Many species or species complexes are known in which there are internal breeding barriers, i.e. where different races are genetically unable to inter breed, at least to form fertile hybrids. In most instances, the underlying cause is chromosomal—usually a difference in chromosome number, but sometimes in chromosome structure {structural hybridity). Several examples of pairs or groups of phenetically similar taxa different in chromosome number are given in Chapter 5. Others are Empetrum nigrum (2n = 26) and E. hermaphrodi-
tum (2n = 52), Monotropa hypophegea (2n = 16) and M. hypopitys (2n = 48), Eleocharis palustris subsp. microcarpa (2n = 16) and subsp. vulgaris
(2n = 38), Lamiastrum galeobdolon subsp. galeobdolon (2n = 18) and subsp.
montanum (2n = 36), and Veronica triloba (2n = 18), V. sublobata (2n = 36)
and V. hederifolia (2n = 54) (Fig. 6.8).
In other cases, however, no obvious differences exist in chromosome number or structure, although it is difficult to be certain that there are not underlying minute structural differences {cryptic structural hybridity). The two taxa Anagallis foemina and A . arvensis both have 2n = 40 and are extremely similar phenetically, but they are intersterile. In the Cerastium arvense group it is often more difficult to synthesize hybrids between different races, even within one ploidy level, than to produce hybrids between C. arvense and other species.
2 Inbreeders
Early in this chapter the idea was put forward that inbreeding species tend to generate small intra- but relatively wide inter-populational variation. Figure 6.1 A represents a hypothetical extreme situation in which there is 100%
148 Information from breeding systems
Fig. 6.8 Drawings of diagnostic parts of three semi-cryptic species of the Veronica
hederifoiia complex: the diploid V. triloba (T), the tetraploid V. sublobata (S), and the
hexaploid V. hederifoiia (H). Taken from Fischer.134
inbreeding, and where each genetically distinct entity (biotype) is a pure line. Whereas one can demonstrate Very low levels of outbreeding (e.g. less than 0.01% in Vulpia microstachys), it is, of course, impossible to prove that any taxon is exclusively inbreeding. In practice the extent of inbreeding varies from 0% (or very close—even in most dioecious species very rare monoecious variants are known) to 100% (or very close). Accordingly the variation pattern varies in all degrees between those depicted in Figs 6.1 A and 6.1C. Jain 226 listed seven evolutionary consequences of inbreeding, several of which bear upon the degree of intra- and inter-populational variability.
In many species the pattern lies much closer to that of Fig. 6.1 A than to that of Fig. 6.1C, and the different population-types are often recognizable as distinct morphological entities. It is hardly surprising that in many genera these entities have been given taxonomic status as species or infra-specific variants. The taxa so recognized are usually capable of hybridizing, and the products of hybridization are mostly fertile and exhibit randomly recombined parental characteristics. Nevertheless, their innately low level of interbreed ing is frequently reinforced by marked ecological or geographical separation, for each biotype is usually well adapted to its local habitat, and the field botanist can become quickly acquainted with them.
As examples of predominantly inbreeding species composed of local populations which have each been given specific rank, one may quote Vulpia
Semi-cryptic species 149 microstachys in western North America (c. 8 taxa named), Senecio vulgaris
(c. 12 taxa named in Britain), Capsella bursa-pastoris (c. 70 taxa named in
north-western Europe), and Erophila verna (over 200 taxa named in western Europe, mainly France). These taxa of a lower level differ in small characters of the type whose inheritance has been much investigated in classical genetical studies: hairiness of various parts, degree of branching, shape of pod, degree of leaf-dissection, etc. In an outbreeding species these characters would be found in all combinations, the precise proportions of which usually change from generation to generation and habitat to habitat. All the above-named species are annuals or ephemerals of short-lived habitats and several of them are very successful weeds. They are all very largely inbreed ing, and in the case of the Vulpia the flowers are cleistogamous (not opening to allow cross-pollination). Clearly the ability to recognize taxa of a lower order is related to the low levels of recombination which occur in these plants, and it is each species in its wider sense which is the ‘equivalent’ of an outbreeding species.
The taxonomic recognition of entities within inbreeding species was first seriously suggested by the Frenchman A. Jordan241 in Erophila verna (Fig. 6.9), which he subjected to extensive cultivation experiments. Consequently these numerous minor taxa, for which the term microspecies is appropriate, became known as Jordanons to distinguish them from the Linnaeons or ‘normal’ species of a larger order. The Jordanons of Erophila verna were found by Winge482 to represent single biotypes or groups of similar biotypes, some of which are marked by only one or two characters. Comparable work in the genus Capsella was carried out by Shull.378 The situation in Erophila is complicated by the existence of a dysploid series with chromosome numbers
Fig. 6.9 Photographs of herbarium specimens of four microspecies of the Erophila
of 2n = 14 to 2n = 94, but this is not so in any of the other three species mentioned above.
The case against recognizing the microspecies as taxonomic species is that there are unmanageable numbers of them, and they are not usually constant over long periods. Sooner or later cross-fertilization will occur and recom binant types which themselves become new microspecies are produced. ‘There is no established taxonomic principle, however, dealing with the recognition of species in inbreeding groups’,98 and the fact that today Jordanons are mostly not recognized as species is simply a reflection of the current taxonomic fashion. In fact in different genera the microspecies are recognized at different levels, from species (rarely) to forms, or riot at all. In the Erophila verna group a compromise is often made, whereby the many microspecies are loosely aggregated into a few groups which are recognized at the species or subspecies level.
Although the taxonomy of inbreeding species has been much neglected in recent years there is every reason to believe that its placement on a firm basis would result in much benefit to those who study weedy species, as has the description of microspecies in many apomictic groups (q.v.). Population geneticists are paying detailed attention to many inbreeding species at the moment. Species which are discussed above are generally recognized as
r-strategists (species allocating a high proportion of their resources to repro
ductive activity), as opposed to K-strategists (species allocating a high proportion of their resources to vegetative vigour and competition), but these two concepts are relative rather than absolute.147
In non-vascular plants the taxonomic consequences of inbreeding versus outbreeding have been scarcely studied,382 although there are many genetical data available. A far higher proportion of non-vascular than vascular plants are obligate outbreeders, for in plants with a free-living gametophyte dioecism is common. It is present in about 80% of all liverworts and over 50% of British mosses.358 In pteridophytes gametophytic dioecism is rare, and sporophytic dioecism unknown (as it is in bryophytes).
3 Apomictic taxa
Apomictic taxa are those which reproduce by apomixis—the habitual repro duction by non-sexual means. In flowering plants this can take two forms—
vegetative apomixis, where reproduction is asexual or vegetative (N.B. only
where it replaces sexual reproduction is it, by definition, apomictic), and
agamospermy, where seeds are formed by pseudo-sexual means. Some
authors restrict the term apomixis to mean the same as agamospermy, and in pteridophytes and lower plants, where the term agamospermy is inappropri ate, apomixis is generally used to exclude vegetative apomixis. Many different mechanisms of agamospermy are known and the subject has been thoroughly reviewed by Gust afsson171,172,173 and more briefly by Stebbins418 and G rant.164 In all cases the embryo is formed from entirely maternal tissue, so that the offspring are genetically identical with their female parent.
Vegetative apomixis occurs in a great variety of situations, mostly where sexual reproduction is difficult or impossible. Examples include dioecious species where only one sex is present (e.g. Elodea canadensis in Europe), species where propagules such as bulbils occur in the place of flowers (e.g.
Semi-cryptic species 151
proliferating or pseudo-viviparous forms of Poa alpina and Allium vineale), and species which reproduce vegetatively but are sexually sterile for genetic reasons (e.g. hybrid taxa in Potentilla, Circaea, Mentha, Opuntia, Pota-
mogeton, Hyacinthus and many ferns) (Fig. 6.10). In eastern North America
four fern genera exhibit vegetative apomixis of the gametophytes, which have become dispersed independently from the sporophyte and in some instances occupy far larger geographical ranges.149 In many bryophytes sexual repro duction is rare, often because the species is dioecious and only one sex occurs in the locality. In about 14% of British mosses and a higher proportion of liverworts sporophytes have not been found in Britain, and in some taxa they are not known anywhere.358 Vegetative apomixis is probably also very common among algae.
Fig. 6 .10 Circaea alpina (A), C. lutetiana (L) and C. intermedia (I, the sterile hybrid between them), adapted from Raven.339 Although sterile, C. intermedia is far more widespread in Britain than is C. alpina, and is recognized as a species in its own right.
It is worth introducing the main terms associated with agamospermous reproduction in flowering plants. The embryo may be formed directly from sporophytic maternal tissue such as the nucellus (i.e. adventitious embryony), or via the formation of a diploid gametophyte by the circumvention of meiosis (i.e. gametophytic apomixis). The circumvention is achieved by either arch- esporial cells (by diplospory) or somatic cells (by apospory) developing directly into a gametophyte. This gametophyte develops by either an egg-like cell (parthenogenesis) or some other cell (apogamy) growing into an embryo.
Adventitious embryony is best known in various cultivated Citrus taxa. Gametophytic apomixis occurs in an extremely wide range of flowering plant families from Ranunculaceae to Poaceae, being particularly prevalent in the Rosaceae (e.g. Rubus, Alchemilla, Sorbus) and Asteraceae (e.g. Taraxacum,
Hieracium). Although a male parent never contributes to the embryonic
tissue, pollination is nevertheless necessary for the successful agamospermous development of the seed in some apomictic plants, for example Rubus
plants 1 cm
flowers 0*2 cm
fruticosus and Ranunculus auricomus. This phenomenon, known as pseudo- gamy, has various bases, but it is most often associated with the need for a
male nucleus to fuse with a female nucleus in order to produce a functional endosperm which nourishes the developing embryo. In practical biosystema- tic terms it greatly complicates the proof of sexuality as opposed to apomixis by the demonstration that emasculation and the exclusion of pollen prevent seed formation; lack of seed production clearly does not prove sexuality.
Apogamy occurs in about 10% of ferns which have, been adequately studied.459 In these cases diploid (2n) gametophytes are formed, and these give rise directly to new sporophytes of the same ploidy level without sexual fusion occurring. The diploid gametophytes are usually produced by normal meiosis which is preceded by a chromosome division without segregation of the chromosomes into separate nuclei, so that the restitution nucleus so- formed exists temporarily at the tetraploid (4n) level. Apogamy has been especially well studied in the genera Pteris and Dryopteris,
Apospory (the development of vegetative sporophytic tissue into a diploid gametophyte) can be easily induced in both pteridophytes and bryophytes by wounding. In the moss Phascum cuspidatum it may be accompanied by apogamy, as in the ferns above, but little is known of the natural occurrence of non-vegetative apomixis in bryophytes. Parthenogenesis and other forms of apomixis are well known in various algae.
The different mechanisms of apomixis outlined above give rise to a common end-result—the preservation and propagation of pure lines by essentially vegetative propagation, which produces a pattern of variation like that illustrated in Fig. 6.1 A for obligate inbreeders. In this way, therefore, apomixis could be looked upon as a most extreme form of inbreeding, since its taxonomic consequences are as might be predicted from such a breeding system. However, there is a major difference, in that pure lines of inbreeders are usually quite highly homozygous, whereas lines of apomicts are often very highly heterozygous, especially because in many cases they are hybrid in origin. Apomicts are usually composed of ‘populations’ (better called ‘aga- modemes’) of extremely uniform individuals. In some cases these may represent a single genotype, but this is certainly not always and perhaps only rarely so (see below).
Differences between these agamodemes are mostly small, but they are usually quite constant and with practice a taxonomist can easily recognize them individually. Needless to say, the different sorts of agamodemes of almost all apomictic species have been given names by taxonomists, but it is a matter of debate as to the level at which these taxa should be recognized. Because they are far more stable than the microspecies or Jordanons of inbreeding taxa, the temptation to recognize apomictic microspecies (agamo-
species) as taxonomic species is far greater. In fact for the majority of
apomictic vascular plants it is the fashion to name each of the agamospecies at the specific level, contrary to the situation regarding Jordanons. In the British Isles alone about 300 agamospecies of blackberry (the Linnaean species
Rubus fruticosus), about 250 hawkweeds (Hieracium murorum), about 250
dandelions ( Taraxacum officinale), 20 whitebeams (Sorbus aria) (Fig. 6.11), 13 lady’s-mantles (Alchemilla vulgaris) and 9 sea-lavenders (Limonium
binervosum), besides others, are currently recognized. More species have
Semi-cryptic species
153
C
Fig. 6.11 Leaves of six representative microspecies of the S o r b u s aria aggregate, taken
from Warburg.465 A, S. e m in e n s; B, S. h ib e rn ic a ; C, S. p o rrig e n tifo rm is ; D, S. ia n ca s- trie n s is; E, S. ru p ic o la ; F, S. v e x a n s .
been described in the genus Hieracium (about 10 000) than in any other genus in the world. In mosses the nine microspecies in the Bryum erythrocarpum group80 have probably arisen as the result of vegetative apomixis. The group is dioecious but rarely produces sporophytes; it reproduces mainly by means of gemmae produced on the rhizoids. In some apomictic groups, however, the agamospecies have usually been given infraspecific status, for example as subspecies within the Ranunculus auricomus group, but this reflects no more than the preferences of the specialists concerned. The use of special categor ies (e.g. forma apomictica, or f.ap.) has often been advocated, but scarcely ever used by practising taxonomists.
The recognition and naming of agamospecies can be justified on the grounds that the taxa so delimited differ as constantly as many sexual species, and they often have equally distinctive geographical distributions and ecolo gical preferences. In the sense that plant names are headings under which information concerning plants can be stored, the recognition of microspecies provides a finer and therefore more useful cataloguing of information.
The distribution patterns of agamospecies vary almost as much as those of sexual species, some being very widespread and others very restricted. This is partly based upon the dispersal mechanisms of the plants concerned. Aga-
mospecies of Taraxacum and Hieracium, which have well-adapted wind- dispersed propagules, are in general far more widely distributed than agamospecies of Ranunculus auricomus and Limonium binervosum, which have poorly developed dispersal mechanisms. Moreover the first pair are weedy genera colonizing open ground, whereas the second pair have more demanding ecological requirements and their habitats are themselves re latively uncommon and disjunct. Most agamospecies of these two genera are known from only one locality, presumably close to where they arose, yet genetically they may be of the same status as the widespread agamospecies of
Taraxacum and Hieracium. ,
More recently evolved agamospecies are also more likely to have very restricted distributions. In Rubus, Newton306 described eight grades of distribution; taxa within the more restricted grades (in extreme cases a single bush) are in general more sterile and probably of more recent origin. Newton suggested giving binomials only to plants in the first five grades and to some of category 6 (defined as containing ‘taxa which may or may not turn out to be deserving of subsequent description’).
The genus Rubus also represents in extreme form a factor which greatly complicates the issue of recognizing agamospecies as taxonomic species. Whereas many taxa in which apomixis occurs are always non-sexual (<obligate
apomicts) , some may be apomictic or sexual (facultative apomicts) . In the
genus Rubus some taxa are facultatively apomictic, producing good pollen and a mixture of apomictic and sexual carpels, whereas a few are wholly sexual and many wholly apomictic.185 In Taraxacum different plants of a single microspecies may be either wholly apomictic or wholly sexual (not both on one plant), but most of the microspecies are wholly apomictic and a few are wholly sexual.356 In ferns, too, apogamous species usually possess functional antheridia on their gametophytes and can act as male parents in hybridizations. All apogamous ferns are, however, obligate apomicts on the female side and their gametophytes bear archegonia which are all non functional.
Facultative apomixis causes great taxonomic difficulties, because it gives rise to plants which are variously intermediate between the erstwhile distinct but closely similar agamospecies. In genera such as Rubus the problem is intractable, and a rule-of-thumb as suggested by Newton is the only practical solution. The plants in Newton’s groups 7 and 8 are recent hybrids, some of which will die out and others of which will become, in time, successful agamospecies in their own right.
The situation seen today in the north-west European Rubus flora is highly dynamic and relatively unusual. It may well be, however, a stage which all apomictic groups undergo at some time in their evolution, and perhaps a stage which is undergone not once but successively, in many cases in a cyclic manner similar to that described previously as a hybridization-stabilization cycle. The two ideas are, in fact, very closely related, since the majority of (perhaps all) agamospecies are hybrid in origin. Apomixis represents, for hybrids, an escape from sterility, which in sexual terms is, in many cases, inevitable. All the agamospecies in genera such as Ranunculus, Rubus,
Hieracium and Taraxacum are polyploids, often with odd levels of ploidy
(triploids, pentaploids, etc.) and usually with very irregular meiosis. In
Semi-cryptic species 155 Limonium binervosum most of the agamospecies are tetraploids lacking one
chromosome (In = Ax — 1 — 35). Apogamous ferns in Dryopteris and other genera are also hybrids which would otherwise be sterile, for example D .
pseudomas, a triploid with In = 123.
It has been suggested that unlike the hybridization cycle of sexual plants the apomictic cycle might cease to operate when the taxa become obligately apomictic and when all related sexual species die out. In those cases apomixis can be considered a reprieve rather than an escape from sterility. However, there is some debate as to whether or not this is the usual situation. The old idea that agamospecies are single genotypes, each the product of a separate hybridization followed by the onset of apomixis, has been abandoned. In
Alchemilla it has long been known that there is much variation within many
agamospecies.461 In Lim onium , the L. binervosum group is entirely obligately apomictic, yet the pattern of variation is markedly hierarchical, just as would be expected from a sexual breeding system.224 This pattern, which must have arisen in Post-glacial times, is interpretable only by a small number (perhaps only two) of hybridizations followed by the onset of apomixis and further