No adults emerged during the second year without artificial chilling (Fig. 1). However, after being chilled in the third year, 71 males and 79 females emerged (proportion of males = 0·47; binomial test, P = 0·57). In the fourth year, 18 males and 15 females emerged (0·55, P = 0·73). After the fifth emergence season, the remaining 189 seeds contained no live larvae. Thus experiencing low temperatures was essential for larvae to terminate diapause and pupate. The pronotum width relative to seed size was smaller in the third-year emergence group in the experiment than the second-year emergence group in the control, both in males (ancova, F1,174 = 0·138, P = 0·71 for slopes; F1,175 = 32·52, P = 0·0001 for intercepts) and in females (F1,201 = 0·0004, P = 0·98 for slopes; F1,202 = 25·62, P = 0·0001 for intercepts; Fig. 4, left). A similar tendency was observed in adults emerging in the subsequent year: the third-year emergence group in the control vs the fourth-year emergence group in the treatment, both in males (ancova, F1,24 = 0·03, P = 0·86 for slopes; F1,25 = 7·35, P = 0·012 for intercepts) and in females (F1,19 = 0·254, P = 0·62 for slopes; F1,20 = 8·56, P = 0·008 for intercepts; Fig. 4, right). Thus experimentally prolonged diapause of one additional year decreased adult body size. The rate of reduction in adult body size increased as seed size decreased, with one exception of males compared between the control third-year emergence group and the experimental fourth-year emergence group. Because prediapause body size is usually correlated with seed size, smaller individuals may lose a greater proportion of their mass (Fig. 4).
Male eye span exhibited body size-dependent changes: it was nearly maintained in larger individuals but greatly decreased in smaller individuals when diapause was extended for an additional year (ancova, F1,174 = 11·38, P = 0·0009 for slopes). In females, eye span decreased according to the duration of diapause in any body size (F1,136 = 1·04, P = 0·30 for slopes; F1,137 = 8·19, P = 0·005 for intercepts; Fig. 5, left). In the subsequent year (control third-year emergence group vs experimental fourth-year emergence group), the pronotum width–eye span relationships did not differ in males (ancova, F1,24 = 1·153, P = 0·29 for slopes; F1,25 = 0·52, P = 0·47 for intercepts). In females, although based on a small sample size, the pronotum width–eye span relationships showed a weak tendency for decreased eye span in smaller individuals (F1,19 = 4·62, P = 0·045 for slopes; Fig. 5, right).