|
In
the last [1999 BWARS] Newsletter, Murdo Macdonald (1999)
presented valuable new data on the distribution of the
bumble bees B. lucorum
and 'B. magnus'. The question is: what is their
status? Are they two separate species, or perhaps parts
of a single, highly variable species? The possibility
that they represent peculiarly variable populations
of B. lucorum in some regions might, if anything,
be even more intriguing. So how would we tell? I believe
that we still need more information (Williams, 1998:
130).
.
Deciding whether
lucorum and magnus are separate species
may seem straightforward, but it conceals philosophical
and practical problems that are not easily solved. To
answer the question, first we need a clear concept of
what a species is. Unfortunately, there are many different
views (e.g. Claridge et al., 1997),
some of them mutually incompatible. Second, having chosen
a species concept, we may need to ask how this would
lead us to identify criteria by which individual bees
could be judged to belong to separate species or not.
For some concepts, these criteria are not easily applicable
to what can actually be observed in the field.
.
In one of
the more straightforward approaches, Jim Mallet (1995,
1997) has argued recently for minimising the number
of assumptions built into species concepts, so that
there is more freedom to discover which processes are
involved. He suggests that two nominal taxa should be
considered conspecific until it can be demonstrated
that data for several characters distinguish the same
subgroups of individuals with few or no intermediates
(the 'character-cluster' concept of species). Mallet
recognised that his prescription differs little from
common practice and that it is already in use as the
practical application of the biological species concept.
The problem with the cluster concept is how to decide
on a threshold for permissible numbers of intermediate
individuals between taxa for them still to be considered
separate species. How does this apply to magnus?
.
Magnus
[photo]*
is usually distinguished from lucorum [photo]*
by its broader and paler yellow bands, and primarily
by the way the pale 'collar' extends down the sides
of the thorax beneath the wing bases. Other external
characteristics are often associated with this distinction,
including yellowish or brownish-white hair on the 'tail'
(Krüger, 1954),
larger body size (Vogt, 1911;
Baker, 1996), and differences
in microsculpture (Krüger, 1954;
Rasmont, 1984). However,
as a starting point, it is worth concentrating on the
yellow thoracic collar, because this is what is used
most widely as the key character to recognise magnus.
.
[ * These
images can be displayed simultaneously with this text,
but may disappear behind other windows - ensure all
windows are sized and positioned appropriately.]
.
Murdo lent
me a series of 32 queens of this group collected by
him in 1995. If we assume that this is a representative
sample from Murdo's region of Scotland, where both nominal
taxa occur together, then it can be used to look at
variation among the local population of these bees.
In the graph below, I have attempted to measure (from
the left lateral aspect) first, how far the yellow collar
extends (dorso-ventrally) below the tegula (at the wing
base), and second, the maximum breadth (antero-posteriorly)
of the yellow collar below the tegula. These measurements
are only approximate, because it is difficult to measure
the size of a hair patch precisely, particularly where
pale and black hairs are intermixed (better methods
are needed). To reduce the effect of variation in body
size, measurements have been converted from simple distances
to shape ratios, by dividing them by the distance between
and including the tegulae (from the dorsal aspect).
Pierre Rasmont has probably examined more specimens
of the lucorum-group than anyone. He has kindly
named these specimens, as shown on the graph (he recognised
none of these specimens as another problematic taxon,
often confounded in Europe, B. cryptarum). To
this sample, I have added a few queens from the other
end of Britain, in Kent: three from my back garden,
and two from near Dungeness.
.
Plot showing
variation in the principal character used to identify magnus:
the extension of the pale thoracic collar below and behind the
wing bases. These measures are expressed relative to thoracic
breadth in order to reduce the effect of variation in body size
on the comparisons.
.
The graph
shows that there is substantial variation among the
Scottish bees. Clearly we need to know more about variation
among bees of the lucorum-group from other parts
of Britain, in order to assess regional patterns in
the variation. Small samples from upland areas elsewhere
in Scotland, Wales and Devon (not shown here) show a
similar scatter of points. More intriguing is that,
although most of the few included specimens from Kent
are close to the Scottish examples of lucorum,
one unusual specimen from near sea level at Dungeness
appears from these colour characters to be magnus.
.
The graph
confirms that the magnus recognised by Pierre
always have the thoracic collar extending further down
the sides of the thorax than his lucorum (y
axis), despite extensive overlap between the two in
the breadth of the lower part of the collar (x
axis). But although the bees in this sample differ in
length of the collar, there is no obviously large gap
separating a cluster of back triangles from a cluster
of open triangles on the graph. In other words, the
pattern of variation appears to show almost a continuum
of variation between the extremes (although future work
could show this sample to be biased). Even at a pragmatic
level, finding a continuum within the population would
be important, because it is likely to result in different
people drawing the distinction at different collar lengths.
Thus, within this Scottish sample, Pierre recognised
as magnus six queens (open triangles), Murdo
provisionally recognised one, and from the graph above
I might have recognised two (lower right). This is not
to say that any of us is right or wrong - if there were
a continuum of variation, then the precise point where
the distinction is made would be an essentially arbitrary
choice.
.
Finding a
lack of an obvious large gap in variation for this,
the principal character used to discriminate the two
taxa in practice, is crucial in the context of Jim Mallet's
cluster concept of the species mentioned above. Taken
at face value, these data provide little support for
recognising two species with this species concept. For
every other morphological character that has been measured,
the overlap between the two nominal taxa is even more
extensive (Løken, 1973;
Pekkarinen, 1979; Rasmont,
1984; Baker, 1996;
Pamilo et al., 1997).
If there were indeed many intermediates generally in
areas of co-occurrence, then according to the cluster
concept they should be treated as subspecies (as is
currently the case with B. muscorum
and 'smithianus' [= B. muscorum bannitus],
which have intermediate individuals on Skye).
.
Of course,
this sample of Scottish bees is small and from just
one region. Variation of the pale collar and the other
characters needs to be looked at much more closely,
providing many opportunities for field work that could
be enlightening. For example, it would be useful to
know more about variation of queens within colonies
as well as among them, particularly from areas where
lucorum and magnus occur together (studying
only the more extreme forms from distant areas where
they may not co-occur, such as the Outer Scottish Isles
and London, is less informative). What are the patterns
of inheritance of this colour variation? Variation might
also be explored in relation to transects by altitude
and longitude, as discussed by Murdo. For this kind
of analysis, it is essential that bees are collected
unselectively. A potentially insidious problem is that
people may be tempted to collect 'good' specimens of
magnus or lucorum, ignoring others with
possibly less convenient, intermediate colour patterns.
Selective collecting definitely does occur, unwittingly
or not, as described by Astrid Løken (1973).
In this case, dark forms of B. hortorum
from Scandinavia were much more frequent in museum collections
than they were in random samples from the field.
.
That
the bees vary and that this variation shows geographical
and ecological pattern is clear. But if discovering
whether or not light and dark bees are parts of the
same species is not straightforward, then seeking good
evidence to discriminate among possible explanations
for the origin and maintenance of divergence between
them (in terms of factors that might include habitat
specialisation, competition, unique events in history,
etc.) will be even more difficult.
.
[See
update by Williams et al., 2012
[pdf].]
.
Home
|
