I was going to leave this until I had some more data, but I might as well post what I have learned so far. Firstly, let me say that I have limited first-hand practical experience in this subject (but it is on the way), but I have done a little reading recently. There has been a great deal of research into ester formation in the brewing journals, which are all in general agreement with one another. I shall blatantly plagiarise from the one I consider to be the best summary(1). I suggest that anyone interested should get hold of the original article(s).
Their concentration in beer is much too great (about 1000 times) to be made (uncatalysed) in this manner when one considers the duration of fermentation and the rate of reaction. It has been suggested that they may be formed in S. Cerevisiae (ale yeast) by a biochemical pathway involving alcohols, fatty acids and co-enzyme A (CoASH).
The fatty acids are activated by combining with CoASH to form acyl CoASH. The most common of these is acetyl CoA. These activated acids then react with alcohols to form esters, catalysed by ester synthesising enzymes (there may be more than one, depending on the activated acid concerned) but as most work has been concentrated on acetate esters (they are the most abundant), the most studied of these enzymes is alcohol acetyl transferase (AAT). If a wort contains several individual activated acids and alcohols, competitive inhibition could take place. ie. some esters could be formed in preference to others. In addition to ester synthesising enzymes, there also exist ester hydrolysing enzymes (esterases), which may break down esters to alcohols and acids. The hydrolysis of acetate esters by esterase is much slower than for other esters such as ethyl octanoate. This could help explain why acetate esters are so abundant in beer in proportion to other types.
(the following paragraph is conjecture)
As you can see, it is complicated! The important points from the above are
that alcohols, acids, CoASH and activation enzymes are required for ester
synthesis. Esters formed may inhibit synthesis of other types, as can the
concs of different alcohols. I think most would agree that ethyl acetate
(solvent, nail polish remover) can be very undesirable in high concs.
Iso-amyl acetate (banana) is also not very nice in high levels other than
weizen. (I should point out that either near threshold will provide fruity
overtones without dominating, and can be OK in some styles) Some of the
higher esters (formed from fusel alcohols) have fruity traits which *are*
desirable in many ale styles. If we could arrive at a wort composition that
inhibited synthesis of the first two by aiding synthesis of the higher
esters, we might be on the right track (for ales anyway).
(back to conventional wisdom)
As acetyl CoA is clearly related to the level of acetate esters, any factors
which increase the intracellular pool of acetyl CoA will elevate ester
production provided that a pool of (fusel) alcohols exist. Acetyl CoA plays
a key role in cell growth (amino acid/lipid/fatty acid biosynthesis/ TCA
cycle), hence any restriction of cell growth will elevate acetate esters, by
increasing the availability of acetyl CoA for ester synthesis. ie. a
deficiency of oxygen will lead to slower cell growth (O2 is required for
sterol/fatty acid biosynthesis) and hence greater availability of acetyl CoA
for ester biosynthesis.
This is not just theory, and is confirmed by results from many different experiments, all of which show greater ester levels with decreased wort oxygenation. For example(3):
8ppm oxygenated wort 4ppm low O2 wort
total esters (mg/l) 24.2 34.6
lipid compounds
C16:0 2010 1540
C16:1 1480 1080
C18:0 1500 1360
C18:1 1010 580
C18:2 880 510
ergosterol 1250 350
conditions:0.1% trub solids by wet volume, OG=1.048, 75% malt, 25% glucose,
S. Cerevisiae, pitching rate =1.5e7 cell/ml, 12-18C fermentation temp.
Notice the trends in sterol levels (in particular) and esters. These figures just refer to original DO levels. Controlled oxygenation during fermentation can lead to even more striking results(2).
It has been shown(2) to be particularly important in high gravity brewing. ie. the higher the gravity, the lower the O2 levels, the greater the esters. Therefore, as high gravity brewing is used extensively in commercial operations (to be diluted post-fermentation), controlled aeration is also used as a matter of course to bring ester levels down to an acceptable value. This is one reason there has been so much professional research on this topic (ie. it means money!).
YEAST STRAIN: Each yeast produces its own ester profile, and some strains produce considerably more than others.
WORT COMPOSITION: In an all-malt brew, where there is abundant FAN, the yeast generally thrives until the O2 supply is diminished. However, as FAN is still available, metabolites, including acetyl CoA, are still formed, stimulating ester synthesis. ie. Adjunct brews (low FAN, hence lower acetyl CoA levels) will tend to have lower esters than all malt brews under similar conditions.
The lipid content of wort can also affect ester levels. In particular, the unsaturated fatty acids, linoleic (C18:2) and linolenic (C18:3) acids cannot be synthesized by yeast and are derived largely from trub(3). They aid fermentation and decrease ester levels. ie. trub rich worts are associated with lower ester levels and better fermentation (fatty acids are associated with poor beer stability however).
WORT AERATION: results in increased yeast growth since O2 is essential in
the biosynthesis of sterols and unsaturated fatty acids. Increased yeast
growth results in additional demand for acetyl CoA for that purpose,
resulting again in lower ester synthesis. Even a low rate of aeration during
fermentation can strongly inhibit ester formation(1,2,3,4,5,6).
OPEN FERMENTATION: As David Burley comments, open fermentations aid DO
levels. They have been shown to also decrease ester levels (see aeration) (4).
PRESSURE: There are conflicting reports, but pressurised fermentation
appears to decrease acetate ester levels. This conflicts with the theory,
since yeast growth is also retarded (shows the theory is not perfect).
PITCHING RATE: When pitching rate is increased by a factor of 4, ester
synthesis is reduced.
SUSPENDED SOLIDS (eg. trub, diamataceous earth, bentonite, activated
carbon). Tends to reduce dissolved CO2 levels by acting as nucleation sites.
High CO2 is toxic to yeast, hence solids lead to better fermentation, lower
esters.
TEMPERATURE: increases esters.
(back to conjecture)
I think that basically science attempts to explain things we observe in nature. If science does not explain our observations, we are possibly measuring the wrong thing. It appears to me that most all of the research has been directed at smaller acetate esters (the most predominant), and ale yeasts, and is most likely correct in its measurements and theories concerning these. However, that same research says that ester biosynthesis can inhibit the formation of others: hence it is possible that some esters other than acetates are formed largely under conditions detrimental to acetate production! Science also says that yeast growth leads to greater fusel alcohol levels. Higher fusel levels may also form different ester profiles! (the most common ester, ethyl acetate, is formed from ordinary old ethanol and acetic acid). Hence aerated fermentations could still lead to an overall big reduction in ester levels (ie. much lower ethyl acetate and iso-amyl acetate levels) but higher levels of other esters that then become more obvious in the flavour profile(6). This could help explain the empirical evidence as reported by Tracy.
So, in short, the jury is still out as far as I am concerned. I shall continue to research this topic and report here if I find anything interesting. In the meantime, we're homebrewers, not science labs, so go try a few things and taste the results! This is why I'm interested in hearing results from individuals.
(2) R. Anderson et al. Quantitative aspects of the control by oxygenation of acetate ester concentration in beer obtained from high-gravity wort. JIB. v81. pp296-301. 1975.
(3) A. Lentini et al. The influence of trub on fermentation and flavour development. Proc 23rd Conv. Aust. IOB. pp89-94. 1993.
(4) A. Palmer et al. Ester control in high gravity brewing. JIB. v80. pp447-454. 1974.
(5) D. Quain. Studies on yeast physiology-impact on fermentation performance and product quality. JIB. v95. pp315-323. 1988.
(6) Hough et al. Malting and Brewing Science vol II 2n edition. table 22.8.D. Esters p793. also pp607-608.
Isn't it always the way? You do as much research as you can and post it off,
only to find a significant new piece of information immediately after. I
felt I had better add this tidbit, but this is it, I promise!
I discussed this ester thing with some brewers I know (from Hahn here in
Sydney), and they told me "aeration produces esters". They have practical
first hand experience with this as when they repitch yeast from their yeast
propagator, the resulting batch is always very estery and needs to be
blended out with subsequent batches. (the yeast propagator is a
mini-cylindroconical, maybe about 1-2hL, which is continually aerated to
produce vast amounts of healthy yeast).
This was interesting by itself, but one of them gave me a recent Brauwelt to
read (no. 1 /96) p18, "Zymotechnical factors and beer quality" by S.
Donhauser and D. Wagner, from Weihenstephan. The Weihenstephan yeast bank
supply yeast to commercial breweries. They tested their most popular one
(strain 34/70 - maybe there is a Wyeast lager equivalent but I don't know
which one) and varied pitching rate, fermentation temperature and aeration.
They measured such things as attenuation, fermentation rate, fermentation
byproducts (including two esters, ethyl acetate and isopentyl acetate) and
also DLG sensorics. They made no attempt to explain possible theories, but
just give raw data.
Well guess what? Aeration 24 hours after pitching *increased* the measured
ester concs.
The figures are (mg/l).
max 1 ppm 9 ppm 9ppm
9ppm with yeast
w/o yeast w/o yeast w/o yeast with yeast
and second aeration
ethyl acetate 13.3 16.4 13.4 14.1
16.5
isopentyl acetate 1.2 1.5 1.2 1.3
1.5
DLG score 30.2 31.4 33.2 35
32