words, the bedding material is an important source of bacteria load, and the material used at dairy facilities influences the S. aureus load on cows' udder and teats. Later, a negative correlation coefficient of 0.13 was found for PT efficiency , which means this parameter can reduce the S. aureus contamination on cows' udders. For the sensitivity analysis, the values of these three input parameters were changed towards a worst and optimal scenario by selecting the lowest and the highest values. In an optimal scenario, the value of S. aureus concentration after treatment was reduced to 7.9 × 10 −2 CFU lactating cow −1 year −1 , compared to the value obtained in the baseline scenario 3.02 × 10 7 CFU lactating cow −1 year −1 . Conversely, in the worst scenario, the S. aureus concentration value increased to 3.08 × 10 8 CFU lactating cow −1 year −1 compared to the baseline scenario. 4. Recommendation This research aimed to predict the future influence of climate change on milk losses due to bovine mastitis caused by S. aureus. The results make it possible to establish the magnitude and direction of the expected consequences on milk production under the future effects of climate change. If unchecked, a potential increase in the prevalence of mastitis is anticipated due to global warming, resulting in milk loss increases, which are anticipated to be more pronounced in the Mediterranean and Atlantic regions of Europe. The predictive model and the results of this research will serve as a framework for farmers to design corrective measures to reduce milk losses due to mastitis. The model may also help farmers and policymakers make the right decision for future adaptation plans to counter climate change scenarios. Global warming is expected in the near future by 2050 (IPCC, 2021); however, dairy farmers can modify on-farm parameters to cope with this potential increase in the prevalence of mastitis. This study demonstrates the positive effect of farm improvements in reducing bovine mastitis. According to the (IDF, 2018), one way to reduce mastitis will be to develop new mastitis treatments. However, medicaments to prevent and treat mastitis can enter in the environment (Guo et al., 2021) and potentially lead to an environmental impact (Guzmán-Luna et al., 2021). Other alternatives to reduce the prevalence of mastitis on farms in the future also need to be considered. Based on the sensitivity analysis of this study, bedding material significantly influences the concentration of S. aureus in raw milk. The reason is that bedding material carries mastitis pathogens to which the udders are exposed, causing bovine mastitis. In this context, the use of low moisture and clean bedding materials could become more important and be an effective strategy to help control potential contamination on cows under climate change conditions. In addition, periodic and more frequent cleaning of animal waste and hygiene practices on the farm are recommended. Bovine mastitis will mean lost profitability on farms due to the loss of raw milk due to the inefficient use of resources used for its production (Halasa et al., 2007). Mastitis in dairy farms has been shown to have a significant environmental impact. For instance, the prevalence of this disease leads to Fig. 3. Effect of input parameters in the exposure assessment model used to calculate the environmental S. aureus at dairy facilities. (Spearman Rank correlation coefficient. P. Guzmán-Luna et al. Science of the Total Environment 833 (2022) 155149 8 increased greenhouse gas emissions per unit of product (Hospido and Sonesson, 2005;Mostert et al., 2019;Vida and Tedesco, 2017). In addition, the dairy sector looks at adapting to the future effects of climate change, but another challenge is to reduce its emissions. This sector is part of climate action and is committed to reducing GHG emissions from the European Union by 2050 (European Dairy Association, 2019). Therefore, in addition to designing adaptation plans to climate change that guarantee animal welfare and food quality, the environmental consequences must also be considered. 5. Conclusions The present research aimed to predict the annual raw milk losses due to bovine mastitis considering several climatechangescenariosusingariskassessment approach. Therefore, a stepwise probabilistic model was developed and proposed. To the best of our knowledge, this is the firsttimethatthecomplete contamination pathway of S. aureus is covered by including all the onfarm production stages and the different sources of S. aureus exposure. Also, this is the first probabilistic model that can explain the influence of each input parameter of the process on the overall outcome. One of the challenges in developing the present model was the lack of studies that partially cover the contamination pathway of S. aureus from the farm to the BTM. Instead, they only tracked the S. aureus concentration of different on-farm spots and unit operations without covering all of them in the same study. The predicted annual raw milk losses varied across regions and scenarios, being the Mediterranean the region that experiences the largest percentage lossinthebaselineandclimatechangescenarios.Forthebaselinescenario, the annual losses ranged from 1.06% to 2.15% across all regions. For the future climate change scenarios, the annual raw milk losses varied from 1.17% to 3.21% across all regions when no on-farm improvements were assumed, whereas the losses dropped from 0.37% to 2.45% when on-farm improvements were considered. Bedding material contributes significantly to the concentration of S. aureus on-farm, and it is an important parameter to focus on for mastitis control programmes. The main cause of raw milk losses was not attributed to losses associated with exceeding quality standards but a reduction in the subclinical and coloniser cows' milk yield. The largest milk yield reduction was found in cows with clinical mastitis. Cows with coloniser and subclinical mastitis present lower milk yield reduction. However, unlike cows clinically infected, these cows are commonly milked, and their milk may enter the BTM due to the lack of infection physical symptoms. Thus, cows with subclinical and coloniser mastitis contribute to the two types of raw milk losses found on the farm. The outcome of the present study will allow farmers to assess and predict potential raw milk losses under climate change scenarios and help identify adaptation plans to reduce the impact of climatechangeonmilkyieldloss. CRediT authorship contribution statement Paola Guzmán-Luna: Conceptualization, Methodology, Data curation, Data analysis, Visualization, Investigation, Writing-original draft. Rajat Nag: Conceptualization, Supervision, Writing-review &editing, Visualization. Ismael Martínez: Supervision and data provision. Miguel Mauricio-Iglesias: Conceptualization, Writing-review &editing. Almudena Hospido: Conceptualization, Writing-review &editing. Enda Cummins: Conceptualization, Supervision, Writing-review &editing. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We gratefully acknowledge the help of the Research and Development Department of CLUN for his contribution of data representative of the selected regions. This project is part of the PROTECT ITN (http://www. protect-itn.eu/), funded under the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 813329. P. Guzmán-Luna, M. Mauricio-Iglesias and A. Hospido belong to a Galician Competitive Research Group (GRC), a programme co-funded by the FEDER (EU). Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2022.155149. References Baranyi, J., Roberts, T., 1994. A dynamic approach to predicting bacterial growth in food. Int. J. Food Microbiol. 277–294. Black, R.A., Taraba, J.L., Day, G.B., Damasceno, F.A., Newman, M.C., Akers, K.A., Wood, C.L., McQuerry, K.J., Bewley, J.M., 2014. The relationship between compost bedded pack performance, management, and bacterial counts. J. Dairy Sci. 97, 2669–2679. https://doi. org/10.3168/jds.2013-6779. Bórawski, P., Pawlewicz, A., Parzonko, A., Harper, J.K., Holden, L., 2020. Factors shaping cow's milk production in the EU. Sustainability 12, 1–15. https://doi.org/10.3390/ SU12010420. Borreani, G., Tabacco, E., 2010. The relationship of silage temperature with the microbiological status of the face of corn silage bunkers. J. Dairy Sci. 93, 2620–2629. https://doi.org/ 10.3168/jds.2009-2919. Bradley, A.J., Leach, K.A., Green, M.J., Gibbons, J., Ohnstad, I.C., Black, D.H., Payne, B., Prout, V.E., Breen, J.E., 2018. The impact of dairy cows' bedding material and its microbial content on the quality and safety of milk –a cross sectional study of UK farms. Int. J. Food Microbiol. 269, 36–45. https://doi.org/10.1016/j.ijfoodmicro.2017.12.022. CESFAC, 2007. Guía para el desarrollo de normas de Higienización de Piensos. CLUN, 2021. CLUN. Cooperativas Lácteas Unidas [WWW Document]. URL https://clun.es/. Coffey, R., Cummins, E., Ward, S., 2009. Exposure assessment of mycotoxins in dairy milk. Food Control 20, 239–249. https://doi.org/10.1016/j.foodcont.2008.05.011. Driehuis, F., Spanjer, M.C., Scholten, J.M., Te Giffel, M.C., 2008. Occurrence of mycotoxins in feedstuffs of dairy cows and estimation of total dietary intakes. J. Dairy Sci. 91, 4261–4271. https://doi.org/10.3168/jds.2008-1093. European Commission, 2003. Opinion of the Scientific Committee on veterinary measures relating to public health on Staphylococcal enterotoxins in milk products, particularly cheeses. European Commission, 2004. Regulation (EC) No 853/2004 of the European Parliament and of the Council of 29 April 2004 laying down specific hygiene rules for food of animal origin. European Dairy Association, 2019. The Dairy Sector &the Green Deal. European Environment Agency, 2017. Climate Change, Impacts And Vulnerability in Europe 2016. European Environment Agency. European Environment Agency, 2019. Climate change adaptation in the agriculture sector in Europe, EEA Report. European Food Safety Authority, 2009. Scientific report on the effects of farming systems on dairy cow welfare and disease. Report of the Panel on Animal Health and Welfare. EFSA J. 1143, 152–284. European Food Safety Authority, 2020. Climate change as a driver of emerging risks for food and feed safety, plant, animal health and nutritional quality. EFSA Supporting Publ. https://doi.org/10.2903/sp.efsa.2020.en-1881. Eurostat, 2021. Milk and milk product statistics. [WWW Document]. URL https://ec.europa. eu/eurostat/statistics-explained/index.php?title=Milk_and_milk_product_statistics#M ilk_production. Feliciano, R., Boué, G., Mohssin, F., Hussaini, M.M., Membré, J.M., 2021. Probabilistic modelling of Escherichia coli concentration in raw milk under hot weather conditions. Food Res. Int. 149. https://doi.org/10.1016/j.foodres.2021.110679. Giorgi, F., Lionello, P., 2008. Climate change projections for the Mediterranean region. Glob. Planet. Chang. 63, 90–104. https://doi.org/10.1016/j.gloplacha.2007.09.005. Gonçalves, J.L., Kamphuis, C., Martins, C.M.M.R., Barreiro, J.R., Tomazi, T., Gameiro, A.H., Hogeveen, H., dos Santos, M.V., 2018. Bovine subclinical mastitis reduces milk yield and economic return. Livest. Sci. 210, 25–32. https://doi.org/10.1016/j.livsci.2018.01. 016. Guo, X., Akram, S., Stedtfeld, R., Johnson, M., Chabrelie, A., Yin, D., Mitchell, J., 2021. Distribution of antimicrobial resistance across the overall environment of dairy farms –acase study. Sci. Total Environ. 788, 147489. https://doi.org/10.1016/j.scitotenv.2021. 147489. Guzmán-Luna, P., Mauricio-Iglesias, M., Flysjö, A., Hospido, A., 2021. Analysing the interaction between the dairy sector and climate change from a life cycle perspective: a review. Trends Food Sci. Technol. https://doi.org/10.1016/j.tifs.2021.09.001 (In press). Halasa, T., Huijps, K., Østerås, O., Hogeveen, H., 2007. Economic effects of bovine mastitis and mastitis management: a review. Vet. Q. 29, 18–31. https://doi.org/10.1080/ 01652176.2007.9695224. Heikkilä, A.M., Liski, E., Pyörälä, S., Taponen, S., 2018. Pathogen-specific production losses in bovine mastitis. J. Dairy Sci. 101, 9493–9504. https://doi.org/10.3168/jds.2018-14824. Hogan, J.S., Smith, K.L., Todhunter, D.A., Schoenberger, P.S., 1990. Bacterial counts associated with recycled newspaper bedding. J. Dairy Sci. 73, 1756–1761. https://doi.org/ 10.3168/jds.S0022-0302(90)78853-4. P. Guzmán-Luna et al. Science of the Total Environment 833 (2022) 155149 9