The yucca moth Tegeticula yuccasella is the sole pollinator of Yucca e
lata, and relies on yucca fruits to complete its life cycle. A high pe
rcentage of pollinated flowers aborts, killing the T. yuccasella eggs
and larvae in them. We examined patterns of fruit production and abort
ion in Y. elata, and related these patterns to vegetative characterist
ics, moth abundance, and environmental conditions. We studied 38 inflo
rescences throughout their flowering period, during one season in sout
hern New Mexico, USA. Each night we recorded the number of flowers ope
ning, the number of fruit formed, the relative abundance of yucca moth
s, and climatic conditions. We monitored 11786 flowers, resulting in 6
99 mature fruit. Large inflorescences produced more, but proportionate
ly fewer fruit than small inflorescences. Inflorescences flowering lat
e in the season produced proportionately more fruit than inflorescence
s flowering early. Only 6.6% (extremes 1.4-15.1%) of flowers produced
mature fruits. Hand-pollination of all flowers on inflorescences did n
ot significantly increase the proportion of flowers that developed int
o mature fruit. Fruit production appeared to be resource-, not pollina
tor-limited. Ninety per cent of observed moth-pollinated flowers abort
ed (N = 31), resulting in the death of moth eggs and larvae laid in th
e flowers opening during a 'window' of consecutive nights, lasting for
five nights on average (36% of the flowering period of an inflorescen
ce). The timing of the window of fruit production was highly variable
and unpredictable: mature fruits were produced from flowers opening at
the beginning, in the middle, or at the end of the flowering period o
f an inflorescence. The results for Y. elata do not support existing h
ypotheses that attempt to explain patterns of selective fruit producti
on. High rates of abortion of initiated fruit, and the apparently unpr
edictable pattern of fruit production by individual inflorescences, ma
y stabilize the mutualistic interaction by preventing yucca moths from
hyper-ovipositing flowers that have a high probability of developing
into mature fruits. A risk-spreading strategy of oviposition is likely
to be more successful for the moth than multiple ovipositions per flo
wer.