^{ 1 }Physiologie, Pathologie et Génétique Végétale, PPGV, Université de Toulouse, INP - Purpan, 31076 Toulouse, France

^{ 2 }Laboratoire de Chimie Agro-industrielle, LCA, Université de Toulouse, INRAE, 31030 Toulouse, France

^{ 3 }Institut Français de la Vigne et du Vin pôle Sud-Ouest, IFV Sud-Ouest, 81310 Peyrole, France

^{ * }

^{ 1 }

^{ 1 }

^{ 3 }

^{ 1 }

^{ * }corresponding author: olivier.geffroy@purpan.fr

Wine grape breeding for fungal resistance has been very dynamic worldwide over the last decade. The quick phenotyping of genotype quality traits, including aroma composition, remains challenging. Selected ion flow tube mass spectrometry (SIFT-MS) could be particularly valuable for this usage. Due to the high number of seedlings to phenotype and the low availability of berries, the sampling strategy can hardly rely on time-consuming destructive methods such as the measurement of classical maturity parameters (i.e., sugar concentration). To investigate the impact of the sampling time on the SIFT-MS fingerprint, berries from _{2}
^{+} as reagent ion. This fingerprint has proved to be stable from 28 days after mid-veraison. This finding greatly simplifies the sampling procedure for future berry phenotyping, which can only rely on non-destructive data (lapse of time after mid-veraison date). For most

SIFT-MS, fingerprint,

Despite the identification of quantitative trait loci (QTLs) for berry and wine quality (

Selected ion flow tube mass spectrometry (SIFT-MS) is a technology commercially available since 2008 that has the advantage of offering real-time headspace analysis and high sensitivity (_{3}O^{+}, NO^{+}, O_{2}
^{+}, NO_{3}
^{-}, NO_{2}
^{-}, O^{-}, O_{2}
^{-} and OH^{-}) can analyse a sample headspace and determine relative abundances in Selected Ion Monitoring (SIM) or scan mode (

A recent study highlighted that SIFT-MS could be valuable for discriminating the volatile composition of _{2}
^{+}. The use of this latter single reagent ion which has the highest ability to ionise most organic compounds was particularly relevant to reduce the time of analysis to 3 minutes. The SIFT-MS technology enabled discrimination of low and high aroma producers and to connect cultivars, in most cases, according to their parentage relationship. In this former research, grape varieties were sampled at three different dates according to their theoretical timing of veraison (

However, in the perspective of further use of this methodology for the quick phenotyping of new varieties, the impact of the sampling date on the SIFT-MS volatilome would deserve to be investigated deeper. This would enable one to establish a reliable berry sampling strategy based on non-destructive phenological data (lapse of time after mid-veraison date). Indeed, the adaptation of the date of sampling and measurement to typical maturity parameters such as sugar concentration is hardly implementable due to the high number of individuals to phenotype and the low quantity of grapes available for each genotype, at best a couple of clusters borne by one single plant. The aim of this research work was to study the impact of seven sampling times over maturation on the SIFT-MS fingerprint of

The 0.51-ha vineyard from where the grapes were sourced was located in the southwest of France (lat. 43° 50′ 25′′ N; long. 01° 50′ 57′′ E) and was typical of the area with 2.20 m × 1 m vine spacing. The block was planted in 2002 with Syrah, the most widely grown cultivar in the vineyards of Occitanie, according to FranceAgriMer (www.franceagrimer.fr). It was grafted on Gravesac rootstock and was trained with vertical shoot positioning on a single Guyot pruning system. The orientation of the vine rows was north-east to south-west. The soil was mechanically managed under the vines and by grass cover in the inter-row area. Samples composed of 100 berries were first collected every third day from the end of July to the beginning of August 2020 to determine mid-veraison (50 % of soft berries), and then in triplicate at seven times during maturation to investigate the impact of sampling time on SIFT-MS fingerprint. Grapes were sampled on 6, 20 and 28 August 2020, 3, 10, 17 and 25 September 2020 which corresponds to mid-veraison (50 % ver.), 14 days after mid-veraison (50 % ver.+14d), 22 days after mid-veraison (50 % ver.+22d), 28 days after mid-veraison (50 % ver.+28d), 35 days after mid-veraison (50 % ver.+35d), 42 days after mid-veraison (50 % ver.+42d) and 50 days after mid-veraison (50 % ver.+50d), respectively. The commercial harvest of the vineyard took place on 15 September. Samples were always collected from the same fifty vine plants spread over three rows, from both sides of the row and several parts of the bunch (50 berries from each side of the row). Crop load was estimated at around 3 kg per vine (150 kg for the whole sampling area), which indicates that the whole amount of grapes harvested over the seven sampling dates (2100 berries) is unlikely to impact crop load or leaf area to fruit ratio for each sampled plant and therefore should not induce any bias.

For each 100-berry sample, 50 g were used for SIFT-MS analysis and the rest for the determination of physico-chemical parameters. In this latter subsample, the number of berries was first counted to determine berry weight. Grape samples were then crushed, the juice was centrifuged for 1 min at 5600

As climatic conditions over the sampling period are likely to impact physico-chemical parameters and particularly berry weight, rainfall and mean daily air temperature were also monitored daily since 2005 by a CimAGRO weather station (Cimel Electronique, Paris, France) placed within 200 m of the experimental site. These data were used to calculate the average mean temperature and cumulative rainfall between 6 August and 25 September for 2020 and for the 2005-2020 period.

Sample preparation and SIFT-MS measurements were performed according to the protocol proposed by

After crushing, 50 g of grapes were transferred into a 1 L Schott bottle (Verres Vagner, Toulouse, France) sealed with a Teflon-secured screw cap. Then, it was kept for 6 h at room temperature and transferred to a water bath for 40 min at 40 °C. These conditions that did not saturate the device analysis potential were determined in previous research (

SIFT-MS measurements were conducted using a Voice 200 Ultra model (Syft Technologies, Christchurch, NZ) in full scan mode (from _{2}
^{+ }as a reagent ion. The injection was conducted using N_{2} flow as a carrier gas (Alphagaz, Air Liquide, 99.9999 %, Paris, France) with a nitrogen flow rate set at 2.0 TorrL/s. The sample headspace was introduced by a calibrated capillary at a sampling flow rate of 0.3 TorrL/s. The analytes reacted with the selected precursor in the flow tube kept at 119 °C and 0.06 kPa

Instrumental repeatability was estimated at 7 % and reproducibility at 10 %. LabSyft 1.6.2. software (Syft Technologies) was used for data acquisition and analysis.

SIFT-MS data were pre-treated by removing masses with an

Then SIFT-MS data, together with physico-chemical parameters, were subjected to a one-way analysis of variance (ANOVA) treatment using XLSTAT software (Addinsoft, Paris, France). Fisher’s least significant difference (LSD) test was used as a post-hoc.

A principal component analysis (PCA) was performed on SIFT-MS significant variables (

).

Results show a steady evolution of the measured physico-chemical parameters over the sampling period (Table 1). As could be expected, sugar concentration increased through accumulation in hexoses while TA decreased through malate catabolism (

titre du tableau
50 % ver.
13.0 ± 0.1 e
16.95 ± 0.36 a
2.80 ± 0.02 e
1.31 ± 0.11 bc
50 % ver.+14d
18.3 ± 0.6 d
15.90 ± 1.31 a
3.12 ± 0.03 d
1.57 ± 0.17 a
50 % ver.+22d
20.0 ± 0.9 c
12.75 ± 1.40 b
3.25 ± 0.02 c
1.21 ± 0.10 c
50 % ver.+28d
21.9 ± 0.6 b
11.25 ± 0.90 b
3.25 ± 0.02 c
1.31 ± 0.12 bc
50 % ver.+35d
22.5 ± 0.4 b
8.44 ± 0.15 c
3.32 ± 0.03 b
1.40 ± 0.08 ab
50 % ver.+42d
24.9 ± 0.2 a
4.77 ± 0.16 d
3.46 ± 0.04 a
1.37 ± 0.06 bc
50 % ver.+50d
25.1 ± 0.7 a
4.90 ± 0.28 d
3.49 ± 0.06 a
1.27 ± 0.11 bc
< 0.0001
< 0.0001
< 0.0001
< 0.05

As a consequence of these warm conditions, the sugar concentration was already high at 50 % ver.+42d, reaching 24.9 ± 0.2 °Brix. Between 50 % ver. and 50 % ver.+14d, and to a lesser extent, between 50 % ver.+28d and 50 % ver.+35d, an increase in berry weight was noticed. Despite that xylem is known to be dysfunctional from veraison and that berries become less sensitive to soil moisture (

Among the 150 ions monitored by SIFT-MS with

For the seven sampling dates, several groups of masses were observed with the highest abundances around _{7}H_{5}O^{+} for

The aroma of Syrah grapes and wines has been the subject of much research worldwide (_{2}
^{+}, ionisation with this reagent ion is always known to generate one molecular ion (Smith and Panel, 2005). If this were the case for β-damascenone and rotundone, a signal would have been expected at

The PCA plot shows that the volatile composition of berries determined by SIFT-MS measurements greatly varied from 50 % ver. to 50 % ver.+28d but remained stable from this latter sampling date (Figure 2). Such a finding is in accordance with previous work highlighting a high similarity in SIFT-MS fingerprint between Sémillon samples harvested at three different times from 40 days after mid-veraison (

In most cases, a decrease in abundance was noticed during maturation (Table 2). Large changes in berry volatile composition involving translocation, accumulation, or metabolism mechanisms have been previously reported during this period (

titre du tableau
4090 a
2620 bc
3679 ab
1904 d
1561 cd
820 d
1061 d
< 0.0001
13428 a
7 752 b
11686 a
5388 bcd
5610 bc
2193 d
3673 cd
< 0.0001
1098 a
664 bc
832 b
400 de
456 cd
187 e
286 de
< 0.0001
4462 a
5441 a
2478 b
1491 bc
1882 bc
2438 bc
1297 c
< 0.0001
998 b
1273 a
591 c
241 d
388 cd
506 cd
310 d
< 0.0001
3382 b
6280 a
1648 c
779 c
1469 c
3677 b
1322 c
< 0.0001
213 bc
381 a
131 cd
93 d
104 cd
254 b
79 d
< 0.01
8024 b
13606 a
2662 c
1001 c
3658 c
8379 b
2379 c
< 0.0001
376 b
633 a
128 c
51 c
149 c
398 b
143 c
< 0.0001
170 d
1028 b
220 d
203 d
579 b
448 bc
222 cd
< 0.0001
139 c
1584 a
86 c
140 c
188 bc
451 b
189 bc
< 0.0001
298 a
242 ab
112 c
101 c
102 c
171 bc
92 c
< 0.0001
537 ab
391 b
600 a
200 c
222 c
132 c
174 c
< 0.0001
5643 a
4074 b
3631 b
1356 c
1463 c
836 c
900 c
< 0.0001
1933 a
1676 a
1559 a
528 b
829 b
699 b
506 b
< 0.0001
12938 a
8112 bc
11308 ab
4593 d
4866 cd
2529 d
3179 d
< 0.0001
900 a
464 bc
669 ab
260 cd
159 d
157 d
151 d
< 0.0001
4087 a
3602 ab
2366 cd
2590 cd
1787 cd
2684 bc
1680 d
< 0.01
280 a
230 ab
181 bc
197 bc
104 d
149 cd
102 d
< 0.01
83 ab
120 a
20 c
16 c
44 bc
77 ab
34 bc
< 0.05
329 a
292 a
50 b
112 b
50 b
71 b
70 b
< 0.0001
573 a
311 bc
426 ab
202 cd
152 cd
67 d
118 d
< 0.0001
453 ab
619 a
314 bc
108 d
208 cd
130 d
110 d
< 0.0001
256 a
196 b
170 b
57 cd
91 c
30 d
41 cd
< 0.0001
463 a
347 a
210 b
91 bc
111 bc
110 bc
62 c
< 0.0001
1389 a
1224 a
451 bc
139 c
440 bc
643 b
254 c
< 0.0001
132 ab
171 a
52 c
33 c
46 c
94 bc
33 c
< 0.01
632 bc
987 ab
239 cd
80 d
359 cd
1204 a
201 cd
< 0.01
112 ab
132 a
20 c
30 c
29 c
106 ab
42 bc
< 0.05
110 c
747 a
82 c
31 c
138 c
459 b
102 c
< 0.01
23 a
107 a
10 a
70 a
39 a
87 a
46 a
ns
177 c
2392 a
118 c
248 c
423 bc
1882 ab
337 bc
< 0.05
33 c
177 a
1 c
68 bc
46 c
130 ab
33 c
< 0.01
400 a
267 ab
218 bc
103 cd
81 cd
74 d
52 d
< 0.01
256 a
167 ab
147 bc
60 c
112 bc
47 c
53 c
< 0.01
2101 a
1378 b
1226 b
284 c
496 c
229 c
301 c
< 0.0001
724 a
828 a
568 a
123 b
257 b
124 b
157 b
< 0.0001
10311 a
11031 a
5700 b
1207 c
3060 bc
1526 c
1529 c
< 0.0001
1281 a
1468 a
767 b
193 c
454 bc
290 c
221 c
< 0.0001
12703 a
15696 a
6170 b
1552 c
4238 bc
3392 bc
2254 c
< 0.0001
1212 b
1517 a
494 c
148 d
384 cd
291 cd
216 cd
< 0.0001
3527 b
6938 a
1331 c
657 c
1559 c
3887 b
1100 c
< 0.0001
282 b
581 a
99 c
37 c
48 c
328 b
66 c
< 0.0001
47 b
101 a
14 bc
1 c
22 bc
26 bc
27 bc
< 0.01
31 ab
53 a
11 b
9 b
20 b
16 b
10 b
< 0.05
82 ab
89 a
30 b
47 b
24 b
42 b
30 c
< 0.01
132 a
60 b
50 b
68 b
13 b
18 b
28 b
< 0.05
100 a
53 bc
71 ab
30 cd
14 d
9 d
14 d
< 0.01
233 a
192 ab
126 b
44 c
43 c
19 c
23 c
< 0.0001
242 a
248 a
131 b
57 c
49 c
66 c
38 c
< 0.0001
516 a
369 ab
459 a
71 c
111 c
223 bc
140 c
< 0.0001
87 ab
130 a
61 bc
52 bc
57 bc
67 bc
22 c
< 0.05
42 c
331 a
42 c
82 bc
131 bc
156 b
91 bc
< 0.01
69 a
70 a
48 a
24 a
40 a
14 a
12 a
ns
114 a
71 b
14 c
6 c
7 c
32 c
6 c
< 0.0001
36 b
137 a
24 b
22 b
30 b
6 b
14 b
< 0.01
22 bc
83 a
71 a
18 bc
50 ab
26 bc
6 c
< 0.05
161 a
124 a
177 a
30 b
37 b
17 b
7 b
< 0.01
248 a
169 a
262 a
27 b
28 b
11 b
24 b
< 0.01
276 a
183 a
242 a
29 b
41 b
16 b
22 b
< 0.01
123 a
103 a
96 a
18 b
18 b
21 b
17 b
< 0.0001

ns = non significant.

Our results are only valid for a season characterised by warm climatic conditions during the maturation period. They might not be generalisable and transferable to vintages with cooler conditions. However, it must be pointed out that most of the newly developed genotypes are generally grown during the first years in greenhouses under semi-controlled environmental conditions. Under these growing conditions, temperatures are expected to be warm, and seasonality is likely to have a weaker impact in comparison with field-grown vines.

Our work highlighted that the SIFT-MS fingerprint of Syrah berries was stable from 50 % ver. +28 days under warm conditions of climate. This result is particularly relevant for the future high throughput phenotyping of berries under warm conditions of maturation as it enables to simplify the sampling strategy greatly. The proposed strategy only relies on phenological data and does not require accurate monitoring of physico-chemical parameters. In most cases, a decrease in abundance was observed over the maturation period, which could be the consequence of volatilisation or an increase in glycosidically-bound compounds that are not volatile and cannot be detected through SIFT-MS measurements. Additional research would be necessary to test this approach over more than one season or in greenhouses and to improve the model to get access to the full grape aroma potential through preliminary acid or enzymatic hydrolysis preparation step.

This study was carried out with financial support from the Occitanie region through the funding of Thomas Baerenzung dit Baron PhD thesis. We are grateful to Leticia Vitola Pasetto of Toulouse INP-Purpan for her assistance in the SIFT-MS analyses.