Problems in elm and rose Identifications
Unfortunately no universally accepted, practical field keys exist to separate the genus into species/
sub-species in a reliable way. Although much work has been earned out by botanists, (eg Rackham
1980; Richens 1983; Stace 1997) the classifications differ and there is much variation. The elms in
this study area, even when close together, and expected to form part of a clone, were very variable.
In November 1995, a visit was made to the London Natural History Museum, looking through all
the relevant herbarium sheets on elms. This proved an inconclusive exercise. Specimens with the
same name looked quite different and many sheets had been annotated in pencil with comments such
as "not so" or "cannot agree". Although we took specimens with us to compare, we were not able to
identify them.
Dr J.V. Armstrong spent several years looking at the elms of East Anglia, developing a taxonomic
treatment of the British elms with recognition of each distinctive form as a microspecies. She had
recognised sixty microspecies, which were often found occurring in distinct localities, characterised
by unique morphologies. She found the variation pattern in elms to be very similar to that in Rubus
and Hieracium (Armstrong pers. comm.). However, this study was ultimately unable to use the new
microspecies method.
Instead, for this study the morphological differences were recorded by taking rubbings of
representative leaves from short shoots. More than one hundred leaves were measured for length,
width, offset at the leaf base, petiole length and number of veins on the longest side (see Fig. 2), and
whether they were smooth or hairy. The location was also recorded so that any affinity might
identify particular clones.
Analysis of the data produced a normal distribution. Perhaps elms have been mixing and recombini ng
their genes for such a long time that it is now foolish to try to be too precise in their identification?
Nevertheless, we wanted to compare the elms from different parts of the area, using radar charts
(Figs 3a-e) prepared from four of the parameters extracted from the spreadsheet. These were: the
leaf-base offset in mm, petiole length, number of veins on the longer side and the ratio of blade
width/length x 10. Although these diagrams cannot be used for identification, they did enable us to
draw certain conclusions about the pattern of distribution, using the 29 charts produced, five of
which arc included below as examples. From the complete set we deduced that:
1 Three distinct groups of elms appear to be represented in the study area, A-C.
2 'A is the most distinctive and represents U. glabra or a slightly diluted form. 'B' is the largest
and has two subsets that are very slightly different, but not different enough to constitute separate
groups. 'C is a small group possessing no measurable offset.
3 The differences are small and would seem, apart from 'A' (U. glabra), to point to various
genetic mixings of U. glabra, U. procera Salisb, and U. minor Mill, with, perhaps, some U.
plotii Druce.
A further method was used to compare leaves from selected areas. The width/length (B/L) ratios
were as before (this time not multiplied by 10). Each ratio was then multiplied by the mean value of
the number of pairs of veins, giving the arbitrary scale used along the base-line of the graph in
Figure 4. The standard deviations for each area were calculated and arc shown. The numbers -3 to
+3 represent one, two and three standard deviations each side of the mean value. N - represents the
mean or normal value.
Essex Naturalist (New Series) 18 (2001)
149