Responses to climatic and pathogen threats differ in biodynamic and conventional vines

Viticulture is of high socio-economic importance; however, its prevalent practices severely impact the environment and human health, and criticisms from society are raising. Vine managements systems are further challenged by climatic changes. Of the 8 million hectares grown worldwide, conventional and organic practices cover 90% and 9% of acreage, respectively. Biodynamic cultivation accounts for 1%. Although economic success combined with low environmental impact is widely claimed by biodynamic winegrowers from California, to South Africa, and France, this practice is still controversial in viticulture and scientific communities. To rethink the situation, we encouraged stakeholders to confront conventional and biodynamic paradigms in a Participative-Action-Research. Co-designed questions were followed up by holistic comparison of conventional and biodynamic vineyard managements. Here we show that the amplitude of plant responses to climatic threats was higher in biodynamic than conventional management. The same stood true for seasonal trends and pathogens attacks. This was associated with higher expression of silencing and immunity genes, and higher anti-oxidative and anti-fungal secondary metabolite levels. This suggests that sustainability of biodynamic practices probably relies on fine molecular regulations. Such knowledge should contribute to resolving disagreements between stakeholders and help designing the awaited sustainable viticulture at large.

Of the 8 million hectares devoted to viticulture worldwide, conventional and organic practices cover 90% and 9% of acreage, respectively. Biodynamic cultivation accounts for only ca. 1% (http://www.demeter.net/ contacts-di), and this practice is still controversial. Studies on biodynamic or/and organic cultivation focused on economy and soil composition 1 , on soil structure, soil biodiversity and microbiological activity 2 , and on fertilizer effects 3 . Studies of grape yield and quality 4,5 , microbial communities in grapes and leaves 6,7 , and wine biochemistry 8 have also been reported. However, the results have not resolved the controversy 9 about the 'anthroposophical paradigm' 10 at the heart of biodynamics practice. To date, the dissenting viticulture communities have not reached a collective plan to reduce the impact of viticulture practices on the environment and human health 11,12 . We have tentatively addressed this problem holistically, bringing together the different stakeholders in a Participative-Action-Research 13 . The workshops highlighted the dissensus among biodynamic and conventional winegrowers, as well as among the non-government organizations Alsace Nature/France Nature Environment, viticulture advisors, technicians, and research scientists. The dissensus stems from lack of distinct proof of the claims made, and from epistemic conflicts 13,14 . Relying on a collective epistemology 13 , the group arrived at the most sensitive and controversial issues. Biodynamic winegrowers lacked experimental evidence for their claim that their practices stimulate plant defense mechanisms. The conventional winegrowers have shown that synthetic pesticides deter pathogens (albeit with a high environmental impact) but there were no data on possible effects on plant defenses. Focusing on plant responses to pathogen and climatic threats, we chose a holistic approach to compare the consequences of the differing practices. We chose Pinot Noir as it is used worldwide for wines and champagne, its genome has been fully sequenced 15 , and its sensitivity to climatic disorders is well documented 16,17 . Our trial consisted of 14 plots of Pinot Noir vines grafted onto the SO 4 rootstock that had been grown under conventional management (8 winegrowers, 8 plots of 21,413 m 2 ) or biodynamic management (3 winegrowers, 6 plots of 9,756 m 2 ) for more than 20 years in the same climatic conditions (Fig. S1). Over a 4-year period, we monitored vine management, plant physiology, and the levels of infection with virus, downy mildew, and powdery mildew. For plant defense responses, we analyzed secondary metabolite content as well as steady-state mRNA levels of 30 immunity and silencing genes.

Results and Discussion
Soils, climatic conditions and vine management. When comparing soils in conventional/biodynamic management, after chemical and physical analysis, values were higher for pH, Calcium and soil components above 2 mm in conventional, and higher for Manganese and penetrometer assays in Biodynamic, suggesting steeper horizons (Mann-Whitney at *P ≤ 0.05; Table S1). Still, there was no clear-cut picture between soils from biodynamic and conventional plots, in contrast to other studies 1,2,4 . The conclusions of the latter authors, however, were based on additional biochemical and biological parameters, and here, we cannot exclude that further analysis, such as in microbiology, may reveal differences as illustrated in organic and conventional management 7 . We characterized pest management practices after interviews with winegrowers 13 . All winegrowers in the study relied on copper and Sulfur treatments. In addition, conventional viticulture employed synthetic fungicides, whereas biodynamic cultures were complemented with preparations such as cow manure (500, 500 P), minerals such as finely ground silica (501), decoctions of nettle (504), willow, horsetail, valerian, and lemon oil. Conventional growers followed the guidelines of suppliers, and biodynamic growers followed the Demeter principles (4, www.demeter.net). To better compare the two managements, we modified TFI 18 by subtracting Copper and Sulfur from conventional products and built up a modified index (mTFI) (Fig. 1). The mTFI values and application timing changed significantly in response to high pathogen pressure, as in summer 2016 (Fig. 1). As synthetic fungicides are often systemic, they persist within plants for about 2 weeks. In contrast, all biodynamic preparations sprayed on plants were washed off by rains and were repeated. This may explain (though not completely) the higher mTFI for biodynamic cultures (7, 10, and 13.3) in 2014, 2015, and 2016, respectively (Fig. 1). In addition, biodynamic applications began earlier in the spring and stopped earlier in the summer (Fig. S2), with more Sulfur than conventional cultures (with the exception of 2014). When summed up for all years, mTFI and Sulfur applications were higher in biodynamic management. For copper, beside 2016, the mean quantities applied did not differ in conventional and biodynamic, remaining lower than the norm of 4 kg/hectare/year (Demeter).

Pathogens loads in vines.
To evaluate the consequences of different management systems on vines, we examined the forth/fifth leaves from the vine apex, which are the most sensitive to pathogen attack. This developmental stage is associated with a physiological change from sink to source 19 , when the green arms reach the developmental stages H in May and K in July 20 . Molecular analysis by qRT-PCR allowed detection of downy mildew, powdery mildew, as well as viruses such as grapevine fanleaf virus, grapevine leafroll virus and grapevine vitivirus A (GFLV, GLRaV 1-3 and GVA, respectively). Interestingly, 94 to 100% plants were pathogen-free  in spring 2014-2016 (Fig. 2). In July 2014-2015, the proportions of pathogen-free plants decreased, notably in biodynamic management. In 2016, facing high pathogen pressure due to humid and warm conditions (Fig. 3), 78% of plants grown conventionally remained pathogen-free and 49% in biodynamic cultures (Fig. 2). In the pooled data (2014-2016), 305 samples were positive for downy mildew and powdery mildew (36 samples with mixed viral and fungus infection not included in calculations). Fungi were more abundant on plants from biodynamic than conventional cultures (18.28% and 7.09%, respectively; independence test χ 2 P ≤ 0.001). The titers for downy mildew were higher for biodynamic than conventional plants (medians of Δ.ΔCT = 11.60 vs 8.29, Mann-Whitney, P ≤ 0.001). For powdery mildew, the Δ.ΔCT medians did not differ between practices (6.46 vs. 6.94, Mann-Whitney P = 0.21). However, for both managements, none of the harvested leaves showed the visible symptoms described in viticulture, such as a powdery leaf surface due to conidiophores of powdery mildew, an oil-spot leaf response, or associated sporangiophores typical of downy mildew 21 . Thus, both pest managements appeared to hinder the progression of fungal infection in planta. The low organic matter and nitrogen contents of the soils (Table S1) may also have reduced pathogen distribution within plots and the multiplication of pathogens on leaves 19,22,23 . With respect to viruses, we observed characteristic symptoms only in autumn 24 . However, molecular analyses of 263 samples collected across 2014-2016 detected infections by one or more of the most frequent grapevine viruses (such as GFLV, GLRaV 1-3 and GVA) already in the spring, suggesting that the viruses were present before the project started. In the case of pathogen threats, whilst both cultivation practices hindered the progress of infections, the wide application of conventional fungicides clearly reduced the frequency of mildews infection of leaves, and the pathogen contents, more than biodynamic management. However, this again is offset by the environmental impact of synthetic pesticides, whereas copper is the only questionable input in biodynamic practice.
Vines responses to abiotic stress. To characterize vine responses to abiotic stress, we analyzed defense gene mRNA levels in 2044 pathogen-free samples. Variations in mRNA levels of housekeeping genes actin, actin7, GAPDH and UBQ were 15.95%, 13.17%, 8.97% and 6.32%, respectively, and boxplot analyses of normalized Ct values showed the lowest variability of data for GAPDH and UBQ. Therefore, these two genes were chosen as controls for the study. Firstly, mRNA levels of apoplastic amine oxidases (AOS), endochitinase 4 C (CHIT4C), lipase enhanced disease susceptibility (EDS1), ETR1, flavonone 3 hydroxylase (F3H), glutathion S transferase (GST1), HSR, lipoxygenase (LOX), transcriptional activators of the salicylic acid pathway (NPR1-1, NPR1-2), phenyl alanine ammonia lyase (PAL), pathogenesis related proteins (PR1, PR6, PR10-1), superoxyde dismutase (SOD), and stilbene synthase (STS1) were analyzed. These contribute to MAMP-triggered immunity, effector-triggered susceptibility, or effector-triggered immunity 21,25 . Comparing all values from 2014-2016, CHIT4C, ETR1, F3H, STS1, LOX, AOS, NPR1-1, NPR1-2, HSR, SOD transcript levels were higher in pathogen-free samples from biodynamically grown vines (Fig. 4A, H samples). In addition, we analyzed transcript levels of RNA-dependent RNA polymerases (RdR-1, RdR-2, RdR-6), microRNA-generating (Dicer like These results suggested overall higher levels of gene activation following biodynamic rather than conventional practices. We then compared mRNA levels of all RNAi genes for each season with a Principal Component Analysis (PCA on the ranks of Δ.ΔCT). Plants grown biodynamically showed repeatedly distinct levels of expression of RNAi genes between 2014 and 2016 (Fig. 5B). In 2016, the hydric balance (difference between rainfalls and evaporation potential ETP-Penman) was positive (Fig. 3) and thus favorable to vine development. However, by the end of June 2016, dry and warm conditions characteristic of semi-continental/semi-arid climate set in. Thus, we hypothesized that, in conditions of low or very low abiotic stress, and in pathogen-free samples, the activity of the silencing machinery was at a low level (May 2016, Fig. 5B) but increased in response to warmer and drier summer conditions (Fig. 5B). Winegrowers confirmed abiotic stress of vines in summer 2016. Moreover, they pointed out that vines suffered from abiotic stress already in spring 2014, associated with an unfavorable hydric balance and high temperatures (Fig. 3). At the molecular level, the transcript levels of silencing genes were higher in spring 2014 than spring 2016 (Fig. 5B) and a summer-shift was observed, mainly in biodynamic plants. In 2015, when the annual rainfall reached only half of ETP-P and very dry/warm weather persisted from mid-May until the end of July, vines faced severe abiotic threats to the extent that numerous winegrowers across the Alsace region (not participants in the Participative-Action-Research) removed grass with herbicides or by ploughing in order to save their vines. Interestingly, vines in that year exhibited the highest silencing gene transcript levels observed in this study, especially in biodynamic management. Taken together, our data suggest a molecular signature specific to vines grown biodynamically and experiencing a more intense response to abiotic stress, with high expression of silencing and immunity genes ( Fig. 5A-B) in pathogen-free leaves. If this reflects enhanced expression of plant defenses then, conversely, the data imply lower plant defense responses in vines subjected to conventional practices. A possible critical difference in plant managements lies in the use of manures and tisanes versus synthetic fungicides. Yet, differences between biodynamic and conventional managements were lower in summer, but also in May and July 2016 (Fig. 5A,B). Importantly, mTFI values as well as amounts of copper and sulfur sprayed were significantly higher in both practices during these seasons ( Fig. 1 and Fig. S2). Therefore, smaller differences between management practices in pathogen-free samples, notably in 2016, may be a response to more favorable climatic conditions or/and to excessive spraying impairing plant molecular responses.

Vines responses to pathogens threats.
To compare the outcome of managements in the presence of biotic stress (2014-2016), we analyzed defense gene mRNA levels in pathogen-containing vines grown conventionally or biodynamically. The earlier observations in pathogen-free plants resembled, with higher transcript levels in biodynamic in virus-containing leaves in all RNAi genes but RDR6, AGO7, HEN1. In fungus-containing samples, only AGO7 differed with higher levels in conventional management ( Fig. 4B; H, V, F samples). From 2014 to 2016, the comparison between biodynamic and conventional leaves was less clear-cut for immunity genes. Transcript levels of EDS1 and PR1 were higher in biodynamic leaves and ETR1, AOS, NPR1-1 in conventional, in fungus-containing leaf-samples (n = 305). Transcript levels of ETR1, F3H, LOX, NPR1-1, NPR1-2, SOD, were higher in biodynamic practices and PR1, PR6, PR10-1, GST1, in conventional practices for virus containing samples (n = 263) ( Fig. 4A; H, V, F samples). Altogether, the data suggest that silencing genes in vines grown in biodynamic management are more prone to activation by biotic stresses, probably because they are already expressed at higher levels in biodynamic than conventional management in the absence of infection, a phenomenon that resembles priming 31,32 . During meetings and interviews, conventional growers criticized the yellowish, unhealthy appearance of vines grown biodynamic. In reaction to this observation, we evaluated chlorophyll content as an indicator of plant fitness. In addition, we evaluated flavonols and anthocyans in 3988 leaf samples from all 14-vineyard plots from 2015 to 2017. Both contribute to the capture of and protection from light, as well as to defense reactions 33 . Chlorophyll contents were higher in conventional vines in all samples except May 2016 (Fig. 6). In contrast, flavonols were higher in vines grown biodynamic in all seasons and years. This was also true for anthocyans in July 2016-2017. The results are compatible with the appearance of leaves from biodynamic vines and it is possible that the increase in flavonoids also influenced the chlorophyll-associated green color of the leaves. Interestingly, increase in flavonols and anthocyans may indicate a more effective response to abiotic and biotic stress in biodynamic vines 34 . Previous research indicates that silicon ("501 preparation") influence the Arabidopsis transcriptome after fungus infection, and creating a physical barrier on plant leaves 35 . In rice, silicon was reported to alleviate pathogen effects on plant carbon metabolism and cytokinins through priming [36][37][38] . However, the 501 preparation is applied only once or twice a year, thus other manures from Equisetum arvense, Salix alba, Achillea millefolium, or Valeriana officinalis may also contribute to elevated defense responses by providing salicylic acid, iron, minerals, and other metabolites. Unfortunately, the constituents and activities of such preparations are poorly documented and they remain a controversy among the vine community. Anyway, causality with plant responses cannot be established firmly here.
Vines responses to pathogens at the biochemical level. Pathogen-free leaf samples collected in July of 2014 to 2016, from two plots per practice, were analyzed further by UHPLC-MS. Of the 880-chromatogram peaks, only twenty-seven molecules were already described in vines ( Fig. 7 and Fig. S3). The levels of 18 of them increased in response to abiotic stress in a manner resembling the expression of the immunity and silencing genes (Fig. 4), notably with the highest levels in 2015. Of these metabolites, the levels of two were higher in biodynamic than conventional vines in 2014, and 7 in 2016 (Fig. 7). The contents of nine metabolites did not change significantly (Fig. S3) (Fig. 7). Interestingly, these flavonols and pro-anthocyanidins have both anti-oxidative and anti-fungal properties [39][40][41][42] and may have enhanced the response to biotic threats, especially in biodynamic vines. In addition, our data confirmed the plasticity of Pinot Noir 17 , particularly when cultivated in biodynamic management. Given the enhanced responses of vines grown biodynamic, the question arose of their energetic cost. As yield is primarily determined by management decisions regarding thinning, pruning weights may better reflect overall energy storage of vines. The 1120 data points from 14 plots of mean pruning weights showed no difference between biodynamic and conventional vines in 2016, a favorable year for plant growth (Fig. 8). In 2017, when vines faced intense drought, as in 2015 (Fig. 3), pruning weights of vines grown conventional decreased dramatically, whereas the values remained stable in biodynamic management ( Fig. 8; Mann Whitney, P ≤ 0.0001). Thus, the enhanced responses to threats in vines grown under biodynamic management were not detrimental to biomass accumulation. On the contrary, it seems that increased resistance to intense climatic stress is associated with maintenance of plant reserves, which contribute significantly to flowering and the grape yield of the following millesimal 43 .  Holistic studies are hampered by the complexity of the interactions between plants, the environment and human practices and reasoning. Biodynamic cultivation still relies on many unvalidated and unrecognized assertions 9 , which hinder acceptance of this form of vineyard management by winegrowers. As a consequence, dialogue between the biodynamic and conventional worlds is still limited. Here we show that involving all of the actors in Participative-Action-Research may contribute resolving the disagreements, firstly by co-constructing questions. Subsequently, all participants were involved in collective reasoning 13,14,44,45 in a workshop where a consensus was build up from raw data. This led to the main conclusion of this paper. Beyond, the group prioritized its further questioning. Namely, instead of looking for causality of 'biodynamic management-specific' plant response to stress at first, the question of 'when' such properties are acquired by plants upon change from conventional to biodynamic management was chosen.

Conclusion
The Participative-Action-Research conducted on this territory lacked winegrowers in organic management. Therefore, we cannot exclude that conclusions raised for biodynamic management would not apply to organic, neither did we resolve fully the controversy about biodynamic practices. However, overall, his project unlocked disagreements between stakeholders by shedding light on unexpected diversity within conventional and biodynamic management, and by characterizing a "biodynamic management-specific" elevated response to climatic and pathogen threats. This suggests that sharing expertise, within a scientific frame, may diminish management intensities, and ultimately, lower environmental and human health impacts of viticulture.

Vineyards plots.
Plots were selected so that they all had the same rootstock, planting density and pruning method, upon proposal of winegrowers. The complete trial consisted of 8 vine plots cultivated conventionally and 6 following biodynamic practices, all for more than 20 years. The plots were planted with Pinot Noir vines grafted on the SO4 rootstock at a density of 4,500 plants/hectare. Winter pruning followed the 'double-Guyot' guidelines with 1-2 arms (depending on the vigor of the plant) bearing 8-10 latent buds per arm. Summer 'green pruning' was conducted to limit vegetative development to c.a. 1 m 2 foliage/kg bunches, i.e. within the wired-frame characteristic of ' Alsace' and according to AOC viticulture guidelines. In each plot, after exclusion of plants from the two first border rows and the first and last five plants of each row, 4 blocks of 10 plants were defined. These blocks were positioned as far from each other, with at least 10 plants between 2 blocks, when situated within the same row, and with at least one row distance between two blocks-containing rows.
Characterization of soils. All data were collected once, in spring 2015. Penetrometer assays were conducted in the 4 blocks of each vine plot (2 measurements/block) with penetrometer (Penetrometre compactometre sol à cone statique: 6120, SPECTRUM Technologies). The first horizon ended when the pressure gauge for penetration reached 200 psi, the second horizon when it reached 300 psi, and the third horizon when a lower pressure was found beyond the second-horizon endpoint. Analysis of structure and physical and chemical composition was carried out on soil samples (c.a. 250 ml) collected in the 0-30 cm horizon in two places in each block. Then all samples were mixed to reach a single sample per vine plot. The following were measured: pH (NF ISO 10390; extraction with 1/5 distilled water); organic matter (sulfochromic method followed by colorimetric assay, NF ISO 14235); nitrogen dosage (extraction with 1 M KCl followed by colorimetric assay; NF ISO 14256-2); phosphorus, potassium, magnesium, calcium, iron, zinc, manganese (10 g soil + 50 ml distilled water, followed by paper filtration and estimation by atomic spectrophotometry (except phosphorus by colorimetric assay) according to BIPEA (https://www.bipea.org/fr/).  Primers. Primers were designed according to Trouvelot el al 48 and Chong et al. 49 for immunity genes, for powdery mildew, and for viruses GVA and GLRaV1-3 [50][51][52] . All other primers were designed, optimized for Tm, and adapted to Fluidigm technology, in the course of this study. Amplified products were cloned and sequenced for alignment according to the corresponding genomes of pathogens 48