453
J._IAnn. Soc. (Boe.), 56, 368, p. 453
With 1 plate and 1 text-figure
PfiffUd in Great Britain
Epidermal characteristics in the diploid subspecies
of Dactylis glomerata L.
BY MARTIN BORRILL, PH.D., F.L.S.
Welsh Plant Breeding Station, Aberystwyth
INTRODUCTION
It is now generally accepted that epidermal characters are of value in grass classification at all levels from the tribe to the species. This has been made abundantly clear from
the classical researches of Grob (1896) and Prat (1932). More recently Church (1949)
has shown that the possession of intercostal long-cells with smooth walls serves to
distinguish Glyceria R.Br. from Puccinellia Pari. and Torreyochloa Church in which the
corresponding cells have ripple walls. In addition, it is possible to distinguish between
sections within these genera according to the occurrence and distribution of papillae.
Borrill (1956) found that the species of Glyceria in Britain can be distinguished by the
type of epidermis covering the leaf sheath nerves. S0renson ·(1953) has shown that the
epidermis provides features which are of value in identifying species of Puccinellia.
Recent studies in the genus Dactylis L. have shown that ten or more diploid subspecies
occur. These are natural groupA, each with a characteristic morphology, and limited
geographical distribution (Borrill, 1961 b). The nature of the epidermis has been studied
in these clearly defined forms to see whether substantial differences occur which might
justify the application of the method to the whole tetraploid complex, which is difficult
to classify. So far, only one natural tetraploid group Dactylis marina Borrill, has been
recognized, in which the epidermal cells differ in shape from those of other tetraploids,
and bear conspicuous papillae (Borrill, 1961 a).
In presenting the results, the names for the diploids proposed by Stebbins & Zohary
(1959) will be used.
MATERIAL AND METHODS
Epidermal characters in the leaves were studied on living plants or herbarium material
of the eight diploid subspecies shown in Table l. Four leaves were taken from each population and these represented four genotypes except where dried material was used, when
the number of plants present was not known. The leaves were not selected in any particu.
lar way except to ensure that they were fully grown and not diseased or withered. A preparation of the upper epidermis was made, using the lactic acid method described by
. Clarke (1960). The resulting layer of cells was usually l cm. long, and extended for the
whole width of the leaf. Photomicrographs were made of the area halfway between the
mid-vein and the leaf margin. This allowed for the gradation in cell size which occurs
in this direction. Each photomicrograph included two or three inter-nerve areas, and
from 50 to 250 cells, depending on the subspecies used. From this a.rea twenty cells were
selected as far as possible at random, and the following measurements were made:
(1) cell length, (2) width of cell at the centre, (3) width of cell at the end, (4) stomatal
length, (5) number of stomata per unit area. The data 'are shown in Table l.
RESULTS: QUANTITATIVE DIFFERENCES BETWEEN SUBSPECIES
An analysis of variance revealed large and significant differences between the ten
populations. For characters such as cell length and width to be of use in identifying
these groups, the means obtained from repeat samples of four genotypes should fall
}
Ira.n, Arak
Portugal, Sintra
-
e (non-santai)
do.
g,judaica
h, woronowii
j, luaitanica
k, aschersoniana
f,
Collector
Kew Herbarium
Prof. G. L. Stebbins
(herbarium material)
Kew Herbarium
F.A.O. (No. 3168)
F.A.O. (No. 3215)
F.A.O. (No. 3328)
Prof. G. L. Stebbins,
Da.vis, Ca.lifom.ia.
F.A.O. (No. 3697)
Prof. A. Ca.ma.ra., Est&;a.o
Agronomics. N39ional,
Sa.oa.vem
)3ota.nic Garden, Uppsa.la.
23·34
25·15
0·25
21·58
26·36
21·81
24·92
29·51
29·32
30·85
25·61
23·80
286·6
212·0
151·3
155·0
173·4
204·8
252·0
237·2
213·6
212·6
. 204·3
30·8
23·48
22·64
82·9
.176·4
17·91
18·19
0·19
13·73
18·24
17·31
19·86
19·53
19·77
18·84
17·03
17·08
14·66
14·38
1·30
1·37
-
-
1-39
1·57
1·44
1·26
1·25
1·50
1·47
1·63
1·50
1·61
1·57
ll·8
ll·2
20·8
ll·7
8·7
7·8
8·8
10·8
10·8
13·9
12·4
5·6
12·2
(Analysis of variance on 10 populations, 4 genotypes, 20 cells from each; linear dimensions in p.)
Significant differences P < 0·01
lsra.e1, Kirya.t Shmuel }
}
Spain, Sierra Nevada
Algeria, Reliza.ne
Algeria., Oued Chiffa.
Algeria., Bouira.
. c, juncinella
d, santai
Canary Islands
Balearic Islands
Looa.lity
·a, smithii
b, ibizenais
Subspecies
32·0
31·9
0·21
37·0
36·3
38·7
36·2
42·8
41·4
46·9
33·0
36·7
27·7
30·8
4954
5148
6191
5597
3298
3859
51 Hi
6001
7762
6072
5084
1948
3987
108
83
82
-
82
87
211
142
130
241
125
Cell shape ratios
r--'----.
Cell
Median
Epidermal cell data
area. Stomata.
Length width Guard length x per
Median Cell end
end
end
cell
median unit
Length
width
width
width width length width
area.
Table l. Epidermal cell data in diploid Dactylis (see Text-fig. 1)
"'Ot"
E
0
b:1
z~
"'"
Epidermal characteristica of Dactylis
455
within the range of variation permissible at P <: 0·01. Measurements were therefore made
on a further four plants of subspecies judaica, g; lusitanica, j; and aschersoniana, k.
Those ofjudaica were from the same population, and this applies also to aschersoniana,
because the plants were derived from a further batch of seed obtained from the same
mother plants in the Botanic Garden, Uppsala. The repeat sample of lusitanica, however,
originated from a different wild population.
Table l shows to what extent the figures for· the duplicate samples agree with the
original means. The most consistent values are shown by the means for cell length, and
width of the cell end, which do not differ significantly from the original observations,
whether the duplicates derive from the same population or not. The repeat data for
median width and guard cell length suggest that these characters would be less reliable
as a means of identification.
In Text-fig. l the sample means are plotted, using cell length and width as ordinates.
There is no correlation between these characters from population to population, although
the data show a very close relation between the length and width of individual cells within
each group. The rectangles surrounding the means show the limit of variation at P =
0·01. Although two pairs of groups overlap, d and k, g andf, the others can be separated
using these characters. Similarly, Martin (1954), using cell length and width in leaf
cuticles, found it possible to distinguish between certain species, including grasses such
as Festuca pratensis Huds. and F. rubra L.
Stomatal length appears to vary independently, since no correlation has been found
with stomatal frequency, cell length, or any of the other features studied either within
or between groups.
The number of stomata per unit area varies considerably, and scatter diagrams were
prepared to test the relation between frequency, and cell length, cell size (area), and cell
shape, both within and between populations. A highly significant negative correlation
with a coefficient of -0·945 was obtained between populations for stomatal frequency
and cell length. It seems probable that other factors, such as the number of veins, will
also influence this frequency, since in many cases the stomata are arranged in rows
closely associated with the nerves.
Variations in cell shape can be assessed quantitatively in two ways,' by finding the
ratios length: end width and median width: end width. The former is a retia ble composite
character in identification. The latter, while giving a quantitative assessment of the
range from square to hexagonal cells in the populations, is not so satisfactory, due to
the variation in median width encountered between duplicate samples.
RESULTS: QUALITATIVE DIFFERENCES BETWEEN SUBSPECIES
In addition to the differences shown by the numerical data, there are others,, mainly
qualitative in character, which are of value in separating the diploid Dactylis groups.
These are (Pl. 1 and Text-fig. 1):
(1) There is a wide range in cell shape of which two aspects have been studied. The
first is the degree to which the cells are hexagonal or square. This is measured by the
ratio width at centre of cell:width of cell end. Amongst the populations, a, subspecies
smithii, has markedly hexagonal cells, and at the other extreme are the populations d, e,
and f, from Algeria with square cells. The second aspect of cell shape is measured by the
ratio length: width of cell end. In this respect the most extreme forms are subspecies
juncinella in which the cells are 20 times as long as wide, and subspecies smithii in which
length exceeds width by only 5 times.
(2) The anticlinal cell walls can be classified as three main types, which are sinuous,
beaded, or smooth, respectively. In the subspecies smithii and ibizensis, the walls are
sinuous in varying degree, this being most pronounced in smithii (Borrill, 1961 a). Two
456
MARTIN BORRILL
of the Algerian popula.tions, e andf, have beaded cell walls; this shows most clearly in e,
(PI. 1). The remainder have smooth walls.
{3) The distribution of the stomata which lie in rows flanking the nerves, except in
subspecies amithii and ibizensi8 where they are more uniformly distributed.
All these features of the groups are illustrated in Text-fig. 1. From the point of view
of identification, the most difficult pairs are first, Algerian d and aBchersoniana k, and
secondly, Algerian f and judaica g, which have similar cell dimensions. The first pair
can be separated by cell shape, cubical in d, hexagonal in k, and the second pair by the
type of cell wall, beaded in f, smooth in g, and, to a lesser extent, by cell shape and
stomatal frequency.
c
300
250
~
h
~ ·~~ B
82
b
~
~
.
..
.
e
::
:x:
:.
.:
.
.
150
~
1
125
: :1)(:
.. j •
130: :
.:
i11
H2
.
.
100
..
241
150
175
Cell width
200
(~-~.)
Text-fig. I. Epidermal cell characteristics of diploid Dactylia subspecies (data in Table 1).
Key: a, smithii; b, ibize1111ia; c, juon<nneUa; d, AJgeria.n subsp. santai; s (non-santai);
J, Algerian (non-aantai) ; g, judaioa; h, woronowii; j, luaitanioa; k, ascheraoniana.
cella hexagonal ~
cella cubical
11
cell w&lla: Brnooth /
sinuous/
beaded/
stomata: scattered::·:
'
...
in rows along nerves : :
The figures indioote the number of stomata per unit area.
In the group with more or less hexagonal cells, the subspecies smithii and ibizensis
have sinuous cell walls and uniformly distributed stomata. Subspeciesjuncinella has very
long slender cells. Populations g, h, j and k which complete this group can be separated
by cell width, and g: h (subspecies woronowii :judaica), and g :j (subspecies woronowii:
lu.sitanica) by cell length as well.
Epidermal charactemtics of Dactylis
457
The data reveal some diversity amongst the populations from Algeria all of which
have more or less square cells; d has smooth-walled, rather long cells, and a low stomatal
frequency, whereas e andf have cells with beaded walls. These latter two groups differ
markedly, e having much narrower cells and higher stomatal frequency than f.
DISCUSSION
The results of this epidermal survey show that the method of sampling employed has
proved capable of distinguishing between the taxa, particularly in regard to cell size.
The cells were selected from a constant position on the leaf, thus taking into account the
gradation in size which occurs from mid-vein to margin, and to a lesser extent from ligule
to leaf-tip (Soper & Mitchell, 1956); but cell size, notably length, can also vary according
to the position of the leaf on the shoot (Borrill, 1959). Greater precision in sampling
could therefore be obtained if a particular leaf, say number 6, was selected on the main
shoot of seedlings grown together in the same environment.
Such a method might be useful to detect small differences in epidermal pattern
between populations in taxonomic groups such as subspecies. It is now being applied
to herbage strains to assess the order of difference found among tetraploid populations.
In the group of populations which make up the diploid Dactylis subspecies, natural
selection appears to have led to physiological and genetical adaptation to diverse, and
sometimes extreme, environments. The resulting large differences in epidermal phenotype
must in part represent functional integration in relation to these environments.
Amongst the cell characters studied the following are of value in grouping populations:
(1) The extent to which the cell wall is smooth, sinuous, or beaded; (2) the type of
stomatal distribution; (3) the extent to which the cells are hexagonal or square. This
suggests that these characters may have been more stable during the evolution of the
taxa, because less subject to environmental selection. With regard to the third character,
there is clear evidence from Cretan tetraploids (Borrill, unpublished) that although the
ratio cell length: cell end width shows a cline of variation in relation to altitude, all the
populations have cells which are hexagonal to a similar degree.
When the subspecies are placed in groups based on these characters the arrangement
conforms with their geographical distribution.
Subspecies smithii and ibizensis possess uniformly distributed stomata, and the walls
of the longitudinal cell rows are more or less sinuous. Both are endemic in the Canary
and Balearic Islands respectively. Detailed studies have shown that the form of these
cells is related to the production of epidermal papillae. This is exemplified by the related
tetraploid Dactylia marina and its hybrids, in which the form of the cells can range
from long, hexagonal, a.s is usual in Dactylia, through shorter, intermediate types with
bulging sinuous longitudinal cell walls, to cells bearing a large spherical papilla. This
character is under polygenic control. In both these diploid subspecies the cells are mainly
of the intermediate type, although in subspecies amithii the genetic background is such
that in a few of the cells the wall bulges sufficiently to form a papilla (Borrill, 1961 a).
The populations from Algeria, distinguished by more or less square cells, are heterogeneous in their gross morphology, and d appears to fall within the limits of subspecies
aantai (Stebbins & Zohary, 1959). The others, e and f, characterized by beaded cell
walls, are phenotypically rather similar (Borrill, 1961 b), but differ in cell size. In f the
beading of the cell walls is less pronounced and the general form of the epidermis shows
some similarity with subspecies judaica or woronowii. Both e and/ originated in a cultivated area with arable cropping, and it is possible that man's activities have provided
opportunities for hybridization between the groups of diploids in this region.
Further sampling of the diploids in North Africa is required in order to clarify their
taxonomic status.
The remaining group of subspecies judaica and woronowii, luaitanica, juncinella and
458
MARTIN BoRRILL
aschersoniana possess hexagonal cells. Subspecies jun.cinella, endemic at high altitudes
in the Sierra Nevada. in Spain, is a somewhat isolated form, with a distinctive pattern
of very long .epidermal cells. All these types can be separated from one another by
differences in epidermal cell size.
These observations on Daetylis add to the growing body of evidence indicating the
value of epidermal studies in plant classification. There is reason to suppose that quantitative studies of cell size will be very useful in this field when considered in relation to the
qualitative characters on which the main- emphasis has hitherto been placed.
ACKNOWLEDGEMENTS
I am indebted to Professor P. T. Thomas, Director, Welsh Plant Breeding Station, for
providing the facilities to carry out this work, to Mr A. R. Beddows, Mr C. E. Hubba.rd
and Dr C. R. Metcalfe for advice during preparation of this paper; also to the Director,
Royal Botanic Gardens, Kew, for permission to use herbarium material. I wish· to thank
Mr John Clarke for technical assistance, and Mr Harold Richards for preparing the
photographs.
REFERENCES
BoruuLL, M. (1956). A bioaystematio study of some Glyceria species in Britain. I. Taxonomy.
Wataonia, 3: 291-298.
BoRRILL, M. (1959). Infloresoenoe initiation and leaf size in some Gramineae. Ann. Bot., Lond., N.S.,
23: 217-227.
BoRRILL, M. (1960a). Daotylis marina Borrill sp.nov., a natural group of related tetraploid forma.
J. Linn. Soc. (Boi.). 56, (368): 431-439.
BoRRILL, M. (1960b). The pattern of morphological variation in diploid and tetraploid Dactyli.s.
J. Linn. Soc. (Bot.). 56, (368): 441-462.
CHURCH, G. L. (1949). A cytotaxono~ostudy ofGlyoeriaandPuccinellia. Amer. J. Bot., 36: 155-165.
CLARXE, J. (1960). Preparation of leaf epidermis for topographic study. Stain Tech., 35: 35-39.
GROB, A. (1896). Beitrlige zur Anatomie der Epidermis der Gramineenblii.tter. Bibliotheca Botanica,
36,
MARTIN, D. J. (1954). Features on plant cuticle. Tram. Bot. Soc. Edinb., 36: 278-288.
PnAT, H. (1932). L'epiderme des gra.minees; etude anatomique et systematique. Ann. Sci. Nat.,
Ser. 10, 14, (1): 117-3~4.
BoPER, K. & MITOHELL, K. J. (1966). The developmental anatomy of perennial ryegrass (Lolium
perenne L.). N.Z. J. Sci. Tech., 37: 484-504.·
80RENBON, T. (1963). A revision of the Greenland species of Puccinellia. Parl. Medd. Grflnl., 136,
(3): 5-146.
STEBBINB, G. L. & ZOHARY, D, (1959). Cytogenio and evolutionary studies in the genus DactyliB.
l. The morphology, distribution, and interrelationship of the diploid subspecies. Umv. Calif.
Publ. (Bot.). 31, (1): 1-40.
EXPLANATION OF PLATE
PLATE 1
The upper leaf epidermis of diploid DactyliB aul>species. a, amithii; b, ibi~::ensis; c, juncinella;
d, Algerian (subep. 8antai); e, Algerian (non-8antai); f, Algerian (non·8antai); g, judaica; h, woronowii;
j, lUBitanica; le, aBChersoniana. 1 division
IOO,u. Enlargements 1 division
50,u.
=
=
.Journ. ]>inn. Soc. Hot. J'of . .Jii, No.
:ws
l'lat.L' I
b
,.
~~
--
MARTIX BORHILL
(i"aciny p. J 5N)