Candidate loci for phenology and fruitfulness contributing to the phenotypic variability observed in grapevine

Distribution of the phenological and fertility traits in the core collection and the QTL mapping population

We observed non-normal distribution of the data for all traits and seasons, except for total fertility in 2009 (in both populations; Fig. 1a). The six phenological events defined in this study varied between cultivars of the “phenological collection” (core P), as well as in the progeny of the Syrah × Pinot Noir mapping population (SY × PN). The window of time in which each event typically occurred in the phenological collection among the analysed years was greater for véraison beginning and véraison end (both 75 days), as well as ripening (64 days), than for budburst (22 days), flowering beginning (19 days) and flowering end (16 days; Table 1). These values in the mapping population were as follows: 17 days for budburst, 12 days for flowering beginning, 14 days for flowering end, 39 days for véraison beginning, 47 days for véraison end and 31 days for ripening. Syrah was the earlier parent for budburst, while Pinot Noir was the earlier parent for flowering, véraison and ripening. Most of the progeny showed on average the timing of phenological events situated between both parents, but transgressive segregation was observed as well (Fig. 1b–e).

Open image in new window

Table 1

Descriptive statistics and comparison of the phenotypic data from the grapevine “phenological core collection” (Core P) and the F1 progeny of Syrah and Pinot Noir (SY × PN) gathered during the same years (the timing of main phenological stages in DOY-days of year)

Trait	Years	Core P									SY × PN
		N	Min		Max		Range	Mean		SD	N	Min		Max		Range	Mean		SD
BB	2008	162	96	Apr 07	123	May 02	27	109	Apr 18	5	168	100	Apr 09	117	Apr 26	17	110	Apr 19	4
	2009	162	93	Apr 03	114	Apr 24	21	102	Apr 11	4	168	96	Apr 06	114	Apr 24	18	100	Apr 10	3
	2010	161	99	Apr 09	116	Apr 26	17	106	Apr 16	4	169	99	Apr 09	116	Apr 26	17	106	Apr 16	4
							22	106	Apr 16							17	106	Apr 16
FB	2006	163	147	May 27	167	Jun 16	20	157	Jun 06	4	148	148	May 28	165	Jun 14	17	156	Jun 05	2
	2007	162	134	May 14	155	Jun 04	21	141	May 21	3	167	131	May 11	142	May 22	11	139	May 19	4
	2008	162	145	May 24	165	Jun 13	20	156	Jun 04	4	166	151	May 30	162	Jun 10	11	155	Jun 03	2
	2009	162	138	May 18	155	Jun 04	17	146	May 26	3	164	143	May 23	154	Jun 03	11	145	May 25	2
	2010	161	144	May 24	161	Jun 10	17	154	Jun 03	3	168	151	May 31	159	Jun 08	8	154	Jun 03	2
							19	151	May 31							12	150	May 30
FE	2006	163	163	Jun 12	174	Jun 23	11	168	Jun 17	2	148	155	Jun 04	174	Jun 23	19	167	Jun 16	2
	2007	162	145	May 25	162	Jun 11	17	152	Jun 01	5	167	134	May 14	156	Jun 05	22	150	May 30	3
	2008	162	153	Jun 01	168	Jun 16	15	162	Jun 10	2	166	158	Jun 06	165	Jun 13	7	160	Jun 08	1
	2009	162	141	May 21	160	Jun 09	19	153	Jun 02	3	164	147	May 27	160	Jun 09	13	150	May 30	2
	2010	161	151	May 31	167	Jun 16	16	159	Jun 08	3	168	154	Jun 03	163	Jun 12	9	157	Jun 06	2
							16	158	Jun 07							14	156	Jun 05
VB	2006	163	190	Jul 09	253	Sep 10	63	220	Aug 08	14	148	204	Jul 23	239	Aug 27	35	223	Aug 11	7
	2007	162	183	Jul 02	262	Sep 19	79	207	Jul 26	16	166	194	Jul 13	221	Aug 09	27	208	Jul 27	6
	2008	161	186	Jul 07	263	Sep 19	77	219	Aug 06	14	166	188	Jul 06	241	Aug 28	53	220	Aug 07	7
	2009	162	183	Jul 02	264	Sep 21	81	211	Jul 30	15	160	190	Jul 09	237	Aug 25	47	209	Jul 28	8
	2010	160	193	Jul 12	269	Sep 26	76	218	Aug 06	13	166	200	Jul 19	232	Aug 20	32	214	Aug 02	6
							75	216	Aug 04							39	215	Aug 03
VE	2006	163	204	Jul 23	276	Oct 03	72	243	Aug 31	16	148	228	Aug 16	247	Sep 04	19	241	Aug 29	5
	2007	162	199	Jul 18	261	Sep 18	62	233	Aug 21	17	167	201	Jul 20	236	Aug 24	35	222	Aug 10	8
	2008	161	203	Jul 21	281	Oct 07	78	241	Aug 28	19	164	212	Jul 30	274	Sep 30	62	238	Aug 25	12
	2009	161	197	Jul 16	276	Oct 03	79	233	Aug 21	23	159	198	Jul 17	267	Sep 24	69	227	Aug 15	15
	2010	160	198	Jul 17	284	Oct 11	86	232	Aug 20	18	165	207	Jul 26	258	Sep 15	51	226	Aug 14	9
							75	235	Aug 23							47	231	Aug 19
R	2006	163	215	Aug 03	276	Oct 03	61	253	Sep 10	14	88	249	Sep 06	262	Sep 19	13	257	Sep 14	5
	2007	162	212	Jul 31	275	Oct 02	63	246	Sep 03	19	143	248	Sep 05	267	Sep 24	19	250	Sep 07	5
	2008	162	225	Aug 12	294	Oct 20	69	263	Sep 19	18	166	233	Aug 20	263	Sep 19	30	247	Sep 03	8
	2010	160	237	Aug 25	299	Oct 26	62	258	Sep 15	18	88	237	Aug 25	299	Oct 26	62	247	Sep 04	14
							64	256	Sep 13							31	250	Sep 07
TF	2007	163	0.00		2.89		2.89	1.64		0.42	170	0.00		3.00		3.00	1.52		0.60
	2008	163	0.00		2.60		2.60	1.39		0.48	170	0.00		3.00		3.00	1.67		0.54
	2009	163	0.00		2.40		2.40	1.37		0.48	170	0.00		2.86		2.86	1.26		0.58
							2.63	1.47								2.95	1.48

BB budburst, FB flowering beginning, FE flowering end, VB véraison beginning, VE véraison end, R ripening, TF total fertility index)

Further evaluation of these two populations in the analysed period of 3 to 6 years indicated that in core P budburst occurred on average on April 16, with a range between the earliest and the latest variety of 23 days (Magaratch and Garganega, respectively; supplementary Table S1). The highest year-to-year budburst variability was observed for varieties Datal, Goyura and Terret Noir (SD ± 8), and the lowest for Fertilia, Humagne, Morrastel, Neretta cuneese and Ortega (SD ± 1). In the SY × PN mapping population, the average budburst occurred also on April 16.

Flowering started in the phenological core collection on May 31 on average, with a range of 15 days (the earliest was Léon Millot and the latest Bombino bianco). The among-years variability in flowering beginning was the highest for Cegled szepe (SD ± 11) and the lowest for Beogradska Rana, Bombino bianco, Dunkelfelder, Kanzler Feld O, Maiolica, Petit meslier and Zweigelt blau (SD ± 4). In the SY × PN progeny, on average, flowering began on May 30.

Flowering ended in the core P on average on June 7, with a range of 13 days (the earliest was Léon Millot and the latest Airen, Alarije, Albana, Bombino bianco, Coda di volpe bianca, Parellada and Trebbiano Toscano). The highest year-to-year flowering end variability was observed for varieties Arnsburger, Ehrenfelser, Muscat delecta, Ortega, Sicilien and Turan (SD ± 9), and the lowest for Bombino bianco and Parellada (SD ± 4). In the SY × PN mapping population, flowering ended on average on June 5.

The average date of véraison beginning occurred in the observed 163 varieties on August 4, ranging for 68 days (the earliest accession was Turan and the latest Ohanés). The highest among-years variation of véraison beginning was observed for Malvasia di candia aromatica (SD ± 19) and the lowest for Coda di volpe bianca, Fertilia, Jacquère, Maiolica and Monja (SD ± 3). In the SY × PN mapping population, véraison started on average on August 3.

Véraison ended in the core collection on August 23 on average, with a range of 65 days (the earliest was Madelaine angevine × Calabrese and the latest Raboso Piave). The year-to-year variability in véraison end was highest for Ohanés (SD ± 28) and lowest for Léon Millot (SD ± 3). In the SY × PN progeny, véraison ended on August 19 on average.

Ripening (harvest) dates for the observed varieties of the phenological collection averaged on September 13, with a range of 54 days between the earliest variety Nektar and the latest variety Ohanés. The highest variation among years in this stage was observed for Cegled szepe and Léon Millot (SD ± 31), and the lowest for Petit Meslier (SD ± 4). Dates of harvest for the SY × PN progeny averaged on September 7.

Intervals between the main phenological events are also an important measure of developmental timing. The 163 grapevine varieties in core P revealed an average interval from budburst to flowering end of 54 days. The shortest interval between these two events was 44 days (in accessions Blauer Gelbhoelzer and Garganega) and the longest 60 days (in varieties Early Muscat and Perlon). The interval between budburst and véraison beginning was 110 days on average, with a range of 62 days from the shortest average interval of 88 days in varieties Beogradska Rana and Turan to 150 days in the variety Ohanés. The period from flowering beginning to véraison beginning averaged 64 days, with a range of 64 days for the analysed years (the variety Turan had the shortest average interval of 40 days, while Ohanés had the longest average interval of 104 days). The time from flowering beginning to ripening for the 163 cultivars averaged 102 days. This interval varied from the shortest for Beogradska Rana, Contessa and Nektar (77 days) to the longest for Aspiran noir and Ohanés (126 days). The total ripening stage from véraison beginning to harvest had an average of 38 days, with a range of 53 days. Coda di volpe bianca and Pinot meunier had the shortest average interval of 20 days, while Boglarka had an average 73 day interval. The length of the interval from budburst to ripening for the region studied covered the period from early April to late October and averaged 147 days across cultivars in the collection. This interval characterized the time needed for each plant to ripen and varied by 51 days on average (from 123 days for varieties Nektar, Pinot meunier and Sicilien to 174 days for Dattier noir).

The total fertility coefficient was estimated as the number of flower clusters per number of shoots. The average fertility index value in the phenological core collection was 1.47. The highest average fertility index value 2.2 was obtained for varieties Charmont, Schiras Samling and Segalin, and the lowest average fertility index value 0.39 was obtained for the variety Braghina. The highest year-to-year variability of this parameter was observed for cultivars Malvar and Montonico bianco (SD ± 1.0), and the lowest for Coarna neagra and Contessa (SD ± 0.0). In the mapping population, the average fertility index in these same growing seasons (2007–2009) was 1.48 and ranged from 0.33 to 2.53.

QTL detection

Candidate genes for development and fertility

Phenotypic evaluation

Phenology and fertility were evaluated in two populations of cultivated grapevine with a potentially different level of variation: the germplasm collection and the offspring derived from a cross between two cultivars Syrah and Pinot Noir (the QTL mapping population). While we observed that the distributions of total fertility, budburst and timing of ripening had a similar shape in both populations, the distribution of flowering and véraison time in the mapping population generally displayed narrow peaks with many individuals finishing the particular developmental stage close to the average timing (Fig. 1b–e). Differences in the range of timing between the core collection and the QTL mapping population were more pronounced with the dates of véraison and ripening than with budburst or flowering (i.e. the DOY ranges of véraison beginning, véraison end and ripening were around twice larger for the core collection than for the mapping population, while the dates of budburst and flowering were similar in both populations). This could be explained by high phenotypic diversity in the core collection, resulting from the presence of specific accessions, such as the early ripening Nektar and the late ripening Ohanés (Supplementary Table S1).

The pairwise Spearman correlations between traits in this study confirmed that as the plants continue the growth cycle, each next event is more strongly correlated to the previous event. In former studies, it was observed that correlations between véraison and ripening dates were very high (Jones et al. 2005; Bock et al. 2011; Tomasi et al. 2011). However, in those studies budbreak was not significantly correlated with successive stages of development and thus it was regarded as an event influenced by the variable weather conditions early in the season. We, in turn, observed that the timing of budburst in both analysed populations was significantly correlated with the timing of flowering and the start of véraison (Table 2). This may suggest that genes underlying these traits are located within the same QTL regions.

The total fertility index in the present research was obtained by dividing the total number of fruit clusters by the total number of shoots per plant. It seems that climatic conditions during the observed growing seasons had an effect on the values of this parameter for all individuals in the core collection and in the segregating progeny of Syrah × Pinot Noir. Former comparative studies of fertility among grapevine varieties indicated that the differences in fruitfulness could be due to variation in cultivars, environmental factors (especially air temperature), as well as grafting and training methods (Sommer et al. 2000). Furthermore, this trait may be subjected to the growing conditions of the plant during the previous season. For instance, compared to well-exposed shoots, shoots which develop in dense shade are more likely to have nodes with less fruitful shoots during the following season (Sánchez and Dokoozlian 2005). Fertility, however, may also be affected by other factors, such as the number of flower clusters on the plant and the number of buds which were left after dormant pruning (Sansavini and Fanigliulo 1998; Morris and Main 2010). Based on our observations in the present survey during three and six growing seasons (in the core collection and the progeny population, respectively), it can be concluded that the studied trait is not genetically stable and depends on external conditions. All the plants in our experimental fields were pruned uniformly in each season; nevertheless, it may be noticed that there were quite high differences in rainfall and temperature means as well as flowering time during the analysed years, which might have affected fruitfulness (Supplementary Tables S1 and S2). The significant differences among seasons were revealed for both populations by the ANOVA test (data not shown).

QTL detection and selection of candidate genes

The QTL mapping methods typically rely on the assumption that the phenotype follows a normal distribution for each QTL genotype. In our case, almost all phenotype datasets displayed a non-normal distribution. In general, the interval mapping procedure (including the multiple QTL model and cofactor selection) is quite robust against deviations from normality (van Ooijen 2009). Therefore, we performed this method together with a maximum likelihood mixture model and the permutation test based on the actual data, rather than assuming normality. We further tested if the results of interval mapping were not influenced by non-normal distributions of the data using the non-parametric Kruskal–Wallis analysis. With these approaches, we were able to detect nine QTLs on chromosomes 2, 3, 7, 15, 17 and 18.

The QTL for flowering time on chromosome 7 has already been found using the progeny from a cross between Gewürztraminer and Riesling (Duchêne et al. 2012). This locus contains several genes implicated in the flowering process, such as VvFT (FLOWERING LOCUS T) and VvSVP1 (SHORT VEGETATIVE PHASE 1) (Carmona et al. 2007; Diaz-Riquelme et al. 2009). Other QTLs for this trait were found in progenies from different biparental crosses on chromosomes 1, 2, 6, 14, 15 and 18 (Costantini et al. 2008; Carreño Ruiz 2012; Duchêne et al. 2012). Likewise, the QTLs for véraison beginning and véraison end on chromosome 2 have been discovered previously in different studies, along with other QTLs on chromosomes 1, 3, 5, 6, 16 and 18 (Costantini et al. 2008; Carreño Ruiz 2012; Duchêne et al. 2012). The region discovered on chromosome 2 co-localized with the locus responsible for berry colour, which carries genes VvMybA1 and VvMybA2 involved in the regulation of anthocyanin biosynthesis (Kobayashi et al. 2004; Fournier-Level et al. 2010). The QTL on chromosome 17 has also been detected in the progeny of Ruby Seedless × Moscatuel as related to véraison and berry colour (Carreño Ruiz 2012). Here, we focused on the QTLs which have not been investigated yet. Below, we discuss several candidate genes underlying the QTLs for budburst and véraison located on chromosome 15, and for fertility on chromosomes 3 and 18 (Table 4).

On chromosome 15 we identified two overlapping QTLs related to véraison beginning and véraison end. These two intervals overlapped as well with the QTL for the start of budburst. Some other QTLs for budburst have recently been detected on chromosomes 4, 12 and 19 in different biparental populations (Carreño Ruiz 2012; Duchêne et al. 2012).

Among the genes on chromosome 15, we found several transcription factors implicated in bud and fruit development. For example, a plant-specific scarecrow-like transcription factor 6 (SCL6) is a member of the GRAS gene family, which controls a wide range of developmental processes, including hormone signalling and bud formation (Llave et al. 2002; Unver et al. 2010; Schulze et al. 2010). Another gene HDG1 belongs to the class IV HD-ZIP gene family. Some genes of this family are involved in epidermal development and accumulation of anthocyanin in the shoot (Nakamura et al. 2006).

Cell elongation is an important process in plant development and it is driven by cell wall loosening and turgor pressure. Cell wall remodelling depends on a complex association of physical, chemical and enzymatic processes that are controlled by hormones and environmental factors. One group of genes regulating cell wall architecture is the expansin family. Two expansin genes are located in this QTL region: VvEXPB3 and VvEXPB4 (Cosgrove 2000; Dal Santo et al. 2013b).

Other candidate genes for véraison time in this interval include VvCHS2, which catalyzes the first step of flavonoid biosynthesis (Parage et al. 2012), and GSTZ2 coding for glutathione S-transferase, which may be involved in the transport of flavonoids (Edwards et al. 2000; Braidot et al. 2008).

Several QTLs for fertility (fruitfulness) in grapevine were previously detected on chromosomes 5, 8 and 14 (Fanizza et al. 2005; Doligez et al. 2010). Here, we identified two QTLs for fertility, which were stable in at least two growing seasons: on chromosomes 3 and 18. Among the genes on chromosome 3, we found several with highly enriched GO terms, such as a cluster of genes coding for serine carboxypeptidase-like peptides (SCP, SCPL) (Table 4). These genes have been isolated from several plant species and some of them function as acyltransferases and lyases. They may be involved in a broad range of biochemical pathways, including those of secondary metabolite biosynthesis, herbicide conjugation and germination-associated degradation of seed protein reserves. Thus, they may be vital for normal plant growth and development, synthesis of compounds that protect plants against UV light and pathogens and resistance to natural and artificial xenobiotics (Lehfeldt et al. 2000; Fraser et al. 2005; Bienert et al. 2012). In grapevine, genes from this cluster on chromosome 3 have been identified as candidate glucose-acyltransferases (VvGAT-like) acting in the proanthocyanidin biosynthetic pathway. Proanthocyanidins are secondary metabolites belonging to flavonoids, which play a major role in plant protection against biotic and abiotic stresses (Carrier et al. 2013). As such, variation in these genes might influence grapevine fertility.

This QTL region harbours another key gene required in many aspects of cell wall biosynthesis: XTH, a member of xyloglucan endotransglycosylases/hydrolases. These enzymes are active in xylem and phloem fibres at the stage of secondary wall formation. They reconstruct primary walls, probably by creating and reinforcing the connections between the primary and secondary wall layers, and these cross-links may play an important role in preventing further cell expansion (Bourquin et al. 2002).

Other candidate molecules in this region include the cell cycle arrest protein BUB3, which functions in a molecular complex controlling cell division (Caillaud et al. 2009) and a precursor of phytosulfokine (PSK2), which is a sulfated peptide hormone required for the proliferation and differentiation of plant cells (Motose et al. 2009).

When searching amongst the 210 genes on chromosome 18, we found several candidate molecules potentially contributing to grapevine fruitfulness, such as transcriptional factors. For example, one candidate gene in this region belongs to a CCCH-type zinc finger family and members of this family have been shown to play diverse roles in plant developmental processes and environmental responses, e.g. in the physiological control of female fertility at the level of early embryonic development (Schmitz et al. 2005; Wang et al. 2008).

There are several genes in this interval which are involved in the response to abscisic acid (ABA). ABA is a hormone that controls the overall plant response to environmental stresses. This molecule influences also the onset of grape berry ripening (Gambetta et al. 2010). Transcription factor AREB3 is an ABA response element DNA-binding protein, which is required in initiating the long-term changes in gene expression induced by ABA (Kline et al. 2010). Calmodulin binding protein IQD32 is a phosphopeptide significantly altered in response to ABA treatment. Calcium signalling plays an important role in plants for coordinating a wide range of developmental processes and responses to environmental change. The frequently predicted nuclear localization of IQD proteins suggests that they link calcium signalling pathways to the regulation of gene expression (Abel et al. 2005; Kline et al. 2010). For example SUN, a member of the IQD gene family in tomato, influences floral and fruit morphology (Wu et al. 2011).

Also cytokinins are essential plant hormones that control various aspects of plant growth and development, such as cell division and flower and fruit formation. These molecules are involved in berry set and in growth promotion, and tend to inhibit ripening. One of the candidates located on chromosome 18 codes for a cytokinin dehydrogenase, active in the maintenance of optimal cytokinin concentration via their degradation (Fortes et al. 2011).

Another key regulator of cellular events, which is located in this QTL, is the LRR receptor kinase CLAVATA1 (CLV1). This gene and its homologues are involved in plant development and environmental responses. CLV1 may function as a signal transduction component that acts in the communication of cell division. In Arabidopsis, this kinase plays a critical role in the maintenance of the stem cells in shoot apical meristems and regulates fruit development (Clark et al. 1997; Durbak and Tax 2011).

The last proposed candidate in this interval, Endo-(1,4)-β-glucanase, has been implicated in the breakdown of cell walls during processes observed in normal growth and development, including floral abscission and fruit ripening (Nunan et al. 2001; Buchanan et al. 2012).