VOLCANISM AND DIVERSIFICATION 103
organic rain allowed a benthic bloom of sessile
suspension-feeders to develop until oxygenation
was depleted below tolerable limits.
The total duration of these bloom events is
reconstructed, based on a comparison with
modern bloom events (e.g. Gallardo et al. 1977)
and post-seismic sedimentation rates (Goff
1997), as five to ten years, with up to one year for
the plankton bloom (Botting 2000). The area
affected depends on the extent of ash dispersal.
Downwelling flows would decay rapidly through
lateral entrainment of fluid, unless concentrated
by continued ash input from above. Distal
deposits in deep shelf settings would thus have
been unable to host benthic blooms, although
the plankton bloom would be unaffected. This
would have provided elevated organic rain and
encouraged benthic anoxia at the ash dispersal
margins, where the high organic input was not
offset by oxygenated downwelling (Fig. 2).
While large eruptions affect wide areas, a typical
pyroclastic event may induce benthic blooms in
areas of a few hundred to a few thousand square
kilometres. Coarse ash and crystal-tuff beds are
less likely to initiate vertical circulation, since
the critical particle density for mass flow is less
likely to be exceeded, and individual particle
settling is much more rapid and more varied.
Population genetics
The presence of repeated regional bloom
events, following the removal of small or sessile
benthos, potentially influenced population
genetics in several ways. For general discussion
of relevant concepts, refer to Harwood & Amos
(1999), Amos & Harwood (1998), and refer-
ences therein. Primarily, bloom events encour-
age the retention of novel characters, via an
increased surviving proportion within each
generation. Although the survival probabilities
of such mutations are not elevated after the
bloom maximum, the reduced juvenile mortality
during the bloom growth phase increases the
chance of persistent variations. For example,
beneficial genotypic changes that require two
mutations, one neutral or detrimental, have very
low probability of arising under normal popu-
lation conditions. During a bloom increase
phase, carriers of each mutation are much more
abundant, allowing carriers of both to arise, and
thereafter become preferentially selected. Ford
& Ford (1930) described abundance and vari-
ation fluctuations in an isolated butterfly popu-
lation (of duration 10-20 years); variation
was maximized during a population increase,
with the subsequent maximum and declining
populations being more homogenous. The final
population was morphologically distinct from
the initial. While a single bloom event is prob-
ably insufficient for speciation to occur from a
homogeneous population, the result is to
produce genetic heterogeneity on a subregional
scale. This may allow distinct species to appear
rapidly under subsequent blooms, as disparate
parts of an already heterogeneous population
are brought into immediate contact during
recolonization (discussed below). Although
some intraspecific homogenization would be
expected to occur during intervening quiescent
periods, migration of genetic characteristics
through a stable population is much slower than
physical migration of a population into a barren
region.
Following the eradication of small mobile and
sessile benthos by ash deposition, recolonization
occurred by some combination of exponential
population increase of rare survivors, and immi-
gration from the surrounding area. The delayed
onset of the sessile benthic bloom relative to the
mobile bloom, interpreted above as resulting
from low food supplies inhibiting suspension-
feeders, may also suggest a gradual immigration
In reality, the bloom population was almost cer-
tainly derived from geographically separated
lateral immigrants, combined with a remnant
endemic population; lateral separation was
probably tens to hundreds of kilometres,
depending on the eruption size. The genetic
diversity of the benthic bloom biota should thus
be much greater than where the population is
derived from a single endemic community. This
disparity is further exaggerated by the process
discussed above, whereby previous bloom
events induced strong regional genetic variabil-
ity. The populations at opposite edges of the
affected area would be expected to differ more
than in a stable environment, with these popu-
lations becoming immediately adjacent during
recolonization. Within the resultant population,
hybridization and segregation between end-
members would result in a higher rate of specia-
tion than in more homogeneous communities.
Hybridization would further promote the
appearance of individuals with distinctive
characteristics by providing novel gene combi-
nations, while the initiation of non-interbreed-
ing subpopulations is a necessary first stage in
sympatric speciation. In a gradually varying
population, there is little incentive for segrega-
tion, but behavioural divisions can exist immedi-
ately when disparate parts of that population are
brought into contact.
The general situation of colonization of an
ecologically depauperate region promotes
genetic variation under many circumstances
