Concentrations of starch in roots of seeder species of Erica from the Cape Floristic Region, South Africa were found to be considerably less than in resprouters. Shoot starch was highly variable but mean values were similar in both seeder and resprouter species of Erica. Three distinct patterns of starch storage in roots were recognized. All seeder species fell within definitions of Categories 1 (narrow major and minor parenchymatous rays, one to two cells wide with no inter-ray storage) or 2 (thick major rays up to seven cells wide and thin minor rays with small amounts of inter-ray storage) whereas resprouter species were consistently within Categories 2 or 3 (broad major and minor rays, up to eight cells wide and conspicuous inter-ray starch storage). Results are discussed in light of similar studies of the related Epacridaceae. ‘Mixed’ species (i.e. with seeder or resprouter individuals present, often in distinct populations) were always classified as belonging to Category 2. Studies of populations of three ‘mixed’ species confirmed that seeder forms had consistently lower amounts of root starch than resprouters. Rays of xylem parenchyma were the main sites for starch storage in roots of both seeders and resprouters and greater proportions of cross-sectional area of roots were consistently devoted to such storage tissues in resprouter forms of the three ‘mixed’ species. Analyses of a number of seeder and resprouter Erica coccinea populations showed that differences in amounts of realised and potential root starch storage are best explained by the effect of regeneration behaviour rather than by among-population variability.

Among citrus rootstocks, higher specific root length (SRL, root length/d. wt) has been linked to several specific morphological and physiological traits, including smaller average root diameter, higher root hydraulic conductivity and higher rates of root proliferation. In this study, thickness of the outer tangential exodermal (hypodermal) wall and its suberin layer, number of passage cells, presence of epidermis, and stelar anatomy were examined and related to variation in root diameter of field roots of known maximum age. We also compared root morphology and anatomy of young roots in the field with those of potted rootstock seedlings in the glasshouse. Fibrous roots were measured separately from pioneer (framework) roots. Among the fibrous roots, only the first-order (root links having a root tip) and second-order (root links bearing first-order roots) laterals were examined. Among first-order field roots, larger root diameter was caused by larger rather than more numerous cells in the cortex. Root diameter of first-order roots was positively correlated with both number of passage cells in the exodermis and thickness of the secondary walls of the exodermis in both field and potted plants.

Exodermal walls were about 80% thicker in field- than pot-grown roots. In the field, in more than 50% of the first-order roots examined less than 30% of the root surface was still covered by epidermis, with few differences among rootstocks. In contrast, in roots of 19-wk-old glasshouse plants generally 70–100% of the epidermis was still intact. There was no evidence of secondary xylem development in second-order fibrous roots in the field; in seedling, pot-grown rootsystems, 75–97% of second-order roots had formed secondary xylem despite their small diameter (<0.8 mm).

It is argued that there can be suites of physiological, morphological and anatomical traits in roots that co-vary with specific root length. Investigations of how root morphology and anatomy are linked to root function, moreover, need to recognize trait variability and the potentially important differences between field- and pot- grown (seedling) roots.

Summary 373

I. Introduction: Arum-types and Paris-types 374

II. Possible functional implications 375

III. Extent of the two classes in the plant kingdom 377

1. Bryophytes and Pteridophytes 377

2. Gymnosperms 379

3. Angiosperms 379

IV. Is the distinction between classes useful? 383

V. The structural basis 383

VI. The role of the fungal genome 384

VII. Physiology revisited 384

VIII. Conclusions 385

Acknowledgements 386

References 386

This review describes diversity in the structure of (vesicular)–arbuscular (VA) mycorrhizas, i.e. endomycorrhizas formed by Glomalean fungi. In particular, we consider the extent in the plant kingdom of the two classes first described by Gallaud (1905). These are: (1) the Arum-type, defined on the basis of an extensive intercellular phase of hyphal growth in the root cortex and development of terminal arbuscules on intracellular hyphal branches; (2) the Paris-type, defined by the absence of the intercellular phase and presence of extensive intracellular hyphal coils. Arbuscules are intercalary structures on the coils. However, there have been many reports that in Paris-types arbuscules are relatively few in numbers, small, or absent altogether.

A survey of the literature has revealed that Paris-types occur more frequently in the plant kingdom than Arum-types and predominate in ferns, gymnosperms and many wild angiosperms. The cultivated herbs that are the subject of much experimental work are mostly Arum-types. Although evidence is still limited, there are differences at the family level. In 41 angiosperm families there are records of only Paris-type VA mycorrhizas and in 30 families records of only Arum-types. Another 21 families have examples of both classes, or intermediates between them. Accordingly, we consider whether the original division into two classes is still useful. We conclude that it is when considering the physiology of the symbiosis and especially the issue of whether different fungus/host interfaces have specialized roles in transfer of inorganic nutrients and organic carbon between the partners. If there is no such specialization between hyphal coils and arbuscules, then the latter might not be necessary for the function of Paris-types. This would account for reports of the infrequency or absence of arbuscules in this class. The control exerted on structures by the genomes of host and fungus, and possible reasons (anatomical and physiological) for the existence of the VA mycorrhizal structures, are discussed. The presence or absence of extensive intercellular spaces and differences in the wall structure of cortical cells might be particularly important in determining which type of VA mycorrhiza is formed.