Starter advice is abundant and mostly anecdotal. A more useful starting point is to look at what measured starters actually look like, because the pattern of variation tells you which of your habits matter and which do not.
What is living in there
A survey characterising traditional starters reports bacterial counts of 8.26 plus or minus 0.57 log colony forming units per gram against yeast counts of 6.52 plus or minus 0.62. That is roughly a hundredfold more bacteria than yeasts, which matches the figure usually quoted.
But the individual samples scatter enormously. Ratios in that survey ranged from 2 to 1 in one sample to 241 to 1 in another. The literature it cites reports traditional sourdoughs spanning 10 to 1 up to 10,000 to 1, and a separate review gives a typical value of 100 to 1 with a range from 1 to 1 up to 1000 to 1.
So the hundredfold figure is a central tendency, not a specification. A functioning starter can be dominated by bacteria to an extreme degree or run close to parity with its yeasts.
The species identified are the familiar ones: Fructilactobacillus sanfranciscensis, Lactiplantibacillus plantarum, Companilactobacillus paralimentarius and Lactiplantibacillus pentosus among the bacteria, with Saccharomyces cerevisiae plus Candida and Kazachstania species among the yeasts. One survey of twenty-five samples from Sardinia identified twelve lactic acid bacteria species, with L. pentosus dominant in most and F. sanfranciscensis present in only a few.
That last detail is worth pausing on. F. sanfranciscensis is the organism most associated with sourdough in popular accounts, and in that regional survey it was the exception rather than the rule.
pH is stable, acidity is not
The most practically useful finding in the survey is a contrast between two measurements that are often treated as interchangeable.
Across thirteen traditional starters, pH ranged from 4.06 to 4.64. Titratable acidity, measured as millilitres of 0.1 molar sodium hydroxide required to bring a suspension of 10 grams of starter in 90 millilitres of water to pH 8.3, ranged from 8.6 to 17.1.
Expressed as coefficients of variation, pH varied by 3.9 per cent across the set while titratable acidity varied by 22.6 per cent. Acidity content is nearly six times more variable than acidity intensity.
The reason is buffering. pH measures the concentration of free hydrogen ions; titratable acidity measures the total acid present, including the large fraction buffered by flour proteins and other components. Two starters can sit at the same pH while one contains twice as much acid.
For a baker, that means a pH reading is a poor guide to how much acid a starter will contribute to a dough. It is a good guide to whether the starter is in its normal operating range.
Dough yield varies more than recipes admit
The survey also reports dough yield, the ratio of total dough weight to flour weight expressed as a percentage, ranging from 135 to 200 across the samples. A dough yield of 135 is a stiff starter at 35 per cent hydration relative to flour; 200 is a liquid one at 100 per cent.
Individual samples illustrate that the three variables move independently. Sample SD42 had pH 4.64 with titratable acidity 15.0 at a dough yield of 150. Sample SD105 had pH 4.16 with acidity 14.7 at a dough yield of 200. Sample SD107 had pH 4.06 with acidity 12.0 at a dough yield of 150.
The lowest pH in that trio came with the lowest acid content, which is exactly what buffering predicts and exactly the opposite of what intuition suggests.
Temperature has a measured effect on timing
For maintenance temperature, a review reports traditional starters being kept at ambient temperatures of 20 to 30 degrees Celsius, with five to fifteen days of continuous refreshment needed to bring a new starter to maturity.
On what temperature does to the fermentation, one quantified comparison is available: reaching a target pH took 11 hours at 25 degrees Celsius against 9 hours at 35. That is an 18 per cent reduction in time for a 10 degree increase, which is a more modest acceleration than most people expect.
The same body of work reports an ideal fermentation quotient, the molar ratio of lactic to acetic acid, of 2.0 to 2.7, and notes that temperatures above 30 degrees tend to favour bacterial activity and organic acid production. That second statement is qualitative; the review it comes from says explicitly that much remains to be studied.
What the evidence does not settle
I looked for controlled data on the two variables bakers argue about most, refreshment frequency and hydration of the starter, and their effect on the microbial population and acidity. The study best positioned to answer it exists, addressing flour type and refreshment frequency, but I could not read its results.
So the honest position is that refreshment schedule advice, whether daily or twice daily or weekly with refrigeration, does not currently rest on measurements I can point you to.
What follows for practice
Four conclusions rest on the numbers above.
Your starter's ratio of bacteria to yeasts is probably not something to worry about. The measured range spans two orders of magnitude among starters all considered functional.
Measure pH to confirm the starter is in range, roughly 4.0 to 4.6 by the survey, and do not use it to estimate acid contribution. Titratable acidity is the measurement for that, and it is the one that varies.
Record your starter's dough yield and keep it consistent, because at 135 versus 200 you are working with fundamentally different materials, and a recipe written for one will misbehave with the other.
And expect temperature to change timing modestly rather than dramatically. Nine hours against eleven for a ten degree difference is real and useful, but it will not turn an overnight schedule into an afternoon one. The variable that reliably shifts a schedule is how much starter goes into the dough, not how warm the room is.
