‘Seed heteromorphism’ is a broadly- and loosely-defined term used to describe differences in size/mass, morphology, position on mother plants and ecological function (e.g. dispersal, dormancy/germination) of two or more seeds or other diaspores produced by an individual plant. The primary aim of this review paper was to characterize via an in-depth classification scheme the physical structural design (‘architecture’) of diaspore monomorphism and diaspore heteromorphism in angiosperms. The diaspore classification schemes of Mandák and Barker were expanded/modified, and in doing so some of the terminology that Zohary, Ellner and Shmida, and van der Pijl used for describing diaspore dispersal were incorporated into our system. Based on their (relative) size, morphology and position on the mother plant, diaspores of angiosperms were divided into two divisions and each of these into several successively lower hierarchical layers. Thus, our classification scheme, an earlier version of which was published in the second edition of ‘Seeds’ by Baskin and Baskin, includes not only heteromorphic but also monomorphic diaspores, the Division to which the diaspores of the vast majority of angiosperms belong. The scheme will be useful in describing the ecology, biogeography and evolution of seed heteromorphism in flowering plants.

A cleistogamous species consists of individuals that produce both chasmogamous (open, CH) and cleistogamous (permanently closed, CL) flowers, which facilitates a mixed-mating system. In contrast to what one might expect, CL (obligately selfed) seeds and the plants derived from them can be more fit than CH (potentially outcrossed) seeds and the plants they give rise to. Our aim was to review some theoretical aspects of mixed mating in relation to retention of both CH and CL in cleistogamous species and to determine if data on germination support the notion that CL is advantageous over that of CH. Based on germination (or seedling emergence) of CH vs CL seeds in 29 species in 21 genera and 11 families of monocots and eudicots, CL seeds germinated better in 107 and equally well as in 64 of 252 case studies as CH seeds (67.9%), and the (CH < CL):(CH > CL) ratio was 107/81 (1.32). We conclude that our study lends support to the notion that production of CL seeds by cleistogamous species is advantageous over that of CH seeds. Retention of CH by CL species may be due to the need to prevent complete selfing (s= 1.0) and thus total inbreeding depression (δ), which theory predicts would decrease reproductive success. Some caveats concerning the results of comparative studies on the germination biology of CH vs CL seeds of amphicarpic sensu stricto Fabaceae species and Commelina benghalensis and the aerial cleistogamous grass Danthonia spicata are discussed.

The success of Japanese stiltgrass as an invader may be due to its ability to respond to stochastic events (e.g., by sexual reproduction via chasmogamous [CH] flowers) and to maintain a beneficial genetic make-up (e.g., by self-fertilizing via cleistogamous [CL] flowers) when conditions are stable. This paper evaluates the importance of Japanese stiltgrass seed type (chasmogamous seeds, cleistogamous seeds, and seeds originating from forest-interior [F-I] plants) in terms of seed mass, viability, and germination across variable moisture regimes (three regions in West Virginia) and at two light levels (roadside and forest interior). Seeds from nine populations were sampled in three site types in 2005 and 2008 and stored at 5 C until testing in April 2009. Seeds were tested for viability using a dye test. Seeds were germinated under both constant and fluctuating day/night temperatures. Additional samples of CH and CL seeds collected in 2008 were tested for viability again in September 2010 for a measure of seed longevity. CL and F-I seeds were smaller in mass than CH seeds. Seeds from the drier sites were smaller in mass than seeds from the more mesic sites. CL seeds, followed by F-I seeds, were less viable than CH seeds in 2005 and 2008. CL and F-I seeds had lower germination rates than CH seeds for each site type in 2005, but germination rates of the seed types did not differ in 2008. Differences in seed longevity for 2008 seeds were lower for CL compared to CH seeds, but only in the drier sites. Japanese stiltgrass' longer-lived and larger CH seeds from the roadsides may ensure population survival over the long term. Younger CL and F-I seeds differ less from CH seeds in terms of germination than older seeds, which may help Japanese stiltgrass to maintain populations under relatively stable conditions in the short term.