Combining innovation in vineyard management and<br/>genetic diversity for a sustainable European viticulture

Project Information

INNOVINE

Grant agreement ID: 311775

Project closed

Start date 1 January 2013

End date 31 December 2016

Funded under

• Specific Programme "Cooperation": Food, Agriculture and Biotechnology

Total cost € 8 489 665,00

EU contribution € 5 999 990,00

Coordinated by

INSTITUT NATIONAL DE RECHERCHE POUR L'AGRICULTURE, L'ALIMENTATION ET L'ENVIRONNEMENT
France

Final Report Summary - INNOVINE (Combining innovation in vineyard management andgenetic diversity for a sustainable European viticulture)

Executive Summary:
INNOVINE project aimed at developing knowledge, tools and genetic resources necessary to better adapt viticulture to climate change and to drastically reduce the use of pesticides in the vineyards. The aim of the project was to disseminate its findings to three different categories of end users: academics, technical advisors/service providers and growers. Most of these activities are described in details in the deliverables D6.8 D6.9 and D6.10 and are summarized below, highlighting the potential impacts and within three main topics: (i) adaptation to climate change, (ii) reduction of chemical applications in the vineyard and (iii) assistance to growers for adapting their viticulture systems towards higher sustainability.
INNOVINE globally led to a better understanding of the impact of vineyard practices and of various abiotic stresses on grapevine physiology and berry composition in the context of climate change. The development of two grapevine models will allow to simulate and predict these impacts in various climatic scenarios. Further developments are needed now to have these models take into account differential impacts according to different genotypes. Methods for screening germplasm for plasticity or for identifying key molecular pathways of adaptation to stress have been proposed. Several non-destructive phenotyping tools based on fluorescence, reflectance, thermal imaging, hyperspectral imaging have been experimented and validated in several work packages of INNOVINE with the goal of monitoring the physiological status of the canopy, as well as the berry content or the onset of downy mildew attacks. The data collected can be used as inputs in decision making tools or as screening tools to investigate clonal or varietal diversity. Different adaptive strategies were tested in diversified environmental conditions and feed the INNOVINE’s tool box for adapting the vineyards to climate change: sets of clones better adapted to various environments, a screening method for plasticity of the varieties with regard to environmental stresses, a better understanding of the effect of various practices aiming to mitigate environmental stresses on berry quality, various monitoring tools and associated DSS.
Researchers from different disciplines have developed foreground that will allow to carry out strategies for a sustainable control of diseases in the vineyards. The most important lever for the diminution of pesticides was found to be the use of resistant varieties. A very important effort has been carried out in INNOVINE project for the screening of yet uncharacterized germplasm collections for resistance to diseases and will be made available through publication in papers and in the European Vitis Database. However, it was also shown that the populations of downy and powdery mildews have the capacity to slowly adapt on resistant varieties and overcome these resistances. The current disease models have been improved to take into account grapevine physiology and genetic diversity. Finally, the partners of INNOVINE showed that canopy management practices that impact the berry size have an impact on botrytis incidence. INNOVINE project has on the whole provided diverse and complementary tools and knowledge that will be useful for a more environment-friendly control of diseases and that can be tested by the growers: level of resistance of tolerant varieties in field conditions, a set of strategies allowing to reduce the number of chemical treatments, improved monitoring and DSS systems.
INNOVINE project has started to work on better assisting grower’s decisions towards a more sustainable viticulture in two directions that will have to be further addressed in the future. The first one is the need of diversification of the varieties planted and the design of viticulture systems aimed at maximizing their quality in relation to the « terroir » where they will be planted. The second is the necessity to fully integrate monitoring tools and the current DSSs into services that are directly useful to growers to support their decisions and improve the management of environmental impacts in their viticulture systems.
INNOVINE’s results have already been disseminated through 59 peer reviewed papers, 154 presentations or posters in conferences and fairs and 65 press and web releases, demonstrations and courses.
Project Context and Objectives:
The European Union (EU) is by far the largest wine producer, consumer, exporter and importer in the world. Total EU-27 wine production is estimated at 156 Million hectoliters (MhL) for the marketing year 2010/2011 (GAIN report IT1105). In the EU, the vineyard area totals 3.64 million hectares, of which, 96 % is dedicated to wine production. The EU has almost half of the world total vine-growing area and 60 % of the global wine production. At present, the EU wine sector faces important challenges with the increasing world supply, the decreasing domestic demand, and the diminishing public support.
To face this increasingly fierce competition, the European wine sector must combine structural improvements with competitiveness reinforcement by replacing low-quality vines with better and diversified quality vines. Grape quality has a huge influence on the composition and the quality of the final product. Better grape quality is influenced by many parameters such as vineyard management, sanitary quality, viticultural practices and adapted varieties. Each aspect needs to be adapted and improved to optimise the composition of the wine and reach the quality desired by customers.

Adaptation of viticulture to climate change
According to the Climator study funded by the French National Research Agency before the start of the INNOVINE project, (http://www.international.inra.fr/research/green_book_of_the_climator_project) climate change will likely result in the following consequences on grapevine:
• Negative impact on the berry quality as warming accelerates all phenological stages, bringing maturation to hotter and dryer periods, making it necessary to change or even reverse practices aimed at increasing the heat undergone by grapevine. These changes should be supported by genetic adaptation to maintain or improve the berry quality.
• Reduction of the water availability with relatively minor degradation of the vine’s hydric comfort, which is not necessarily negative for berry quality. In some cases, irrigation is a way to maintain the hydric comfort. Without irrigation, one has to make use of density shift, which impacts quality and yield.
• As a result of the combination of these two phenomena, the impact of climate change on yield proves to be extremely variable according to the viticultural system considered.
• Contrasted effects regarding pests and diseases depending on the pathogen, the grape variety, and the geographical situation considered.
In conclusion, the report suggests, among other topics, to:
• Further understand the impact of high temperatures on vine photosynthesis, growth, and development
• Evaluate the adaptation of agricultural practices such as soil management, fertilisation, summer pruning along with adaptation of grape varieties
• Assess the role played by the socio-economic context, notably but not exclusively, regulation frame, cultural heritage, and profitability

Reduction of the application of pesticides in the vineyards
The use of chemical pesticides after World War II has allowed a spectacular increase in crop yield and the development of intensive farming. The aftermath of the intensive use of pesticides is a widespread contamination of the environment, the emergence of pest resistance to these products, the presence of pesticides residues in food, and the negative effects on human and wildlife health (IFEN, 2006, www.epa.gov www.cape.ca).
Vine is a crop particularly dependent on pesticide use. In average, twenty treatments were applied annually in Europe. The total dosage of plant protection products (PPP) used in EU viticulture is 21.4 kg active substances (AS)/ha per year, from which 19.5 kg AS/ha are fungicides, 1.3 kg AS/ha herbicides, 0.3 kg AS/ha insecticides and 0.3 kg AS/ha other PPP. In Europe, 69,819 tons of pesticides were employed each year to protect grapevine (500 €/ha per year) with a total cost of 1.9 billion € in EU27 (Eurostat, 2007). The main fungicide treatments are required to control two important diseases, powdery mildew and downy mildew, which generate 70% of the total grapevine growing costs (INRA report, 2006). In countries such as France, 20% of agricultural pesticides are used in vineyards while vine represents only 3% of the cultivated areas (Aubertot et al., 2005). The geographic concentration of grapevine growing, often in regions bordering rivers, exacerbates the hazardous effect of pesticide on the environment, both locally (soil, air and subterranean waters) and globally (running waters and air). According to a European project, of 24 food products tested, grapevine is the 4th crop exceeding the authorised residues limitation. More than 25 molecules were found in wine after analyses including copper, folpet, dimetomorph, azoxystrobin, iprovalicarb, metalaxyl, benalaxyl and imazalil. Nowadays, grape growers (about 2.4 million people over EU27) are confronted with an increasing social demand for environmental preservation and food safety followed by increasingly restrictive national and international regulations. To tackle the need of a reduction in the use of pesticides across Europe, the Council of the European Union recently adopted the so called ‘pesticide package’ (2009). The introduction of this Pesticide package will mean a fundamental change in European agriculture, with each Member State responsible for hammering out plans to ensure that farmers use non-chemical effective alternatives for crop protection whenever possible and adopt some form of Integrated Pest Management (IPM), starting from 2014.
In this context, the grape growing sector is increasingly concerned with environmental, toxicological and health issues. New solutions are needed to assist the vineyard sector to change practices that combine better environment protection (pesticide use reduction, use of organic agricultural practices and/or biological control) with long-term economic performance (grape quantity and quality).

Strategic objectives of INNOVINE project
The strategic goals of INNOVINE project were to help the European vineyard sector facing the increasingly global competition by meeting:
• Consumers’ demands for diversified high quality wines and concerns for food safety
• Citizens’ requests for environment-friendly production systems involving decreased or no use of pesticides and spare of not renewable natural resources
• Producers’ needs of plant material, tools and methods to help them cope with the negative impacts of climate change while responding to demands for quality, environmental friendliness and needs of profitability
To reach this overall goal, the project will:
• At the plant level, improve and design agricultural practices (canopy management, irrigation, fertilisation, training systems, pest and disease control, etc.) aimed at maximising berry quality, durable resistance to pests and diseases, and adaptation to climate change (higher CO2, drought, UV light, and higher temperatures)
• At the vineyard level, design, develop and test innovative agronomic systems integrating new agricultural practices and taking into account the variability of constraints met by European vineyards grown under a wide range of environments
• At the breeding level, diversify grapevine varieties with regards to desirable adaptive traits building on tools and knowledge developed through international breeding and genomic initiatives.
The project will combine short, medium, and long-term approaches to respectively conceive innovative viticulture systems, design and test novel agronomic practices and decision support systems, and exploit the genetic diversity of grapevine that all together will ensure a progress towards sustainable viticulture.

Workplan objectives of the INNOVINE project
The project was broken down in five research and development work packages, one work package devoted to dissemination of the project’s results and one work package for to the project’s coordination and management.
WP1 (Effects of vineyard practices and of environment on grapevine physiology and berry quality in relation with climate change) objectives were to:
• understand and model berry composition in relation with different stress conditions (individual and combined effects of CO2, drought and temperatures, i.e. the main components of climate change) and vineyard management practices in view of tailoring berry quality in response to climate change
• develop relevant phenotyping methods and tools to better dissect the response to abiotic stress.
WP2 (Designing optimized vineyard practices to reduce pesticides) objectives were to:
• develop knowledge on the durability of resistance genes to powdery and downy mildews
• determine the efficiency of indirect resistance (e.g. improved cluster morphology for lower susceptibility to bunch rot), sanitation, exploitation of beneficial effects of biodiversity) for drastically reducing pesticide use.
• improve predictive disease models and epidemics monitoring to guide defense schedule in viticulture
WP3 (Exploiting the genetic diversity in grapevine) objectives were to:
• screen for new resistances to diseases and pests (mildews, black rot and Phylloxera) and adaptation to abiotic stresses in the grapevine germplasm
• study the genotypic diversity and plasticity in response to abiotic stresses
• provide a basis for a European strategy for the pyramiding of resistance genes in elite material
WP4 (Conception and test of sustainable viticulture practices or systems) objectives were to:
• test the practices developed within WP1, WP2, WP3 and WP5 and evaluate their impact on berry content and/or on vineyard health
• assess their impact on environment
• evaluate their technical feasibility and economic costs
• develop integrated management systems for a viticulture combining low environmental footprint and high quality in different agro-ecological conditions
WP5 (Implementation of decision support systems towards a sustainable viticulture) objectives were to develop decision-support systems and user-friendly vineyard monitoring tools for:
• abiotic stresses, disease management, and canopy management
• adaptation to various agro-ecological conditions of vineyards in Europe
WP6 (Dissemination and technology transfer) objectives were to:
• produce guidelines of agricultural practices and cropping systems management and training
• manage intellectual property rights (IPR)
• disseminate to the scientific community
• communicate to the general public
• organise training courses and workshops
WP7 (Consortium coordination and project management) objectives were to provide:
• strategic steering of the project (scientific, societal and industrial survey, contingency plan, etc.)
• project monitoring (project progress against milestones and deliverables, resources consumption, etc.)
• administrative, logistical, and financial coordination
Project Results:
WP1 – Effects of vineyard practices and environment on grapevine and berry in relation with climate change
The objectives of WP1 were to understand and model berry composition in relation with:
• different stress conditions in view of tailoring berry quality in response to climate change
• the use of genomic tools to dissect the response to abiotic stresses
• vineyard management practices in view of tailoring berry quality

We first designed a common set of sampling and analytical procedures (D1.1) including the number of plants needed for greenhouse, field and FACE experiments, the number of clusters and berries sampled per plant, as well as their exposition, and all the procedures related to assessment measurements of phenology, environment, growth conditions, vegetative development, and fruit development and composition. These procedures were used throughout the consortium, especially in WP1 and WP4, to gather a common set of data that could be used to test the effects of environment and viticultural practices on grapevine physiology and berry composition (T1.1) and for modelling purposes (T1.3). These sampling and analytical procedures may be used for other large scale and long term experimental vineyards. In parallel, new phenotyping tools (mostly non-destructive) were tested and improved in T1.2.

Phenotyping tools
Several types of non-destructive phenotyping tools (Multiplex, hyperspectral imaging, thermal imaging, reflectance) have been tested in different locations and on different varieties. Their range of use and reliability were assessed.

Fluorescence based measurements. The Multiplex (Force A) was used to monitor the status of the canopy and berry ripening. Standardization procedures of the punctual data treatments were designed. The amount of berry anthocyanin was determined and compared to sensor data. A high correlation (0.88) between sensor data and anthocyanin content (mg/berry) was found. Multiplex sensors proved to provide good estimate of anthocyanins and flavonol contents, in agreement with biochemical measurements. Field measurements turn out to be rather simple, but some environmental factors might interfere with the measurement, making the use of this technology still more robust when it is performed in the laboratory. The use of the Multiplex Force A sensor associated to the design and validation by the partners allowed to consolidate a calculation method (CUBA anthocyanin converter) delivering the anthocyanin content on a berry weight basis (instead of surface).
Concerning the canopy, plot mapping work was partially performed on the foliar indices of the mounted Multiplex.
The results of WP1 and WP5 were used to set up practical application with end users in WP4. Within the frame of WP4, selective harvest based on plot segmentation through the association of leaves and grapes maps were conducted at Château Couhins in 2015 and 2016.

Reflectance spectroscopy. 16 white berry accessions were used to compare wet analysis to reflectance spectra in three berry phenological stages, for the quantification of chlorophylls and carotenoids content. An intensive defoliation and exposition of grapes were done in two moments of the day, in pre- and post-veraison stages, to study the physiology of berry sunburn and the varietal susceptibility. Reflectance measurements provide a cheap and non-destructive way of assessing chlorophyll and carotenoid contents on berries, also with a better accuracy and ease with lab measurements than with field samples. They showed that the radiative excess appeared to have a central role in sunburn symptoms, while temperature seemed to have a secondary effect. They also showed that water deficit produced spectral variations in both rootstocks and scions. Reflectance spectroscopy could find applications for drought tolerance screening. On solid reactions appeared as a promising approach for compositional characterization. Stems seem to play an important role in plant drought adaptation.

A non-invasive portable NIR spectrometer was also used successfully to determine total soluble solids of berries under field conditions.

Thermal imaging showed that Water Stress Index (CWSI) and delta T are correlated with leaf physiological parameters (leaf water potential and stomatal conductance to water vapour) in cvs Aragonez and Touriga Nacional. Interactions are currently developed for the use of hyperspectral imaging (J. Tardaguila, WP4). A four channel multispectral camera was also tested to find out whether a camera mounted to a helium filled balloon can be used as an accurate phenotypic tool.

A new, easy-to-use, smartphone application (VitisFlower®) for the automated actual flower estimation in grapevine inflorescences in the field was designed based on artificial vision techniques and mathematical modelling. The application can bring valuable information for yield forecast.

Environmental components
The effects of environmental parameters (high CO2, high temperature), innovative management practices (irrigation regimes, source limitation (early and late), and genetic diversity (varietal and clonal) of grapevines grown under different terroirs have been assessed.

Water. Sustained Deficit Irrigation (SDI) and Regulated Deficit Irrigation (RDI) with early and late deficits were compared in a field trial on cvs. Aragonez and Touriga Nacional in Portugal. The irrigation regime had no effects on vigor, yield and berry composition but induced some differences in berry cuticular permeance and wax content and on vine physiology (leaf water potential, photosynthesis rate, stomatal conductance and canopy temperature) mainly during the most stressful days. Irrigation strategy and cluster position in the canopy (east vs west side) modulate berry cuticle structure, permeance and biosynthesis.
Another irrigation experiment was set up in Rioja on Tempranillo Blanco. Irrigation had no effects on yield components but impacted berry composition. A large variation in yield among locations, and yield reduction relatively to Tempranillo Tinto and a reduction in seed number were observed.
Anti-transpirant treatments were combined with defoliation treatments to alter the water and carbon status. On the Nebbiolo variety, anti-transpirant treatments significantly reduced yield, while increasing total soluble solids and reducing bunch compactness. Overall, the effects of anti-transpirant pre-bloom spray, early main leaf removal at pre-bloom or early main and lateral leaf removal at pre-bloom on phenolic composition were mild, whereas berry sunburn and bunch rot incidence were very sensitive to the amount of leaf cover, especially laterals, from flowering onward. Late season apical leaf removal reduced yield and retarded ripening especially when applied pre-flowering on cv. Ortrugo. However, at harvest, titratable acidity of the two defoliation treatments did not differ from control’s level, suggesting that cv Ortrugo is likely more sensitive to late season climate pattern than to canopy manipulations. In particular, under a warm climate, the drop in malic acid seems to be very difficult to be decelerated. Transcriptomic analysis validated six markers of the early defoliation treatment in Sangiovese. Furthermore, candidate molecular markers were also obtained in other three different grapevine genotypes (Ciliegiolo, Nero D’Avola and Ortrugo), subjected to the same level of pre-bloom defoliation over two consecutive years in their distinct areas of cultivation.

CO2 level. A FACE (Free Atmospheric Carbon enrichment) system was set up to study the impact of atmospheric CO2 increase on berry composition of cv. Riesling and Cabernet Sauvignon. Trials conducted under high CO2 (550 ppm) showed that growth and vigour increased for both varieties tested. This will be of particular importance since growth and vigour impact on vine´s microclimate and hence interfere with fruit composition and fruit health. However, for the red variety tested, high CO2 in field conditions had no major effect on anthocyanins and did not affect single berry weight in the short term, but might increase it in the long term due to a cumulative effect.

Heat stress. Moderate (1-2°C) but long term temperature increase were imposed locally on Syrah clusters, and berry temperature, PAR irradiance and composition were monitored. The data, obtained with cv. Syrah, were used for calibration of FERARI multiplex sensor.

Combined stress. More integrative approaches of the effects of environment were developed either by combining and controlling different stresses in greenhouse conditions, or by studying the effects of terroir under different natural environment.
White and red cultivars of Tempranillo were used to study the effects of different abiotic stresses in greenhouse conditions (combining elevated CO2, elevated temperature and drought acting simultaneously or in interaction). White Tempranillo was less productive than the red one, especially under elevated temperature and current CO2.
The effects of terroir on the performance of a new variety (Tempranillo Blanco, TB) were also investigated. A large variation in yield of Tempranillo Blanco among different locations, and yield reduction relatively to Tempranillo Tinto (TT), especially under elevated temperature and current CO2, and a reduction in seed number were observed. Interesting interactions of different traits with temperature and altitude were discovered for TB.

Canopy manipulations
The effect of changes in the source/sink ratio (partial defoliation) imposed at two stages of development on cv. Sangiovese were investigated in two field experiments. Pre-bloom defoliation of cv. Sangiovese reduced fruit set and berry weight but increased Brix and colour. After late season source limitation, no significant differences were detected on yield components, but the more intense defoliation decreased solid soluble solids and whole-vine net CO2 exchange rate. In another defoliation experiment in Italy with Sangiovese, the pre-bloom defoliation reduced fruit set and berry weight but increased Brix and colour.
Potted grapevines were subjected pre-véraison to a limiting vs. non limiting crop load by leaving 3 and 12 primary leaves per shoot. Input data sets required to validate the sugar and colour accumulation model developed at INRA Bordeaux were acquired along with seasonal assessment of whole-canopy gas exchange rates in both treatments.

Genetic component
The lower yield observed in Tempranillo Blanco (TB) compared to Tempranillo Tinto (TT) cultivated under the same experimental vineyard was due to differences in cluster weight but not with cluster number per vine. Reduced fruit set (ranging between 19 to 29% reduction depending on the year) and increased millerandage (24 to 42% increase) appeared as the main causes of cluster weight differences considering that seeded berries reached similar weight and size in both Tempranillo lines. Concurrently, pollen viability and seeds developed within seeded berries were also reduced in TB by 29-65% and to 0.7 to 1.1 seeds per berry, respectively. The performance of TB analyzed in a network of 9 TB experimental vineyards of the same age, cultivated under similar management practice, but at different altitudes showed that low temperature and rainy conditions around flowering time correlated with reduced fruit set, increased millerandage rate, and reduced pollen viability, whereas number of seeds per berry remained more stable. Collectively these results suggest that reproductive dysfunction in TB results in looser clusters and lower yield. This genetic-driven dysfunction is more highly expressed under environmental conditions that are unfavorable for gametophyte development and fertilization. The genetic dysfunction of TB that affects gamete viability very likely originated from a chromothripsis event which provides the first known example in plants.
A comparison of the phenology and berry composition was conducted for 52 varieties all grafted on the same rootstock and grown under natural conditions in Bordeaux (INRA) and revealed large differences among them, with, along the years, phenology correlating with the climatic data. Among these 52 varieties, three were selected for further comparisons with other locations run by other partners: Cabernet Sauvignon, Tempranillo and Riesling. The data collected during the 4 years of the project are under analysis.

Transcriptome-metabolome correlations
A set of data obtained on berries from Cabernet Sauvignon plants that have received different levels of nitrogen supply was used with special focus on secondary metabolism. For technical reasons, and because we also wanted to deal with situations that often lead us to compare heterogeneous samples, the integrative analysis was carried out using the metabolomic data related to skin samples while the transcriptomic dataset is related to the entire berry including seeds. The integrative analysis of this heterogeneous set of samples derived from the same berries allowed to design a strategy that allowed (a) to avoid the bias due to sample heterogeneity (b) to identify metabolite-transcripts correlations that are expected from literature data, this showing the validity of the method (c) to assign a putative function to some transcription factors whose biological function is still unknown, based on their high correlation with specific metabolites. The putative biological function of the transcription factors will have to be further demonstrated with adequate biological tests.
The pipeline thus may be applied to other data obtained in INNOVINE or elsewhere.

Modelling
Berry composition has been modelled from the biochemical parameters that have been collected throughout the consortium for at least two years. Modelling was based on the collection of a full set of phenotypical and berry composition data (total C and N, sugars, organic acids, amino acids, anthocyanins, polyphenols, K+, and in some cases, aroma precursors) for a wide range of different varieties cultivated in different locations and under different vineyard and canopy management practices (type of pruning, training systems, defoliation, irrigation). This also extended to comparisons of varietal and clonal diversity of grapevine plants.
Two functional-structural grapevine models, respectively "HydroShoot" and "GrapevineXL", were developed in this framework to simulate the impact of environmental stimuli on grapevine functions and delivered (D1.3 and D1.5). Both models have a common base which couples gas-exchange processes to those of xylem transport, but serve two complementary objectives: HydroShoot estimates the impact of canopy structure on plant Water Use Efficiency (WUE) while GrapevineXL aims at stimulating growth and sugar concentration of the berry. Finally, a method for modeling the correlations between transcriptome data and metabolome data was delivered (D1.4).
The GrapevineXL model couples the water transport (adapted from Tardieu-Davies module), leaf gas exchange (an extended Farquhar-von-Caemmerer-Berry module), and a berry growth module. The xylem water potential and phloem sugar concentration were simulated dynamically based on environmental conditions (e.g. CO2, light, temperature, humidity, soil water content etc.), and were fed into the berry growth model to simulate berry growth and sugar concentration. The model was calibrated with detailed data sets of leaf gas exchange, leaf water potential, ABA and hourly transpiration and validated with a separated dataset. The model reproduced well the effects of radiation, temperature, CO2, VPD, soil water content on leaf photosynthesis, transpiration, leaf temperature and stomatal conductance and leaf water potential, demonstrating its robustness. By varying the hydraulic conductance along the pathway (e.g. soil, root and leaves) and the sensitivity of stomatal conductance to ABA and leaf water potential, the model can produce virtual genotypes with different water use behaviours (iso- and anisohydric). The model was also calibrated and validated for simulating berry growth (as berry fresh weight and dry weight). It can reproduce well the experimental results under various source-sink ratios. Model simulation showed the effects of water stress and light on berry growth and sugar concentration. This validated model will provide a valuable tool for assessing climate change effects on berry growth.
The models can serve as a framework for investigating the effects of ambient and below-ground factors, as well as the effects of hydraulic conductance along the water transport pathway, on plant water use through their combined effects on leaf temperature, photosynthesis, transpiration rate, xylem and leaf water potential, and berry growth. A first scientific paper about the GrapevineXL model is under review. Other manuscripts on GrapevineXL and HydroShoot will be submitted soon. Once published, the program of the models will be made available for public access to GroIMP and OpenAlea platforms.

Conclusions
WP1 designed several important tools that will help to improve European viticulture: standardized sampling and analytical procedures of berries, improvement and/or validation of several phenotyping tools allowing to characterize the physiological status of grapevine plants and the berry content, correlation analysis for transcriptome and metabolome, modelling of berry composition in response to environment.

WP2 – Designing optimized vineyard practices to reduce pesticides
Fed by the fact that, in EU 28, grapes occupy about 3.5% of the arable land in front of absorbing over 60% of the total pesticide application, WP2 was conceived to design vineyard practices allowing a reduction in pesticides use while assuring profitable plant protection. In other words, WP2 was compromising between environmental and economic sustainability. Moreover, reduction in the use of pesticide is socially perceived as a primary goal.
This aim was pursued by integrating four different pillars based on: assessing degree and durability of resistance primarily to downy and powdery mildew in a number of pre-existing accessions (task 2.1 with links to WP3); testing reliability and repeatability of new vineyard or canopy management techniques aimed at inducing higher indirect tolerance to bunch rots through modifications of canopy microclimate as well as of cluster and berry morphology (i.e. looser and then more ventilated clusters, berries with a thicker skin having less contact points with adjacent berries (task 2.2 with links to WP1 and WP4); with special reference to organic viticulture, task 2.3 investigated methods suitable to reduce inoculum of either downy and powdery with special attention to the use of bio-control agents; finally task 2.4 focused on the approach gravitating about the use of modelling for either predicting vineyard features typically correlated with disease occurrence and severity (i.e. phenological stages and vegetative growth) and loss of yield and grape quality due to incidence and severity of certain diseases). This task had direct connections with WP4 and WP5.
Next paragraphs will summarize how each task has fulfilled the question given in the WP title.

Task 2.1. A new generation of partially-resistant grapevine cultivars is now available allowing low-input viticulture to emerge (up to 90% reduction of fungicides). However, there is a need to define strategies for the deployment of these partially-resistant cultivars in order to maintain their efficiency over time. Partial resistances are indeed prone to select for increased aggressiveness of pathogen’s populations (i.e. quantitative pathogenicity) leading to a decrease in the efficiency of the resistance (i.e. resistance erosion). The aim of the task was therefore to develop knowledge on the durability of resistance genes to powdery and downy mildews. With these aims, we have evaluated fungal epidemics grape varieties showing varying levels of partial resistance to downy and powdery mildews in a first step. In a second step, we have tested the adaptive potential of powdery and downy mildew pathogens to partial resistance. The results obtained showed that the partially resistant varieties currently available, although providing a good control on downy and powdery mildew alone (up to 90% of sporulation less), are still susceptible to many other fungal diseases such as black rot that can potentially be very destructive for the harvest. Moreover, powdery and downy mildew are able to adapt to partial resistance leading to the emergence of isolates that have a higher level of pathogenicity. Therefore, natural resistance of grapevine need to be protected in order to prevent the emergence of disease isolates of downy and powdery mildew and to control for other fungal pathogens. This can be achieved by using new decision rules based on low-input management strategies (including organic ones) that are adapted to the partially resistant varieties. By contributing to an accurate understanding of fitness trajectories of pathogen populations facing partial resistance, these innovative results provide essential data for devising durable management strategies of resistant grapes.

Task 2.2. Grapevine is a crop particularly dependent on pesticide use and requiring a treatment index markedly above ten. Nevertheless, fungal infection can cause extensive fruit quality and economic losses in almost any grape growing region around the world. Despite integrated pest management programs, fungicide application is the most common strategy to control fungal development. Most of the sprays usually applied in the vineyard are targeted at powdery (Erysiphe necator) and downy (Plasmopara viticola) mildews and bunch rot, caused by Botrytis cinerea. The aim of this task was to assess practical solutions to reduce fungicide input in viticulture by applying phytosanitary and cultural methods – without detrimental losses in profitability or quality.
With all varieties tested, the risk of bunch rot was mainly reduced by decreasing cluster compactness. Changes in leaf area to fruit weight ratio led to reduced fruit-set, loosening cluster compactness, reduced cluster weight, crop load and the susceptibility to bunch rot as well as impacting on the ripening process. Results also hint to the point that pre-flowering leaf removal, heavy shading and pruning systems with high bud load and afterward adjusted through bunch thinning are the best techniques to increase indirect tolerance to bunch rots. In normal seasons, a combination of these strategies seems to be an excellent tool to decrease bunch rot risk in the vineyard. Notably, the proposed techniques have two additional advantages: i) under very few exceptions they are fully mechanisable thus reducing the impact of production costs while reaching a higher number of potential users; ii) regardless of their positive effect on restraining incidence of bunch rots, some of the features promoted by this approaches (i.e. smaller berries with a more favorable skin-to berry weight ratio) are usually conducive to better grape and wine quality.
Whilst cluster compactness and berry morphology play a key role on bunch rot development, there is no immediate relation between cultural practices and the severity of downy and powdery mildew. Under severe infection conditions an intensive fungicide application schedule with short intervals is strongly required for Vitis vinifera cultivars to overcome negative impacts in concern of quantity and quality.

Task 2.3. Both downy and powdery mildews, caused by the fungi Plasmopara viticola and Erysiphe necator, respectively, represent, major diseases for grape growers throughout Europe and are also the main target of chemical applications. More than 45% of active ingredients sprayed on vineyards each year are sulphur and copper (Eurostat, 2008) used to cope with these two diseases, including in organic viticulture systems.
Within INNOVINE project, and in particular in WP2, a specific focus was therefore dedicated to study sanitation techniques based on different methods (physical and by mean of bio control agents) and enhance the efficacy of an already known bio control agent for controlling powdery mildew. Moreover, a specific section was devoted to the study of biodiversity within a vineyard managed according to different strategies and different intensity of chemical usage.
The different sanitation treatments, aimed to limit infections of downy mildew, included: i) fire ii) bio control agents (such as Trichoderma and Bacillus spp.) and iii) composts from leaf debris. For two years, infected leaves containing oospores were collected before leaf fall and then prepared and treated according to different methods. Then, in the following seasons oospores germination dynamics were indirectly observed by means of the floating disk test bioassay. Downy mildew oospores showed a high variability in their germination rates both in 2014 and 2015 seasons for all the treatments so that no significant differences were observed among treatments and in comparison to the untreated control.
Concurrently, different laboratory experiments were carried out to enhance the efficacy of the application of an Ampelomyces based product for sanitation on powdery mildew. In particular, the efficiency of pre-germination was studied as well as its applicability with the last copper spray in the season against downy mildew. In the field, sanitation was performed in 2014 and then evaluated in 2015 when it was integrated into an overall strategy aimed at merging reduction of primary inoculum across seasons and control of secondary infections. Results obtained showed that a pre-germination period of 12 hours could speed up by about 6 hours the germination of 50% of the Ampelomyces conidia, so that they are more promptly effective when sprayed into the field. On the other side, the negative effect of copper salts was highlighted on both quantity and quality of Ampelomyces conidia germination leading to exclude the possible combination of the two treatments (copper against downy mildew plus Ampelomyces as sanitation against powdery mildew). Finally, the sanitation treatment combined with an advanced organic management strategy (based on sulphur application according to a decision support system) proved to be effective for the control of the disease and a sustainable option when the disease level is, for any reason, particularly high at the end of the season.
Finally, field experiments were conducted by partner GRC at Geisenheim over two years in a vineyard of cv. Riesling treated using different spraying strategies: a) reduced minimal copper regime; b) organic crop protection as well as c) integrated crop protection. During the development of the berries, samples were taken after and mid-veraison and close to harvest. The natural population of yeasts present on the berry skin was than assessed and periodically quantified. At the first sampling date, the dominant genera in reduced copper and organic protection was Aureobasidium and for the integrated regime Rhodotorula; both genera are without any relevance for the fermentation process. At the second sampling date, the genera of Aureobasidium remained for the reduced copper regime. In the integrated protection, Hanseniaspora sp. became dominant both at mid-veraison and close to harvest. Hanseniaspora sp. was the leading yeast genera of the carpospheres close to harvest for all spraying strategies and hardly any other genera were found.

Task 2.4 The experiments carried out over three years on twelve resistant varieties (Calardis blanc, Bronner, Solaris, Reberger, Fleurtai, Cabernet volos, Villaris, Felicia, Regent, Johanniter, Calandro), one reference (Merlot), at three timing during the season, for downy mildew (both on leaves and bunches) and for powdery mildew (on leaves) allowed to perform a complete assessment of the resistance on single infection components. Components such as the ability of pathogen to cause infection, the incubation period, the latency period, the production of sporangia/conidia per sporulating area and the ability of sporangia/conidia to cause new infections were studied. For downy mildew, the area under the disease progress curve (AUDPC) on leaves confirmed that some varieties showed highly significant differences from the V. vinifera cv. Merlot reference and other varieties. Merlot Khantus, Bronner, Johanniter, Solaris, Cabernet Volos, Villaris, Calandro and Regent showed a very low AUDPC, whereas Palava, Rkatsiteli and Reberger were similar to the Merlot control. Merlot Khorus, Fleurtai, Felicia and Calardis Blanc recorded an intermediate value of AUDPC, in any case lower than Merlot. All the resistant varieties expressed a significant resistance on bunches compared to the V. vinifera control. For powdery mildew, interesting differences were observed between resistant varieties and the V. vinifera control (cv. Pinot noir) for the number of conidia produced on inoculated leaves. In particular, Bronner and Johanniter showed a resistant expression that it will be useful to deeply investigate.
The phenological model was successfully calibrated for the 19 selected varieties with Italian and Spanish data. Mean Absolut Error were more than satisfying in almost all the 57 cv/year and 57 cv/BBCH cases: higher MAE values were obtained in the early developing stage (bud break), while, once started, the growing of shoots, i.e. the development of new leaves, is better simulated by the model. The cultivar specific model calibration here present will be validated in the next years in order to test robustness.
The results obtained by UCSC confirmed that there is a different behavior of the same pathogen population on different grapevine varieties. The simulations performed with the model prototype developed during INNOVINE project confirmed a large variability in the quality variation between clusters apparently affected by the same level of the pathogen.
Obtained results allowed developing a model prototype for the relationship between disease severity and yield loss: it was built in the form of a polynomial equation (i.e. y=ax3+bx2+cx+d) and allows estimating the reduction/increase of sugar content, pH, titratable acidity and polyphenols in different varieties according to the disease severity on bunches. Thus, the regression equations produced during this activity represent a useful tool in order to estimate possible variation of oenological characteristic of the juice starting from a field observation of incidence and severity of downy and powdery mildew infection on bunches.

Concluding remarks from WP2 activity
Extensive work carried out in this WP2 on the epidemiology and pathogenicity of several partially resistant varieties to downy and powdery mildew has confirmed the genetic tool as the most powerful for drastically reducing fungicide use in vineyard, up to -80-90% of normal dosage. It was confirmed that at least two “maintenance” fungicide sprays are needed early in the season to prevent or slow down resistance erosion and to avoid damaging late season disease outbreaks. However, under very high disease pressure, the resistance of these genotypes might turn out to be insufficient. Moreover, the occurrence of other diseases such as black rot or Colomerus (Eriophies vitis) is often higher than in susceptible V. vinifera cultivars. In order to cope with these limitations, WP2 is proposing new decision rules based on low-input management strategies (including organic ones) that are adapted to the partially resistant varieties and to the climatic features of the specific site.
In addition, WP2 has disclosed and reinforced the importance to act in the vineyard in order to induce indirect higher tolerance to bunch rot. Manipulations of the source-sink balance can be used, on purpose, to induce more favorable bunch characteristics which, under low to medium disease pressure, can totally replace the need for spraying against bunch rots. The most powerful, readily available and fully mechanisable techniques are pre-flowering basal defoliation and pre-veraison apical to the cluster defoliation. Sustainability of the proposed techniques also relies to the fact that while carrying out positive modification in cluster morphology (i.e. looser bunches, smaller berries with a thicker skin) they can concurrently help solving other viticultural problems such as excessive yield, incomplete ripening or, in the case of the late, apical leaf removal, too fast ripening.
While the results gathered in terms of effectiveness of bio-control agents have confirmed that they can function as effective surrogates to a main protection strategy, results obtained in task 2.4 of WP2 encourage the use of modelling for increased effectiveness of the protection strategy. Epidemiological modeling now embraces cultural features such as the prediction of phenological stages which is a further fine tuning for the release or reliable alarms. The most notable outcome is, indeed, between cultivar variability in terms of yield and must composition response to an increasing severity of powdery and downy mildew attack. This is precious information since depending upon the genotype chosen a given fraction of disease incidence can be tolerated with no detriment for vineyard efficiency and, under extreme circumstances, mild attacks can also turn out to be useful to modify grape composition toward a desired trend (e.g. less sugar and increased acids).

WP3 – Exploiting the genetic diversity in grapevine
INNOVINE WP3 was searching for genotypes that are superior for specific traits and show new characteristics in terms of biotic and abiotic stress. New sources for resistance against powdery and downy mildew, black rot and phylloxera were identified in yet scarcely explored germplams. Within the abiotic stress task partners looked for cultivars and clones with different behaviour under heat and drought conditions.

The objective of the Task 3.1 was the screening for powdery (PM, Erysiphe necator) and downy mildew (DM, Plasmopara viticola) resistance in accessions of domestic and wild Vitis vinifera from Eastern Europe. Resistance evaluation was performed by leaf disc assays and disease assessment under controlled conditions as well as by experimental inoculations of cuttings in the greenhouse. In parallel, genotyping by SSR markers for accession identification and genotyping of the most interesting accessions by markers linked to the REN1 locus were done. A testing with the Multiplex-330 sensors as a new tool for mildew evaluation hashas been included.

Downy mildew. 1707 tests were made on 1187 grapevine accessions and controls. The degree of resistance to P. viticola, assessed by leaf disc test, revealed 39 resistant accessions (OIV scoring > 6, from medium-high to very high). Thirty-five accessions originate from Caucasus (Georgia), 1 from Balkan region, and 3 from central Asia. The Multiplex 330 sensor was shown to be able to monitor the DM infection time course. Diseased vines could be differentiated from healthy vines with the first significant signals detectable 5-6 days after inoculation.

Powdery mildew. 1179 tests involving 910 accessions have been completed and 282 accessions reached a degree of resistance to E. necator with scores > 6 (from medium-high to very high). Most of them (249) originate from the Caucasian region (including 197 from Georgia). Twenty accessions are derived from Central Asia, 10 from North Black Sea, 2 from Balkans, and 1 from the Near East. They belong to Vitis vinifera ssp. vinifera (253) and ssp. sylvestris (29). Genotyping of PM resistant accessions has been completed. In particular, markers have been tested that are specific for the REN1 resistance locus, already identified in the V. vinifera genotype Kishmish vatkana. A few accessions turned out to carry the REN1 locus. Others showed no indication for REN1 and are therefore novel sources of resistance and highly interesting for further use in grapevine breeding and research.

Task 3.2 evaluated the resistance of grapevines to black rot (BR, Guignardia bidwellii). From germplasm collections in Hungary and Germany, potential resistance sources against BR were tested. One group of 207 accessions was chosen based on known BR resistant wild progenitors in their pedigrees, the other group consisted out of 459 seedlings of two independent F1 progenies. Accessions were classified into resistance grades on a 9 grade scale. Clearly sensitive accessions were discarded. Remaining plants were preserved for verification. Quantitative Trait Locus (QTL) analysis for BR has been performed in a genetic background of ((Bacchus x Seyval) x Villard Blanc) (151 F1 plants). Several QTLs have been identified among them a major QTL on chromosome 14. This locus could be considered as a meta-QTL since it was also found in different populations. Both parental genotypes are very promising sources for further crosses. Eighteen accessions rated > 5 and the top of this list can be used for further elaboration of the genetics of black rot resistance and breeding.

The objective of Task 3.3 was to identify novel sources of phylloxera resistance among wild Vitis or related species. Muscadinia rotundifolia and Vitis aestivalis are especially targeted. In a first step, common phenotyping procedures were defined by partners and plant material was exchanged for control checking. Each partner has developed phenotyping facilities and plant material is under evaluation for nodosity development after artificial inoculation with phylloxera. As phenotyping for phylloxera resistance is a difficult task, it is a long lasting process. In parallel a Muscadine x V. vinifera progeny of 92 individuals was genotyped using SSR markers and a genetic map was developed. The map presents a good coverage, although the density of markers is less on the female map. It is characterized by a good synteny, by zones of high segregation distortion and the lack of polymorphism of some markers for the female map. Up to now reliable QTLs for nodosity are to be identified. All together these new data will improve our understanding of phylloxera resistance and contribute to develop tools for rootstock breeding.

In Task 3.4 abiotic stress was in focus. Screening several collections of Tempranillo (syn. Aragonez) clones revealed variation for surface leaf temperature and ripening time under heat and drought stress. A short list of selected clones with extreme phenotypes was developed to perform transcriptional and phenotyping analyses. Genetic variation for plasticity of the berry composition under drought has also been studied in a core collection of V. vinifera cultivars. The comparison of the transcriptomes of isohydric and anisohydric cultivars under water stress conditions allowed the identification of candidate genes encoding master regulators of the grapevine water stress response using an integrated approach based on the analysis of topological co-expression network properties. The genome-wide transcriptional study indicated that the isohydric behavior relies upon the following responses: i) faster transcriptome response after stress imposition; ii) faster abscisic acid-related gene modulation; iii) more rapid expression of heat shock protein (HSP) genes and iv) reversion of gene-expression profile at re-watering. Conversely, reactive oxygen species (ROS)-scavenging enzymes, molecular chaperones and abiotic stress-related genes were induced earlier and more strongly in the anisohydric cultivar. In parallel, a combination of genetic and genomic approaches allowed to identify genes and gene variants associated to the phenotypic changes observed in Tempranillo short and long ripening cycle clones. Some of them could be at the origin of such changes. Further transcriptome analyses are required to confirm these possibilities. Further analysis on clones with contrasting responses to stresses are still underway in order to identify predictive markers of adaptation.

On the whole, these new sources of V. vinifera varieties or clones better adapted to abiotic stresses can be directly used for new plantations in relevant areas. Together with the germplasm identified as resistant to diseases, they also represent a highly valuable plant material for the breeding programs that aim to develop varieties improving the sustainability of viticulture. Adaptation to abiotic stresses can rely on different mechanisms and WP3 gives tools (e.g. molecular markers), methods (e.g. how to screen for plasticity) and knowledge (e.g. impact on molecular or physiological mechanisms of drought stress in clones or varieties that show different behavior under stress) that will allow to develop more efficient breeding strategies.

WP4 – Conception and test of sustainable viticulture practices or systems
The overall aim of WP4 was to design and assess under real vineyard conditions the cost-benefits relationships of new practices and management systems aiming at adapting viticulture to climate change, at reducing pesticide application and at optimizing precision viticulture practices. These assessments included not only economic but also environmental impacts and technical feasibility.

TASK 4.1.
For the first and second objectives, innovative viticulture systems that integrate new agricultural practices while taking into account the variability of constraints met by European vineyards, were developed and tested.

Objective 1: Effect of practices aiming at reducing pesticide application
Temperatures in the northern hemisphere have increased, particularly over the last 20 years. These climatic changes have led to an advancement of key phenological stages and the harvest dates in Europe. Changing the fruit to leaf area ratio, was tested as an option to maintain and enhance fruit and wine quality and reduce phytosanitary treatments.
Trials to test improved viticultural practices from WP2 under end-users´conditions, including organic management regime, were performed by SCHLOSS and GCR. The overall response of partial defoliation and severe summer pruning conducted prior to veraison suggested that it can be used as a management practice to alter berry ripening speed and grape composition, influencing wine quality and phytosanitary responses.

Objective 2: Effect of practices aiming at tailoring the quality with regards to climate change
The effect, on the berry quality, of different treatments under end-users conditions were under test by the partners: irrigation regimes, canopy management, yield management, plant density, pruning types.
Three different pruning strategies (vertical shoot position: VSP, mechanical pruning: MecP, minimal pruning: MinP) were tested by INRA in combination with a control of irrigation in southern France on 3 grapevine varieties (Viognier, Marselan, Petit Verdot) over 3 vintages (2013, 2014 and 2015). In 2013, irrigated and non-irrigated treatments were implemented. For 2014 and 2015, the objective was to reach a more uniform plant water status all over the experiment. To achieve this, water supply depended on the block and was monitored. In general no differences were found on harvest dates between MecP and VSP. On the contrary, significant differences in harvest date occurred between VSP and MinP with delay up to 2 or 3 weeks.
Two markers of grape aroma potential were analysed by ENTAV-IFV (PDMS and b-dam potential) and completed polyphenols analysis were also done by INRA (anthocyanins, flavanols, hydroxycinnamic acids, flavonols). Comparing variety, vintage and training system effects, on those 6 berry quality indicators, the variety was found to be the main factor of variation before vintage and the training system the third. Minimal pruning presented some advantages: lower pruning cost, higher yield, later harvest, maintaining or even increasing berries quality. The main disadvantage however was a higher water demand. Mechanical pruning showed an intermediary agronomic behavior between VSP and minimal pruning. However, no positive effect was observed on berry quality.

In a climate change context, bodegas TORRES established field trial assays all aiming at mitigating the negative effects of climate change particularly of heat waves during summer. Three plots with three different varieties were tested in order to compare the current, traditional practice with innovative practices, referring to canopy management, plant density, training and pruning systems. Measures concerning phenology, productivity and physiology of the plants, from different treatments, were collected. Microvinifications allowed to analyse the musts and wines. Climatic data and disease incidence were noted and measured.
Changing the training system from a cordon to a palmeta (figure 1) lowered the yield and led to more alcoholic wines, richer in polyphenols and more coloured.

Regarding plants density trials, the lower the density, the higher the number of bunches and the final yield were. The wines were also preferred by the tasting panels. Reducing the number of buds retained after pruning (from 12 to 8) showed also positive effects on field behavior and wines, however there was a slight decrease of yield. Considering the climatic conditions of the zone (low rain and extreme temperatures), TORRES concluded that decreasing the density of plantation of its vineyards would allow to better adapt the physiology of the vine while maintaining or increasing the quality of the wines and yield.
Finally, no shoot tipping led to a higher incidence of Botrytis compared to the control treatment, a higher vigor (pruning weight) while all the other parameters were similar.

The main goal of the VINZAVOD Asenovgrad was to test various viticulture practices in the continental vineyards of South Bulgaria (village of Pravishte) aiming at optimizing productivity and improving the wine quality. Green pruning, norming and defoliation practices were tested in combination with a conventional and an organic system (Figure 2). The experiments were done with the Mavrud variety, which is an old and emblematic variety of the area. Like in the other countries, the meteorological conditions were quite variable across years.
Green pruning with 2 spurs, one fruit cane and 8 dormant buds with norming and defoliation, gave the best quality results in both organic and conventional systems however with a lower yield. It seems however a good compromise. Indeed, at the other extreme, the highest yield was achieved by increasing the number of spurs to 4, the number of canes (2) and the number of dormant buds (24) but with a strong diminution of the quality.

Objective 3: Exploitation of genetic diversity
Here, the objective was to evaluate the interest of new clones in relation to adaptation to climate change and of new cultivars in relation with the reduction of the use of pesticides in real production conditions. In terms of results, some Tempranillo clones have been identified for their adaption to climate change by PROVEDO and RODA while VCR obtained a consistent level of resistance to both powdery and downy mildew in the tested new varieties compared to the control, reducing the need in treatments.

The objective of PROVEDO was the evaluation and selection of the Tempranillo clones showing the best quality, in three different areas of La Rioja (Spain) showing different climatic and soil conditions. For that purpose, 9 clones were selected by a clonal selection from centenary vineyards (call VP + number clone), and planted with reference clone RJ-43.
Genetic differences on Tempranillo clones shown that is possible to mitigate the negative effect of climate change, planting the best adapted clones in different areas. On the whole, the different clones did behave differently in the 3 areas for various traits related to yield and it was possible to propose a few best clones for each one (Figure 3).

The main results achieved in INNOVINE by Bodegas RODA were (i) the study of a core collection of 40 Tempranillo biotypes (clones), selected among the 532 existing in Bodegas RODA´s Genebank (created in 1999), and (ii) the pre-selection of 8 biotypes adapted to climate change. The traits that have been studied are related to the phenology (as a lever of adaptation to climate change) as well as other viticultural and oenological parameters.
Short cycle and long cycle biotypes were observed in the core collection while most of the biotypes showed an intermediate phenotype. The intra-varietal variability observed for all measured traits showed that the core-collection was well capturing the variability of the entire collection (Figure 4) and is therefore now in itself a good reservoir of Tempranillo’s variability from la Rioja.

VCR assessed characteristics and performances of six innovative varieties named Ud.30-080 (Sauvignon Maris), Ud.31-120 Ud.31-103 Ud.34-111 (Fleurtai), Ud.34-113 (Soreli) and Ud.76-026 (Sauvignon Kretos), which carry resistances against downy and powdery mildews. These varieties were obtained through conventional breeding. During the project INNOVINE, agronomical and oenological investigations have been carried out in high-fungal pressure sites of Fossalon di Grado (site 1) and Rauscedo (site 2), located in north-eastern Italy, and in a lower-fungal pressure site of Ripa teatina (site 3) in Central Italy. Resistant varieties (innovative practice) were compared with four traditional varieties (common practice), Merlot, Tocai Friulano, Sauvignon blanc (site 1 and 2) and Montepulciano (site 3), grown in the same vineyard. The potential for mildews infection was estimated using HORTA’s DSS Vite.net that predicts the number and timing of expected infections. Disease incidence was assessed by visual inspections of non-treated controls of sensitive varieties. A low-input pest management was applied to resistant varieties in sites 1 and 2; no treatments were applied to resistant varieties in site 3. The level of disease resistance has been evaluated by visual inspections and scored using OIV descriptors, according to a protocol shared with INNOVINE partners. Sensory evaluations of wines produced in the previous vintage (2015) have been carried out.
Resistant varieties displayed good-to-excellent foliar and bunch resistance to downy mildew, in the field conditions of the three experimental sites. Powdery mildew resistance was adequate for the investigated sites. Damage of powdery mildew on bunches was scored on Sauvignon Kretos (76-026) under highly conductive conditions. Resistant varieties always required fewer treatments than conventional varieties (2-3 treatments on resistant varieties, 12 to 18 treatments, on average, with peaks of 22 treatments in 2014 for conventional varieties). Environmental and human toxicity score as well as economic and environmental impact were substantially lower for resistant varieties (Figure 5). Yield of resistant varieties was comparable to conventional varieties, stable over the years (except for Sauvignon Maris that requires cross-pollination and benefits from good weather conditions during blooming). Wines from resistant varieties were generally appreciated, sometimes getting higher scores than their parentals.
All the involved actors could get their profits and advantages from resistant varieties: growers, consumers and citizens.

Objective 4: Precision vitivulture
The experiments carried out by CSIC-University La Rioja and FORCE-A, aimed at testing the utility of non-invasive technologies under field conditions in order to predict wine colour and to assess the spatial variability of total soluble solids, titrable acidity, leaf density, chlorophylle, flavonols, nitrogen status and nitrogen uptake to allow a better monitoring of the quality at harvest.
CSIC-University La Rioja was mainly involved in testing the effectiveness of tools that monitor berry maturity at the intra-vineyard level for precision viticulture. New, non-invasive sensing technologies could be applied in precision viticulture. Some applications and tools, based on computer vision techniques and spectroscopy were developed. NIR spectroscopy was used for grape composition assessment and showed its efficiency for determining total soluble solids using a spectrometer under field conditions. In parallel, computer vision systems were used to assess several features of the plant, such as the grapevine yield components. A new smartphone-based phenotyping tool was improved and successfully assesses the number of flowers per inflorescence from a low-cost RGB image analysis. Similarly, the number of berries per cluster was also assessed in the field by means of non-invasive computer vision techniques. One step forward was made and a computer vision set up was installed on a moving vehicle and used to assess on-the-go the main features of the canopy within a vineyard. Using this methodology, our results indicate that good estimations of the percentage of exposed clusters, the canopy porosity and total wood pruning weight can be obtained from information extracted from images acquired on-the-go under field conditions. All these viticultural parameters were georeferenced, enabling their use in precision viticulture. Moreover, additional several non-invasive sensors could be embedded and mounted on a vehicle or in a robot for in field high-throughput plant phenotyping, allowing the assessment of multiple viticultural features simultaneously.
In the frame of WP4, the aims of FORCE-A were to test selective harvesting at the plot level to optimise harvest for premium wine production. The available input parameters were grape anthocyanins, foliar porosity, leaf chlorophyll, leaf nitrogen.
For the first period FORCE-A worked with CSIC to study the ability of Multiplex indices to estimate the anthocyanin content of red varieties. Experiments were organised by the CSIC in La Rioja, on Tempranillo grapes in a commercial vineyard and were successfully performed.
First results were able to show the temporal variability of the Multiplex indices FERARI (increasing index during ripening) and ANTH_RG (decreasing index during ripening) related to berry anthocyanin content along 5 dates during ripening 2013. In addition to the temporal variability of the indices during ripening, the spatial variability was studied too along the same dates. This preliminary study was an introduction to larger experiments ideally conducted in commercial vineyards in order to adapt FORCE-A diagnostic models to business cases.
Concerning period 2, part of the works done in WP4 and WP5 are in interaction, as FORCE-A & ENTAV-IFV Sud Ouest work done in task 5.2. In 2014, focus was done on plot selection based on anthocyanin content with wine tasting validation. 30 commercial plots of Fer Servadou were followed along the ripening, 3 groups were discriminated (low, medium and high anthocyanin potentials) in which 3 plots were selected (per group) for micro-vinifications and wine tasting (in June 2015).
The effect of practices aiming at reducing pesticide application was also studied by GRC. The 2014 experiments focused on Riesling in the commercial vineyard of VOLLRADS with different canopy manipulations through the use of the on-the-go Multiplex. The sensor allowed to follow the canopy evolution according to 5 indices: foliar density, chlorophyll, flavonols, nitrogen and nitrogen uptake. The Mutiplex data were recorded at 11 dates (from end of June to the end of September), they were analysed and maps were provided for the 11 dates. Multi-clustering analyses were performed during vegetative growth (from flowering to véraison) to characterise the plot according to the 5 indices.
Referring to period 3, objectives were to exploit WP1 and WP5 results (on sensors, data acquisition methods and mapping algorithms) to show the operational usefulness of plot segmentation based on the association of vigor and grape mapping to improve wine quality. The plot segmentation and wine quality analysis have been done in 2015, and replicated in 2016 to study the potential impact of the vintage effect. Partners for this work were INRA (Château Couhins, Bordeaux) in two plots with Sauvignon blanc and one Merlot.
Data acquisitions have been achieved with FORCE-A proximal sensors:
• On-the-Go Multiplex sensor for the foliar nitrogen status mapping around bunch closure.
• Hand-held Multiplex sensor for the assessment of anthocyanins in Merlot bunches and of flavonols in Sauvignon Blanc bunches close to the harvest.

Four different modalities corresponding to 4 spots per plot have been selected thanks to the association of the maps of foliar and grape indices. These 4 modalities led to 4 different wines (Figure 6). Berry tasting, must analysis, wine analysis and wine tasting have been achieved for the 4 modalities.
Results are known for 2015 trials. The analyses of 2016 wines are in progress, with wine testing planned in December 2016.

TASK 4.2. & TASK 4.3.
The aim of ENTAV-IFV was to achieve with the partners of the project the design durable viticultural systems. Combining different practices in consistent systems is the best way to reach major modifications in production systems. It was planned to design innovative vineyard systems to face new climatic conditions and to optimise energy, carbon and nutrient flows that conserve natural resources and maximise production for different European pedo-climatic zones.
The task aimed to define sets of objectives and constraints (SOCs) for the main types of viticulture in Europe; design system prototypes during workshops; assess the systems in order to choose which systems will be experimented in task 4.3. The SOCs were defined during the first period. The SOCs already established for France were translated in English, and completed. Some more were established for Spain, Italy, Germany and Bulgaria.
The second objective, was based on some of these SOCs, to design new prototypes of viticulture systems. A first workshop took place during the second annual meeting in Geisenheim, and two system prototypes were designed:
• Continental vineyard, Rhine valley, Riesling production.
• Mediterranean Vineyard, High hydric stress conditions. This prototype was refined with the local team involved (INRA Pech-Rouge).
During the third annual meeting in Plovdiv, two different workshops were organized, in order to answer the following questions:
• What benefit could we expect in using on the same plot several new technologies (sensors / image treatment / DSS / ...)?
• What system can we build around a resistant variety?
Also, another workshop took place in Sophia in order to design of a system adapted to Bulgarian conditions.

Referring to deviations from the planned work schedule: Systems produced are not precise enough to start assessing and experimenting them. Refining them would need more work in the future. We underestimated the time needed to get enough results from the other WP, and to implement them in systems. Furthermore, the workshops were very fruitful, but rise many questions. We thought it was more relevant to perform some ad-hoc experiments before experimenting the systems. The tasks 4.2.3 and 4.3 were therefore not performed during the project.

TASK 4.4.
Different protocols were defined by HORTA to collect data necessary to evaluate the management options compared. These protocols were adjusted to the target in terms of agronomical assessment and to regular grower’s management practices. The objective of Task 4.4 was to assess in terms of environmental and socio-economic impacts, the innovative practices designed within the project and tested in experimental trials within Task 4.1.
To do so, at the beginning of the project, different documents were prepared: i) a common protocol for experimental design and data collection (Deliverable 4.1) to ensure homogeneity among partners testing INNOVINE management options in experimental trials within WP4 and allow statistical significance when comparing the innovative practices to the current (farm) technique; ii) a file excel for the collection of all input data necessary to perform the cost/benefit analysis and to calculate the environmental impacts of the novel vineyard management strategies tested in WP4; and ii) a web survey to draw a map of the viticulture management system presently in use, in order to better identify the critical points and the priority axes of improvement

The files were circulated to all WP4 partners and data were collected for the grapevine growing seasons 2013-2016. Data collected were inserted into the DSS vite.net through the functionality “Field book” and, thanks to the new algorithms for the environmental indicators developed and implemented into the system (respectively in Task 5.2 and 5.4) were used to successfully calculate the indicators values of 6 case studies that represent the different innovations applied in the WP4 trials. The type of innovation that mainly reduced both economic and environmental impacts was the use of resistant variety, followed by the use of new clones with high yield production. The other innovations did not affect significantly the environmental impacts.
Raw data about the measurements performed in all the experimental trials were instead used to perform a global statistical analysis with the aim to compare common practice to the innovation tested. For each trial the % differences between the innovative practice and the common one (Δinn) was calculated and exploratory statistical analysis on data distribution and calculation of summary statistics performed. Finally, the significance level of the differences between the two distributions (innovative and common), as well as their normality, were tested by using non-parametric tests (Mann-Whitney U and Kolmogorov-Smirnov Z) and the Shapiro-Wilk Test, respectively. The results allowed to identify the innovation types tested within the INNOVINE project that had some effect on vine yield and those that impacted must and wine characteristics.

The web survey prepared was translated in 5 languages and sent to thousands of stakeholders. The sample of respondents to the web survey (174 valid answers) well represented the vine growing production structure of Europe. Very interesting indications were obtained, such as number and kind of plant protection products applications usually performed, and it was confirmed that treatments against downy and powdery mildew are the most common actions performed. The relevance/interest of the themes addressed by INNOVINE project was also investigated and the respondent rated highly interesting the following themes: “Effects of innovative vineyard management practices on berry composition”, “Crop protection techniques to replace/reduce pesticides”, and “Efficiency and durability of grape resistance to major fungal diseases”. Finally, the survey allowed to create a contact with a large number of wineries, in the most relevant European wine regions, interested in technologies studied in the project.

WP5 – Implementation of decision support systems towards a sustainable viticulture
The general objective of WP5 was to develop decision support systems (DSSs) for sustainable viticulture, able to support European vine growers in taking optimal decisions about: (i) management of abiotic stresses (e.g. frost injuries, water stress); (ii) control of pests and disease; and (iii) management of vine canopy.
The approach used to reach this objective was based on a logical sequence of activities: in particular in task 5.1 and 5.2 monitoring and data collection devices, as well as DSS for vineyard management were developed; these new devices and functionalities were then tested (task 5.3) implemented (task 5.4) and finally evaluated by end users (task 5.5).
Partners involved in WP5 worked on different tools/DSSs/vineyard management aspects, but following this same approach.

HORTA and UCSC
An extensive review on palm top devices available and having the potential to be connected to the DSS vite.net® was performed in such a way to analyse their pro and con. Based on the review evaluation and the experience gained within other EU projects, it was decided to develop a stand-alone scouting module that can be used from either the personal computer through the DSS vite.net® or portable devices (through a web application). The new scouting activities module was designed and implemented following the rational scheme:
1. Definition of a scouting plan (step by step selection of the observations to be performed, i.e. diseases, insects, plant, abiotic stresses, etc.)
2. Selection of the crop units created in vite.net® for which the scouting plan is activated
3. Input of scouting data
4. Visualization of scouting data
Moreover, during the project HORTA and UCSC developed new and improved already existing functionalities to be implemented in the web-based DSS Vite.net®:
• Model for predicting grapevine bud infection by powdery mildew (UCSC): starting from the date of first unfolded leaf, the model uses daily mean temperature, rainfall, mean air humidity and leaf wetness duration as input data, and provides the cumulate daily infection rate by powdery mildew for each developing bud as output. The model was run using weather data and vine phenology of different locations in Italy during grapevine growing season 2014 and 2015 and finally implemented in the DSS vite.net with specific designed outputs (colours from green to red are used to indicate low to high infection rate values) that shows the “Infection rate of the first 10 buds” and the “Seasonal dynamic of the infection rate”.
• Model for estimation of the optimal dose of fungicides (HORTA): once defined canopy dimension and vigour of the canopy at full flowering stage, the model calculates the daily percentage of filling of this virtual parallelepiped, combines it with the disease risk in that particular day and the amount of rain forecasted in the following days and provides the optimal dose to be applied. The model was implemented into the DSS vite.net at the beginning of 2014 and was tested by selected users by comparing the disease management based on i) the use of phytosanitary bulletins (public advices), ii) the use of vite.net to schedule treatment at label dose, and iii) the use of vite.net to schedule treatment at the dose suggested by the model for optimal dose.
• Modeling of leaf development in different grapevine training systems and Model for early detection of drought stress (UCSC): tracking previous knowledge gained during the MoDeM_IVM project (EU-FP7, Research for SMEs, GA n° 262 059) a model estimating plant growth validated over traditional vertically shoot positioned (VSP) systems was extended to a single hire wire (SHW) trellis having free vegetation and no foliage wires for supporting the growing shoots. Moreover, the algorithm for water stress simulation, based on the resolution of the continuity equation applied to the monolayer reservoir given by the soil layer explored by roots, developed in the same previous project, was improved in such a way to connect it with the model for leaf development in the different grapevine varieties and training systems.
• A set of sustainability indicators concerning vineyard management practices were developed with the final purpose of assessing impacts of the grape production. The selected environmental issues to be addressed by these indicators refer to: Human Health, Air, Soil, Water, Energy and Biodiversity aspects. In total 20 indicators referring to the different activities at the individual vineyard scale were developed. When implementing the indicators onto vite.net efforts were made to make them easy and clear to understand by end-users: only few (as minimum as possible) and easy-to-find input data are requested to calculate them and clear-to-understand outputs were designed. To do so, different databases and a rating system were developed for each indicator, ranging from 0 (no impact) to 5 (very high impact). The indicators were tested by using data collected in experimental trials carried out within task 4.1 for three grapevine growing seasons.
Finally, in order to allow the evaluation of the DSS Vite.net by experts in viticulture, weather stations in very diverse grape-growing areas around Europe were connected during the project to the system and access and training was given to growers, advisors (both private and public), weather station providers, and researchers. A questionnaire was prepared to gather the interest and suggestions by these experts on the usefulness and user-friendliness of the DSS.

ENTAV-IFV
ENTAV-IFV aimed at identifying iso-susceptibility zoning of a vineyard for the elaboration of an Optimized Treatment Plan® (OTP). It was based on the precision viticulture techniques currently available. Several information layers collected on vineyards were used to set a database showing interactions between elements that can explain epidemic diseases on a vineyard. The final purpose of the OTP is to reduce the use of chemical inputs taking into account the high disease variation at plot scale. Such a tool should also take into account several diseases and propose a global posology varying at the intra-plot scale. Considering that treatment decision is taken at the scale of the winery, a DSS proposing to adjust the spraying dose would have to be built at this scale.
Experiments were conducted on a 100 ha vineyard in Bordeaux area. The first step was to design the best zoning of a vineyard, reflecting the occurrence of fungal diseases (downy mildew, powdery mildew, black-rot). This has been done by crossing a data layer representing the structure of the soil (by using a soil resistivity sensor) and another one the vigor of the plant (by using remote sensing NDVI). With an appropriate classification, this resulted in a new combined data layer showing an a-priori zoning composed of 6 classes. These 6 different classes were called PBU (Physiological Behavior Unit).
A disease and phenology monitoring was conducted every weeks during 3 growing seasons on 36 untreated plots each composed of 3 contiguous plants and well representing the PBU classes. These plots remained untreated throughout the growing season, by putting a plastic tarpaulin on vegetation during the spraying activity.
A statistical study was conducted. The spatial distribution of diseases showed different levels of attack with different outbreaks locations depending on diseases and the years. The variance analysis showed that the PBU explained the variations of disease incidence in a varying extent, depending on the disease and the year. Soil Factor or Plant Factor can be predominant depending on the years, diseases and the organ (leaves or bunches). The whole results showed that although the PBU concept is a relevant reading key to explain variability at a spatial level and an intra-plot modulation of spraying would be useful, there was no consistent effect neither within a season nor from one year to another with the current method of zoning.
The aim is now to test the use of different factors (age, cultivar, climate ...) for strengthening the proposed zoning.
In parallel, an equation was developed to calculate the amount for fungicide dose cep by cep. As soon as improved, the zone and the vine by vine dose method of calculation will be integrated into the PTO® (Optimized Treatment Plan) DSS, able to pick up information from the Epicure system (disease risk modeling) and from Optidose (its dose reduction module) to assist GPS-guided spraying programs in vineyards.

NOVELTIS
At the beginning of the project, NOVELTIS investigated what aerial and satellite images would be useful for characterizing the vineyard, by satisfying both the end-user needs and the technical requirements (in terms of quality, spatial resolution, spectral resolution, temporal resolution...) of the processing chain that had to be developed in the project. Suppliers for both satellite and aerial imagery (including UAVs) were identified. The pre-processing chains for three kinds of images (satellite, aerial and UAVs) were identified and their development was initiated. These pre-processing chains included image-to-image co-registration, geo-referencing, radiometric correction and atmospheric correction and ensured the best geometric and radiometric quality of the data. Moreover, NOVELTIS interacted with the end-users involved in the project (SCHLOSS, VINZAVOD and TORRES) and with other potential European end-users to analyze their needs regarding precision viticulture using aerial images and to determine test sites.
NOVELTIS has also consolidated several processing chains (extraction of the plots, analysis of the plot characteristics, formatting of the outputs) that characterize the vineyard by using remotely sensed imagery. The consolidation was mainly related to the flexibility of the processing chain in terms of spatial resolution and number of spectral bands. Regarding the precision viticulture needs, NOVELTIS focused on the use of UAV images because of their very high resolution. The pre-processing chains for UAV images have been developed and the processing chains improved to face artefacts caused by very high resolution. This has allowed to develop and implement a new method for the missing plot detection. New functionalities have been added such as computing each grape variety separately, generating vigor maps of the inter rows or classifying the vigor maps of the vine and the vigor maps of the inter rows to define homogeneous areas.
NOVELTIS has tested the developed pre-processing and processing chains on three different vineyards. The first one, Herdade Do Esporão, is located near Lisbon, Portugal and was conducted in cooperation with AGRICIENCIA. The tests have been conducted in 2005, 2007 and 2010 using aerial images at 50 cm of resolution. This test allowed to validate, thanks to the wine grower feedbacks, the existing processing chain using aerial images. Another test was conducted using UAV data at Leoville Las Cases (Bordeaux area, France) vineyard in cooperation with IFV. This test enabled to validate the new method for missing plot detection. The third test was conducted over IFV vineyards in the South West of France. The campaign was led in 2015, 2016 and 2017, and evolutions of the different functionalities and outputs was discussed with the wine-growers.
During the project, NOVELTIS asked the end-users which support would be the more convenient to disseminate the products as a Decision Support System: KML files, web mapping interface, mobile app... As a result of this investigation, a web mapping interface accessible 24/7 and password protected has been developed. The DSS is now accessible online at the following address: http://INNOVINE.noveltis.fr. The DSS allows users to upload data and to ask for processing them. It is also possible to visualize the following outputs per vineyard and per date: vineyard boundaries, rows, missing crops, vigor maps, base images and synthesis of the main outputs of the vineyard. The developed DSS has been presented and validated with different end users, and a demonstration version has been provided to several vineyards. The DSS has been updated following the feedbacks of the users (contains, web usability...) and is today ready for operational exploitation.

FORCE-A
Within INNOVINE WP5, FORCE-A successfully fulfilled the objectives of developing or improving:
(i) its data collection devices, in particular the On-the-Go Multiplex sensor and the associated FA-Box electronic box for foliar vigor mapping, and the Dualex leaf-clip with calibration for optimum vigor assessment,
(ii) its data processing and display functionalities, including the mapping algorithms and the FA-Server web platform, in order to accelerate and simplify the process of data acquisition, processing and display, and their applications to the four following topics with several partners as indicated in table 1.

The main improvements on the devices functionalities are the following:
• Proximal On-the-Go Multiplex sensor and FA-Box: easy mounting and improved real-time data filtering,
• Dualex leaf-clip: validated thresholds for optimum vigor at bunch closure, for the calibration of Multiplex vigor maps,
• Mapping processing algorithms:
o Automated generation of maps using the shapefiles,
o Automatization of zoning along the rows and customized grid display
o Foliar diagnostic decision matrix: coupling of vigor (NBI) maps and foliar density maps
o “Premium harvest” diagnosis map: coupling of foliar maps and anthocyanin maps
• FA-Server web platform with upgraded functionalities: dynamic threshold for anthocyanin zoning and classification, display of GPS points, rainbow maps, diagnostic maps.
Moreover, the project allowed determinant progress in the demonstration of early detection of downy mildew in the field with the fluorescence-based Multiplex 330. The whole combination of sensors and web platform has been operated for selective harvesting on Sauvignon Blanc and Merlot plots at Château Couhins in 2015 and 2016 in the scope of WP4.
Thanks to these results, FORCE-A organized their grapevine services into three product lines: FA-vigor, FA-vendange, FA-wood. Each of them is operated at a different time of the year and grapevine phenological stage in order to help grape growers to optimize their plot vigor and harvest.

AGRICIENCIA
Aiming at improving the system for collecting data at farm level, during the first year of the project, several innovative devices monitoring the vineyard environment (weather and soil) and the plant status were installed in a commercial vineyard that was located in the estate “Herdade do Esporão” (Alentejo winegrowing Region, Portugal). Within the same experimental plots used by ISA and ITQB partners for WP1, four “islands” of sensors were installed, two at the variety Touriga Nacional plot and two at the variety Aragonez plot, one island per water regime (RDI – Regulated Deficit Irrigation and SDI – Sustained Deficit Irrigation), as used in WP1 by ISA and ITQB partners. Each “island” was equipped with an autonomous system using solar panel and battery and consisted of the following sensors: 1 soil moisture sensor with 3 capacitance probes (20, 40 and 60 cm depth) 1 dendrometer (trunk diameter variations), 1 sap flow sensor, 1 humidity sensor (leaf wetness duration) and 9 thermocouples (3 for leaf and 6 for berry temperature) located in different canopies positions. Climate data (global radiation, rainfall, wind speed, air temperature, relative humidity and dew point temperature) was collected every 30 minutes from a meteorological station located within the same vineyard. The data were collected during consecutive seasons along the time of the project (2013-16) using an advanced data logger system and GSM for data communication with the central server. A web service was developed to collect data from vineyard´s weather station and to integrate the data into the warehouse that supports the dashboard accessed via web browser. The data was transferred automatically and inserted in the database via executable procedure that was scheduled in the central server. To facilitate the interpretation of these large volumes of data, a Business Intelligence approach was developed for the manipulation and delivery of useful information. The data was extracted from the sensors, loaded, stored and processed in a database (data warehouse) and finally delivered in a digital dashboard, accessed via web browser that was created to visualize and manipulate the collected data. This digital dashboard is a real-time user interface, easy to read, showing a graphical presentation of the status (snapshot) and historical trends enabling instantaneous and informed decisions to be made at a glance.
The test and validation of those devices and functionalities were done in cooperation with the technicians from Esporão estate and with the ISA and ITQB INNOVINE Portuguese research teams. Furthermore, in cooperation with partner HORTA, the data collected by the Esporão weather station was connected to the DSS developed by HORTA (vite.net ®), to test the functionalities with Herdade do Esporão weather data.
During the third season of the project (2015), the technicians from the end-user Esporão have used the dashboard for irrigation management and have tested successfully the use of soil humidity data to set irrigation starting and scheduling on the vineyard plots contiguous to the experimental plots. Finally, an alert system for smartphone and smartwatch with possibility to set up berry and leaf temperature thresholds were developped to be used as alerts for deficit irrigation management during heat events, along with the monitoring of real time values.
In collaboration with other Portuguese partners (ISA and ITQB) some dissemination activities were performed (oral presentation and proceeding in scientific events) and relationships between sensor data and physiological data collected on the same vineyard were analyzed.

IRSTEA
IRSTEA has designed its first prototype of SmartGrappe just at the launch of INNOVINE project. This prototype uses a palm top device to help growers to follow the maturity and other growing stages of the berries. The initial prototype consisted in a plastic support in which the mobile device used for taking pictures of the berries is inserted. A first version running under android OS was developed at first. However, to facilitate the software development, this solution was abandoned for a more standard solution on PC computer. The idea was to test the performance of the device on a large collection of images and extend the functionalities to INNOVINE project demands. Indeed, IRSTEA’s long experience in such development showed the difficulties to run efficient algorithms without a long testing period in a large variety of situations.
During the project IRSTEA improved the design of the support to allow the insertion of any commercial mobile smartphones. This new design was necessary to enable a large testing panel and used to improve the algorithm. This was achieved through sequential activities:
• Sharing and testing the device in the field
• Collecting a maximum of images to improve algorithms
• Trying to respond to new demands coming from the project partners
• Changing and adapting the algorithms to these demands
• Repeating again the process until getting a stable result
Moreover, during the project IRSTEA has developed new functionalities and improved the existing ones. These functionalities were implemented and tested on the SmartGrappe device.

Automatic berries detection algorithm:
The challenge was to develop a robust algorithm able to locate each berry on each image, regardless the grabbing condition and quality of the image. This algorithm was hard to tune due to numerous unexpected cases collected during the life of the project. This algorithm was tested on a very large image data base produced by project partners (IFV, INRA) and external winery companies like Moët & Chandon, Fruitions Sciences or CIVC. Indeed, during the project more than 8000 images were collected, but not all of them where tested. The final device was tested randomly on more than 1000 images without any bugs.

Artificial intelligence embedded:
To enhance the performance of software, an intelligent classifier capable to discriminate berry from non-berry after the detection was embedded. This algorithm uses a semi-supervised approach to learn the discrimination. The final result proved better in robustness and performance, but the final program also grew in complexity, with currently a 3s speed treatment per image on a standard dual-core computer.

Finally, in order to let professionals in viticulture evaluate the SmartGrappe device, 15 prototypes were produced and used to collected and share information with growers, advisors and researchers. IRSTEA has also engaged a technical joint with a small French company for a commercial exploitation and is discussing the license fees.
Potential Impact:
The INNOVINE project aimed at developing knowledge, tools and genetic resources necessary to better adapt viticulture to climate change and to drastically reduce the use of pesticides in the vineyards. The aim of the project was to disseminate its findings to three different categories of end users: academics, technical advisors/service providers and growers. Most of these activities are described in details in the deliverables D6.8 D6.9 and D6.10 and are summarized below, highlighting the potential impacts, under three main topics:
• Adaptation to climate change
• Reducing chemical application in the vineyard
• Assisting growers for adapting their practices and viticulture systems towards more sustainability

1.4.1 Adaptation to climate change
1.4.1.1 Potential impacts on future researches in the field
INNOVINE globally led to a better understanding of the impact of vineyard practices and of various stresses alone or, which is newer, in combination, on grapevine physiology and berry composition in the context of climate change. The functional-structural grapevine models, "HydroShoot" and "GrapevineXL", allow to simulate the impact of environmental stimuli and viticultural practices on grapevine functions and berry composition. They will be further developed and integrated to exploit genetic diversity and the response of virtual genotypes to different climatic scenarios. This will provide guidance for selection and breeding of new grapevine genotypes better adapted to future climate. The standardized sampling and analytical procedures (see also paragraph 1.4.1.2 below) that have been designed to collect the data necessary for the validation of the models and made publicly available will allow to collect in the long term big data set on different varieties grown under different climatic environment and viticultural practices during many vintages, which would enable in the future understanding and manipulating the most relevant parameters controlling berry composition. Methods for screening germplasm for plasticity or for identifying key molecular pathways of adaptation to stress have been proposed. Finally, long term experiments and platforms (e.g. FACE experiment, Vitadapt experiments, transcriptome data) will provide a support for future researches on the impact of climate change related environmental conditions on grapevine physiology and berry composition. The partners were very active in term of publications in peer reviewed journals (36 to date), dissemination in scientific conferences or lectures (63 presentations and posters) and outreach to growers, stakeholders or larger audiences (13 presentations or events). Several reviews of the current knowledge of different aspects of the impact of climatic factors on grapevine and berries have also been published by the partners of the project (Carvalho et al 2014; Costa et al 2016; Zarrouck et al 2016a), making the whole consortium quite visible in the field. Some of the main results that can be further exploited are highlighted below.
Two to three years of data on grapevines under canopy management that altered the source/sink balance with the aim to cope with climate change have been obtained. These studies showed the diversity of responses to similar stresses depending on grapevine varieties (Poni et al 2014, Merli et al 2014, Martinez-Lüscher et al 2015, Gatti et al 2016), but also of effects on berry ripening and final composition (Genebra et al 2014, Gatti et al 2014, Martinez-Lüscher et al 2015, Gatti et al 2016).
Temperature was shown to affect the colour of white berries through photo-oxidation mechanisms (Friedel et al 2015). Deficit irrigation significantly affects berry ripening, anthocyanin accumulation and hormonal balance, while temperature is an important variable determining the improvement (moderate temperatures) or impairment (high temperatures) of berry quality by the deficit irrigation regime (Permanhani et al 2016, Chaves et al 2016) and underlying physiological and molecular mechanisms are starting to be deciphered (Texeira et al 2013, Conde et al 2014, Genebra et al 2014, Noronha et al 2014, Fernandes et al 2015, Zarrouck et al 2015, dal Santo et al 2016, Zarrouck et al 2016b, 2016c, Jardal Jamoussi et al 2016, Chavez et al 2016; Martinez-Lüscher et al., 2016).
Interesting links between stresses (namely elevated CO2, elevated temperature and water deficit) and yield factors were found in Tempranillo. Although variation of fruit set was found correlated with pollen viability, yield was also influenced by the number of flowers per cluster indicating an ambient-dependent response of flower-setting and gamete success in this cultivar. The whole results obtained in INNOVINE project suggest that reproductive dysfunction in Tempranillo blanco compared to Tempranillo nero results in looser clusters and lower yield. This genetic-driven dysfunction is more highly expressed under environmental conditions that are unfavorable for gametophyte development and fertilization. The genetic dysfunction of TB that affects gamete viability very likely originated from a chromothripsis event which provides the first known example in plants (WP1 unpublished results).
A first attempt to combine experimental results and mathematical modeling by UCSC, Univ. Bordeaux and INRA allowed to better understand the physiological process underlying experimental observations. A study on late-season leaf removal was conducted on cv. Sangiovese potted grapevines that were subjected pre-veraison to a limiting vs. non limiting crop load by leaving 3 and 12 primary leaves per shoot. Carbon limitation led to a strong imbalance between sugars and anthocyanin and affected anthocyanin profiles. Mathematical analysis showed that under carbon-limited conditions, the grape berry can manage the metabolic fate of carbon in such a way that sugar accumulation is maintained at the expense of secondary metabolites (Bobeica et al 2015). INNOVINE made an important contribution in terms of new validated models through the combination of modelling approach and experiments and calibration of the models with different varieties. Future extension of this work will help to better understand and predict the effect of environmental stresses or viticulture practices on different cultivars.
Indeed, two functional-structural grapevine models, respectively HydroShoot and GrapevineXL, were developed to simulate the impact of environmental stimuli on grapevine functions. By varying the hydraulic conductance along the pathway (e.g. soil, root and leaves) and the sensitivity of stomatal conductance to ABA and leaf water potential, the model can produce virtual genotypes with different water use behaviours (iso- and anisohydric). Model simulation showed the effects of water stress and light on berry growth and sugar concentration. The models will be made available for public access through the GroIMP and OpenAlea platforms.
Grapevine adaptation to environmental conditions in relation to climate change requires the identification of genetic variation suitable for its genetic improvement: genetic resources carrying adaptive traits, markers allowing to follow favorable alleles in breeding programs, better understanding of these complex traits and targeted screening methods. INNOVINE has contributed to the improvement of such a tool box in several ways (see also paragraph 1.4.1.2).
A method was proposed for the screening of a panel of V. vinifera cultivated varieties, maximized for its diversity, for plasticity of traits related to yield (quantity and quality) under drought stress. A first list of 20 varieties which berry traits seem poorly affected by drought (or fairly stable even under dry conditions) was obtained. This ranking needs to be confirmed by additional experiments but the method could be re-used for a screening for plasticity for any trait.
Genes and gene variants that could be used as markers related to stress tolerance in grapevine were searched for, specifically targeting drought and temperature stresses that are supposed to increase with climate change. To identify these markers and genes, two approaches were proposed. In the first one, the transcriptome of grapevine cultivars having different strategies to cope with water stress was analysed and the results allowed the identification of several drought specific responses in cultivars that behave as isohydric and anisohydric as well as the selection of specific genes involved in those responses (Dal Santo et al, 2016). The second strategy exploited transcriptomics analyses to try identifying those genes and gene variants responsible for intra-cultivar variation in Tempranillo (syn. Aragonez) clones selected for variation in surface leaf temperature (SLT) and the rate of berry ripening. It was possible to identify somatic genetic variation for the two traits analyzed in collections of Tempranillo / Aragonez accessions. This variation could be useful in the improvement of the ability of cultivar Tempranillo to stand stress conditions and extend the ripening cycle to better adapt to climate change. Altogether these results improve our understanding of stress responses in grapevine and provide clues for its genetic improvement of stress tolerance.

1.4.1.2 Potential impact on a better monitoring of yield and quality in viticulture in relation with climate change
Several non-destructive phenotyping tools based on fluorescence, reflectance, thermal imaging, hyperspectral imaging have been experimented and validated in WP1, WP4 and WP5 to monitor the physiological status of the canopy (nutritional and water status), as well as the berry content (sugars, anthocyanins, chlorophyll, carotenoids) or the onset of downy mildew attacks. These tools may be used either as hand-held devices or mounted on mobile equipments (quad, tractors). The data collected can be used to draw high resolution maps of the vineyard and result in decision making tools (such as selective harvesting) or to assess the sanitary status of the vineyard and drive viticulture practices/treatments that would limit the spread of the disease. They can also be used as screening tools to investigate clonal or varietal diversity. While these approaches still need to be made more publicized and probably adapted in a more user-friendly way, they offer a strong potential for dissemination. Indeed, many different actions of dissemination have been carried out during the course of the INNOVINE project: publications in peer reviewed journals (10), participation to scientific conferences, workshop or fairs (38) and a lot of efforts given to the organization of testing days, demonstration in the field or master courses (16). Highlights of the main outputs in this domains are given below.

Different tools were developed in order to monitor stresses in the field (see deliverable D5.1). For instance, Multiplex 330 sensor was able to monitor the DM infection time course from 5-6 days after inoculation (Latouche et al 2015). Thermal imaging measurements of canopy were correlated with leaf water potential, stem water potential and stomatal conductance, suggesting that thermal imaging can be feasible to assess crop conditions and may contribute to speed up field measurements (Costa et al 2015). A rapid, non-invasive, and low-cost method based on reflectance spectroscopy was developed to assess berry sunburn symptoms. This new method could support grapevine variety characterization with respect to sunburn susceptibility (Rustioni et al 2014&2015) as well as study of the physiological processes involved in the symptoms’ appearance through the quantification of chlorophyll a, chlorophyll b and total carotenoids berry skin content (Rocchi et al 2016). The technique also gave promising results to screen for stem traits related to water conductance (Rustioni et al 2016, Grossi 2016).
At the canopy level, the fluorescence based Multiplex measurements (Leaf Cholorophyll, leaf flavonols, Nitrogen Balance Index) are sensitive indicators of the physiological status of the plant and of the soil heterogeneity, both in terms of intra- and inter-plot comparisons. At the berry level (Lafontaine et al 2015), comparison of Multiplex data on berries and chemical analysis of berry content showed a very good correlation between the two types of results and allowed to precise the reliable range of utilisation. Grape berry content was also assessed with a destructive but quick method based on FTIR on a range of 50 different varieties. Results obtained by FTIR were highly correlated with reference analyses for Brix, total acidity and malate (R² values > 0.97). Based on such results, FORCE-A improved the acquisition system for phenotype selection, developed a new FA-BOX for foliar management that is able to collect automatically all data and improved the firmware for berry selection (Cerovic et al 2014). The Multiplex firmware for anthocyanin accumulation/plot selection and anthocyanin mapping developed in the previous period was further improved and an internet-based software application for berry anthocyanins units’ conversion was developed to express the anthocyanin content according to different units. Various aspects regarding the calibration and implementation of predictive models of total soluble solids (TSS) in grape berries using laboratory and in-field collected NIR spectra were developed (Urraca et al 2015).
Several other tools have been developed or improved during the INNOVINE project and have been transferred to researchers and extension services for further test and validation. ICVV, University of La Rioja developed a new, easy-to-use smartphone application for the automated estimation in the vineyard of the number of flowers in grapevine inflorescences. The application, called VitisFlower®, is offering a user-friendly interface and should be a valuable tool contributing to yield forecast. In parallel, IRSTEA developed a smartphone based tool called SmartGrappe that allows to determine the size, number, colour and surface defect of berries in the vineyard. It can be used for experiments and IRSTEA is in discussion with a French company for commercialization. The data collected with these applications could be integrated in larger databases to get a complete picture of the vineyard status, yield and berry composition in response to terroir, environment, viticultural practices, and disease incidence but these databases remain to be identified.
Finally, other tools allow to acquire and exploit geo-located measures provided by a series of sensors or aerial imaging to generate zones in the vineyard for differential practices or harvest are a predictable output of the project (Debord et al 2014, WINEO software developed by NOVELTIS). The exploitation of data from various sensors combined with geographical informa¬tion systems to generate and cross different layers of measures corresponding to different variables of the production system and to elaborate a priori zones within the vineyards, likely to explain variations of the physio¬logical development of vines and possible differences of the plants susceptibility to fungus diseases.

1.4.1.3 Potential impacts on adapting vineyard practices to climate change
Different adaptive strategies were tested in diversified environmental conditions and provide evidence that seasonal canopy management practices and varietal or clonal diversification give opportunities to adapt to weather conditions or in a long term to climate change:
• Sets of Tempranillo (syn Aragoñez) clones better adapted to various environment that can already be proposed for planting (see deliverable D3.1)
• A method for testing for plasticity of the varieties with regard to environmental stress that could be implemented in the future for further screening (see deliverable D3.2)
• A better understanding of the effect of various practices aiming at mitigating environmental stresses on berry quality (see deliverables D4.4 and D4.5)
• Various monitoring tools that should facilitate the following of the physiological state of the vine in the field and DSS (see deliverables D5.1 and D5.4)
All these results have now to be combined for the design of new viticulture systems adapted to the constraints of each given « terroir » as it was sketched during the workshops aiming at prototyping innovative systems (see deliverable D4.3). Results have been disseminated in conferences and workshops by the partners (9 presentations and posters) and field demonstrations or wine tasting (6). They are still being disseminated across European vineyards and some of them are already tested in private wineries. Highlights of the main transferable results are given below.

Using the anti-transpirant Vapor Gard (VG) coating either at post-véraison or on a whole-season basis (PFPV) was quite an effective way of delaying sugar accumulation with no significant detriment to accumulation of anthocyanins, regardless of specific seasonal weather.
Leaf removal, either pre-flowering or pre-véraison, can have a significant impact on yield performance and must composition. Pre-bloom defoliation is a canopy management practice that has high potential to replace cluster thinning in order to control grape yield in high-yielding varieties with benefits for berry health. Yet the carry-over effects on node fruitfulness commends a cautionary use of this practice in non-irrigated Mediterranean vineyards because the lack of leaf area compensation might induce an excessive yield reduction. Different canopy management practices give the opportunities to reduce the ‘velocity’ of berry maturation. Of utmost interest is the new and fully mechanisable technique of apical-to-cluster late leaf removal, where some functional leaves are removed from the top of the canopy to slow down the sugaring process at no detriment of phenolic maturity since the microclimate around the fruit zone is not modified. Decreasing the density of plantation of vineyards allowed to better adapt the physiology of the vine while maintaining or increasing the quality of the wines in dry and very hot climatic conditions and reducing the number of buds conserved after pruning had positive effects on yield, quality and plant vigor.
Precision viticulture has been enabled or improved by INNOVINE’s findings. FORCE-A and CSIC-University la Rioja have started to deliver georeferenced services for selective harvest and disease prediction, The WINEO application of NOVELTIS that allows to develop georeferenced vineyard maps for identifing missing plants, mapping zones with differential vigour, etc...Many on-the-go monitoring applications described in the former paragraph can be tested now for precision viticulture applications.
Tempranillo clones (syn. Aragonêz) were screened for adaption to abiotic stress conditions to obtain a list of tolerant genotypes. Considering together data from all studied seasons, clones 56, 232, 807 and 1084 were selected to study the molecular basis of the ripening delay independently of yield as well as their presumable adaption to high ambient temperatures. As contrasting material, clones 1048 and 1089 might be good candidates to search for yield-independent genetic causes of hastened ripening. These results will guide future planting strategies.

1.4.2 Reducing chemical application in the vineyard
1.4.2.1 Potential impacts on future researches in the field
INNOVINE project has contributed to make researchers from three different disciplines (genetics, physiology and pathology) working together towards the same goal: provide foreground for the development of strategies to sustainable control foliar diseases in the vineyards (i) more durable over long periods of time, (ii) allowing to diminish drastically the amount of pesticides used in the vineyards and (iii) still allowing economical sustainability. The most important lever for the reduction of pesticides was found to be the use of resistant varieties. A very important effort has been carried out in INNOVINE project for the screening of yet uncharacterized germplasm collections for resistance to powdery and downy mildews and to black rot, a secondary pathogen that becomes a real threat when the treatments against the mildews are decreased. As a result, an important list of very useful genetic resources for breeding for resistance has been delivered and will be available in the European Vitis Database (www.eu-vitis.eu). That includes information about the repository from which repository the grapevines of interest can be obtained for further study. In addition, the description of the notation scale for black rot tolerance, not yet present in the international descriptor lists, will be submitted to the OIV (Organisation Internationale de la Vigne et du Vin). The standardized protocols for screening mildews, black rot and phylloxera developed in the project will be used for future screening of grapevine material.
However, the partners have also shown that without any corrective measures, the populations of downy and powdery mildews will likely be able to slowly adapt on resistant varieties and overcome these resistances. Secondary diseases and durability of resistance will have to be taken into account by the breeding programs for new grapevine varieties. The genotypic data obtained by the partners will be deposited in databases and, for some of them, will facilitate the pyramiding of resistances in new varieties.
Another important development of the INNOVINE project was the detailed study of epidemiological traits on a set of varieties susceptible and tolerant to downy mildew and the use of these data to improve the current disease models. Finally, the partners of INNOVINE showed that canopy management practices impacting berry size have an impact on botrytis incidence. The partners have already started to disseminate their results in peer reviewed journals (9 to date) and to communicate in conferences and fairs (15 to date). Highlights of the main results are summarized below.

INRA and PTE collected 131 P. viticola isolates from partially-resistant varieties planted in French and Hungarian experimental vineyards and the collection is stored at INRA in liquid nitrogen. The controlled cross inoculation of susceptible and tolerant grapevine varieties with virulent and non-virulent downy mildew isolates was undertaken at INRA (Delmas et al 2016) and showed a slow, yet significant erosion of resistance.
Studies from UCSC, ITQB and GRC showed that canopy management, and especially leaf removal carried out at specific severities and timings, can help to reduce the incidence of botrytis mainly through an effect on cluster morphology, skin/berry ratio and wax content of the skin (Molitor et al 2015; see also deliverable D2.1).
Disease models have also been improved (see deliverable D2.4) by calibrating them for resistant varieties (PhD work supervised by UCSC), modelling the effect of disease on yield (unpublished work of UCSC). The modelling work is under achievement and should be one of the major outputs of INNOVINE both to improve predictions of epidemics for scientific purposes and for DSS to help advisors and growers.
An impressive screening (yet partly published) for resistance to mildews of eastern European/middle eastern grapevine varieties and wild Vitis vinifera accessions has been achieved in INNOVINE (deliverable D3.4): 1179 and 883 non redundant accessions respectively tested for resistance to downy and powdery mildew. This work results in a better knowledge of resistance to these diseases both in the wild and cultivated compartment of Vitis vinifera, opening new questions about the co-evolution between the plant and its pathogens. The precise genetic determinism of these resistances will have to be deciphered by the scientific and breeding communities. Indeed, more practically, INNOVINE project brings to grapevine breeders 26 new V. vinifera accessions tolerant to downy mildew and 33 to powdery mildew, possibly carrying so far unknown genes, as shown by genotyping with markers linked to known genes. These sources of resistance are particularly interesting because there are already showing a good quality.
Another important output of INNOVINE was the set-up of a reliable method for producing Guignardia bidwellii inoculum for black rot resistance tests and for the test itself (deliverable D3.4). With this method, 298 accessions from different origins (Eastern Europe V. vinifera varieties, a few accessions from other Vitis species and interspecific hybrids) have been tested. Thirty-seven of them showed moderate to high level of resistance (D3.4). In parallel, the genetic determinism of three sources of resistance has been studied, showing in all cases a control by several genes (at least two), some with strong effect. A major QTL has been detected on chromosome 14. Again this work will be the ground for further genetic studies and for the improvement of grapevine breeding strategies. Indeed, black rot re-occurs in vineyards that are not treated against mildews and INNOVINE provided the first comprehensive work on this secondary disease of growing importance in relation with the use of varieties tolerant to downy mildew.
Finally, the same kind of work was also carried out for resistance to Phylloxera (deliverable D3.4). The screening work focused on a set of V. aestivalis accessions and the genetic mapping of the resistance from a Muscadinia rotundifolia accession and again, the screening method was improved. The INNOVINE project delivered 7 V. aestivalis accession with an original resistance source compared to the already known pool and a set of V. vinifera/M. rotundifolia (BC2) individuals, highly resistant to phylloxera.

1.4.2.2 Potential impacts on adapting viticulture practices for reducing the use of pesticides in the vineyard
The information collected by VCR when performing assays with varieties tolerant to diseases showed that this innovation has the greatest economic and environmental impacts (see deliverable D4.4) while it has no specific impacts on yield, must and wine composition (see deliverable D4.5). This was strengthened by the observations made on a set of resistant varieties by INRA, UCSC and JKI. INNOVINE project has on the whole provided diverse and complementary tools and knowledge that will be useful for a more environment-friendly control of diseases in the vineyards that can be tested by the growers:
• Knowledge on the level of resistance of tolerant varieties in field conditions (WP2, WP4; see deliverable D2.3)
• A first set of strategies aiming at reducing but not suppressing the number of treatments on tolerant grapevine varieties (2-3 treatments instead of 7-13) showed to be efficient in controlling secondary diseases. It will be a way also to manage in the long term the resistance and avoid adaptation of the populations of mildews and progressive inefficiency of the resistance (deliverable D2.3).
• Alternative practices to chemicals (see deliverable D2.2)
• Results of tests of canopy management practices aiming at modifying the cluster architecture for a better tolerance to botrytis (see deliverable D2.1)
• Improved DSS systems that allow to be more efficient in the decisions of when spraying and what quantity of active compound (see deliverables D2.4 and D2.3)
• Finally, improved monitoring tools for downy mildew infection (Latouche et al, 2015).
These results have been already quite well disseminated by publications in peer reviewed papers by the consortium (10), but also by communications in 18 conferences and posters and through 11 targeted field demonstrations and tastings organized by the partners. The main results are highlighted below.
INRA, JKI and UCSC assessed the prevalence (incidence, severity, and sporulation) on leaves and clusters of two major grapevine diseases (downy mildew and black rot) on more than 10 partially resistant varieties (see deliverable D2.3). On the whole, the partially resistant varieties currently available, although providing a good control on downy and powdery mildew alone (up to 90% of sporulation less), are still susceptible to many other fungal diseases such as black rot that can potentially be very destructive for the harvest. Moreover, powdery and downy mildew are able to adapt to partial resistance leading to the emergence of isolates that have a higher level of pathogenicity.
Overall, the experiments performed in WP1, WP2 and WP4 confirmed that changes in leaf area to fruit weight ratio impact grapevine phenology, and that those reducing in fine crop load also reduced the susceptibility to bunch rot. However, canopy manipulation brings no protection against mildews for traditional susceptible V. vinifera cultivars (deliverable D2.1).
An Ampelomyces-based sanitation method allowed to reduce the impact of powdery mildew in combination with DSS-based treatments (deliverable D2.2).
Disease models (deliverable D2.4) implemented in the Vite.net have been improved by calibrating (i) the underlying growth model for phenological parameters of 19 varieties, (ii) the infection parameters for a set of resistant varieties (under progress) and (iii) the impact of mildew diseases on yield and quality. This makes Vite.net a better tool for assisting the decision of the growers than before, as its disease predictions are better fitting the reality. In parallel, HORTA designed and implemented other new DSS functionalities, such as the “Sustainability Indicators” concerning vineyard management practices and the model for estimation of the optimum dose of fungicides. Thanks to the implementation into the DSS, these new results will directly reach HORTA’s stakeholders (about 300 vite.net users on 10.000 hectares and other around 500 contacts) and have a real impact on the vineyard management, optimizing it and making it more sustainable. For instance, it was demonstrated that the use of vite.net with its model for optimal fungicide dose enabled an average reduction of the Treatment Frequency Index by 37%.

1.4.3 Assisting growers for adapting their viticulture systems for an increased sustainability
INNOVINE project has enlightened two major subjects on which researchers, extension services, technical advisors and service providers will have to focus their future work to better assist grower’s decisions towards a more sustainable viticulture. The first one is the need of diversification of the varieties planted. Indeed, INNOVINE’s work strongly suggests that diversifying the genetic catalog has the greatest impact for adaptation to biotic and abiotic stresses (D4.4 and D4.5). However, especially regarding varieties resistant to diseases, there is a lot of work ahead to design viticulture systems maximizing their quality in relation with the « terroir » where they will be planted. These management systems will also have to address the durability of the resistance to disease through the management of the populations of pathogens (see deliverable D4.3). The second is the delivery of an integrated set of user-driven services assisting grower’s decisions all along the season. The growers attending INNOVINE final conference showed a particular interest in these questions, through their questions and the success of the tasting and demonstration session (see deliverable D6.10). These future developments will hopefully be stimulated and supported by the important efforts of dissemination towards producers, growers, advisors and policy makers by the partner of INNOVINE project (see deliverable D6.9). Indeed, more than 40 events were organized all along the project for them with a peak of activities in 2015, when results obtained within the projects started to consolidate. The majority of the events specifically targeted advisors and viticulturists (about 70% of the workshops) while the participation of scientists and policy makers was also effective in about one third of these events, enriching the discussions from different standpoints and improving further networking. 43% of the workshops and trainings were open or targeted to students, which is a way to impact the new generations of growers, scientists and technical advisors.

1.4.3.1 Towards a diversification of the varieties in the European viticulture
INNOVINE has contributed to improve the toolbox of the breeders (see paragraphs 1.4.1 and 1.4.2): new sets of clones and varieties that can be used as sources of trait of adaptation, improved screening procedure (for plasticity, for adaptation to drought, for tolerance to mildews, black rot and phylloxera), molecular markers that can serve in marker assisted strategies. The evaluation data on genetic resources made in INNOVINE will provide the basis for a grapevine improvement initiative enabling a concerted European pre-breeding effort of scions and rootstocks. The novel resistances will be introgressed in elite breeding lines using the high scored sources as parents in crossings. The method of marker assisted selection (MAS) has been shown to clearly accelerate the entire grapevine breeding process. Specific markers closely linked to genetic loci of biotic or abiotic resistances will be elaborated building on tools and knowledge developed through international breeding and genomic initiatives. Those markers help grapevine breeders to select varieties adapted to biotic and abiotic stress conditions. They will allow to pyramid multiple resistance loci within a single plant and improve and design durable resistances to pests and diseases. The accessions and clones with no apparent biotic resistance or interesting abiotic plasticity screened for in this project will be carefully kept in germplasm repositories to make them available for the future. This material can be exploited regarding other traits not yet analyzed in following research and development projects.
However, as it has been worked out along the centuries for traditional varieties, a lot of work the best viticulture systems including these varieties will have to be defined for each “terroir” or soil-climate condition. Moreover, with regard to disease resistance, these viticulture systems will have to enhance its durability. Here again, INNOVINE has developed a tool box that now needs to be fully used and completed: improved DSS systems for better targeted treatments, canopy management practices to reduce the impact of botrytis, biocontrol method to improve the control of powdery mildew (see paragraph 1.4.2).
There is now a need to complete these tool and knowledge boxes by studying in more details the impact of combination of biotic and abiotic stresses on grapevine physiology and response to pathogens. This should be facilitated by the models developed in INNOVINE project (WP1 and WP2) and by the experimental platforms set up by the partners.

1.4.3.2 From monitoring the grapevine physiological state in the vineyard to assisting the grower’s decisions
Another necessity is to fully integrate monitoring tools into services that are directly useful to support growers’ decisions and improve the management of environmental impacts in their viticulture systems. INNOVINE project has made a step in this direction by working on web interfaces transforming the output of the models into simple and visual messages or monitored data into more classical indicators, already manipulated by growers (vite.net WINEO, CUBA, FA-server AGRICIENCIA). Vite.net tended towards a comprehensive set of indicators: assisting the diminution of pesticide application (less in number and less in dose) and evaluating different kinds of impacts related to sustainability of the whole production system. This approach has been tested along the project within the consortium by producers but the web survey performed to draw a map of the viticulture management system presently in use mostly in Europe (task 4.4) and the contacts taken during INNOVINE final conference has already proved useful to create or increase a network of potential “end-users” of all the innovative management strategies developed. An important effort of dissemination and demonstration to growers and grower’s technical advisors has been carried out during INNOVINE project with 58 actions of dissemination all along the project to end-users by the partners and collaborations within the project between the developers and producers (see also deliverable D6.9). This allowed to improve the maturity of the products already available to the market and to bring new ones to this stage. More details are given below.
The collection of raw data collected from the experimental trials of WP4 was important to compare, from both economic and environmental point of view, common practice to the innovation tested and give some practical figures about the real applicability of these innovations by grape growers. Moreover, the data collected were used to test environmental impact indicators developed within WP5 and guarantee their robustness when implemented into the DSS vite.net.
During the project, UCSC developed new models and modified the following existing ones: model for bud infection by powdery mildew, leaf development in different grapevine training systems and early detection of drought stress). UCSC shared these improvements with HORTA that implemented them in the DSS vite.net. In parallel, HORTA designed and implemented other new DSS functionalities, such as the “Sustainability Indicators” concerning vineyard management practices and the model for estimation of the optimal dose of fungicides. In order to disseminate project results and to demonstrate to growers and advisors as well as other stakeholders (around 300 persons) the benefits of using vite.net 14 workshops were organized in different Italian grape growing areas and access to the system was given to expert in viticulture to gather feedbacks and possible improvements suggestions.

ENTAV-IFV’s results on the PTO concept® (Optimized Treatment Plan) that were disseminated at INNOVINE final conference, showed the interest of capitalization of exhaustive data stemming from proxy-detection sensors, operated within the framework of emergent precision viticulture techniques. The approach made on a single vineyard has however to be generalized and validated by testing in other climatic areas. When achieved, such information will improve the already existing disease modeling and dose reduction modules of the Epicure DSS. The possibility to propose different doses in relation to spatial coordinates will allow to assist future intelligent sprayers.
NOVELTIS developed unique, reliable and innovative processing chains and products for characterizing the vineyard plots by using the latest technologies of aerial and satellite remote-sensing. A graphical user interface based on new web-mapping techniques was also developed for distributing products attractive to end-users. Indeed, all the activities performed by NOVELTIS in INNOVINE were designed and validated in close cooperation with several end-users involved in the project from different European countries (France, Germany, Spain, Bulgaria and Portugal). This allowed to ensure the maturity, accuracy, generality and usefulness of the products developed. To-day these products, wrapped together in WINEO, are operational and ready for commercial deployment in Europe and at the global scale. The market analysis and commercial prospection have already started by NOVELTIS and the first market returns for the exploitation of INNOVINE results are very promising.
FORCE-A successfully fulfilled the objectives of developing or improving its data collection sensors and its data processing and display functionalities including the mapping algorithms and the FA-Server web platform, in order to accelerate and simplify the process of data acquisition, processing and display.
FORCE-A’s exploitable findings are integrated into several commercial products and services:
• On-the-Go Multiplex with improved performances and easy mounting,
• FA-Box with upgraded functions: easy data classification according to winery shapefile,
• Calibrated vigor maps with Dualex leaf-clip, and validated thresholds for optimum vigor at bunch closure,
• Improved zoning display along the vineyard rows,
• FA-Server platform with upgraded functions: display of GPS points, rainbow maps, zoned maps with dynamic threshold, diagnostic maps,
• CUBA software for the conversion of Anthocyanin units.
The commercialization of diagnostic services is ongoing by FORCE-A in several wine estates in France and abroad and scientific dissemination has been achieved through four publications on the following topics.
AGRICIENCIA has improved an information collection system for viticulture (soil, weather and plant data) which combines innovative software and hardware to collect and analyze data from multiple sources dynamically integrated in real time. The system is independent of the hardware brands and fully interoperable with different software solution even with other DSS available, such as for example, the DSS prototype for deficit irrigation management complemented with an alert system for smartphone and smartwatch developed by other Project partners. The potential impact of the obtained results is difficult to evaluate precisely at this stage. However, from the high interest demonstrated by the technicians of Esporão Estate (450 ha of vineyards in Portugal) and by all the participants of the several dissemination activities performed a high impact on the wine industry is expected, particularly for irrigated winegrowing regions like for instance the Alentejo region (> 20,000 ha).
List of Websites:
1.5 Project public website and contacts

Website
www.INNOVINE.eu

Contacts

Dr. Anne-Françoise ADAM-BLONDON (Project Coordinator)
Institut National de la Recherche Agronomique (INRA)
Unit of Research in Genomics-Informatics INRA
Route de Saint-Cyr, 78026 Versailles cedex
France
Phone : [phone]
Fax: [phone]
e-mail: [e-mail]

Dr. Enric BELLES-BOIX (WP7 Leader)
INRA Transfert
3, rue de Pondichéry
F-75015 Paris, France
Phone: [phone]
e-mail: [e-mail]

Ms. Elodie TAN (Project Manager)
INRA Transfert
3, rue de Pondichéry
F-75015 Paris, France
Phone: [phone]
e-mail: [e-mail]


1.6 Project logo
In attachment.

Related documents

• final1-innovine-final-report-figures.pdf