PATTERNS OF PHENOTYPIC AND GENETIC CORRELATIONS AMONG MORPHOLOGICAL AND LIFE-HISTORY TRAITS IN WILD RADISH, RAPHANUS RAPHANISTRUM

Abstract

Genetic correlations can have profound effects on evolutionary change (Lande and Arnold, 1983; Mitch-ell-Olds and Rutledge, 1986). Present patterns of genetic correlations in an organism may be caused by pre-existing pleiotropic and developmental relationships among traits and may produce constraints on evolution by natural selection (Cheverud, 1984; May-nard Smith et aI., 1985; Via and Lande, 1985; Futuy-ma, 1986; Mitchell-Olds and Rutledge, 1986; Barker and Thomas, 1987; Clark, 1987a; Zeng, 1988). Alternatively , selection may directly alter the patterns of genetic correlations, especially in cases in which two or more traits interact to perform a given function (Cheverud, 1984; Lande, 1984; Clark, 1987a. 1987b). In this paper we examine the patterns of phenotypic and additive genetic correlations among 10 morphological and life-history traits in wild radish plants. We hypothesized that some ofthese correlations have been influenced by selection. We predicted that floral and vegetative traits would be uncorrelated and that correlations among the lengths of the corolla tube, pistil and stamens would be higher than the rest of the floral correlations. The results were consistent with most of these predictions. Wild radish, Raphanus raphanistrum (Brassicaceae), is an annual weed of disturbed areas. The hermaph-roditic flowers of wild radish are almost entirely self-incompatible (Sampson, 1964; Stanton et aI., 1989) and this species does not propagate vegetatively, so virtually all reproduction depends on successful insect pollination. Wild radish is pollinated by a variety of insects, mainly bees, butterflies, and flies (Kay, 1976; 3 Address for correspondence. Stanton et aI., 1989; Conner and Jennetten, unpubI. data). Pollination success is affected by floral morphology in the closely related R. sativus (Stanton et aI., 1991). Therefore, patterns of correlations among floral and vegetative morphological traits may reflect past selection on these traits caused by differential pollination success. Selection can theoretically increase or decrease correlations, and our results suggest that both may have occurred in wild radish. First, selection could increase the correlation between functionally related traits, that is, traits that interact to perform a given function. Therefore, positive correlations may evolve under selection to increase the functional integration of trait groups. Our results show that the correlations among the lengths of the filaments and corolla tube (which together determine anther placement) are much higher than the correlations among the rest of the floral traits; this higher correlation could be due to selection for effective pollination. Alternatively, selection could reduce pre-existing correlations if functional independence between traits or groups of traits increases fitness. Berg (1960) found that the phenotypic correlations between floral and vegetative parts were reduced in insect-pollinated plants relative to wind-pollinated plants. Berg hypothesized that selection to maintain a proper fit between the flower and its pollinators reduces the correlation between the flower and the rest of the plant. Our results show that wild radish also fits the Berg pattern. MATERIALS AND METHODS Experimental Procedures Seeds were collected from a population of R. ra-phanistrum in an alfalfa field near Binghamton, NY,