POLYPLOIDY, HYBRIDIZATION, AND THE INVASION OF NEW HABITATS G. LEDYARD STEBBINS' ABSTRACT Experiments are described showing that most artificial autopolyploids derived from native or in-troduced perennial grass species from California are far inferior in behavior under field conditionsthan their diploid ancestors. In a single species, Ehrharta erecta in two out of 22 localities, theautotetraploid maintained itself for, respectively, 19 and 39 years, but remained either in the exactlocality of planting or in nearby localities having very similar ecological conditions. The controldiploids, direct descendants ofthe progenitors of the autopolyploids, spread more widely and evolvedmore variation with respect to growth habit among each progeny. These results, along with otherevidence derived from several literature sources, strengthen the hypothesis that successful polyploidsamong natural populations are usually or almost always the result of increased heterozygosity accom-panying either interracial or interspecific hybridization. The hypothesis that polyploids succeed becauseof their greater tolerance of severe ecological or climatic conditions is again rejected, and that whichpostulates secondary contacts between previously isolated populations as the principal cause for theirhigh frequencies in some groups ofangiosperms is favored. The unusually high frequency ofpolyploidsin the Gramineae is attributed to the probable fact that habitats to which they are best adapted havechanged in extent and position repeatedly during the geological periods since the initial evolution ofthe family. The family Gramineae contains higher per-centages ofspecies and races or cytotypes ofpoly-ploid origin than any other large family of an-giosperms. More than 80% of its species have undergone polyploidy some time during their evolutionary history. Polyploidy is expressed by four different kinds of numerical series, as fol-lows: (1) Multiples of an original low basic number.Examples: Triticum, Bromus, and Festuca thatinclude species having somatic numbers of 2n =14, 28, 42, ..., basic number x = 7. (2) Multiples of a secondary basic number,that was itselfderived from the original numberby an earlier cycle of polyploidy. Examples: Poascabrella complex, 2n = 42, 84, x = 21 (Har-tung, 1946); Bothriochloa saccharoides complex,2n = 60, 120, 180, x = 30 (Gould, 1966); andAustralian species of Danthonia, 2n = 24, 48,72, x = 12 (Brock & Brown, 1961). (3) Multiples of a basic number that is thelowest in its genus, but was probably derivedfrom that ofpreexisting genera by a cycle ofpoly-ploidy in the remote past. Examples: Oryza,2n = 24, 48, x = 12 and Tripsacum, 2n = 36,72, x= 18 (Fedorov, 1969). (4) Aneuploid series that most probably rep-resent successions of alloploids based upon dif-ferent basic numbers. Example: Stipa, 2n = 22,24, 28, 32, 34, 36, ... 82. All of the numberslisted in Fedorov (1969) can be derived fromvarious combinations ofthe basic gametic num-bers x = 5, 6, and 7. The existence of these four kinds of situations indicates that polyploids in this family include, in addition to those that have originated recently during the past few thousand years, other poly-ploids that are intermediate in age from one to several million years and still other genera that acquired a secondary polyploid number during the early evolution of the family 50-70 million years ago. The original basic number for the fam-ily has been the subject ofmuch speculation, but no convincing evidence has been presented for any of the various hypotheses. In this writer's opinion, the basic numbers x = 5, 6, and 7 are almost equally probable. They could, in fact, have all been acquired by the species complex from which the family first arose. Moreover, the pres-ence of x = 11 in Streptochaeta, a genus that is very much isolated with respect to morpholog-ical characters, and at the same time is a mosaic of very primitive characteristics along with oth-ers that are highly specialized, suggests that poly-ploidy, along with trends toward aneuploidy and a high degree of specialization with respect to some morphological characteristics, evolved quickly and soon after the family first became differentiated. Accordingly, the early stages ofgrass evolution must have produced many speciesSDepartment of Genetics, University of California, Davis, California 95616.ANN. MISSOURI BOT. GARD. 72: 824-832. 1985.