Preslia, Praha, 53(3): 289-304, 1981
Chemotaxonomic studies in the family Rosaceae and the evolutionary origins of the subfamily Maloideae
James Challice

Quillaja has since been transferred
to a family of its own, Quillajaceae.

CHALLICE, J. (1981): Chemotaxonomic studies in the family Rosaceae and the evolutionary origins of the subfamily Maloideae. — Preslia, Praha, 53: 289-304.
The hypotheses of botanical taxonomy and cytology relating to the evolutionary origins of the subfamilies of the Rosaceae have been briefly reviewed and a tentative scheme of phylogenetic relationships has been produced. The scheme is based upon the supposition that evolution proceeded from a spiraeoid-like ancestral group of x = 9 which gave rise to a group of x = 8 (Prunoideae ancestor). Allopolyploidy then took place between these two closely related groups to give an ancestral group of x = 17 with free carpels which eventually evolved into the present-day Maloideae (Sax, Stebbins and Gladkova). The Spiraeoideae arose directly from the ancestral group of x = 9 without any change in basic chromosome number but the Rosoideae genera of x = 7 derived from ancestors of the present-day Rosoideae genera of x = 9 which in turn derived from the spiraeoid-like ancestral group (Gajewski). The affinities of aberrant genera such as Dichotomanthes, Quillaja and Exochorda have been considered. All available chemotaxonomic evidence which has significance within the Rosaceae (principally phenolic constituents) has been collated and shown to be generally consistent with the evolutionary scheme which is presented. Although many taxonomists consider that Maloideae could have arisen directly from primitive Spiraeoideae without the involvement of primitive Prunoideae, the chemotaxonomic evidence clearly shows that Maloideae has strongest affinities with Prunoideae, and then (to a lesser extent) with Spiraeoideae. There is no chemotaxonomic evidence which indicates any exclusive affinity between Maloideae and Rosoideae (x = 7).
Long Ashton Research Station, University of Bristol, BS18 9AF, England.


The Rosaceae is a family of exceptional horticultural significance, containing many economically important fruit-bearing plants, ornamental trees and shrubs. The chemotaxonomic aspects of this subject were last reviewed some years ago (CHALLICE 1974) and the present paper1) updates this review, an updating which is particularly necessary because of a number of important developments in this subject area.

Tab. l. — Basic chromosome numbers in the tribe Quillajeae (Spiraeoideae)

Genus   n   suggested re-assignment
Lindleya 17 } Maloideae?
Vauquelinia 15
Kageneckia 17 } New subfamily collateral with Maloideae?
Quillaja 14
Exochorda   8 Prunoideae?
Lyonothamnus 27 (triploid) Spiraeoideae?

The family is generally sub-divided into four subfamilies (e.g. REHDER 1940, MELCHIOR 1964): Spiraeoideae, Rosoideae, Prunoideae and Maloideae (formerly known as Pomoideae). Two former subfamilies, Neuradoideae and Chrysobalanoideae are no longer included in the Rosaceae but are each given separate familial status (MELCHIOR 1964).

The Maloideae itself is a group extremely well-defined and standing apart from the rest of the family and may be worthy of family status. Prunoideae, Spiraeoideae and Rosoideae are also treated as separate families by some authors. The Spiraeoideae is undoubtedly the least important from a horticultural point of view; a few ornamental shrubs (notably Spiraea, Sorbaria and Exochorda) are the only representatives normally encounterec. horticulturally.

Inevitably this means that the Spiraeoideae has received least attention from natural product chemists, which is a pity because the limited phytochemical data which is available indicates that this subfamily is of considerable chemical interest.


The basic chromosome numbers of the subfamilies are: Maloideae (x = 17), Prunoideae (x = 8), Spiraeoideae (x = 9) and Rosoideae (x = 7) (SAX 1931, 1932, 1933) but the basic chromosome numbers of the Spiraeoideae and Rosoideae are subject to certain exceptions. Table 1 shows some recent chromosome counts in the tribe Quillajeae of the subfamily Spiraeoideae by GOLDBLATT (1976), together with his suggestions for the re-assignment of the genera to other subfamilies.

The diversity of chromosome numbers in the tribe Quillajeae has led Goldblatt to suggest that this taxon is not a natural alliance and should be dispersed as indicated. The count of n = 17 for Quillaja by BOWDEN (1945) has now been shown to be in error. ROWLEY (1978) has recently referred Quillaja to the Maloideae but presumably this was due to the erroneous chromosome count of n = 17; following DARLINGTON et WYLIE (1955) who did the same. Later in this review it will be seen that the chemotaxonomic data to some extent supports the suggested transfer of Lindleya to the Maloideae and the removal of Quillaja from the Spiraeoideae but not the transfer of Exochorda to the Prunoideae. On morphological grounds, the transfer of Exochorda to the Maloideae is impossible, however, because its fruit is a. capsule; in fact Exochorda is aberrant in any of the subfamilies on account of the nature of its fruits.

Tab. 2. — Rosoideae genera with basic chromosome numbers higher than 7

Tribe Kerrieae (monotypic genera)  
  Rhodotypos n = 9
Tribe Dryadeae (part)  
  Cowania n = 9
Tribe Potentilleae (part)  
  Alchemilla n = 8 (?)
Tribe Adenostomeae (monotypic)  
  Adenostoma n = 9

Table 2 shows some chromosome numbers in the Rosoideae which are higher than 7 (DARLINGTON et WYLIE 1955; GAJEWSKI 1957, 1959; FEDOROV 1969).

The genus Dichotomanthes (n = 17) has been considered by different authorities as belonging to the subfamilies Maloideae, Prunoideae and Spiraeoideae in turn, and separate subfamilial status has been proposed by GLADKOVA (1969).


The Rosaceae is a family whose existence poses some interesting phylogenetic problems and the chemotaxonomic data now available would seem to be of some relevance in the consideration of these problems. However, before discussing the chemotaxonomic evidence, the hypotheses of nonchemical botanical taxonomy will be briefly discussed.

A number of conflicting hypotheses have been advanced to account for the origins of the Maloideae (recently reviewed by KOVANDA 1965, GLADKOVA 1972), but the one which has found most favour recently was first formulated by SAX (1931, 1932, 1933) and subsequently elaborated by STEBBINS (1950, 1958). Here, the Maloideoe are postulated to have arisen by allopolyploidy between different primitive forms of Rosaceae, one of x = 8 (a prunoid ancestor) and the other of x = 9 (a spiraeoid ancestor). It is a necessary part of this hypothesis that the respective primitive forms (in the Cretaceous era) were more alike than are the present-day forms, natural allopolyploidy between contemporary members of the subfamilies Prunoideae and Spiraeoideae would be unthinkable. Whether the recently developed techniques of genetic engineering and protoplast fusion could achieve this task would appear to be an intriguing, if remote, possibility.

It is worthy of note that STEBBINS (1958) had Exochorda in mind as a possible living relict of the primitive prunoid ancestor, although traditionally this genus has been placed in the Spiraeoideae. It will be recalled that GOLDBLATT (1976) suggested the transfer of Exochorda to the Prunoideae. GLADKOVA (1972) has maintained that it is unnecessary to postulate a prunoid ancestor, because the basic chromosome numbers of x = 8 and x = 9 exist in the Spiraeoideae already; here Exochorda is considered as a member of Spiraeoideae. GLADKOVA goes on to suggest that the apocarpous Quillaja could be a living relict of a precursor of primitive Maloideae.

2) It is of interest to note that the most ancient fossil forms of the Rosaceae are a couple of extinct Prunus-type species of lower Cretaceous origin. Since early Tertiary fossil forms of all 4 subfamilies are known, it seems quite certain that the family is of Cretaceous origin at the latest (KIRCHHEIMER 1940, 1942, GAJEWSKI 1957).

Taxonomists generally consider now that the Spiraeoideae include the most primitive living forms of the Rosaceae2) and that the Prunoideae and Rosoideae are somewhat specialized evolutionary offshoots from a basically spiraeoid-like rosaceous stock (GAJEWSKI 1957, 1959; GLADKOVA 1972; KALKMAN 1965; TAKHTAJAN 1969; EYDE 1975; STERLING 1964a, b). It is the Spiraeoideae that have retained some primitive morphological characters, such as the apocarpic gynoecium not coalesced with the receptacle.

DARLINGTON (1963) has maintained that the most primitive form of the Rosaceae had x = 7 (a rosoid ancestor) and that higher chromosome numbers arose later by stepwise addition in the Prunoideae and Spiraeoideae and by unequal reduplication (7 + 7 + 3 = 17) in the Maloideae. However, GAJEWSKI (1957, 1959) has convincingly argued that because of the specialized herbaceous nature of most Rosoideae (x = 7) and the generally shrubby nature of the few small Rosoideae genera of x = 9, it is most likely that x = 7 was derived from a more primitive x = 9.

Thus, Darlington's hypothesis has fallen from favour and even the recent finding of n = 14 in the primitive spiraeoid Quillaja seems unlikely to revive it. The contention by GOLDBLATT (l976) that the Maloideae, x = 17, is a palaeotetraploid group, arrived at by the doubling of x = 9 and an aneuploid loss of one chromosome is an interesting suggestion. Perhaps at some future time a critical morphologist will be able to say something authoritative on the subject (GOLDBLATT, personal communication). However, it does appear at first sight that morphology neither supports nor contradicts Goldblatt's hypothesis (KOVANDA, personal communication). The finding of n = 14 in Quillaja is difficult to explain, except by a stepwise aneuploid loss from x = 17 or a stepwise gain from x = 9. GOLDBLATT (1976) himself admits that a concurrence with the most common base number of x = 7 in the Rosoideae must be regarded as coincidental, and of no phylogenetic significance.

Fig. 1. — Hypothetical scheme for the evolution of Rosaceae subfamilies. Reproduced, with modifications, from J. Linn. Soc. — Bot. 64: 239-259 (1974).

Fig. 1 shows an attempted synthesis of the phylogenetic views just outlined. It should be emphasised that such a scheme can only represent an approximation to what must have been in reality an exceedingly complex nexus of evolutionary relationships. It is quite common to represent such relationships by means of a tree-type diagram as in Fig. 1, and as long as it is realised that all present-day taxa are at the tips of the branches of this "tree" and the inner parts are hypothetical entities, we may perhaps be excused for attempting such exercises. However, an additional complication is that in reality our "tree" extends into a phenetic hyperspace of as many dimensions as there are characters which serve to differentiate the taxa·. Only the time element may accurately be represented by one single dimension, although even this is subject to the complication that not all taxa have evolved at the same rate: some taxa (or some characters associated with these taxa) must have changed very little, whilst other taxa (or some of their characters) must have changed to a considerable extent during the course of evolution. Nevertheless, there are mathematical techniques available for projecting diagrams in multidimensional phenetic hyperspace onto spaces of reduced dimensionality and tests for determining the extent of information lost in the process e.g. GOWER (1966, 1967), GOWER et ROSS (1969); mathematical procedures which have been used in investigations of the genus Pyrus (CHALLICE et WESTWOOD 1973) so the exercise undertaken in Fig. 1 (albeit speculative) is not entirely without meaning.


Until comparatively recently, Crataegus was the only genus of Rosaceae known to contain the distinctive flavone C-glycosides (FISEL 1965) as opposed to the more common flavone O-glycosides and to the even more common flavonol O-glycosides, but chemotaxonomic surveys (CHALLICE 1974, 1975; CHALLIC'E et KOVANDA 1978, 1980; KOVANDA et CHALLICE 1981) have now shown that these flavone C-glycosides have a much wider occurence within the Rosaceae, especially within the Maloideae. Table 3 gives the distribution within the Maloideae.

Flavone C-glycosides are a class of flavonoid in which the glycosidic moiety is attached directly to the flavonoid skeleton by a carbon-carbon bond, rather than by the more usual carbon-oxygen-carbon linkage, as in the more common flavone and flavonol O-glycosides. These O-glycosides are readily hydrolysed to flavonoid + sugar by hot acid or by enzymic action, whilst C-glycosides under these conditions remain intact. It is generally considered that these C-glycosides are biosynthetically and phylogenetically more primitive than O-glycosides; hence we have here a chemotaxonomic character of considerable potential usefulness in the Rosaceae.

A convenient procedure has been devised for screening large numbers of leaf samples (both fresh and herbarium specimens) for the presence or absence of these flavone C-glycosides (CHALLICE 1974; CHALLICE et KOVANDA 1978). The basic flavone C-glycosides so far encountered in the Rosaceae are:

vitexin (apigenin 8-C-glucoside)
iso-vitexin ( apigenin 6-C-glucoside)
orientin (luteolin 8-C-glucoside)
iso-orientin (luteolin 6-C-glucoside).

Tab. 3. — Distribution of flavone C-glycosides in the subfamily Maloideae


Orataegus* (V + O) Chamaemeles (V)
Pyracantha (V + O) Aronia* (V)
Dichotomanthes (V) Malacomeles (V)
Osteomeles (V) Micromeles* (V)
Hesperomeles* (V + O) Sorbus subgenus Torminaria* (V)
  Sorbus subgenus Aria* (V)
  Sorbus subgenus Sorbus* (V)
  Sorbus subgenus Chamaemespilus (V)


Sorbus subgenus Cormus Stranvaesia Peraphyllum
Cotoneaster Eriobotrya Malus*
Mespilus Rhaphiolepsis Docynia
Photinia Amelanchier Chaenomeles*
Heteromele   Cydonia

* Flavone O-glycosides also present
V = vitexin (apigenin 8-C-glucoside)
O = orientin (leteolin 8-C-glucoside)
N.B. The occurrence of orientin and iso-orientin in Crataegus (C. monogyna and C. pentagyna) has been definitively demonstrated by NIKOLOV (1977).

It has been found that under conditions of hot acid treatment, some interconversion between vitexin and iso-vitexin and between orientin and isoorientin takes place, so under the experimental conditions used the isomeric forms cannot be accorded separate chemotaxonomic status. It is interesting to note from Table 3 the following points:

  1. Both Malus and Pyrus (apples and pears) are -ve, yet the closely related Sorbus (all subgenera except Cormus) and Micromeles are +ve.
  2. There is no apparent correlation between the presence of flavone O-glycosides and the presence/absence of fiavone C-glycosides.
  3. The distribution of flavone C-glycosides supports the hypothesis that the endemic South American Hesperomeles (the only naturally occurring representative of Maloideae in that subcontinent) evolved from primitive North American Crataegus, the two small endemic genera Aronia (North America) and Malacomeles (Mexico and Guatemala) representing surviving relicts of the evolutionary line as it moved southwards.
  4. Malacomeles has been said to have affinities with Malus, Pyrus, Arnelanchier and Peraphyllum (JONES 1945); doubt must be cast upon this statement since flavone C-glucosides, although present in Malacomeles, are absent from the other four genera.
  5. KOEHNE (1890, 1891), on the basis of reproductive morphology, divided the Maloideae into Crataegeae and Sorbeae (Maleae), a division supported by a study of the distribution of stone cells in the fruit (REMER 1905) as pointed out by HUCKINS (1972). It is interesting to note that the group of Maloideae which contains flavone C-glycosides to some extent corresponds with Koehne's Crataegeae. If Cotoneaster and Mespilus contained flavone C-glycosides (which they do not), and all subgenera of Sorbus (except Cormus), Micromeles and Aronia lacked flavone C-glycosides (they do in fact contain them), then the Present/Absent division in Table 3 would correspond with Koehne's Crataegeae and Sorbeae (Maleae). However, it is important to note that none of these genera just mentioned can be transferred because it would not be morphologically feasible. Apparently the chemotaxonomic data are inconsistent with classical methods and any compromise appears impossible. Perhaps a morphological expert might find it worthwhile to attempt to reconcile the flavonoid data with the grouping of Maloideae genera into Crataegeae and Sorbeae  — an apparently natural division, recently well supported by KOVANDA (1965) and KALKMAN (1973).
  6. As mentioned earlier, the taxonomic position of the monotypic Dichotomanthes has been in dispute; the occurence of flavone C-glucosides indicates strong affinity with the Maloideae. Dichotomanthes could be a relict from some primitive group, ancestral to the Maloideae, exhibiting, as it does, certain morphological , characteristics of the postulated ancestors to a greater extent than the other Maloideae genera (GLADKOVA 1969).
  7. It is assumed that the genera containing both vitexin and orientin are, in a chemical sense, more primitive and that the loss of orientin is indicative of a more advanced state. Similarly, the retention of flavone O-glycosides represents a primitive character which has been lost in some genera. The evolutionary sequence seems to be (1) loss of orientin, (2) loss of vitexin and finally (3) loss of flavone O-glycosides.

At this stage perhaps a note of warning should be made: the presence of chemical character might indicate some particular evolutionary origin but the absence of a chemical character could mean one of two situations — either it was lost at some earlier evolutionary stage or it was never there in the first place. Thus more significance is generally given to the actual presence of a particular chemical character, than to its absence.

Tab. 4. — Indicators of phylogenetic affinities between subfamilies of Rosaceae (Flavones)

Subfamily Flavone
Isoflavones Flavone
Maloideae 10 genera1) Malus2) 0 Cotoneaster6) Malus7)
Spiraeoideae Quillaja1) 0 Sorbaria4) 0 Spiraea8)
Prunoideae 0 Prunus3) 0 Prunus3) Prunus9)
Rosoideae ( x = 9) Adenostoma1) 0 Kerria5) 0 0
Rosoideae (x = 7) Agrimonia1) 0 0 0 0
  1. CHALLICE 1974, CHALLICE et KOVANDA 1978, 1979, 1980;
  2. WILLIAMS 1967, 1979;
  3. HASEGAWA 1958;
  4. ARISAWA et NAKAOKI 1969, ARISAWA et al. 1970;
  6. COOK et FLETCHER 1974;
  7. HIROSE 1909, WILLIAMS 1968, 1969;
  8. CHUMBALOV et al. 1975;

Table 4 includes the remaining occurrences of flavone C-glucosides in the Rosaceae; it will be seen that they are not many. Spiraeoideae: Quillaja only, Prunoideae: nil, Rosoideae: Adenostoma (x = 9) and Agrimonia (x = 7). The restriction of flavone C-glycosides within Spiraeoideae to Quillaja alone, supports the opinion that this genus is not easily accommodated within the Spiraeoideae. DARLINGTON et WYLIE (1955) have placed Quillaja in the Maloideae, but this was done solely on the basis of a wrongly determined chromosome number (n = 17). In fact neither Quillaja nor Dichotomanthes fit readily into any of the four subfamilies of Rosaceae and it is perhaps to aberrant genera such as these that we should look for phylogenetic clues. In this connection it is of great interest that BASINGER (1976) has discovered permineralized flowers, from the Eocene of British Columbia, of Paleorosa similkameenensis (Rosaceae), which he considers to combine more primitive features than any living member of the Rosaceae. He comments that Paleorosa probably represents an early group of rosaceous plants that preceded the tribes Quillajae and Sorbarieae of the Spiraeoideae and may signify the incipient development of the Maloideae.


Tables 4-8 list the occurrence of certain other classes of phenolic compounds (together with the non-phenolic cyanogenic glucosides and sorbitol) which appear to indicate affinities betvveen the subfamilies of Rosaceae. Tables 9-12 list classes of phenolic compounds which appear to be restricted to particular subfamilies: it would appear that these phenolics are generally later evolutionary elaborations of more primitive chemical structures. Of particular note is the interesting degree of apparent specialization in the Spiraeoideae. The presence of delphinidin in Quillaja, a phenolic cyanogenic glucoside in Sorbaria, a glycosylated catechin and complex diterpenoid alkaloids in Spiraea, all combine to make this a most chemically distinct subfamily. It will be of interest to note if any of these unusual substances are subsequently found elsewhere in the Rosaceae. The Rosoideae is also of particular interest in its apparent degree of chemical specialization, notably the presence of myricetin in Potentilla and the general loss of sorbitol and cyanogenic glucosides.

All of the compounds listed in both sets of tables have been selected for their potential chemotaxonomic usefulness. There is a vast array of phenolic compounds which occur throughout the Rosaceae in all subfamilies, though not necessarily in all genera or species: there are phenolics such as the various substituted cinnamic acids, C6-C1 phenolic acids, common catechins, leucoanthocyanidins and anthocyanidins, common flavones such as apigenin and luteolin 7-glycosides and common flavonols such as kaempferol, quercetin and isorhamnetin 3- and/or 7-glycosides (CHALLICE 1972). These are sometimes of value at lower taxonomic levels such as that of particular genera or species, but generally their usefulness is somewhat limited and for the purposes of this review they can be disregarded.

Tab. 5. — Indicators of phylogenetic affinities between subfamilies of Rosaceae (Miscellaneous flavonoids)

Subfamily Dihydrochalcones Flavanones Leucopelargonidin Leucodelphinidin
Maloideae Malusl) Docynia2) 6 general), 2),3),4),5) Crataegus7) 0
Spiraeoideae Sorbaria2) 0 0 Quillaja10)
Prunoideae 0 Prunus6) 0 0
Rosoideae (x = 9) 0 0 Dryas8) Kerria9) Neviusia9) 0
Rosoideae (x= 7) 0 0 0 Potentilla9), 10)
  1. WILLIAMS 1966;
  2. CHALLICE 1973;
  3. PARIS et ETCHEPARE 1965;
  4. WILLIAMS 1962;
  6. HASEGAWA 1958, HERGERT 1962;
  7. LEWAK et RADOMINSKA 1965;
  8. PANGON et al. 1964;
  9. BATE-SMITH 1961;
  10. BATE-SMITH 1965.

Tab. 6. — Indicators of phylogenetic affinities between subfamilies of Rosaceae (Substituted Flavonols)

Subfamily Flavonol
8-0-methyl ethers
Maloideae Crataegusl) Crataegus3) Sorbus4) Malus7) 0 Malusl0) Sorbusl1), l2)
Spiraeoideae 0 0 0 Vauquelinia9) 0
Prunoideae Prunus2) Prunus5) 0 Prunus2) Prunus13)
Rosoideae (x = 9) 0 Dryas6) Rhodotypos8) 0 0
Rosoideae (x = 7) 0 0 0 0 Filipendula14) Rosa15) Geum16)
  1. NIKOLOV et al. 1973;
  2. WOLLENWEBER et al. 1972;
  3. BYKOV et GLYZIN 1972;
  6. PANGON et al. 1974;
  7. WILLIAMS 1968;
  8. PLOUVIER 1967;
  9. BATE-SMITH 1965;
  10. WILLIAMS 1969;
  11. BORISOV et ZHURAVL'OV 1965;
  12. CHALLICE 1973;
  14. HÖRHAMMER et al. 1956;
  15. HARBORNE 1961, 1967;
  16. KAMINSKA 1971.

Returning to the chemotaxonomically significant substances, it should be remembered that it is far easier to report a presence than to report an absence of any particular substance. Sometimes the zero, as recorded in the tables 4-8, means that a survey of varying comprehensiveness has failed to detect the particular substance; other times the zero merely means that there are no records of the subfamily having been screened for the particular substance. To some extent this is a fault of the literature itself — all too often. negative results are not reported; in fact pa:1ers which report purely negative results are rarely, if ever, published!

Tab. 7. — Indicators of phylogenetic affinities between subfamilies of Rosaceae (Miscellaneous phenolics)

Subfamily 4-Allyl
Proto- catechuic acid
Arbutin p-hydroxybenzoyl-vanilloyl-
(3 compounds)
Maloideae Pyrus1) Pyrus1) Pyrus2) Pyrus1) many genera4) 0
Spiraeoideae 0 0 Sorbaria3) Exochorda4) 0 Lindleya4) 0
Prunoideae Prunus1) Prunus1) 0 Prunus1) 0 Pygeum5)
Rosoideae (x = 9) 0 0 0 0 0 0
Rosoideae (x = 7) 0 0 0 0 0 many genera6)
  1. CHALLICE et WILLIAMS 1968a;
  2. CHALLICE et WILLIAMS 1968b;
  3. PLOUVIER 1971;
  4. CHALLICE 1973;
  5. BATE-SMITH personal communication;
  6. BATE-SMITH 1961.

N.B. Pygeum has now been incorporated into Prunus subg. Laurocerasus by KALKMAN (1965).

Tab. 8. — Indicators of phylogenetic affinities between subfamilies of Rosaceae (Non-phenolic compounds)

Subfamily Cyanogenic1) glucosides Sorbitol2)
Maloideae 16 genera 14 genera
Spiraeoideae 6 genera 9 genera
Prunoideae 4 genera (incl. Prunus) Prunus
Rosoideae (x = 9) Kerria
Rosoideae (x = 7) 0* 0
  1. HEGNAUER 1973, GIBBS 1974, GERSTNER et al. 1968, CONN et BUTLER 1969.
  2. PLOUVIER 1963, GIBBS 1974.

* The isolated report of cyanogenesis in Geum (x = 7) by GIBBS 1974 should be checked.

However, the format of occurrences in tables 4-12 should enable the chemotaxonomic significance of any subsequently reported phenolic in the Rosaceae to be immediately assessed. For example, although flavanones are generally correlated with woodiness rather than the shrubby or herbaceous habit, there are invariably exceptions and it is quite possible that a detailed survey of the Spiraeoideae and Rosoideae subfamilies would reveal the presence of some type of flavanone in these subfamilies. Nevertheless, it is interesting that there does not appear to be even a single report of any flavanones from the Spiraeoideae or Rosoideae in the literature: if they are present they are probably only rarely present, in contrast with the Maloideae and Prunoideae. Thus, the situation as indicated in Table 5 is probably indicative of a general trend, if not of a clear-cut distribution. The same type of argument could apply to most of the substances which are listed.

It appears to be fairly obvious that when the tribe Kerrieae of Rosoideae (x = 9) evolved from Rosoideae (x = 7) as proposed by GAJEWSKI (1957, l 959), ellagic  acid, sorhitol and cyanogenic glucosides were generally lost. There may be a few exceptions yet to be discovered, but the general trend seems quite clear. The presence of arbutin in Exochorda indicates that this genus should perhaps remain in the Spiraeoideae and not be transferred to the Prunoideae (where arbutin is absent), just because Exochorda has prunoid chromosome number of n = 8. STEBBINS (1958) has speculated that Osmaronia and Exochorda could be relicts of the original prunoid ancestor of the Maloideae; chemical evidence for such a. relationship exists only in the case of Exochorda. It appears that there are no genera which embody strong morphological and chemical affinities with the postulated ancestors of the Maloideae; isolated characters, distributed amongst the present-clay genera, are all that survive from the ancestral genera in question.

Tab. 9. — Chemotaxonomic specialization in subfamilies of Rosaceae (Maloideae)

Flavone 4'-0-glycosyation: Luteolin and apigenin 4'-0-glucosides in Pyrus1) and Sorbus2)
Vitexin 4'-0-rhamnosylglucoside in Crataegus3)
Flavone 5-0-methylation (?): suspected luteolin 5-methylether in Pyrus4) ?
Caffeoylcalleryanin in Pyrus1),5)
2,4,6-trihydroxydibenzoylmethane 2-glucoside in Malus6)
Leucocyanidin 3-0-arabinoside in Eriobotrya7)
  1. CHALLICE et WILLIAMS 1968b;
  2. CHALLICE et KOVANDA 1978, 1979;
  3. FISEL 1965, LEWAK 1966;
  4. CHALLICE 1972, 1973;
  5. CHALLICE et al. 1980;
  6. WILLIAMS 1967a, 1979;
  7. AGARWAL et MISRA, 1980.

Tab. 10. — Chemotaxonomic specialization in subfamilies of Rosaceae (Prunoideae)

Flavanone O-methylation: sakuranetin in Prunus1)
Isoflavanones: padmakastein in Prunus2)
O-methylated coumarins: herniarin3) and 5-OH 6, 7-di MeOH coumarin4) in Prunus
  1. HASEGAWA 1958;
  2. HERGERT 1962;
  3. BATE-SMITH 1961;
  4. HASEGAWA 1969.

Tab. 11. — Chemotaxonomic specialization in subfamilies of Rosaceae (Spiraeoideae)

3',4',5'-trihydroxylated anthocyanidin: delphinidin in Quillaja1)
Phenolic cyanogenic glucoside: 2-β-D-glucopyranosyloxy 4-p-hydroxybenzoyl-3-methylenbutyronitrile in Sorbaria2)
Glycosylated catechin: Catechin 7-0-rhamnoside in Spiraea3)
Complex (diterpenoid) alkaloids*: e.g. Spiradin G and Spiradin F in Spiraea4)
* N.B. Trace amounts of the much simpler pyridine alkaloid, nicotine in Prunus cerasus4)?
  1. BATE-SMITH 1965;
  2. NAHRSTEDT 1976;
  3. CHUMBALOV et al. 1976;
  4. HEGNAUER 1973 and references therein.j
3) Two calleryanin esters (caffeoyl- and protocatechuoylcalleryanin) have also been found in the gymnosperm Podocarpus andina (POYSER et al. 1973): it would thus appear that these particular chemotaxonomic markers are of considerable phylogenetic age, pre-dating the rise of the angiosperms from their putative gymnospermous ancestors. The relatively rare dihydrochalcones also provide a similar link: BHAKUNI et al. (1973) have found α-hydroxyphloretin (nubigenol) in Podocarpus nubigena. CRONQUIST (1968) has suggested the following evolutionary sequence of dicotyledonous plant orders: Rosales → Myrtales → Proteales; interestingly, derivatives of calleryanin provide a common connecting link between these three orders. Calleryanin + four phenolic acid esters in Rosaceae (Rosales), calleryanin 3-methyl ether in Daphne mezereum, Thymelaeaceae (Myrtales) (KOSHELEVA et NIKONOV 1968) and p-hydroxybenzoylcalleryanin in Protea cynaroides, Proteaceae (Proteales) (VANWYK et KOEPPEN 1974).

Returning to the evolutionary scheme in Fig. 1 it will be noted that supposed losses of the ability to synthesize ellagitannins (E) and flavone C-glycosides (C) are indicated here. It is certainly of some significance that the primitive ellagitannins have been found in Pygeum (now transferred to Prunus subg. Laurocerasus by KALKMAN 1965); the evidently primitive p-hydroxybenzoyl-, vanilloyl- and protocatechuoyl — calleryanin esters are restricted, within the Prunoideae, to P. lusitanica which also belongs to the subgenus Laurocerasus.3)

The remaining chemotaxonomic data are generally consistent with this scheme of relationships. It has already been indicated that there is a tendency amongst taxonomists now to discount or minimize affinities between Maloideae and Prunoideae, and to regard the Maloideae as a specialized development from Spiraeoideae alone. This is not supported by the chemotaxonomic evidence which establishes strong affinities between Maloideae and both Spiraeoideae and Prunoideae. On reproductive morphology, however, the Maloideae are most closely related to the Prunoideae. Affinities to the Spiraeoideae appear less distinct.

Table 13 represents an attempt to summarize the evidence from 24 chemotaxonomic indicators of various affinities between the subfamilies of Rosaceae. It will be seen that chemically the Maloideae shows most affinity to Prunoideae, with Spiraeoideae taking second place: we have already mentioned that this is supported by morphological evidence. The affinity between Maloideae and Rosoideae (x = 9) is not surprising, because the latter group of genera are known to be spiraeoid-like and the chemical data supports the contention of some taxonomists that the monotypic genera of the tribe Kerrieae (x = 9) should perhaps be transferred to the Spiraeoideae (e.g. BATE-SMITH 1961). The discrepancies between "simple affinities" and "exclusive affinities” are regarded as indication of a degree of reticulate evolution in the Rosaceae. The fact that there are no chemotaxonomic characters which indicate any exclusive affinity between Maloideae and Rosoideae (x = 7) is taken to be a clear refutation of the hypothesis, mentioned earlier in this paper, that Maloideae did not evolve from primitive Rosoideae of x = 7, the previously regarded ancestral basic chromosome number for the Rosaceae.

Tab. 13. — 24 chemotaxonomic indicators of affinities between subfamilies of Rosaceae

Subfamilies Simple affinity Exclusive affinity
Maloideae → Prunoideae 14 9
Maloideae → Spiraeoideae 7 3
Maloideae → Rosoideae (x = 9) 6 2
Mciloideae → Rosoideae (x = 7) 2 0
Spiraeoideae → Prunoideae 4 1

Spiraeoideae → Rosoideae (x = 9)

4 1
Spiraeoideae → Rosoideae (x = 7) 2 1
Prunoideae → Rosoideae (x = 9) 3 0
Prunoideae- → Rosoideae (x = 7) 2 1
Rosoideae (x = 9) → Rosoideae (x = 7) 1 0

Simple affinity: characters which in some instances are also shared by a subfamily other than the two subfamilies being compared.
Exclusive affinity: characters restricted to the two subfamilies being compared.

The investigation of phylogenetic interrelationships within the Rosaceae is a fascinating  subject and it is hoped that this review of the chemotaxonomic data which is available will not only serve to pinpoint the areas which merit more detailed chemical study, but will also encourage plant morphologists, geographers and geneticists to give further attention to the many phylogenetic problems which remain within this important family.

Journal of Genetics 22: 129-151 (1930)
Primary and secondary chromosome balance in Pyrus
C. D. Darlington & A. A. Moffett

  1. The basic chromosome number in Pyrus is seventeen. Cultivated varieties are all orthoploid. Aneuploid seedlings are poor and abnormal.
  2. The somatic chromosomes in “diploid” Pyrus have four representatives of a long type, in “triploid,” six.
  3. Multiple association occurs amongst the chromosomes of “diploid”Pyrus giving, in extreme cases, seven groups; four quadrivalents and three sexivalents (Table I).
  4. In “triploid” varieties of P. Malus  associations of four, five, six, seven, eight and nine chromosomes have been observed, although trivalents are usually formed (Table II). This means that antosyndesis takes place within each of the three supposed haploid complements.
  5. Instead of giving a binomial frequency or the elimination of intermediate numbers, natural seedlings of “triploid” apples most frequently have numbers of chromosomes approximately to 2n + 7 (Table III).
  6. Thus the pairing, morphology, and breeding results show, directly or indirectly, that the thirty-four chromosomes in the “diploid”Pyrus are of seven types, of which four are represented four times and three are represented six times. Such forms may be described as trebly hexasomic tetraploids (v. diagram, p. 145).
  7. The number seventeen is therefore a secondary (unbalanced) basic number, and the derived series of polyploids (2n = 34, 51, 68) are secondary polyploids.
  8. The establishment of a secondary basic number must mean (on the analogy of all experimental observations) a definite evolutionary step. It is therefore plausible that the Pyrus group owe their special morphological characters (e.g. the pome type of fruit) to this reorganisation of the nucleus. The work is being continued with this consideration in view.

Journal of the Arnold Arboretum 13: 363-367 (1932)
Chromosome Relationships in the Pomoideae
Karl Sax


Sorbaronia alpina, Sorbopyrus auricularis and Malus theifera are triploids. At the first meiotic division in the pollen mother cells there are about 17 bivalents and trivalents, and from 6 to 15 univalent chromosomes. The fact that about 12 univalents are usually found in triploid forms of Pomoideae shows that this subfamily is not an allopolyploid with a basic number of 7 chromosomes as several writers have suggested.

The basic chromosome numbers in the other subfamilies of the Rosaceae are 7, 8, and 9. The Pomoideae may have originated from one or perhaps two of these subfamilies by hybridization between different primitive forms followed by chromosome doubling in the F1 hybrid. Remote chromosome affinities are indicated by secondary association of bivalents in the Pomoideae. True multivalent chromosome pairing rarely, if ever, occurs in "diploid" species. The available evidence seems to indicate that the Pomoideae are allopolyploids.

Genetica 15: 511–518 (1933)
Cytological studies in cultivated pears
A. A. Moffett

  1. Of fifteen cultivated varieties of pears examined, eleven were diploid (2n=34) and four triploid (2n=51).
  2. Secondary pairing occurred at metaphase of meiosis in diploids, but no multivalents were observed.
  3. In triploid varieties, associations higher than trivalents were not observed.
  4. The lower degree of secondary pairing in pears as compared with apples indicates a greater degree of differentiation of the chromosomes in pears.
  5. It is considered that the evidence points to 7 being the primary basic number, and not eight as suggested by SAX.

CybeRose note: It appears that the concept of "diploidized tetraploid" had not yet been accepted. That is, a plant may originate as a tetraploid hybrid, but in time come to behave as a simple diploid with little or no secondaty pairing.

Hutchinson et al: The genetics of the diploidized tetraploid Avena barbata: Acid phosphatase, esterase, leucine aminopeptidase, peroxidase, and 6-phosphogluconate dehydrogenase loci. J. Heredity 74(5): 225-230 (Sept 1983)
Formal genetic studies of six pairs of enzyme loci in Avena barbata, a tetraplold grass with 2n = 4x = 28 chromosomes, are reported. We obtained data from an F2 self-fertilized progeny of an F1 hybrid between two California genotypes and from self-fertilized progenies of heterozygous plants collected in nature. The results indicate that the phenotype scored from our gels for each locus pair was the product of four "homoeoalleles," two carried by one pair and two by the other pair of homoeologous chromosomes of this tetraplold. In each case observed segregations fit expectations for pairs of independently segregating loci with codominant alleles. A. barbata thus behaves genetically as a diploidized tetraplold, i.e., a tetraploid in which pairing is preferential within each pair of homologues rather than at random among the four chromosomes of each homoeologous set. We studied one pair of loci for each of acid phosphatase, esterase, leucine aminopeptidase, and peroxidase and two pairs of loci for 6-phosphogluconate dehydrogenase. One of the acid phosphatase loci is linked to one of the peroxidase loci and both pairs of 6-phosphogluconate dehydrogenase loci are linked to each other and to the pair of leucine amlnopeptidase loci. It also is likely that the second acid phosphatase locus and the second peroxidase locus are linked but this could not be demonstrated because both loci were monomorphic in all of our population samples.

It is also unfortunate that none of the researchers mentioned above were aware of Schoener's hybrids raised from Rosa pomifera pollinated by the Spitzenburg apple. It would have been interesting to see how the chromosomes behaved.

Chromosome Changes in Evolution and Adaptation