Some experiments with yeast strains, MLF and blending

Winemaking Talk - Winemaking Forum

Help Support Winemaking Talk - Winemaking Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.


Sep 10, 2020
Reaction score
Hi all,
I've been meaning to post my findings on a yeast comparison study I did with last year's grapes. I'm a research scientist by trade and find myself turning everything into an experiment. In this case I have some (hopefully) interesting findings and experiments with yeast strains and MLF.

Reds and yeast strain
In 2020 I got 400lbs each of six reds: Barbera, Tempranillo, Zinfandel, Mourvedre, Malbec and Cabernet franc. The first four came from Shake Ridge Ranch winery in Amador and the last two from Slate Creek Hill farm in the Fiddletown AVA of Amador.

For most I split each batch into three fermentors and used a different yeast in each: D21, D80 GRE and D254 were the options (this wasn't a completely controlled experiment.) The evaluation was done around 6 months after fermentation was complete, in March, and all samples were drawn from the vat and put into wine glasses. The samples were not blind, but the people assessing the wine didn't have a close what "D80" meant, so so real bias.

All yeasts are from Lalvin:
The GRE and D-254 are fruit-forward yeasts
The D21 accentuates acidity and fruit
The D80 enhances mouthfeel and tannin

Each taster tried the samples and took notes, comparing the three yeasts on the same varietal each time.

Conclusion (which I agreed with) was that tasters placed the greatest importance on mouthfeel and fullness. The overwhelming favorite was D80, with almost everyone agreeing that (at this stage at least) this was the preferred sample. D254 was also highly rated, but it also enhances mouthfeel, though to a noticeably lesser extent. GRE was hard to gauge because it was only used in two wines. The least favorite was D21, which definitely had a higher perceived acidity, but thin mouth feel and no real additional fruitiness.

Analysis: it's hard to tell if these conclusions would hold in the final wine, and it will be hard to measure because many of the cuvées have already been blended for aging (the intent was to blend the strains to make a more complex and balanced wine, which I think succeeded since the blends do seem better than any one cuvée.) It may well be that in very young wine the glycerine from certain yeasts is smoothing the wine out and making it seem more acceptable at that stage, but that aging will accomplish this for all cuvées anyway.

Whites and yeast strain
The same type of experiment was performed with another set of yeasts for white wines: Roussanne, Viognier and Muscat Blanc. The yeasts used were D47, QA23 and D254 and GRE. Results were very similar. The wines with the best mouthfeel were overwhelmingly rated higher. The best performing yeast by far was D47, which really had a great mouthfeel and clean taste. D47 was also highest rated for aroma and fruit expression. The difference was quite stark with Roussanne and Muscat, with Muscat having an almost gewürztraminer-like rose petal aroma with D47 that was very attractive. Roussanne likewise had a floral and honey aroma that came out very well with D47. Viognier showed a preference for D254. I think this is because the Viognier has a naturally very full body and the fruitiness of D254 made the difference. GRE was not a standout in any way -people put this in the middle for pretty every wine. The surprise was QA23. people did not like what this yeast was doing at this stage. People described the aroma as muddled and indistinct. For muscat people indicated that there was a broad spectrum of aromas in the wine, but that the ones they really wanted to smell were masked by many others they liked less. People REALLY didn't like the mouthfeel of QA23, describing it as coarse and rough.

Again, this is just a snapshot in time of these yeasts, and things could change. For example, at this stage fruit is pretty subdued across the board, because it hasn't really emerged yet. The issue is that many people assemble blends at this stage, and may never see what would happen if the individual cuvées were left to mature. What really surprised me was 1) for all wines, the yeasts tended to make a big difference, 2) that mouthfeel was such a strong motivator in people determining which wines they preferred, 3) the interaction between the varietal and the yeast was not a big factor -people preferred D47 and D80 across the board and not that D80 was great for Cabernet Franc but really not for Mourvedre, for example.

I also did experiments with MLF to try and see what I could do with this to make wines with more freshness and fruit flavor. I find California wines tend to be really lacking in this compared with Europe and the North East. I was hoping to have a set of reds that would be different: more food-friendly, acid-structured and lively.

White ML
I did one 'experiment' with Roussanne. It was not a planned experiment :) I don't generally like MLF with whites but one vat of Roussanne "gifted" me with a spontaneous MLF that I monitored and liked how things smelled so I left it to go on. The resulting wine was quite a surprise, with a much fuller body (not a surprise) and interesting aromatics. It was radically different from the no-ML wines, but in a pleasing way. I have bottled the no-ML Roussanne already and it is actually my favorite white -really nice. I'm letting the ML Roussanne age some more and I'll label it as a "reserve". They will definitely be very distinct wines. I think ML really worked with Roussanne in this case because 1) Roussanne is not a fruity wine, and the aromas imparted by MLF worked well with the spicy/honey/nutty character of Roussanne, 2) this Roussanne was picked with relatively low brix of 22.5 and relative high acid of .70%, so the added mouthfeel of ML and the reduction in acid was quite beneficial. I find that many California whites that are reflexively put through ML (e.g. Chardonnay) just get overwhelmed by the process (which I guess is the whole point for 95% of CA Chardonnay, it's basically a cold-climate grape grown in a very warm climate because it has a famous name, IMO, and dressed up with oak and other flavorings to make is consumer friendly).

Red ML
This was a much more controlled experiment, since it was planned. I held out one 9 gallon demijohn each of two varietals, Barbera and Tempranillo, that tend to have low fruit expression and in the case of Tempranillo, low acid/high pH. This was protected from MLF with SO2 and (obviously) not inoculated. I also co-fermented 4 parts of Zinfandel with 1 part of Muscat Blanc that resulted in 5 gallons of red wine that did not go through MLF. This last batch was not really part of the experiment, but kind of another experiment.

I was quite surprised with the results. Even with Barbera (which was pretty acidic at .70% TA) the wine was not overly tart, and Tempranillo was very well balanced, even when pretty young. The aroma of the no-ML cuvées was much more pronounced and fresh compared with the ML counterparts, similar to wines I’ve had from coastal regions in CA from the central Coast area near SLO and Monterey than the foothills. The malic acid added an interesting “cranberry” component to the wines, especially Zinfandel and Barbera. In general, I was very happy with how they turned out -they really exceeded my expectations and actually accomplished what I’d hoped they would in capturing the grape character and liveliness of the wine.

The wines with ML were quite different, and maybe not in a great way (at this stage at least) they seemed very subdued, flat and earthy, especially Tempranillo. It was clear that ML changes a wine drastically. Interestingly, the body of the post ML wines was not necessarily improved at this stage. The body of the no-ML wines seemed “bigger” surprisingly. I believe this is because of the increased structure in the wine with both acid and tannin contributing to the mouth.

The co-fermentation was also eye-opening, but in a way I didn’t expect. I though this would be dominated by Muscat, because Muscat is um… dominating. Instead the Muscat had the effect of bringing out a huge Zinfandel aroma. I believe the terpenes in Muscat were combining with the aromatic compounds in the Zin and expressing the Zin character over the muscat. Compared with the standard Zin the difference was huge -like a 10X effect.

Since the no-ML wines were supposed to be ready early, we did some blending trials with the three cuvées. To our surprise blending all of them gave us the best wine. The Barbera and Tempranillo complement each other very well -Barbera has high acid and color, while Tempranillo has tannin and vinous mouthfeel. The Zinfandel added the fruit. We ended up with 23 gallons of the blend and bottled it in March. The resulting wine is one of my favorites -it has both tannin and acid structure, complex aroma, and is very lively, refreshing and food-friendly. It reminds me the the wines you taste all over Italy that never seems to make it to the US, because everyone know we only want big, soft wines.
Last edited:
We adjusted tartaric after a 24 hour cold soak for those wines with high pH. Mourvedre, Zin, Tempranillo and Muscat were adjusted to 6.5g/l tartaric. Viognier, Barbera, Malbec, Cabernet Franc and Roussanne were not adjusted.