86 H. A. ARMSTRONG & A. W. OWEN
Baum 1991). Sedimentological evidence from
Dob's Linn, located on the Laurentian slope,
indicates that at least in low latitudes ocean
cooling persisted from the early Ashgill, with
the glacial maximum ranging from the late
Rawtheyan to mid-Hirnantian (Armstrong &
Coe 1997). This glaciation has been considered
a major causal factor in the late Ordovician
mass extinction (see reviews in Brenchley 1988;
Brenchley et al. 1995a,b; Armstrong 1995;
Owen & Robertson 1995). At the same time
plate configurations were changing dramatically
associated with the closure of the Iapetus Ocean
(Scotese & McKerrow 1990; van Staal et al.
1998)
We have conducted an analysis of Late
Ordovician euconodont generic diversity in the
circum-Iapetus region. Different diversity trajec-
tories are found in different biofacies in each
of low-latitude Laurentia and mid-temperate
Avalonia and Baltica, and can be attributed
to different causal mechanisms. Laurentian
diversity appears to have been moderated by
changing climate, associated with the onset of the
latest-Ordovician glaciation, whereas Avalonian
diversity reflects the change from temperate to
tropical conditions as the plate drifted north-
wards. This analysis thus supports the hypothesis
that changing palaeoenvironmental setting is the
primary influence on clade diversity within the
euconodonts.
The method we have applied allows regional-
scale diversity patterns to be elucidated from
existing presence-absence data and is thus
widely applicable. The generic associations
established by the method also provide an
opportunity to analyse biofacies architecture
and stability in a number of different regional
settings and hence examine the palaeoecological
impact of global-scale events (see also Droser et
al. 1997, 2000).
Method
Presence-absence matrices for genera have
been compiled from sections in Laurentia and
Avalonia for the upper serra and anserinus
biozones (Llandeilian-Aurelucian), velicuspis
Chron (equivalent to the upper superbus
Biozone in the British biostratigraphical
scheme), upper ordovicicus Biozone and the
celloni Biozone. Genera are considered taxo-
nomically stable and are more likely to be
identified correctly. Each genus comprises less
than five species; several are monospecific. The
results of a cladistic analysis of euconodont
genera are not currently available but many are
likely to be paraphyletic (P. C. J. Donoghue,
pers. comm., 2001). This does not detract from
our analysis as paraphyletic taxa are more likely
to have had narrower palaeoecological ranges
than monophyletic taxa, in the same way that
many paraphyletic fish genera have highly
restricted ecological distributions (Sepkoski &
Kendrick 1993; Kemp 1999; Miller 2000).
Two major Ordovician conodont faunal
provinces have previously been identified in the
Iapetus region: the Midcontinent Province,
including Laurentia, Siberia and parts of China;
and the North Atlantic Province, including
Baltoscandia and eastern Laurentia (Barnes et
al. 1973; Sweet & Bergstrom 1974, 1984).
Cluster analysis of Jaccard and Dice coefficients
for euconodont faunas of the lower Llanvirn
indicate that the faunal similarities between
Baltoscandia and eastern Laurentia are due
entirely to the presence of a cosmopolitan deep-
water fauna, the Protopanderodus-Periodon
Biofacies (comparable to OB2 herein) (Ras-
mussen 1998). Once this biofacies is removed
from the analysis then the North Atlantic
Province is restricted to Baltoscandia and
neighbouring areas. Genera considered
representative of the North Atlantic Province
(including coniform taxa) and the Midcontinent
Province have therefore been separated in the
presence - absence matrix for each region.
The original matrices were compiled with the
localities arranged with an onshore-offshore
distribution (Figs 1-5). The matrices were sub-
jected to manual constrained sedation with the
onshore to offshore distribution of localities
maintained (Brower & Kile 1988). The result
therefore reflects an offshore deepening
palaeoecological gradient. Generic biofacies,
identified as blocks of taxa within the matrix,
were named, and the included genera were
coded for biofacies for each time slice. The
assumption was made that a genus was more
likely to stay within the same biofacies in the
subsequent time slice. Anomalous occurrences
in the initial coding were corrected by a posteri-
ori revision of the seriated matrix. The generic
coding is illustrated in Figure 6; biofacies diver-
sity counts were made from this matrix for each
time slice and region. The juxtaposition of bio-
facies at each locality enables the construction
of a biofacies architecture diagram (Figs 4b, 5b,
7) for each time slice and region. Biofacies
architecture can then be compared through
time. There is no correlation of the shelf bio-
facies with the described shelly invertebrate
biofacies such as the substrate-controlled
palaeocommunities of the late Llanvirn - early
Caradoc of the Welsh basin summarized by
Lockley(1983).