Intravarietal genetic variability in grapevine varieties

Grapevine varieties are generally grapevine populations with a clear morphological identity, which can be determined by using classical ampelographic methods, or the more modern methods of molecular analysis. The identity of a grapevine variety is usually tied to a name or a description; this makes identification easier and also gives a sense of its sometimes very long history.

Nevertheless identifying a variety does not mean that all individuals within the population are completely alike. Vintners have recognised for a long time that different ampelographical variants exist within a single grapevine variety, and have gone as far as giving these variants their own names, such as Tourigão, Touriga Foufeira, Castelão de Vide Branca, Chasselas Ciutat, Negra Mole Encarrapitude, Fernão Pires do Beco, Tinta Roriz Pé de Perdiz, to name a few.

When clonal selection began in the early 20th century empirical knowledge of the existence of variability within varieties was broadened as a result of specific experimentation, which in turn led to the selection of biotypes with superior properties within the variety, which is, in itself, proof of intravarietal heterogeneity.

The idea of the existence of intravarietal variability was slow to develop, however. Even in the 1920s, an impassioned controversy played itself out between two prominent scientists in this field, Sartorius (1926, 1928) and Bioletti (1928), because whilst the former was a proponent of genetic variability within individual European grapevine varieties, the latter was unable to confirm this as being the case in respect of the same varieties in America. From the 1950s onwards, however, propagation by clonal selection experienced a considerable upswing in the major viticultural countries of the world, which led to a marked increase in the performance of these varietal clones and, inevitably, to the recognition of intravarietal genetic variability, the sole attribute responsible for such improvements. The findings of Sartorius and Bioletti, mentioned above, became more intelligible, since both admitted that the former had been working with the full spectrum of much older varieties, in their historical area of origin, whilst the latter had only a fraction of the number of varieties available for his studies, imported into America much more recently, as they were (Rives, 1961).

At the same time, theoretical advancements in the fields of genetics and evolution offered new, fundamentally ground‑breaking interpretations of the various trials conducted, which explained the apparent contradiction between the expected genetic stability and the actual heterogeneity observed in ancient varieties, as evidenced by vegetative propagation. In fact, from the supposed homogeneity of a given population, natural genetic mutation, associated with all DNA replication which precedes cellular division and multiplication, can, in itself, explain the emergence of genetic variability. Consequently, variability increases in proportion to the rate of vegetative propagation, and depends upon the number of individual plants and the number of successive years of pruning and growth cycles. Accordingly, for conditions not too divergent from the actual population, there will be a relatively close relationship between degree of variability and the evolutional age of these populations.

Thus, almost universal consensus has been reached on the existence of intravarietal variability and also on the parallel between the level of variability and the evolutional age of the populations. One would have expected this to have resulted in systematic experimentation to define the practical application (by clonal selection) of this occurrence, not to mention in research aimed at better understanding the process. Paradoxically, however, this is not what happens in the viticultural world: clonal selection now focuses on grapevine sanitation (clones free from viruses), whilst research into the theoretical aspects of variability is limited and not always supported by the most convincing methods.

Portugal, however, has developed a unique research strategy of its own which facilitates an understanding of diversity, and places great value on quantifying intravarietal variability and determining its geographical distribution, whilst enabling important practical and theoretical lessons to be learned at the same time (Martins et al., 2005):

1) Clarification as to which stock material (variability) is available for selection, and the regions in which it may be found in order to elicit its full variability potential;

2) Targeted planning as to the number of genotypes and locations from which genotypes are to be taken to ensure that genetic diversity is maintained, and to prevent genetic varietal depletion, which is currently occurring at alarming levels; and

3) Clarification of varietal origin, and of the comparative evolutional processes between varieties and between regional populations of the same variety; and cross‑referencing them with historical and archaeological information, in order to gain greater insight into the culture and history of the people.

Quantifying intravarietal variability and methods to determine geographical distribution

Genetic diversity can be investigated by means of molecular biological analysis of genetic material (DNA), or by more traditional methods, which are no less efficient when analysis focuses on the quantitative characteristics of the vine arising from the action of the DNA. Here, the yield of the plant is used as a guideline for determining variability, because it is a feature with a comparatively wide range of variation and is suitable for evaluating even large‑scale field trials.

In order to study variability, one has to begin by selecting representative sample plants from the population (the cultivar population belonging to a particular viticultural region). The selection process must follow the findings from previous field trials and their computer simulation. The sample taken must contain 50 or more plants per production region, which have previously been identified in old vineyards, and then each one grafted (thus becoming a clone) into large experimental populations containing representative samples of a particular grapevine variety from various regions.

If one considers that some varieties have long been cultivated in many different regions, then one can appreciate that populations can easily reach 300, and more, effective clones. Obviously, the presence of so many clones is highly unusual in the context of agronomic experimentation, and can give rise to complex problems in monitoring environmental variation, which we seek to solve by classical techniques of experimental delineation (example of typical design: 250 clones x 5 repetitions (blocks) x 3 plants) and by developing new data analysis models.

Yield data, collected over 2‑5 years from the second year after grafting (on site) were then analysed with a view to separating the observed variations into two categories – genotypic variation and environmental variation – and to creating a stable quantifier, the genotypic coefficient of variation (genotypic standard deviation of the quotient of the population mean). Obviously, the results thus obtained are estimates, and can vary depending on the number of years that data has been collected, on the number of test sites, and on other non‑controlled viticultural factors. In particular, the fact that different populations are tested in different locations presents some difficulties when making comparisons between varieties (but the same is not true relative to separate sub‑populations of a particular variety, because they belong to the same test). These constraints (which will be gradually controlled in the future), however, do not prevent the attainment of a new level of understanding of intravarietal variability of grapevine varieties, something which has never before been achieved by other methods.

It is immediately apparent that the CVG yield values are spread over a very wide range, which confers on this quantifier the great advantage of being able to distinguish between varieties. Moreover, the coefficients of environmental variation are generally moderate and homogeneous, indicating satisfactorily objective results. The positioning of varieties over such a broad range suggests primarily that major differences should exist between them, and between the regional sub‑populations of variety as regards its evolutional age and the clonal material available for selection. It is therefore necessary to reflect on these numbers and find, in each case, possible interpretations and implications based thereon.

Given the lack of space for a comprehensive discussion on this subject, some varieties have been taken as typical examples below, and possible interpretations of the CVG values given.

Seara Nova has the lowest CVG values, which indicates that this variety is evolutionally young. In fact, this is absolutely correct, since this variety was created through crossbreeding (about 50 years ago) and has a very limited spread.

Touriga Franca also has low variability, which likewise suggests an evolutionally young variety. Its considerable morphological affinity with Touriga Nacional lends weight to the hypothesis that the former differentiated from the latter through successive somatic mutations, or propagated via its seed. This theory may be confirmed or otherwise by molecular analysis research in the future. Because of the low genetic variability observed, it was predicted from the outset that gains from selection would be very modest. This is, in fact, what occurred, since the selection achieved a neglible increase in yield of 1.6%. For future reference, it is now known that genetic selection of this variety is not particularly useful and may not even be justified.

Jaen has low CVG values, very similar to those of the recent crossbreed Seara Nova, which seems incompatible with its much greater evolutional age. Given that this variety is cultivated solely in the Dão region, it seemed probable that it was a recent introduction from abroad. The name Jaen has Spanish connotations, which provided strong motivation to find a Spanish ancestral population of this variety. The search lasted many years, until it was indeed found in the Bierzo area (Castile and Leon) as a result of an ampelographic comparison made in 1993.

As a result, a new sampling of genotypes in the Dão and Bierzo regions was performed, and a new experimental clonal population (POP) planted; yield analysis made it possible to prove the existence of greater variability in Bierzo, which supported the hypothesis that Bierzo was the region of origin of the Jaen variety (Table 2).

The Bierzo region is criss‑crossed with “Camino de Santiago” trails taken by pilgrims travelling to Santiago de Compostela, which could well have played a role in introducing the variety to the Dão region. This is one theory which may be verified or otherwise by means of parallel studies in the future.

The Avesso variety is cultivated in a confined area in the Baião community and has a relatively low CVG. The parallel between this situation and that of Jaen (also cultivated in a small area in the interior) is striking, hence the hypothesis that the Portuguese Avesso variety may have been imported from Spain should not be discarded.

Alvarinho. Until very recently, this variety was cultivated in a very small area in Portugal (Monção), a circumstance which lends itself towards moderate genetic variability. The CVG value of 23.3 was therefore much higher than expected, which could be partially explained by the high incidence of exchanges between Monção and its neighbouring region, Galicia, where the variety is much more widespread. Once again, history could enhance understanding of the genetic situation as it currently stands, as well as bolster confidence in historical interpretations, which have hitherto been somewhat fluid.

As a direct result of these analyses, new samples of genotypes in Portugal and Spain have already been gathered, and conserved in a collection of 530 genotypes in Valença, until the opportunity to conduct fresh, more rigorous POP experiments to ascertain the variability of these clones presents itself.

Arinto. This variety is cultivated all over Portugal. It showed a relatively high level of intravarietal variability, which points to both its evolutional age and extensive morphological divergence.

By comparing the four sub‑populations (Table 2), it seems probable that the variety originally established itself in Oeste (Bucelas, Portugal), then travelled to Bairrada and, later, to Vinhos Verdes and Lafões. The fact that the variety is frequently called Arinto de Bucelas [lit. “Arinto from Bucelas”] also suggests that there must have been a time when it existed exclusively in this region, thus revealing how historical detail and the current body of knowledge enrich one another.

Touriga. This variety’s variability is neither high nor low, which makes it difficult to assess its comparative evolutional age. A comparison of the variabilities of the Douro and Dão populations (Table 2) did not give any clue as to the original region of the variety either. The most one can assume is that exchanges of material between the two regions must have begun close to its place of origin, or were repeated frequently over the course of its evolution. Assuming the former, the two regional populations are of similar age, and therefore have comparable variability. Assuming the latter, which is entirely possible given their geographical proximity, the frequent exchanges have ensured the homeogeneity of the two propulations, even if they were once distinct. Whatever the case, these present observations are indicative of a long history of extremely close ties between the Dão and the Douro.

Negra Mole. The present spread of this variety is limited to the Algarve, and even in this region its cultivation is extremely limited. The justification for its high variability has to lie not only in its very early origins but also in its widespread cultivation in the past. It is primarily this extensive distribution historically which would have resulted in the high variability now being observed, something which continues to be the case because the meagre current populations continue to be representative samples of variability which occurred long ago.


The quantification of the coefficient of genotypic variation of ancient varieties is an innovative approach which could add new knowledge on the genetic nature and evolution of grapevine varieties. When applied to the 45 ancient varieties this method is able to confirm that these varieties are distinct in terms of the amount of variability, which, when expressed as a coefficent of genotypic variation, varies between 6.8 and 42.8. The lower CVG values suggest that the varieties are younger ones, whose clonal selection provides only moderate yields, as is the case with the Seara Nova, Jaen, Avesso, Alfrocheiro, Trincadeira and Bical varieties. Those varieties with higher values, on the other hand, suggest the opposite, as observed in the Sercial, Negra Mole and Viosinho varieties.

A more in‑depth investigation into the issue of variability is possible by conducting comparative analyses of the various regional sub‑populations of the same variety. This was how the origin of the Jaen variety was traced to the Spanish Tinto de Mencia variety found in the Bierzo region. In future, similar research could be done on several other varieties.