The Fresh Loaf

A Community of Amateur Bakers and Artisan Bread Enthusiasts.

Initiating a new starter with fruit and veg

greyspoke's picture

Initiating a new starter with fruit and veg

I hadn't really considered this, but I have been fascinated by the question of where all the microbes in a starter come from, particularly in the light of the paper by Landis et al The diversity and function of sourdough starter microbiomes , which has of course been discussed here.

The question of why these starters are different is not really addressed by the research, but in an interview with The Bread Code Landis herself suggested that if someone's starter wasn't doing well, they should buy/borrow one and keep that as she had found that starters were quite persistent, ie a mix of microbes that works will survive in a range of environments.

Landis also wrote a nice summary of the types of yeast typically present  in sourdough on  This particularly interested me, as she says that Kazachstania servazzii is responsible for high CO2 production, yet her final research indicated that this is not present in many souordough cultures, in fact in many the predominant yeasts are types of Saccharomyces Cervisiae, ordinary bread yeast.  Which (S. Cerevisiae) doesn't like acid conditions so much and would not be optimal (according to established lore and probably research also).

Doing some further research I found an article by Ripari, Ganzle and Berardi, Evolution of sourdough microbiota in spontaneous sourdoughs started with different plant materials.  My take-home from this was that, so far as the lactic acid producing bacteria are concerned, some important ones, including L. Sanfranciscensis, are not ubiquitous and so may not arise in a from-scratch starter - but once introduced, such bacteria will establish themselves.  It also mentions using plant and flower material in the initiation stages of the starter to provide a source for the full range of bacteria.  This reminded me of my grandfather feeding his yeast potato peelings.

Putting all this together, it appears to be a reasonable hypothesis that some yeasts also might fall into the category of microbes that might, but might not,  find their way into a from-scratch starter.  Probably the ubiquitous S. Cerevisiae will, not necessarily the others.  Or more generally, lack of availability of microbes in the environment in which a starter is initiated can limit those that are found in the final, stable culture.

This struck a chord with me as my starter, though it makes tasty bread that is not too dense, is not a massive riser.  If I wait until my bulk dough has expanded 50% by volume, it will be way overproofed.  20% or less works for me, though clearly others have different experiences.  Maybe mine is one of those without the gassy sourdough-specific yeasts in it.

So I have decided to kick off another starter using Debra Wink's pineapple juice method, but adding blossom/flowers and skins of fruit (nowhere near ripe pears and plums) from the garden, and also peelings and trimmings from some carrots and a cabbage from our local organic supplier (muck brushed off, but not washed).  For the flour I am using a bit of everything in the cupboard, all in the cause of biodiversity.  

So far, it has behaved a little differently, after a couple of days I was getting behaviour quite similar to a mature starter - rising well in the same sort of times.  It is now at 4 days, I am not able to judge the aroma yet as it is just coming off the pineaple juice and has some vegetably smells as well.  Once it has been going for a week or so and I have stopped feeding it fruit and veg I will try a side-by-side feeding cycle with my existing starter to see if there is any difference and report back.  And if it works, then a loaf test...

(I am posting this now to make sure that I do report back and carry on with the experiment!)

Cheers, TIM


Debra Wink's picture
Debra Wink

I kicked off a new flower-inoculated starter on June 1. It has been really lovely, and yes, more like a mature starter early on. It was strong and would have been ready for the levain build the evening of day 5, and baking bread on day six if I hadn't waited for the negative control to catch up, but I wanted to bake side by side for a comparison. Both made really nice bread, but the two breads had different aromas (as did the ripe starters that went into them), and the flower-inoculated one was noticeably more complex in flavor. Will share more details and compare notes as time permits.

My best,

clazar123's picture

I will await your observations.

I once had a starter I called Wild Child. It rose furiously when fed but when it came time to raise a loaf, it didn't seem able to sustain the rise to a full proof. It died a natural death only a few weeks into its life. I always wondered what wild beast I had captured. I had a basement office in a county government building and that is where I cultured my starter with AP flour brought from home and a bottle of spring water. It lived on my desk so I could stir it multiple times a day for a week before I brought it home.

I have a starter sold as part of a tourist momento from San Francisco (Sourdough Jack) and a starter from a friend who claimed her starter was over 70 yrs old at the time (now it is 85y/o) and was from an ancestor that brought it west in a covered wagon. Might just be family lore but who knows?

My 3rd starter I started from grape skins from my daughters garden. Lovely starter.

All 3 starters have become more similar to each other over the years but still retain unique characteristics, which is why I keep the 3.

At one time, I cultured so many foods (veggies,dairy,bread,fruit,teas) that nothing rotted or got moldy in my kitchen or refrigerator-it fermented. If fresh fruit had a bruise as it sat on the counter, it would start bubbling and fermenting the next day. Micro-organisms are ubiquitous and busy-that's for sure. I look forward to following this post.

greyspoke's picture

...particularly as you have similar sounding experiences Debra nd clazar.  I will indeed keep you informed.  In the meantime, here are a couple of photos of the thing as it was just now, about at its peak.

Debra Wink's picture
Debra Wink

This series may continue or may move to a new location in the future so as not to monopolize greyspoke's. In all photos, the negative control (flour+water) is on the left, and the test (flour+water+flowers) is on the right. I wish you could smell them, because that was the biggest difference throughout. 

A little background first. The fascinating article by Ripari, et al that Tim linked to, came to me last fall by way of Michael Gaenzle, one of it's co-authors, after asking about his work with fruit flies and whether he was able to establish a connection between them and L. sanfranciscensis (now Fructilactobacillus sanfranciscensis). But, they turned out not to be the smoking gun as hoped. However, even though a specific reservoir for this bacterium is still unknown in nature, there does seem to be an insect connection, which this 2016 paper points to.

One arm of the study involved creating a series of new starters from scratch using various plant materials as inocula (particularly flowers and berries), and finding that F. sanfranciscensis was present in a number of them from very early in the process. Some as early as the first fermentation cycle -- something unheard of in those initiated by flour and water only. When I mentioned that many American bakers believe (sometimes righteously so) that only flour and water should be used to start new sourdough, he seemed genuinely surprised, saying that European bakers generally use "stuff" (diverse plant material or even manure) to get started. This paper certainly provides scientific support for that.

Unfortunately, I had to wait until spring to try it for myself, since nothing was in bloom at the time. I have a wild plum tree in my backyard that blooms profusely for 5-7 days in early April, and all kinds of insects visit the blossoms. After I saw that it was all abuzz I waited a few days more to let them interact before collecting flowers and putting them into a paste of unbleached all-purpose flour and water, hoping the flowers would provide all the microorganisms needed "to accelerate the establishment of suitable fermentation microflora." It did speed up the succession of bacteria as noted, so it was a success from that standpoint, but didn't really progress to rising within a few more days, so I abandoned it. Unfortunately, the plum tree had shed its flowers by then, so I had to wait for more plants to come into bloom.

The first time around I had followed the procedure in the paper which was refreshed 1:5:5 (starter:water:flour) every 48 hours at ambient temperature. The initial mix being the same ratio, but with the flowers standing in for starter. Perhaps the flowers didn't have much yeast in them, the yeast wasn't sourdough compatible, or the refreshments were just too big and flushed them out (as well as acidity) before their time. So, for the second trial, I used whole wheat flour to more or less guarantee the presence of yeast, optimized the procedure to get the most milage out of the flowers, and added a negative control in hopes that there would be a noticeable difference between the flower-inoculated and flour-only starters in lieu of formal laboratory analysis. Because, if a difference isn't discernable (and positive), what's the point?

So here's what I did. I went outside and collected a few flowers from just about everything I could find that was in bloom that day. If I can't isolate and ID the microorganisms, I can at least increase the odds of recruiting F. sanfransicensis by collecting from many different sources. On June 1st that was forget-me-not, coreopsis, daisy, bugleweed, white clover, wild blackberry, strawberry and allium, dandelion, phlox, a shrub I've forgotten the name of, plus some immature dogwood berries from a tree that had bloomed in April. A few were landscape plants, but many were just growing wild near and around our home. A small bowl-full turned out to be only 10 grams -- surprisingly light -- but that was enough.

I added 50 grams of water and pulverized the flowers with an immersion blender. It was too small a volume to get very smooth, but it was good enough. When I added the 50 grams of Bob's Red Mill whole wheat flour (hard red spring wheat), the mixture was too stiff, so I added water a little at a time until it reached about the same consistency as the 100% hydration all-purpose starters that I'm used to. Then I mixed the negative control to the same total weight and consistency. Making up for the weight of flowers with 50-50 extra flour and water did the trick. You can see that the Test was chunkier due to the flowers, and it needed a higher percentage of water to make up for their surprisingly low water content.

TUESDAY, June 1, noon

                        Control               Test
BRM ww     100 %   55 g     100 %   50 g
Water          155 %   85 g     160 %   80 g
Flowers          0 %     0 g       20 %   10 g
                               140 g                140 g

0 hours.   pH 5.5  in both.   Lots of gluten.

(I didn't think to record the pH until a few hours later, so that was read at 3:30 pm, but I included it in the stats above  for continuity.)

   Smells like wet flour.

Test:   Smells like raw plant materials (grassy), but with a hint of onion from the wild allium.


WEDNESDAY, June 2, noon

24 hours.   pH 5.5 in both.   Gluten still intact.

   Smells a little malty, with a grassy note that comes out when stirred. No expansion.

Test:   Smells a little like ham, and is slightly greyish on top. Some bacterial expansion.

No refreshment. Stirred down and scraped sides clean to reset. 


THURSDAY, June 3, noon

48 hours.   pH around 4.5 (slightly lower in Test than Control).   Somewhat less gluten today.

Control:   Some bacterial expansion --- has peaked, but not fallen enough yet to leave much of a tide mark. Whitish bottom sediment appears to be mainly starch. 

Test:   Earlier and slightly higher rise than Control --- at least 50% (note the tide mark). Top surface a little more grey today.

No refreshment. Stirred down and reset. 


FRIDAY, June 4, noon

72 hours.   pH about 4.0 in both.   Gluten completely disintegrated now.

No bacterial expansion in either culture in the last 24 hours, but continued acid production and both are thin and loose today. (I know this is what strikes panic in the uninitiated, but this is good!)

They are both now ready for a small refreshment --- 2:1:1 (starter:water:flour)

BRM ww     100 %   35 g
Water          100 %   35 g
Culture       200 %   70 g
                               140 g

Saturday, June 5, noon

24 hours.   pH <4.0 in both.

Control:   Started rising later and had not started to recede. Smelled pasty. 

Test:   Had almost doubled and was receding. Smelled faintly of cured sausage (lactic acid + allium?). Still contains 1/2 of the initial flowers at this point.

Both are showing yeast activity now and are ready for another refreshment. Needs more water than yesterday because they are thicker today.

Refreshment (10:6:5):

BRM ww     100 %   35 g
Water          120 %   42 g
Culture       200 %   70 g
                               147 g


Sunday, June 6, 7:30 am

19.5 hours.

Good rises overnight, and they smelled yeasty. Both need more aggressive refreshing now. Test could have advanced yesterday, but Control was still lagging behind, and I decided to keep them on the same schedule. The less whole-grain appearance is from the stronger lighting. At this hour, there is direct sunlight coming in the east-facing window.

Refreshment 8:00 am (1:5:5, changing flour to King Arthur all-purpose):

KAB ap      100 %   55 g
Water         100 %   55 g
Culture         20 %   11 g
                               121 g

4 hours   (Sunday, noon)

Both on the rise, neck and neck. Started rising less than 2 hours after the morning refreshment.

12 hours   (Sunday, 8:00 pm)   

No photo available, but same as next photo (Monday morning's). Both showed strong rise and had peaked at 3x volume. Ready to make bread.

Control:   Faint orangey aroma. New to me, but I like it.

Test:   Pleasant floral scent. Note that there is less than 0.2 gram of the initial flower material present in the whole thing at this point, and less than 0.02 gram in the levain build (next).

Levain build, 8:00 pm (1:5:5):

KAB ap      100 %   55 g
Water         100 %   55 g
Culture         20 %   11 g
                               121 g

Monday, June 7, 9:40 am

13.7 hours

Levains are both mature and ready to inoculate dough. They still have their lovely orange (Control) and floral (Test) aromas when fully ripened. Since the flowers are essentially gone (less than 2 hundredths of a gram theoretically), the differences in smell can be attributed to the microorganisms at this point.

For the bake I used the formula and process for Vermont Sourdough from Jeffrey Hamelman's Bread, 3rd ed. -- 90% "bread" flour, 10% whole rye, 70% water, 1.9% salt, with 15% of the total flour prefermented in the levain.  After adjusting for the 100% hydration starter and levain, here is my working formula:

KAB ap       88 %      218.2 g
BRM wr      12 %        29.1 g
Water          65 %      160.0 g
Salt            2.2 %          5.5 g
Levain        35 %        87.2 g
                  202.2       500.0 g

There was a slight difference in the doughs at the beginning (at the end of mixing). Test came out of the mixer as a nice smooth, strong dough, and Control came out slightly looser and cottage cheesy. They each got exactly the same times, temperatures, and treatments throughout, but staggered by 45 minutes. At the first fold, the differences noted after mixing had disappeared, and they behaved pretty much the same the rest of the time.

And the bread ... 

They looked the same and were really nice breads, but with different aromas. They both had a wonderfully intense "bready" flavor and crust aroma that anyone could be proud of, but the flower inoculated loaf had something extra. The orange scent didn't carry through as such, but the floral one did in a nice subtle way, and the flower-inoculated bread was noticeably more complex in flavor. 

Monday, June 21

This bake was a repeat of the first loaf (14 days later) to see if the starter's character has changed in that time. I did a third bake on July 15, and after each bake I allowed them to cool overnight and tasted the next day. Then froze half the loaf for a side-by-side tasting and comparison of all three.

I also did an offshoot by converting part of the starter to 50-60% hydration for 7 days after the second bake, and baked a loaf from that on June 29 using the same overall formula as before. The working formula for that one (with a 60% hydration levain) looks like this:

KAB ap       88 %      218.2 g
BRM wr      12 %        29.1 g
Water          72 %      177.4 g
Salt            2.2 %          5.5 g
Levain        28 %        69.8 g
                  202.2       500.0 g

Sunday, July 18

I should first talk about the starter. The aroma did change a bit in between the first and second bake in that the floral notes have almost disappeared behind stronger sourdough smells of increasing, albeit mild, acidity. It retained strong leavening throughout. The floral scent is still there, but it comes out after the stronger smells have had a chance to dissipate, particularly after scraping out the jar and letting it air for a minute or two. That's when it is most aromatic and enticing. I also occasionally detect the same orangey aroma I noted in the control. So I think it has (so far) kept the best microbes of both the flour and flowers. Is one of them F. sanfranciscensis? I have no way of knowing, but if it is there, it will likely persist only as long as I maintain it at room temperature. According to Gaenzle in his recent presentation at the On The Rise symposium, sanfransicensis isn't an inhabitant of refrigerated starters. So that's a bummer.

On to the results of the side-by-side tasting. It really paralleled nicely with the aroma observations. All were very nice, with the same basic flavor. The differences were mainly in acidity. The day-7 loaf had a wonderful complex and bready taste, but no real reaction in in the back of my mouth where the sour taste receptors are. The 21-day loaf had a little sourness in the aftertaste, and the 45-day loaf had a little more. So a trend of increasing acidity where I didn't think there would be a noticeable difference. The day-29 loaf made from the stiff levain offshoot, was almost identical to the very first loaf, evidence that stiff starters (all else being equal) are milder than their liquid counterparts.

greyspoke's picture

Wow that is positively frothy!  Mine never goes like that.  Looks like 100% white flour?  Almost ready to start a side-by-side test with my existing starter.  I thought of starting another one as a control, like you have Debra, but it seemed like too much work.

Debra Wink's picture
Debra Wink

It was all whole wheat through Saturday, then all-purpose after that. Running a control is more work, but is also invaluable :)

DanAyo's picture

I have no input at this time, but am interested in this subject. The reply is to be notified as more info is posted.

DanAyo's picture

What a back and forth race! Each starter falls back, then races ahead. Day 7 seems to produce identical results.

A wise man’s momma once said, “live is like a box of chocolates, you just never know what you gonna get”.

alcophile's picture

What is the risk of toxicity from flowers or fruit using this method? Should only known edible plant materials be used?

There are many garden plants that are toxic and may be unwittingly used by people unfamiliar with a plant's toxicity. I can think of common plants like digitalis, pokeweed (Phytolacca americana), monkshood (Aconitum), and larkspur (Delphinium spp.) that might be toxic.

Mini Oven's picture
Mini Oven

I got some herbs and veggies blooming, think it better to use them.  Orange, rose, marrigold, petunia, clover, basil, marjoram and thyme. 

Debra Wink's picture
Debra Wink

I don't think anyone is advocating the use of poisonous plants, but the flowers are gone within a relatively short time. My first loaf on day 7 had 0.01 grams of the initial flowers (theoretically). That works out to 0.005% by baker's math. Assuming any were toxic, and those toxins didn't break down in the sourdough over time, they would account for only a small percentage of the flower weight. And most people don't bake with a new starter as soon as I did here, so the more feeding, the more flushing. Luckily, I have survived :-)

alcophile's picture

I did not mean to imply that you were advocating the use of poisonous plants. I'm just overly cautious and some members may not be aware of the toxicity of certain flowers and fruit. Some flowers and fruit may be more suitable for use in food and others are not.

Do you think the bacteria and/or yeast may consume the toxic compounds? Some are proteins that could be broken down, but others are alkaloids or other small molecules that may require more harsh conditions (pH < 1 or > 12, high heat, etc.).


Debra Wink's picture
Debra Wink

But at least they would be flushed out by repetitive feeding. How fast depends on the refreshment rate and frequency.

All best,

greyspoke's picture

I see your point.  But as Debra noted, poisons in the plant matter will be heavily diluted by the subsequent feedings, and I did dredge out the big bits by spoon.  In fact, it maintained a slightly green tint for a bit, but that went away.

Bugs that are bad for you themselves would be killed by cooking (eating raw sourdough is probably not a good idea).

The concern would be one of those bugs that itself produces toxins, like the ones that live in rice or ergot fungi (which can infect cereals).  But I am not sure how much of a concern this would be with flower and vegetable matter.

alcophile's picture

I agree that any toxic component of flowers or fruit is diluted through refreshment and use at small percentages in bread dough. But I do note that the authors of the Int. J. Food Microbiol. research article (Table 2 and text) used only edible plant material in their preparation of de novo sourdoughs.

Your sourdough experiment also used only edible plant material; others may choose material that is not edible and, unwittingly, potentially toxic. I just thought it might be prudent to err on the side of caution.

Benito's picture

Very interesting Tim and Debra.  I will follow this with great interest.


albacore's picture

Well done Greyspoke for your research on Kazachstania servazzii and to you and Debra for your ongoing starter experiments. I wish you success.

Out of interest, I had a look in the Puratos sourdough library to see how much Kazachstania servazzii there is in working SD cultures. The answer appears to be not much, at least in Europe - I haven't looked elsewhere.

I found one bakery in France and one in the Netherlands. The French one is Bruno Cormerais. He has two cultures listed and one is exclusively Kazachstania servazzii. The Dutch baker is Bij Robert - mixed culture with S. cerevisiae.




greyspoke's picture

Right, here is the side-by-side starter feeding comparison.  


The veggie starter had been feed 2 or 3 times a day at 1:2:2 and 100% hydration with 50:50 strong white bread flour/ wholemeal bread flour until 10 days in, when it appeared to be behaving in a stable way.  Meanwhile my original starter was put on the same feeding schedule for 2 days. (I stopped putting vegetable matter in the new starter after 4 days).


I noticed during this time that the aroma from the new starter changed from a markedly acetic one, to one more like the existing starter, that is a slightly musty smell.  But when comparing the two, there was a definite different note to teh new one - a bit cheesy and a bit spicy as well.


I then put the two side-by-side in glasses for an easy comparison.  Timings are marked on the sides of the glasses, the old starter is on the left in all the photographs.

almost there...


at the max - I biffed the new one a bit and I think it deflated a little, it was definitely bigger, but not by much.

So, to summarise, a small change in behaviour and a somewhat different aroma.


A bread comparison is underway as I type this, so far so similar...

Debra Wink's picture
Debra Wink

My experience was similar. The flower-inoculated one was ready for business a day sooner than the flour-only one (or would have been, had I not waited for the control to catch up), but the main difference between them was aroma. Both were pretty equal in leavening power. I did not taste them though.

greyspoke's picture

And here is the loaf comparison.  Overall formula:

  • SWBF                                      66%
  • Wholemeal wheat bread flour 10%
  • Wholemeal rye flour                10%
  • Wholemeal spelt flour             10%
  • Vital wheat gluten                     4%
  • Butter                                        4%
  • Skimmed milk powder              3%
  • Salt                                           2.1%
  • Hydration                               78%

I made two 250g (flour weight) loaves in small tins, one using each starter.

As this was an overnight room temperature (23C) ferment, innoculation was 4.5% pff of starter, which was 50:50 SWBF:wholemeal bread flour at 100% hydration.  Simple method - mix it all up roughly (butter was dribbled in to the dry ingredients melted and then rubbed in a bit), leave for 20 mins, mix more thoroughly (all by hand as I don't have a mixer), leave overnight and four in-bowl S&Fs in the morning before it was done - bulk ferment ended up taking just under 11 hours.  Shape and into tins, another 80 mins then score and bake.  I wouldn't have pushed it so far for a free-form loaf (and it would have gone into the fridge after bulk), but I wanted to develop the sourness to see if there was a difference.

This is a favourite everyday loaf as it makes good toast, can be sliced nicely for sandwiches (quite soft crumb) and lasts a good few days.  I fiddle around with the wholemeal ingredients and sometimes stick some grains or seeds in. I find it difficult to get flavour to balance the sourness with low wholemeal levels.  I bake for 15 mins with steam then these took a further 35 mins at 180C - this gets a soft crust.  My oven manages simultaneously not to get really high temperatures and burn things, I find ali foil underneath and a baking tray on top deals with this.

The tins are small (~1lb loaf) tins and I think the loaves look quite cute.  My tin loaves always end up camel-shaped unless I divide the dough into bits as in a Japanese milk bread, or overproof the dough. The tins are not brilliant, being folded not pressed they have sharp corners which stick a bit, also there are gaps with some burn-through which you can see in the photos.  But ideal for this experiment as I have a pair of them.  The loaf with the old starter has a few poppy seeds on top for identification.

Old starter loaf on the left in all pics.  I reckon the new one got a bit more lift, but that could be down to variation in the processes.


After all that, damn near identical loaves to the eye and, I can confirm, to the nose and tongue as well.  The slightly different aromas detectable (or so I thought) in the starters haven't come through in the loaves.  They both taste great and came out well (good enough for me anyhow).

My interpretation of all this is that during the early starter development stages yeasts from the vegetable matter got going early and were responsible for the rise and aroma in the starter, but after a few days with no vegetable additions, they fell away (possibly in response to the lactic acid bacteria colony becoming established) and the same yeasts as are in my original starter came to predominate.  So this appears to be a good way to get a starter going at any rate.

The challenge is to find a natural source of yeasts/bacteria that will survive through to the "flour only" maintenance phase of the starter and give it a different character.  And if such things exist, it ought to be possible to introduce them into an existing starter rather than starting from scratch.  Alternatively, feed with plant matter and use immediately to retain the difference?

Well, it's been fun and I may have another go when I have time. 

Thanks everyone for the helpful and encouraging comments, looking forward to hearing what happens with your starter Debra.


Debra Wink's picture
Debra Wink

The slightly different aromas detectable (or so I thought) in the starters haven't come through in the loaves.

In my case, the differences did come through in the loaves. They had different aromas and flavors. Both had intense "bready" flavor and crust aroma that didn't dissipate in a day. But the flower inoculated loaf had something extra. An additional floral scent and more complex flavor. It really was the better of the two.

I used Jeffrey Hamelman's Vermont Sourdough formula (same day process, not retarded), because it has a simple ingredient list, doesn't have too much whole grain or sourness to compete with and mask the more delicate and subtle differences that we're comparing here.

There is more to flavor than acidity, and it's the production of all the other flavor compounds that vary from one starter culture to the next. That can get lost or hidden behind sourness, whole grains and enrichments. Sourdoughs will all produce lactic and acetic acids to some degree, but it's their many other flavor and aroma compounds that set them apart from each other.