We tested the hypothesis that animals caching acorns for future food u
se may preferentially cache acorns high in tannin content. We used a p
opulation of Florida scrub-jays (Aphelocoma coerulescens) that cache a
corns From four oak species. Tannin activity in acorns was measured by
protein precipitating ability (PPA). Logistic regression models accur
ately predicted frequency of acorn caching among the oak species, usin
g frequency of usable acorns and PPA loss during burial as predictor v
ariables. Among the four oak species, frequency of caching increased w
ith both increasing frequency of undamaged acorns at peak harvest time
, and increasing loss of PPA during burial. Other regression models ac
curately predicted frequency of undamaged acorns for each species, usi
ng green acorn PPA and buried acorn PPA as predictor variables. Among
the four oak species, frequency of usable acorns tended to increase wi
th increasing green acorn PPA, and decreased with buried acorn PPA. We
suggest that a deterrent effect of tannins on insect larvae is respon
sible for the differences in frequency of acorns usable to jays among
oak species, and that jays select acorns for burial by assessing insec
t damage. Species differences in PPA change during burial probably rei
nforce this pattern of scrub-jay caching behavior.