snapdragon seedlings, but from my ob- 
servations of the animals prevalent in 
the vicinity of railroad tracks, I suspect 
that edge feeders, such as woodchucks 
and rabbits, were responsible for much 
of the herbivory. 
Whatever the identity of the herbi- 
vores, they cause a catastrophic mortal- 
ity rate among seedlings: less than one 
seedling in thirty survives long enough 
to set some seed. Plants that do survive 
this early predation are seldom prey to 
herbivores once they have formed ma- 
ture fruits, perhaps because they are 
less palatable than when they were 
younger. In consequence, these plants 
will most likely survive to the end of the 
growing season several months later. 
They may grow to be relatively large 
(eight to ten inches tall, with extensive 
branching), and each may produce 
many thousands of seeds. My present 
record holder is a specimen I found at 
Coming, New York, in July 1979. This 
splendid individual was thirteen inches 
tall, had sixty-two branches (giving a 
total stem length of almost fourteen 
feet), and bore 63 flowers and 383 
fruits. Within its reproductive struc- 
tures, even if no more developed, this 
plant could produce more than 24,000 
seeds! The record holder for smallest 
size at maturity was truly a dwarf snap- 
dragon, just two-thirds of an inch tall 
with a single flower at the top of its un- 
branched stem. 
The finding that lack of sufficient 
water is not the principal cause of seed- 
ling mortality does not eliminate it as a 
source of stress for railroad plants. The 
cinders become so dry and dusty after 
just a few days without rain that water 
shortage must be a severe problem dur- 
ing much of the summer. Since compe- 
tition for the small quantities of water 
present will be particularly intense 
where plants grow close together, I was 
not surprised to find that tall, repro- 
ductively successful dwarf snapdragon 
plants grow well separated from other 
plants and that where the plant grows 
in dense patches of vegetation, only 
small individuals with few reproductive 
structures are found. Dwarf snapdrag- 
on is invariably absent from crossings 
that are heavily overgrown with other 
species of plants. These observations 
suggest that the species is a good colo- 
nizer of bare cinders but either fails to 
grow or is outcompeted by other plants 
in areas of the railbed that hold more 
water, perhaps because of local accu- 
mulations of soil. 
Dwarf snapdragon seeds will germi- 
nate only when water is present in small 
Jay Maise! 
quantities. Seeds placed on moist filter 
paper absorb water and rot; none of 
them ever germinate. The only time I 
have ever succeeded with a laboratory 
germination experiment was when I 
planted seeds in pots containing rail- 
road cinders and watered them very 
sparingly. Much work remains to be 
done, however, before I can understand 
how this plant, once it has germinated 
and taken root, manages to survive 
without rain for long periods, some- 
times several weeks. Its tiny root sys- 
tem, penetrating less than one inch in 
any direction into the cinders, can have 
access to water for only a short period 
after rain or, during a dry spell, from 
the early morning dew. Water loss is no 
doubt minimized by the small, narrow 
leaves, but other anatomical modifica- 
tions common to plants of dry habi- 
tats — sunken stomata or a thick, waxy 
cuticle covering all external surfaces — 
are absent in this species. 
Dwarf snapdragon plants that sur- 
vive the attentions of herbivores and 
are not overcome by a shortage of water 
have only a short period in which to set 
seed before they are killed by herbi- 
cides. I found that all individuals that 
grew to a large size and produced a 
large number of seeds began flowering 
when they were only about one to two 
inches tall and before they had begun to 
branch. I have never found a dwarf 
snapdragon plant taller than about two 
inches that did not have flowers or 
fruits. Clearly, one reason dwarf snap- 
dragon — and no doubt other railroad 
weeds — can be found at a given rail- 
road crossing year after year, despite 
the weed-control program of the rail- 
road company, is that most individuals 
set seed after a short period of vegeta- 
tive growth and before the midsummer 
weedkiller application. 
Before a flower can give rise to a 
mature fruit, it must be successfully 
pollinated. Wind pollination, common 
among many successful weed species, is 
prevented in dwarf snapdragon by the 
enclosure of the male and female repro- 
ductive structures (stamens and pistil) 
of each flower in a two-lipped floral 
tube. Several aspects of the flowers sug- 
gest insect pollination: the colorful ap- 
pearance of the flowers, the arrange- 
ment of the low'er lip of each flower as a 
possible landing platform, and the pres- 
ence of two magenta-colored stripes, 
richly clothed with glandular hairs, on 
the inside of the lower lip. Like a pair of 
fluorescent arrows, these stripes, which 
are sufficiently bright to be visible from 
outside the flower, point the way to the 
nectar and pollen in the interior of a 
flower. Yet, despite long periods of ob- 
servation, days and weeks went by 
without my seeing a single insect as 
much as pause at the sight of the beck- 
oning flowers. My patience was finally 
rewarded, however, one hot and sultry 
July afternoon when I saw a small bee 
in the act of burrowing headfirst be- 
tween the lips of a flower. I watched 
this insect emerge covered with a whit- 
ish dusting of pollen and then fly off to 
other flowers where it repeated the 
process. 
Subsequent observations showed 
that these bees (family Halictidae) fly 
only on the hottest of midsummer days. 
Knowing that the petals of dwarf snap- 
dragon wither and fall to the ground 
within twenty-four hours of opening, I 
reasoned that these insects must polli- 
nate only a very small proportion of the 
flowers that open during the whole 
summer and that self-pollination must 
be the most frequently adopted mode in 
this species. To test the ability of dwarf 
snapdragon to self-pollinate, I tied 
small muslin bags around several flow- 
er buds, thus denying access to would- 
be pollinators. I removed the muslin 
bags after the petals had withered and 
become detached from the ovary of 
each flower. Flowers treated in this way 
formed mature seeds indistinguishable 
in all respects from those produced by 
untreated flowers, evidence that suc- 
cessful self-pollination can occur. Thus, 
dwarf snapdragon has the best of both 
worlds. When growing alone or in the 
absence of insect visitors, an individual 
plant can pollinate itself. With the aid 
of bees, the plant can also exchange pol- 
len with other snapdragons, a less reli- 
able method of reproduction in this 
species but one that insures genetically 
variable offspring. 
While this article was in preparation, 
I learned that the railroad line where I 
carry out my research is to be closed in 
the near future. Although the railroad 
enthusiast in me laments this shut- 
down, as an ecologist I see a rare oppor- 
tunity to study the changes in vegeta- 
tion on the cinders in the absence of 
maintenance activities. What will be- 
come of dwarf snapdragon along this 
line? I predict it will be outcompeted by 
other annuals and by perennials as the 
railbed becomes more and more over- 
grown. Thus, the snapdragon may van- 
ish from an area in which it has held on 
to a precarious existence for many 
years. I will miss this little plant even 
more than the trains that spread it so ef- 
fectively. 
65 
