Content uploaded by Andrew Flynn
Author content
All content in this area was uploaded by Andrew Flynn on Apr 25, 2023
Content may be subject to copyright.
Article
Early Paleocene tropical forest from the Ojo Alamo Sandstone, San
Juan Basin, New Mexico, USA
Andrew G. Flynn and Daniel J. Peppe
Abstract.—Earliest Paleocene megafloras from North America are hypothesized to be low diversity and
dominated by long-lived cosmopolitan species following the Cretaceous/Paleogene (K/Pg) mass extinc-
tion. However, megafloras used to develop this hypothesis are from the Northern Great Plains (NGP) of
North America, and relatively little is known about floras from southern basins. Here, we present a quan-
titative analysis of an earliest Paleocene megaflora (<350 kyr after K/Pg boundary) from the Ojo Alamo
Sandstone in the San Juan Basin (SJB), New Mexico. The megaflora, comprising 53 morphotypes, was
dominated by angiosperms, with accessory taxa composed of pteridophytes, lycophytes, and conifers.
Diversity analyses indicate a species-rich, highly uneven, and laterally heterogeneous flora. Paleoclimate
estimates using multivariate and univariate methods indicate warm temperatures and relatively high pre-
cipitation consistent with a modern tropical seasonal forest.
When compared with contemporaneous floras from theDenver Basin (DB) of Colorado and theWilliston
Basin (WB) of North Dakota, the SJB flora had significantly higher species richness but lower evenness.
Paleoclimate estimates from the SJB were 7–14°C warmer than the estimates for the DBand WB, indicating
a shift froma temperate forest in the NGP to atropical forestin the SJB. These results demonstrate the pres-
ence of a latitudinal floral diversity and paleoclimatic gradient during the earliest Paleocene in western
North America. We hypothesize that the warm, wet conditions in the earliest Paleocene SJB drove rapid
rates of speciation following the K/Pg boundary, resulting in a diverse and heterogeneous flora.
Andrew G. Flynn and Daniel J. Peppe. Department of Geosciences, Baylor University, Waco, Texas 76706, U.S.A.
E-mail: Andrew_Flynn@Baylor.edu,Daniel_Peppe@Baylor.edu
Accepted: 27 June 2019
First published online: 12 September 2019
Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.j0k8370
Introduction
The Cretaceous/Paleogene (K/Pg) boundary
at ∼66 Ma is perhaps best known for the extinc-
tion of nonavian dinosaurs (e.g., Schulte et al.
2010; Brusatte et al. 2015). However, there was
also a major extinction of plant taxa across the
K/Pg boundary, with ∼50–60% and 15–30%
of mega- and microfloral species becoming
extinct, respectively (e.g., Wilf and Johnson
2004; Nichols and Johnson 2008; Vajda and Ber-
covici 2014). The best records of this extinction
in plants are from North America (e.g., Nichols
and Johnson, 2008). The postextinction North
American record documents a major restructur-
ing of terrestrial ecosystems, a destabilization of
terrestrial food webs, and a prolonged recovery
that extended at least into the middle Paleocene
(Hickey 1980; Wing et al. 1995; McIver 1999;
Dunn 2003; Wilf and Johnson 2004; Wilf et al.
2006; Peppe 2010; Blonder et al. 2014). Thus,
reconstructions of patterns of plant community
diversity and composition in the early Paleo-
cene are necessary to fully understand both
local and regional ecosystem recovery following
the K/Pg mass extinction.
Extensive early Paleocene megafloral collec-
tions have been made in North America for
over 150 years (e.g., Newberry 1868; Lesquereux
1878;Brown1962; Hickey 1980; Wolfe and
Upchurch 1987; Wing et al. 1995; Johnson
2002; Barclay et al. 2003; Peppe 2010). The
majority of these studies have focused on the
Northern Great Plains (NGP) of North America
(Fig. 1A) (e.g., Newberry 1868; Lesquereux
1878;Brown1962; Hickey 1980; Barclay et al.
2003; Dunn 2003; Wilf and Johnson 2004;
Peppe 2010), with large collections of floras
from the first 300,000 years of the Paleocene
being primarily from the Denver Basin (DB) of
central Colorado (Barclay et al. 2003; Johnson
Paleobiology, 45(4), 2019, pp. 612–635
DOI: 10.1017/pab.2019.24
© 2019 The Paleontological Society. All rights reserved. 0094-8373/19
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
FIGURE 1. Regional basin map, San Juan Basin (SJB) geologic map, and map of the study area. A, Early Paleocene basins of
western North America (modified from Peppe 2010) showing the location of the SJB. B, Geologic map of the southwestern
SJB showing the geographic occurrences of Late Cretaceous–Eocene deposits and the study area (modified from William-
son et al. 2008). C, Geologic map of the Bisti/De-Na-Zin Wilderness Area with fossil leaf localities indicated (filled shapes,
census collections; open shapes, voucher collections).
SAN JUAN BASIN MEGAFLORA 613
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
et al. 2003) and the Williston Basin (WB) of
North Dakota (e.g., Johnson 1989,2002; Johnson
and Hickey 1990; Wilf and Johnson 2004; Peppe
2010). Early Paleocene plant communities from
the NGP are characterized by low diversity and
are dominated by long-lived, cosmopolitan,
mire-adapted taxa (Hickey 1980; Johnson 2002;
Barclay et al. 2003; Peppe 2010). These floras
were dominated by fast-growing angiosperms
with high assimilation rates and low carbon
investment, suggesting the K/Pg bolide impact
selected against slower-growing evergreen spe-
cies, leading to large-scale ecosystem restructur-
ing (Blonder et al. 2014). These patterns of low
floral diversity and relatively homogeneous
floral composition have been documented across
a distance of >1200 km from southern Alberta
to central Colorado (e.g., Brown 1962; Hickey
1980; McIver and Basinger 1993; Johnson 2002;
Barclay et al. 2003;Dunn2003; Peppe 2010).
Analyses of NGP floras through the Paleocene
indicate that it took several million years for
floras to fully recover and approach levels of
floral diversity common in the Late Cretaceous
(Hickey 1980; Wing et al. 1995;Dunn2003;
Wilf and Johnson 2004; Peppe 2010).
In contrast to the patterns from the NGP (e.g.,
Brown 1962; Johnson 2002; Barclay et al. 2003;
Peppe 2010), fossil megafloral patterns from
the first ∼3.0 Myr of the Paleocene collected
along the Colorado Front Range in the DB are
considerably more diverse and comprise
many species that are interpreted as endemic
to the area (Johnson and Ellis 2002; Ellis et al.
2003; Johnson et al. 2003). While contemporan-
eous floras from the NGP had ∼10–40 morpho-
types (Hickey 1980; Wing et al. 1995; Johnson
2002; Dunn 2003; Wilf and Johnson 2004;
Peppe 2010), the floras from the Colorado
Front Range typically have >30 morphotypes
and are the basis for suggestions that rapid
floral recovery and high rates of speciation fol-
lowing the K/Pg boundary along the margins
of the ancestral Rocky Mountains (Johnson
and Ellis 2002; Ellis et al. 2003; Johnson et al.
2003). This pattern of recovery is in stark con-
trast to patterns of speciation and diversity
documented across the NGP, including floras
from <50 km to the east in the central DB
(e.g., Wing et al. 1995; Barclay et al. 2003; John-
son et al. 2003; Wilf and Johnson 2004; Peppe,
2010), and suggests both latitudinal differences
in diversity and/or the potential for local or
regional diversity hotspots that resulted from
rapid recovery after the K/Pg extinctions.
Further, the high proportion of endemic
megafloral taxa in the Colorado Front Range
(Johnson and Ellis 2002; Ellis et al. 2003; John-
son et al. 2003) suggests that the early Paleocene
flora is not as homogeneous as has been sug-
gested, and instead that there are latitudinal
differences in floral composition across North
America. Studies of pollen and fossil wood
from the latest Cretaceous and early Paleocene
also support a latitudinal gradient in floral
composition across North America and a clear
differentiation in composition of floras from
the mid- to low latitudes and floras from the
NGP (Frederiksen 1987; Nichols et al. 1990;
Lehman and Wheeler 2009). Nichols et al.
(1990) found that floras from mid- to low lati-
tudes had higher diversity and were more
dominated by endemic taxa than floras from
more northern provinces. However, relatively
little research has been conducted on the fossil
leaf floras from southern North America, limit-
ing comparisons of the megafloral record
across latitudes and a comparison of the bio-
geographic patterns identified from both fossil
wood and pollen to the megafloral record.
Here we describe a relatively diverse early
Paleocene fossil megaflora from the San Juan
Basin of New Mexico that occurred within the
first ∼350 kyr of the Paleogene (Figs. 1,2). We
describe the composition and diversity of floral
communities, examine differences in floral
communities between depositional environ-
ments, and estimate mean annual temperature
and precipitation within the San Juan Basin.
Finally, we compare the earliest Paleocene San
Juan Basin megaflora to contemporaneous
megafloras from the NGP of North America
to assess regional differences in floral diversity,
composition, and paleoclimate during the earli-
est Paleocene.
Geologic Setting
The San Juan Basin (SJB), located in north-
western New Mexico and southwestern Color-
ado (Fig. 1A), is a foreland basin formed during
the Laramide Orogeny (Chapin and Cather
ANDREW G. FLYNN AND DANIEL J. PEPPE614
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
FIGURE 2. Generalized geochronology and measured section of latest Cretaceous Naashoibito Member and early Paleo-
cene Ojo Alamo Sandstone and Nacimiento Formations. Age interpretations based on Mason (2013) and Peppe et al.
(2013). North American Land Mammal Age (NALMA) and biozone (Pu1, Pu2) from Williamson (1996). The stratigraphic
position and lithology of all fossil leaf–bearing horizons are shown.
SAN JUAN BASIN MEGAFLORA 615
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
1983; Williamson 1996). The basin preserves a
relatively continuous succession of terrestrial
sedimentary rocks spanning the Late Cret-
aceous (Campanian) through the early Paleo-
cene (Baltz et al. 1966;O’Sullivan et al. 1972;
Williamson 1996; Williamson et al. 2008). Fossil
plant material for this study was collected from
the earliest Paleocene Ojo Alamo Sandstone in
the Bisti/De-Na-Zin (BDNZ) Wilderness Area
(Fig. 1B,C), which varies from 10 to 15 m in
thickness in the study area and is predomin-
antly a gold to yellow-colored, cross-bedded,
medium- to coarse-grained sandstone with
interbedded sandstone and siltstone deposits
and localized carbonaceous shale beds (Supple-
mentary Fig. 1; Baltz et al. 1966;O’Sullivan
et al. 1972). The Ojo Alamo Sandstone is
thought to represent an alluvial plain with
one or more sediment sources in the southern
Rocky Mountains that was deposited during
the onset of tectonism associated with basin
development (Powell 1973; Chapin and Cather
1983;Sikkink1987;Cather2004). The amalga-
mated channel belts, low mud to sand ratio,
and the presence of relatively high flow regime
bedforms correspond with proximal deposits
in a distributed fluvial system (Weissmann
et al. 2013), which supports the hypothesis that
the Ojo Alamo was deposited during a time of
relatively limited accommodation space asso-
ciated with the onset of basin subsidence in the
latest Cretaceous and early Paleocene (Cather
2004). The Ojo Alamo Sandstone overlies the
Maastrichtian Naashoibito Member of the Kirt-
land Formation and underlies the early Paleo-
cene Nacimiento Formation (Baltz et al. 1966;
Lindsay et al. 1978; Williamson 1996;Mason
2013; Peppe et al. 2013). The contact between
the Ojo Alamo Sandstone and Nacimiento For-
mation is complex and regionally varies from
being conformable with the basal paleosols of
the Nacimiento weathering into the upper Ojo
Alamo Sandstone to unconformable with evi-
dence for the contact being erosive (e.g., Wil-
liamson and Lucas 1992; Williamson 1996;
Davis et al. 2016). In the BDNZ, the contact
between the Ojo Alamo Sandstone and Naci-
miento Formation appears to generally be con-
formable (Williamson 1996; Davis et al. 2016).
Differences in nomenclature between the Ojo
Alamo Sandstone and the Naashoibito Member
of the Kirtland Formation have caused confu-
sion when relating the two lithologic units
(e.g., Sullivan et al. 2005; Williamson and Weil
2008; Fassett 2009). The Ojo Alamo Sandstone
and Naashoibito Member have both been
recognized as unique stratigraphic units in
different formations (i.e., the Ojo Alamo Sand-
stone is its own formation and the Naashoibito
is a member of the Kirtland Formation [e.g.,
Baltz et al. 1966; Williamson 1996; Williamson
and Weil 2008; Williamson et al. 2008]) or as dif-
ferent members within the Ojo Alamo Forma-
tion (i.e., the Kimbeto Member and the
Naashoibito Member [e.g., Sullivan et al.
2005; Fassett 2009]). Herein we use the defin-
ition of Baltz et al. (1966) for the Ojo Alamo
Sandstone and recognize it as a unique strati-
graphic formation. For clarity, this is equivalent
to the Kimbeto Member of the Ojo Alamo
Formation proposed by Sullivan et al. (2005).
In addition to confusion about the nomencla-
ture of the Ojo Alamo Sandstone and the Naa-
shoibito Member of the Kirtland Formation, the
age of the units remains contentious. Some
authors have claimed that both the Ojo Alamo
Sandstone and Naashoibito Member of the
Kirtland Formation are Paleocene (Fassett
2009), while others have interpreted the Naa-
shoibito Member of the Kirtland Formation to
be Late Cretaceous and the Ojo Alamo Sand-
stone to be early Paleocene (Sullivan and
Lucas 2003; Sullivan et al. 2005; Williamson
et al. 2008). Palynostratigraphy of the Ojo
Alamo Sandstone and the Nacimiento Forma-
tion indicates that the Ojo Alamo Sandstone is
earliest Paleocene (Anderson 1959; Williamson
et al. 2008). Further, the occurrence of Momi-
pites inaequalis in the Ojo Alamo Sandstone
(Anderson 1959), suggests that it can be corre-
lated to the North American Paleocene paly-
nostratigraphic zones P1 or P2 (Nichols 2003).
Mason (2013) used detrital sanidine dating to
constrain the maximum depositional age of
the Naashoibito Member of the Kirtland For-
mation to the late Maastrichtian and the Ojo
Alamo Sandstone near Cuba, New Mexico, to
early Paleocene in age. Magnetostratigraphy
from the BDNZ Wilderness Area through the
Naashoibito Member of the Kirtland Forma-
tion, Ojo Alamo Sandstone, and lower Naci-
miento Formation indicate that the upper
ANDREW G. FLYNN AND DANIEL J. PEPPE616
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
Naashoibito Member of the Kirtland Forma-
tion, the entire Ojo Alamo Sandstone, and the
basal 4 m of the Nacimiento Formation are all
reversed polarity, indicating deposition within
magnetic chron 29r (Peppe et al. 2013). Based on
the Ojo Alamo Sandstone pollen assemblage and
geochronology of the Naashoibito Member of the
Kirtland Formation, Ojo Alamo Sandstone, and
Nacimiento Formation, we interpret that the
Ojo Alamo floras, which are the focus of this
work, to have been deposited within C29r during
the first ∼350 kyr of the Paleocene (Fig. 2).
Methods
Collection and Classification
Plant fossils were collected from 12 localities
from the Ojo Alamo Sandstone in the BDNZ
Wilderness Area during field seasons in 2013–
2016 (Fig. 1C). Because the upper and lower
contacts of the Ojo Alamo are not flat lying,
the stratigraphic position of each leaf locality
was measured to both from the top of the
underlying Naashoibito Member of the Kirt-
land Formation and to the base of the overlying
Nacimiento Formation (Fig. 2). Localities ran-
ged from 4.65 to 14.5 m above the base of the
Ojo Alamo Sandstone and between 10.0 and
1.0 m below the Nacimiento Formation contact
(Fig. 2). Localities were assigned a two-letter
code for site discoverer, and a field locality
number using a two-digit annual number indi-
cating the year of collection and a sequential
site number (e.g., AF1404 is the fourth site
found by A.F. in 2014 and DP1301 is the first
site found by D.P. in 2013); when more than
one quarry was collected at a locality, sites
were given an additional letterto denote differ-
ent quarries (e.g., AF1409D is the fourth quarry
from the ninth site found by A.F. in 2014). All
collected specimens are stored in the Carlisle
Geology Research Building at Baylor Univer-
sity in Waco, Texas, USA.
Fossil leaves were collected using the bench
and quarry method in which large blocks of
matrix were split along leaf-bearing bedding
planes (following Johnson 1989). Fossil leaf
quarries covered ∼5–10 m
2
of surface area and
generally contained multiple, relatively thin
(∼5–25 cm) leaf-bearing horizons. At all leaf
quarries, the sedimentological features of the
site were recorded, and sites were assigned to
sedimentary facies.
Voucher collections were made at all fossil
localities. These voucher collections were
selectively collected based on preservation. In
addition to collecting the best-preserved speci-
mens, when making voucher collections, we
also collected at least one specimen from each
known morphotype and all reasonably well-
preserved unidentified plant fossils. Census
collections were made at six of these localities,
in addition to the voucher collection. During
census collections, we quantitatively counted
all identifiable plant material. At least 300 iden-
tifiable specimens were tallied in the field dur-
ing census collections, because modern
taphonomic studies have shown that ≥300 spe-
cimens are needed to accurately reflect floral
composition (Burnham et al. 1992). During cen-
sus collections, a new quarry adjacent to the
voucher collection site was dug, and a new col-
lection was made for the census. Representative
samples and/or well-preserved samples from
each identifiable morphotype and all unknown
fossil material were collected. All collected spe-
cimens were given a unique site and specimen
identification number (i.e., the sixth census
morphotype from the site AF1404 census col-
lection was labeled AF1404-C-006 with the
“-C”denoting census collections from sites
with voucher collections). After collection, all
samples were wrapped in toilet paper, labeled,
and shipped to Baylor University for additional
analyses.
Fossil leaves are common in the Ojo Alamo
Sandstone and were collected from two distinct
lithologies, which we have interpreted as dif-
ferent depositional environments (Supplemen-
tary Fig. 1). The first fossil leaf–bearing
lithology is interbedded medium- to fine-
grained sandstone and siltstone beds (Supple-
mentary Fig. 1D). The interbeds generally dip
at relatively low angles and pinch out laterally.
Sedimentary structures, such as ripples and
small-scale cross-bedding, occasionally occur.
Well-preserved fossil leaves are common in
the fine-grained interbeds and rare in the
coarser-grained sandstone beds. Based on the
physical characteristics, we interpret these
deposits to be downstream accretionary bar
forms in a braided river system with the finer-
SAN JUAN BASIN MEGAFLORA 617
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
grained siltstones representing periodic flood-
ing over the predominantly sand bar forms
(Powell 1973; Miall 2010). We classified five
leaf localities collected from this sedimentary
facies as “overbank facies”(42% of sites;
Figs. 1C and 2, Supplementary Fig. 1D).
The second fossil leaf–bearing lithology is
very fine grained, thinly laminated, carbon-
aceous shale beds with locally abundant sulfur
and jarosite. The carbonaceous shales are often
preserved as lenticular deposits ∼15–25 m
across and ∼1–4 m in thickness (Supplemen-
tary Fig. 1F,G). The beds are flat lying, and no
additional sedimentary structures were
observed. Fossil leaves generally occur in very
dense leaf mats in these beds. We interpret
the carbonaceous shale deposits to be low-
lying, abandoned channel fills that were off
axis from the major source of deposition in
the braided stream, allowing ponding of
water and very-fine-grained sediments to be
deposited (Miall 2010). These deposits are
analogous to oxbow lakes in meandering flu-
vial systems (Miall 2010). We classified seven
leaf localities from this sedimentary facies as
“pond facies”(58% of sites; Figs. 1C and 2,
Supplementary Fig. 1F,G).
In the lab at Baylor University, we assigned
all identifiable plant material to SJB–wide mor-
photypes (morphotypes denoted by the prefix
SJ- and their morphotype number, e.g., SJ-13).
Morphotypes are distinct floral morphologies
within a flora with no formal taxonomic assign-
ment but are presumed to represent biological
species (for review of morphotyping method,
see Ash et al. [1999], Peppe et al. [2008], and
Ellis et al. [2009]). When possible, morphotypes
were also assigned to previously described
Linnaean taxa for comparison to previously
published collections (Supplementary Table 1).
The majority of morphotypes from the
Ojo Alamo Sandstone represent apparently
previously undescribed species and are
presumed to represent endemic taxa. Non-
monocotyledonous angiosperm morphotypes
(which will be referred to as dicotyledonous
angiosperms or dicots for the remainder of
the text following common convention) were
described using the Manual of Leaf Architecture
(Ellis et al. 2009). All other plant groups were
described based on their morphology and
distinctive traits. Brief descriptions and illustra-
tions of all morphotypes are included in the
Supplementary Material (Appendix 1), and a
systematic list of morphotypes with their
inferred taxonomy and species names is
included in Supplementary Table 1.
Floral Diversity
The SJB megafloral diversity and compos-
ition were independently analyzed. These ana-
lyses were then compared with age-equivalent
sites (i.e., early Paleocene localities within mag-
netic C29r) from the DB of central Colorado
(Barclay et al. 2003) and the WB of western
North Dakota and eastern Montana (Wilf and
Johnson 2004; Peppe 2010)(Fig. 1A). These con-
temporaneous localities provided a tightly con-
strained interval spanning ∼350 kyr after the
K/Pg boundary (Ogg 2012). The geographic
extent of collections made in the SJB (∼5km
2
)
is similar to the collection area in the DB
(∼1.5 km
2
; Barclay et al. 2003). However, the
WB material was collected over an area
>100 km
2
(Wilf and Johnson 2004; Peppe
2010), which may have affected the diversity
results of the WB when compared with the
SJB and DB.
All diversity analyses were performed using
the paleontological statistical program PAST
3.0 (Hammer et al. 2001). Rarefaction analyses
were conducted on the six megafloral quantita-
tive collections in the SJB, and all quantitative
collections from sites within C29r in the DB
and WB (Barclay et al. 2003; Wilf and Johnson
2004; Peppe 2010) using all vegetative (i.e., non-
reproductive) morphotypes. Diversity analyses
were performed using only vegetative morpho-
types due uncertainty of taxonomic placement
of reproductive organs and the possibility of
double counting morphotypes. Rarefaction
analyses for only dicot angiosperm leaves
were also conducted for the quantitatively col-
lected localities from the SJB, DB, and WB with
≥200 dicotyledonous angiosperm samples.
Total basin rarefaction analyses, in which all
census data from each basin were combined,
were also calculated for all plant groups and
only dicot angiosperms from the SJB, DB, and
WB. Ecological diversity indices were calcu-
lated using the same quantitatively collected
localities from the SJB, DB, and WB to compare
ANDREW G. FLYNN AND DANIEL J. PEPPE618
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
the species richness, evenness, and dominance
of the different floras. The differences in floral
species richness was estimated using the Mar-
galef diversity index (D
mg
) (Margalef 1958;
Magurran 2004). Floral evenness was assessed
using Pielou’s evenness index (J′) (Pielou
1969). Floral dominance was assessed using
the Berger-Parker index (d) (Berger and Parker
1970). All diversity indices were bootstrapped
(N= 10,000) to generate confidence for site
and basin comparisons.
In addition to species richness and diversity
analyses, the influence of depositional facies,
floral heterogeneity, and similarity was ana-
lyzed using the six SJB census collections.
Detrended correspondence analysis (DCA)
was used to assess the influence of depositional
facies (i.e., landscape position) on floral com-
position. Cluster analysis was also performed
using the Raup-Crick similarity index (Raup
and Crick 1979) to identify groupings of local-
ities with all analyses bootstrapped (N=
10,000). Floral heterogeneity between sites
(i.e., β-diversity) was analyzed using the Bray-
Curtis distance index (Bray and Curtis 1957)
between all pairs of census localities. The
total, within-facies, and between-facies results-
were then averaged to obtain total and facies-
specific values of floral heterogeneity.
Paleoclimate and Paleoecology
Mean annual temperature (MAT) and mean
annual precipitation (MAP) were reconstructed
using leaf physiognomic methods. For the SJB
material, digital leaf physiognomy (DiLP), a
multivariate paleoclimate model that uses the
size and shape of fossil leaves, was used to esti-
mate the MAT and MAP (Huff et al. 2003;
Royer et al. 2005; Peppe et al. 2011). All nona-
quatic woody dicot angiosperm leaves were
measured using the protocols of Peppe et al.
(2011) and Royer et al. (2005), which are briefly
summarized here (see Supplementary Fig. 2 for
examples). Fossil leaf specimens were digitally
extracted from the rock matrix using Adobe
Photoshop (Adobe Systems, San Jose, CA).
The leaf size was then reconstructed for suffi-
ciently well-preserved specimens, and the
inferred blade area, inferred major axis length,
and inferred Feret’s diameter were calculated
for each specimen in this subset of fossil leaves.
For any morphotypes with an entire margin
(i.e., lacking leaf margin teeth) for which leaf
size could not be reconstructed, only the mar-
gin state was recorded. Fragmentary toothed
leaves were included if two or more consecu-
tive teeth and at least 25% of the leaf area and
margin were preserved. For fragmentary speci-
mens, the damaged margin was removed and
only the total number of teeth, undamaged per-
imeter length, and undamaged leaf area were
recorded. For toothed leaves too fragmentary
to measure for DiLP, only the margin state
was scored.
Leaf margin analysis (LMA) (e.g., Wilf 1997;
Miller et al. 2006; Peppe et al. 2018) and leaf
area analysis (LAA) (e.g., Wilf et al. 1998;
Peppe et al. 2018) were used to estimate MAT
and MAP, respectively, for the SJB, DB, and
WB to facilitate regional paleoclimate compari-
sons. LMA is a univariate climate model that
uses the presence or absence of toothed
woody dicot angiosperm leaves to estimate
MAT (Wilf 1997; Miller et al. 2006; Peppe
et al. 2018). The standard error for the LMA
MAT estimate was calculated using the uncer-
tainty equation of Miller et al. (2006). LAA is
a univariate climate model that uses the aver-
age leaf size of a flora to estimate MAP (Wilf
et al. 1998; Peppe et al. 2018). Each morphotype
was scored to a Raunkiaer-Webb leaf size class
(Webb 1959) to estimate leaf area. Site mean leaf
area was calculated using the methods of Wilf
et al. (1998).
Leaf mass per area (M
a
) is a proxy for leaf life
span (Wright et al. 2004; Royer et al. 2007,2010;
Riva et al. 2016), and the M
a
threshold between
deciduous and evergreen taxa is ∼129 g m
−2
(Wright et al. 2004; Royer et al. 2007,2010).
The M
a
of fossil leaves was estimated following
the methods of Royer et al. (2007) using the
relationship between petiole width (PW) and
leaf area (A) to M
a
.
Results
San Juan Basin Floral Description
Megafloral collections from the SJB yielded
55 morphotypes: 5 pteridophytes (9.43% of
the morphotypes), 1 lycophyte (1.89% of mor-
photypes), 2 conifers (3.77%), 7 monocotyle-
donous (monocot) angiosperms (13.21%), and
SAN JUAN BASIN MEGAFLORA 619
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
38 dicot angiosperms (71.70%) (Table 1). The 6
census collections produced 2939 specimens
composed of 4.86% pteridophytes, 0.71% lyco-
phytes, 3.27% conifers, 4.83% monocots, and
86.33% dicots (Table 1). The majority of the
morphotypes have not been previously
described (69.81%) and may represen taxa
endemic to the SJB during the early Paleocene
(Supplementary Table 1).
San Juan Basin Floral Diversity
The average site rarefied richness from the six
Ojo Alamo census localities was 12.99 ± 0.86
morphotypes (downsampled to 300 specimens
per locality; Fig. 3, Supplementary Table 7).
With the exception of site AF1402, the site rar-
efaction curves were not asymptotic, indicating
that total species richness at most sites has not
been fully sampled (Fig. 3) (Foote 1992; Ham-
mer et al. 2001). The rarefied richness for the
entire Ojo Alamo flora was 34.52 ± 2.17 mor-
photypes (downsampled to 1000 specimens),
Supplementary Table 7), and the rarefied rich-
ness for only dicot leaves was 21.79 ± 1.90 mor-
photypes (downsampled to 900 specimens).
There was a strong facies effect on floral
composition and diversity within the SJB
flora. The pond and overbank floras had a simi-
lar number of total morphotypes, 28 and 32
morphotypes, respectively (Table 2). However,
the taxonomic composition of the floras from
each facies is notably different. Floras from
both facies share 12 morphotypes despite
being located within a relatively small geo-
graphic area (∼5km
2
;Fig. 1). DCA indicated
that composition and relative abundance of
morphotypes was considerably different
between the pond and overbank facies
(Fig. 3A). Raup-Crick cluster analysis resulted
in two major clusters corresponding to the
overbank and pond localities (Fig. 3B), and
the clustering of overbank and pond localities
into separate groups occurred 100% of the
time, indicating the difference in floral compos-
ition between facies is highly significant. Clus-
ter analysis also indicated that overbank
localities were less similar to each other than
the pond localities, with the two overbank
localities having a 0.25 similarity index, while
the four pond localities had a similarity index
>0.9 (Fig. 4B). The overall mean Bray-Curtis
distance between localities was 0.278, indicat-
ing considerable compositional differences
between the censused floras (Table 3). The aver-
age Bray-Curtis distance was 0.571 between
pond localities and 0.396 between overbank
localities, indicating that the pond localities
were more similar to one another than the over-
bank localities. Additionally, the mean Bray-
Curtis distance was 0.154 between facies and
0.588 within facies. These results are statistic-
ally significant (t= 4.312, p= 0.008, df = 13),
indicating that the β-diversity between differ-
ent facies was significantly greater than within
facies. The diversity and species richness of the
pond and overbank floras were also sig-
nificantly different. When all plant groups
were included, the overbank flora had a greater
site-based rarefied richness at 300 specimens
than the pond flora (17.70 ± 0.51 morphotypes
[n= 2] for the overbank vs. 10.63 ± 1.04 mor-
photypes [n= 4] for the pond; Fig. 4). Further,
TABLE 1. Number of morphotypes and specimens by major taxonomic category and organ type. For detailed taxonomic
information see Supplementary Table 1; for morphotype descriptions and illustrations see Supplementary Appendix 1.
Higher taxon or organ Morphotypes Specimens % Morphotypes % Specimens
Pteridophytes Leaves 4 142 7.55 4.83
Reproductive structures 1 1 1.89 0.03
Lycophytes 1 21 1.89 0.71
Conifers 2 96 3.77 3.27
Monocotyledonous angiosperms Leaves 6 141 11.32 4.80
Reproductive structures 1 1 1.89 0.03
Dicotyledonous angiosperms Leaves 35 2533 66.04 86.19
Reproductive structures 3 4 5.66 0.14
Total 53 2939
ANDREW G. FLYNN AND DANIEL J. PEPPE620
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
when all sites from each facies were combined,
the overbank flora had significantly greater
rarefied richness than the pond flora (Fig. 4).
The total overbank flora also had a significantly
greater D
mg
value than the total pond flora
(overbank flora: 4.157; pond flora: 2.976), indi-
cating greater species richness (Table 2). The
overbank flora was also significantly more
even (overbank J′: 0.6985 ± 0.0287; pond J′:
0.3327 ± 0.0200) and less dominated by the
TABLE 2. Comparison of total number of morphotypes, number of specimens, Margalef’s diversity index (Margalef 1958),
Pielou’s evenness index (Pielou 1969), and the Berger-Parker dominance index (Berger and Parker 1970) for each floral
facies and total flora. The pond flora had fewer morphotypes, lower diversity, and evenness compared with the overbank
flora. Additionally, the pond flora was more strongly dominated by the most common morphotype. Error indicates 95%
confidence interval for estimates.
Facies No. of
morphotypes No. of
specimens Margalef’s diversity
index (D
mg
)Pielou’s evenness
(J′)Berger-Parker dominance
index (d)
Total 53 2941 5.386 0.4488 ± 0.0164 0.6282 ± 0.0176
Pond 27 2275 2.976 0.3327 ± 0.0200 0.7755 ± 0.0174
Overbank 31 666 4.157 0.6985 ± 0.0287 0.1858 ± 0.0219
FIGURE 3. Rarefaction curves using all identified vegetative plant organs from the six census localities in the early Paleo-
cene Ojo Alamo Sandstone; envelopes indicate 95% confidence intervals. Dark gray curves indicate pond localities and
light gray curves indicate overbank localities. Overbank localities have greater rarefied richness at both the site and
study area scales, indicating facies effect on floral species richness.
SAN JUAN BASIN MEGAFLORA 621
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
most common morphotype than the pond flora
(overbank d: 0.1858 ± 0.0219; pond d: 0.7755 ±
0.0174) (Table 2). Taken together, these results
demonstrate that the overbank floras and
pond floras are taxonomically distinct, and
the overbank floras are more diverse and spe-
cies rich than the pond floras.
San Juan Basin Paleoclimate and Paleoecology
The estimated MAT using DiLP with all
measurable dicot angiosperm morphotypes
from the Ojo Alamo Sandstone was 27.4 ± 4.0°C
(n= 30 morphotypes), and using LMA it was
23.5 ± 2.5°C (n= 33 morphotypes) (Table 4,
Supplementary Table 5). The estimated MAP
using DiLP for the Ojo Alamo Sandstone was
154.4 cm/yr (+126.9/−69.6cm/yr, n= 30 mor-
photypes), and using LAA it was 192.8 cm/yr
(+83.3/−58.2cm/yr, n= 33 morphotypes)
(Table 4). DiLP is likely underestimating MAP,
because large fossil leaves within our collection
were often fragmentary and impossible to
include in DiLP analysis (see further discussion
of this general point in Peppe et al. [2018]).
When leaf area estimate from LAA was used
in the DiLP model, the MAP estimate was
201.0 cm/yr (+165.3/−90.7 cm/yr), which likely
represents a more reasonable approximation of
MAP.
To assess the effect of facies on paleoclimate
estimates, the DiLP MAT and MAP were calcu-
lated separately for the overbank and pond
floras. The two different data sets used to calcu-
late these estimates had very little overlap in
their morphotype composition (Supplemen-
tary Table 6), yet their paleoclimate estimates
were similar and overlap within uncertainty.
The MAT estimated using DiLP was 23.6 ±
4.0°C for the pond flora and 27.3 ± 4.0°C for
the overbank flora (Table 4, Supplementary
Table 6). The MAP estimated using DiLP was
156.3 cm/yr (+128.5/−70.5 cm/yr) for the
pond flora and 174.3 cm/yr (+143.3/
−78.6 cm/yr) for the overbank flora (Table 4,
Supplementary Table 5). These similar esti-
mates for MAT and MAP for the overbank
and pond facies indicate that despite differ-
ences in the taxonomic composition, there is
no obvious facies effect with respect to paleo-
climate estimates.
The M
a
estimate for the total Ojo Alamo
Sandstone was 78.1 g/m
2
(+113.9/−55.2 g/
m
2
,n= 11 morphotypes; Table 4), and only
one morphotype had an M
a
greater than
129 g/m
2
. These results demonstrate that the
majority of Ojo Alamo morphotypes had a
FIGURE 4. A, Detrended correspondence analysis scatter
plot of all San Juan Basin (SJB) census localities and mor-
photypes (axis 1 eigenvalue = 0.7338; axis 2 eigenvalue =
0.1302). Localities clustered by their floral facies with the
pond floras located on the left side of the plot and the over-
bank facies in the upper right. There was a closer relation-
ship between morphotypes found in individual overbank
facies localities compared with pond facies localities. B,
Dendrogram of six SJB census localities showing two
major clusters of localities (Cohen correlation = 0.9702).
Dendrogram generated using the Raup-Crick index (Raup
and Crick 1979). Localities clustered together based on
their depositional facies. Percentages at each branching
point indicate percentage of 10,000 iterations in which
that branching point appeared. In 100% of runs, the two
overbank and four pond localities formed separate clusters.
ANDREW G. FLYNN AND DANIEL J. PEPPE622
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
leaf life span <12 months and were deciduous
(Wright et al. 2004; Royer et al. 2007). This dis-
tribution of M
a
is most similar to the distribu-
tion found in riparian habitats in temperate
rain forests and tropical seasonal forests (Fig. 5).
Regional Floral Diversity and Paleoclimate
All species richness indices indicate that
the SJB had a higher species richness than con-
temporaneous floras in the NGP. The SJB flora
had greater species richness (n=55
morphotypes) than contemporaneous floras
from the DB (n= 49 morphotypes [Barclay
et al. 2003]) and the WB (n= 33 morphotypes
[Wilf and Johnson 2004]; n= 24 morphotypes
[Peppe 2010]) (Supplementary Table 7). The
mean site rarefied richness at 300 specimens
from the SJB (12.99 ± 0.86) was greater than
what has been reported for the DB (11.50 ±
0.58 [Barclay et al. 2003]) and the WB (7.64 ±
0.73 [Wilf and Johnson 2004]; 11.06 ± 0.59
[Peppe 2010]), indicating greater α-diversity in
the SJB than in floras from the NGP (Fig. 6A,
Supplementary Table 7). The basin-wide rar-
efied richness downsampled to 1000 specimens
was also greater in the SJB (34.52 ± 2.17) than
what has been reported for the DB (22.53 ±
1.30 [Barclay et al. 2003]) and the WB (20.52 ±
1.47 [Wilf and Johnson 2004]; 18.95 ± 0.20
[Peppe 2010]) (Fig. 6C, Supplementary Table 7).
The difference in both site and basin-wide rar-
efied richness is smaller, but still distinct,
when using only dicot angiosperm leaves
(Fig. 6B–D), because the SJB flora had a higher
proportion of non-angiosperm morphotypes,
in particular pteridophytes, when compared
with the DB and WB floras. The D
mg
for the
SJB (5.386) was greater than what has been
reported for the DB (3.268 [Barclay et al.
2003]) and the WB (3.205 [Wilf and Johnson
2004]; 2.596 [Peppe 2010]), indicating greater
species richness, similar to the pattern observed
from the rarefaction analysis (Fig. 7A, Supple-
mentary Table 7). The differences between the
SJB and the DB and WB were significant
(p= 0.0027 to <0.0001), but there was no
significant difference between the DB and
WB ( p= 0.9509 to 0.2217) (Supplementary
Table 8A).
Diversity indices indicate that while the SJB
floras had higher species richness than contem-
poraneous floras, it was less even and more
dominated by the most common morphotype
than floras from the DB and the WB. The J′
value for the SJB (0.4488 ± 0.0164) was lower
than what has been reported for the DB
(0.6583 ± 0.0139 [Barclay et al. 2003]) and the
WB (0.6425 ± 0.0089 [Wilf and Johnson 2004];
0.6746 ± 0.0205 [Peppe 2010]), indicating the
SJB flora is less even (Fig. 7B, Supplementary
TABLE 3. Bray-Curtis index of similarity: β-diversity proxy. Mean distance = 0.278. Bray-Curtis distance within facies
(0.588) is greater than between facies (0.154) and is statistically significant (t= 4.312, df = 13, p= 0.0008). Average
Bray-Curtis distance between pond localities (0.571) is greater than between overbank localities (0.396), indicating pond
facies had more similar floral composition.
DP1301/01B AF1402 AF1404 AF1407 AF1407B AF1409D
DP1301/01B —
AF1402 0.788 —
AF1404 0.090 0.069 —
AF1407 0.883 0.721 0.081 —
AF1407B 0.350 0.366 0.292 0.316 —
AF1409D 0.158 0.126 0.396 0.133 0.279 —
TABLE 4. San Juan Basin paleoclimate estimates using digital leaf physiognomy (DiLP) (Peppe et al. 2011), leaf margin
analysis (LMA) (Miller et al. 2006), leaf area analysis (LAA) (Wilf et al. 1998), and M
a
(Royer et al. 2007) for the total flora
and each floral facies. Error indicates 95% confidence intervals for each estimate. MAT, mean annual temperature; MAP,
mean annual precipitation.
Site/facies DiLP MAT (°C) LMA MAT (°C) DiLP MAP (cm/yr) LAA MAP (cm/yr) M
a
(g/m
2
)
Total Ojo Alamo 27.4 ± 4.0 24.3 ± 2.3 154.4 + 126.9/−69.6 192.8 + 83.3/−58.2 78.1 + 113.9/−55.2
Pond flora 23.6 ± 4.0 24.5 ± 2.8 156.3 + 128.5/−70.5 142.3 + 61.5/−42.9 74.5 + 107.5/−51.6
Overbank flora 27.3 ± 4.0 22.4 ± 3.0 174.3 + 143.3/−78.6 243.0 + 104.9/−73.3 79.3 + 113.4/−55.5
SAN JUAN BASIN MEGAFLORA 623
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
Table 7). This difference was significant ( p<
0.0001) when the J′permutations from the SJB
were compared to those of the DB and WB,
but there was no significant difference between
the DB and WB ( p= 0.8309 to 0.9974) (Supple-
mentary Table 8B). Finally, the dvalue for the
SJB (0.6282 ± 0.0176) was greater than what
has been reported for the DB (0.2332 ± 0.0184
[Barclay 2003]) and WB (0.3167 ± 0.0136 [Wilf
and Johnson 2004]; 0.2917 ± 0.0273 [Peppe
2010]), indicating the SJB flora was more
strongly dominated by the most common mor-
photype (Fig. 7C, Supplementary Table 7). The
difference in dvalues between the SJB and the DB
andWBwassignificant when the permutations
between the SJB and the other basins were calcu-
lated ( p≤0.0001). Additionally, the dvalue of
the DB flora was also significantly different
than for both WB floras ( p≤0.0001), but there
was no significant difference between both WB
floras ( p= 0.1031) (Supplementary Table 8C).
Paleoclimate estimatestherefore indicate that
then SJB was warmer and wetter than the DB
and WB. MAT estimates using LMA for the
SJB are >6°C warmer than estimates for the
DB and >14°C warmer than those for the WB
(Table 5,Fig. 8). The estimated MAP using
LAA from the SJB is >35 cm/yr wetter than
MAP estimates for the DB and WB, though
these differences are within the uncertainty of
the estimates (Table 5,Fig. 8).
In addition to these LMA and LAA paleo-
climate estimates, there are estimates for MAT
and MAP from the WB using DiLP (Peppe et al.
2011) that can be compared with the ones for
the SJB flora. There are no DiLP MAT or MAP
estimates for the DB. MAT using DiLP is 27.4 ±
4.0°C for the SJB and 15.7 ± 4.0°C for the WB
(Peppe et al. 2011), considerably warmer than
LMA estimates for the same floras (3.1°C and
5.5°C warmer for the SJB and the WB, respect-
ively). The large difference in MAT between
DiLP and LMA is probably related to the fresh-
water margin effect, wherein toothed leaves are
more common in environments with seasonal
water availability because leaf teeth allow for
more rapid leaf growth (Peppe et al. 2011),
which can cause LMA to underestimate MAT.
Thus, we interpret the DiLP MAT estimates to
be a more accurate reconstruction of climate.
However, the magnitude of difference in MAT
between the SJB and WB is similar using LMA
and DiLP (14.1°C and 11.7°C, respectively),
suggesting that the magnitude of differences in
LMA MAT estimates between the basins are
reasonable. The DiLP estimated MAP for the
FIGURE 5. Comparison of Ojo Alamo Sandstone floral leaf mass per area distribution with representative modern sites with
different biomes/environments (modern biome data from Peppe et al. 2011). The leaf mass per area distribution of the SJB
flora is most similar to riparian tropical seasonal forests, in agreement with the paleoclimate estimates, which indicate a
tropical seasonal forest.
ANDREW G. FLYNN AND DANIEL J. PEPPE624
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
FIGURE 6. Regional rarefaction comparison between the San Juan Basin (SJB) (this study), the Denver Basin (DB) (Barclay
et al. 2003), and the Williston Basin (WB) (Wilf and Johnson 2004; Peppe 2010) using plant vegetative organs with envelopes
indicating 95% confidence intervals. A, All census localities using all groups; B, all census localities using dicot angios-
perms only; C, basin-wide using all groups; and D, basin-wide using only dicot angiosperms. At both the site and basin-
wide levels, the SJB flora has greater rarefied richness than the DB and WB floras, but the difference is smaller when only
dicot angiosperms are included, due to the high diversity and abundance of non-dicot angiosperm taxa in the SJB flora.
SAN JUAN BASIN MEGAFLORA 625
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
SJB (154.4 cm/yr + 126.9/−69.6 cm/yr) is lower
than estimates for the WB (175.0 cm/yr +
144.0/−79.0 cm/yr [Peppe et al. 2011]). How-
ever, as discussed earlier the DiLP MAP esti-
mates are probably too low, and when the
average leaf size from LAA is used in DiLP, the
MAP estimates for the SJB are >25 cm wetter
than those for the WB. The MAP estimates
using LAA and DiLP for the SJB and WB overlap
within uncertainty, suggesting that both are
reasonable approximations of MAP.
Discussion
San Juan Basin Flora
San Juan Basin Floral Composition.—The early
Paleocene SJB flora is dominated by dicot
angiosperms, which account for 66.04% of mor-
photypes and 86.19% of specimens collected
during censuses (Table 1), similar to observed
patterns in early Paleocene floras from across
North America (e.g., Brown 1962; Barclay
et al. 2003; Dunn 2003; Wilf and Johnson
2004; Peppe 2010). The majority of dicot
angiosperm leaves were interpreted to be
woody plants (i.e., trees or shrubs; 94.24%) con-
sistent with modern taphonomic studies (Ellis
and Johnson 2013). Interestingly, pteridophytes
are relatively common (9.09% of morphotypes
and 4.86% of specimens; Table 1) and conifers
are relatively rare (3.64% of morphotypes and
3.26% of specimens; Table 1), which is a differ-
ent pattern than observed in contemporaneous
floras from the NGP (i.e., Barclay et al. 2003;
Wilf and Johnson 2004; Peppe 2010;Table 1).
Landscape Floral Heterogeneity.—There is sig-
nificant facies effect on the Ojo Alamo floral
composition and diversity (Figs. 3 and 4,Tables
2and 3). Our interpretation for the depositional
facies (pond vs. overbank) is based on the sedi-
mentological features of the sites (Supplemen-
tary Fig. 1) and is qualitatively supported by
the floral composition of each facies. For
example, the pond flora is more strongly domi-
nated by non-angiosperm morphotypes than
the overbank facies. The modern nearest living
relatives of the majority of these non-
angiosperm morphotypes, such as the ferns
FIGURE 7. Diversity comparison between the San Juan Basin (SJB) (this study), Denver Basin (DB) (Barclay et al. 2003), and
Williston Basin (WB) (Wilf and Johnson 2004; Peppe 2010). A, Margalef’s index (D
mg
), B, Pielou’s evenness (J′), and C,
Berger-Parker index (d). The SJB D
mg
value for the SJB flora was significantly higher than for both the DB and WB floras,
indicating greater species richness. The J′value was significantly lower and the dvalue was significantly greater for the SJB
compared with the DB and WB, indicating the SJB flora was significantly less even and more heavily dominated by the
most common taxa. Bars indicate 95% confidence intervals. For absolute values and pairwise comparisons, see Supplemen-
tary Tables 7 and 8.
ANDREW G. FLYNN AND DANIEL J. PEPPE626
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
Onoclea sensibilis (SJ-56) and the cupressaceous
conifer Cupressinocladus interruptus (SJ-68), are
commonly found in water-saturated environ-
ments such as ponds and/or swamps (e.g.,
Mehltreter et al. 2010; Pittermann et al. 2012).
In comparison, the overbank facies is more
strongly dominated by dicot angiosperms and
the modern relatives of common overbank
morphotypes such as Platanites marginata
(SJ-71) and Browneia serrata (SJ-78) that com-
monly inhabit disturbed lake and streamside
environments (Stromberg 2001; Manchester
and Hickey 2007), as would be expected in
downstream and lateral bar forms in a braided
stream.
The strong facies effect on diversity could be
explained by taphonomic differences between
the facies or lateral heterogeneity in the floral
communities between the environments. If
taphonomy caused the observed facies effect,
we would expect the overbank facies to have
TABLE 5. Comparison of San Juan Basin (this study),
Denver Basin (Barclay et al. 2003), and Williston Basin
(Peppe 2010) mean annual temperature (MAT) estimates
using leaf margin analysis (Miller et al. 2006) and mean
annual precipitation (MAP) using leaf area analysis (Wilf
et al. 1998). Error indicates 95% confidence interval.
Basin MAT (°C) MAP (cm/yr)
San Juan 24.3 ± 2.3 192.8 + 83.3/−58.2
Denver 17.6 ± 3.0 155.0 + 66.9/−46.8
Williston 10.2 ± 4.0 154.4 + 66.7/−46.6
FIGURE 8. Modern ecosystem plots with paleoclimate variables. The San Juan Basin (SJB) flora corresponded with modern
tropical seasonal forests using both digital leaf physiognomy (DiLP) and leaf margin analysis (LMA) + leaf area analysis
(LAA). Both the Denver Basin (DB) and Williston Basin (WB) corresponded with warm and cool temperate forests, respect-
ively, indicating the SJB represents a different biome than previouslystudied localities. The 95% confidence intervals for all
paleoclimate estimates are shown by gray bars.
SAN JUAN BASIN MEGAFLORA 627
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
higher species richness than the pond facies
due to a larger catchment area “collecting”mul-
tiple depositional environments (Ellis and
Johnson 2013). Additionally, we would expect
that because the overbank facies incorporated
morphotypes from other facies, paleoclimate
estimates between facies might differ due to
the incorporation of allochthonous morpho-
types. However, while the overbank facies did
have higher species richness than the pond
facies (Table 2), the two facies shared relatively
few morphotypes. Only 12 morphotypes
(22.64%) are found in both the pond and over-
bank facies, and the majority of the shared mor-
photypes, such as Platanites raynoldsii (SJ-36)
and Averrhoites affinis (SJ-46), occur at the
majority of SJB sites and also have widespread
distributions throughout the Western Interior
of North America during the early Paleocene
(e.g., Johnson 2002; Barclay et al. 2003; Dunn
2003; Peppe 2010). Conversely, most facies-
restricted morphotypes have only been identi-
fied in the SJB. Further, while the pond floras
are relatively homogeneous, there is a large
difference in species composition between the
overbank sites, indicating a considerable vari-
ability in floral composition within the higher-
disturbance overbank facies (Figs. 3A,Band 4).
Within groups, the overbank localities were
more dissimilar to each other than the pond
localities (Table 3), which suggests consider-
able variability in the floral composition
between overbank facies localities and between
the pond and overbank facies. Conversely, two
pond localities that were collected along a tran-
sect from the same carbonaceous shale deposits
and are separated by the shortest geographical
distance (AF1407 and AF1407B; Fig. 1C)havea
relatively low Bray-Curtis similarity (0.316;
Table 3), indicating the occurrence of spatial
heterogeneity within the plant community
even within the facies that is more likely to
preserve relatively autochthonous flora. Anec-
dotally, the overbank deposits preserve many
complete compound leaves of P. raynoldsii
(SJ-36) and P. marginata (SJ-71), as well as
many relatively complete branches of C. inter-
ruptus (SJ-68), which are delicate and are
unlikely to have been preserved over long
transport distances, suggesting the overbank
deposits primarily sampled an autochthonous
or parautochthonous flora. Thus, there is no
evidence to suggest that the overbank facies
sampled taxa from the pond facies. Finally,
DiLP paleoclimate estimates for the two facies
were nearly identical, which suggests both
facies were sampling relatively autochthonous
or parautochthonous floras (Table 4). Based
on the differences in floral composition
between the pond and overbank facies, differ-
ences between floras collected from the over-
bank facies, and the similar paleoclimate
estimates between facies, it is unlikely that the
facies effect that we document is the result of
differences in taphonomy between the over-
bank and pond facies.
Instead, we interpret the facies effect to be the
result of landscape heterogeneity in the Ojo
Alamo Sandstone plant communities. This
strong facies effect, in which the overbank
floras have higher species richness and diver-
sity than the pond floras, is best explained by
the intermediate disturbance hypothesis,
which predicts the highest species diversity
occurs in areas of intermediate disturbance
(Connell 1978). In tropical forest communities,
disturbance creates holes in the canopy that
allow a succession of plant species to colonize
the newly available growing area, leading to
greater species richness (e.g., Brokaw and
Busing 2000; Molino 2001). The overbank
flora was deposited in more active parts of the
depositional system, and thus underwent
more disturbance, which in turn likely resulted
in greater floral diversity. The overbank facies
had clear evidence of repeated depositional
events, likely from flooding and high water
(Supplementary Fig. 1), which would have
regularly disturbed the landscape and likely
killed some of the standing vegetation. The
modern relatives of common SJB overbank
flora species (e.g., P. raynoldsii, SJ-36) are
commonly pioneer species (Stromberg 2001),
supporting our hypothesis that episodic dis-
turbance in these localities created canopy
breaks that allowed colonization by pioneer
species. In comparison, the lower-diversity
pond flora was deposited in a restricted and
less active area of the depositional system that
encountered relatively little disturbance,
which may have allowed a few species (i.e.,
A. affinis, SJ-46) to become dominant. The effect
ANDREW G. FLYNN AND DANIEL J. PEPPE628
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
of disturbance on diversity is also highlighted
by the pond flora being significantly less even
and more dominated by a single morphotype
when compared with the overbank flora
(Table 2). These results indicate considerable
landscape heterogeneity in the SJB flora and
suggest that if a larger spatial area was col-
lected, the species richness and diversity
would likely increase.
Paleoclimate and Paleoecology.—The Ojo
Alamo Sandstone has been interpreted to
represent a braided stream system on an allu-
vial plain subject to seasonal precipitation
(Powell 1973; Fassett 1985; Sikkink 1987). Our
MAT and MAP estimates indicate that the cli-
mate of the SJB flora is most similar to a tropical
seasonal forest biome (Fig. 8), which is typified
by environments with a pronounced dry sea-
son (Whittaker 1975), supporting the interpret-
ation of seasonal precipitation. The M
a
estimates for the Ojo Alamo Sandstone flora
also suggest seasonal precipitation. Deciduous
taxa are commonly found in disturbed environ-
ments and in environments with marked differ-
ences between seasons in rainfall and/or
temperature (e.g., Whittaker 1975; Reich et al.
1992; Lavorel et al. 1997; Yoshifuji et al. 2006;
Poorter et al. 2009). Further, in tropical forest
ecosystems, the degree of deciduousness in
plant communities is controlled by the amount
of precipitation and the length and magnitude
of the dry season (Condit et al. 2000; Pyke
et al. 2001), and modern tropical forests
dominated by deciduous species are marked
by a long dry season with little to no precipita-
tion and a short but intense wet season (Bullock
and Solis-Magallanes 1990). The distributions
of M
a
for the Ojo Alamo flora indicate that
it sampled a riparian temperate rain forest/
tropical seasonal forest (Figs. 5 and 8), which
also suggests that precipitation during the
deposition of the Ojo Alamo Sandstone was
seasonal with a pronounced dry season. This
interpretation is further supported by fossil
wood anatomy from the early Paleocene SJB
that exhibited variations in vessel density and
diameter that may correspond with seasonal
variations in precipitation (Wheeler et al.
1995). Taken together, the sedimentological
and paleontological evidence from the Ojo
Alamo indicates that it was deposited in a
tropical seasonal forest biome with a consider-
able dry season.
Regional Patterns of Vegetation and
Paleoclimate
Regional Patterns in Floral Composition.—Pre-
vious research on early Paleocene floras from
North America has suggested that they are rela-
tively similar across the continent (e.g., Brown
1962; Johnson 2002; Johnson et al. 2003; Barclay
et al. 2003; Peppe 2010) and are dominated by
diagnostic “FU1 taxa”(sensu Johnson and
Hickey 1990; Johnson 2002). In the SJB, the
FU1 taxa P. raynoldsii,B. serrata (SJ-78), “Popu-
lus”nebrascensis (SJ-13), and Paranymphaea
crassfolia (SJ-51) are present. However, the
majority of FU1 taxa, such as the dicot angios-
perms Quereuxia angulata,Mciveraephyllum
nebrascense, and Zizophoides flabella and the tax-
odiaceous conifers Glyptostrobus europaeus and
Metasequoia occidentalis, are absent. Further,
with the exception of P. raynoldsii (SJ-36), the
FU1 taxa present in the SJB are very rare,
while accessory taxa from the NGP, such as
A. affinis (SJ-46) and Ditaxocladus catenulatus
(SJ-102), are common in the SJB. Further, Gingko
adiantoides, which is also common in the NGP
(Johnson 2002; Peppe 2010) is notably absent
from the SJB flora. The SJB was likely too
warm for G. adiantoides and the taxodiaceous
conifers M. occidentalis and G. europaeus,as
their nearest living relatives are primarily
restricted to temperate climates in latitudes
above ∼35°N (e.g., Tralau 1967; Ziegler et al.
1996; Liu et al. 2007).
Though markedly different, the floral com-
position and relative abundance of morpho-
types between the SJB and contemporaneous
localities in the NGP have some similarities.
For example, the most common SJB morpho-
type (A. affinis, SJ-46) is relatively uncommon
or absent in floras from North Dakota and
Montana (Johnson 2002; Peppe 2010), but is a
common morphotype in the West Bijou Site
from the DB (Barclay et al. 2003). Additionally,
the most common morphotype found by John-
son (2002) in the WB flora was P. raynoldsii
(SJ-36), which is a common morphotype from
the SJB, but relatively rare in the DB (Barclay
et al. 2003). Conversely, the most abundant
morphotype in the DB (M. nebrascense; Barclay
SAN JUAN BASIN MEGAFLORA 629
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
et al. 2003) is also common in the WB (Peppe
2010), but absent in the SJB. This suggests
some plant species had extensive ranges during
the early Paleocene, while others, such as the
most common constituents of the FU1 flora
from the NGP, were restricted to more northern
latitudes.
The greater diversity and abundance of pter-
idophyte morphotypes in the SJB were likely
linked to the warmer and wetter climate com-
pared to contemporaneous localities and the
transition from temperate to tropical forests
(Figs. 6B–Dand 8,Table 5) (Barclay et al.
2003; Wilf and Johnson 2004; Peppe 2010).
This suggests a latitudinal difference in pterido-
phyte diversity and abundance across North
America in the early Paleocene. The greater
abundance and diversity of pteridophytes
may have been driven by higher rate of speci-
ation in tropical ecosystems compared with
temperate forests, a pattern that has been well
documented in modern fern communities
(Haufler et al. 2000). Additionally, the climate
of the SJB was likely better suited to pterido-
phytes than the climate of the more northern
basins (Mehltreter et al. 2010;Table 5). There
is evidence that the SJB, in addition to being
warmer and wetter than other basins, experi-
enced seasonal precipitation (e.g., Wheeler
et al. 1995; Davis et al. 2016). The modern rela-
tives of the most common pteridophyte mor-
photype found in the SJB (SJ-57, cf. Anemia)
are drought tolerant and are often found in
seasonally dry and warm habitats (Tryon and
Tryon 1982) similar to those predicted for the
SJB based on the paleoclimate estimates. Add-
itionally, fossil relatives of SJ-57 (cf. Anemia)
from the Lower Cretaceous Crato Formation
in Brazil have been interpreted to be ground
cover in dry, sunny areas with adaptations to
survive drought stress (Mohr et al. 2015).
Taken together, the lack of many diagnostic
FU1 flora species and the relatively high
abundance of pteridophytes in the SJB flora
suggest significant latitudinal variation in floral
composition across western North America
immediately following the K/Pg boundary.
Palynological work from the latest Cret-
aceous and earliest Paleocene (e.g., Frederiksen
1987; Nichols et al. 1990) also supports this lati-
tudinal compositional gradient found in the
megaflora. Frederiksen (1987) found different
palynological floral provinces during the latest
Cretaceous along the continental margin and
the Western Interior. Interestingly, he found
that the SJB palynological assemblage shared
a higher percentage of palynomorphs with
the continental margin than with other local-
ities in the continental interior, suggesting
that the SJB was composed of a different floral
assemblage than other Western Interior basins.
Using palynological collections from the earli-
est Paleocene, Nichols et al. (1990) found three
pollen-based floral clusters that roughly corres-
pond to the southern, mid-, and high latitudes
of North America. The southern group, includ-
ing the SJB and DB, had greater species richness
and greater endemism than the mid- and nor-
thern groups, which include the WB (Nichols
et al. 1990). A “southern”floristic province is
also supported by fossil wood collections
from Mexico, Texas, and New Mexico, which
suggest that the vegetation across these areas
was similar from the Campanian through the
Paleocene (e.g., Wheeler et al. 1995; Lehman
and Wheeler 2009). When combined, our
megafloral analyses and the palynology and
fossil wood record from the Late Cretaceous
and early Paleocene all suggest latitudinal vari-
ation in floral compositions across North Amer-
ica and a marked difference in the SJB flora
compared with floras in of the NGP of North
America in the early Paleocene, and also
possibly in the Late Cretaceous.
Regional Patterns of Floral Diversity.—The
species richness of the SJB flora is 10–55%
higher than those of contemporaneous floras
from North America (Supplementary Table 7).
The SJB flora also has ∼15–20% higher rarefied
richness at the site level and ∼35–40% higher
rarefied richness at the basin-wide level
(Table 5). The increased relative difference in
rarefied richness at the basin-wide level is
driven by the predominance of non-dicot
angiosperm taxa and the large amount of
lateral heterogeneity in the SJB (Fig. 6B–D, Sup-
plementary Table 7). However, while the SJB
flora was more species rich, it was significantly
less even and more strongly dominated by
common morphotypes than contemporaneous
NGP floras (Fig. 7B,C, Supplementary Table 7).
While the most common taxa make up between
ANDREW G. FLYNN AND DANIEL J. PEPPE630
https://www.cambridge.org/core/terms. https://doi.org/10.1017/pab.2019.24
Downloaded from https://www.cambridge.org/core. Baylor University Libraries, on 21 Nov 2019 at 20:10:53, subject to the Cambridge Core terms of use, available at
21.54 and 33.19% of samples from the DB and
WB, the most common taxa make up 62.67%
of samples collected in the SJB (Table 5). The
high richness but uneven flora runs counter to
the hypothesis that the earliest Paleocene floras
had low species richness and were relatively
even (e.g., Hickey 1980; Johnson and Ellis
2002; Barclay et al. 2003; Wilf and Johnson
2004). This suggests notably different patterns
in floral diversity across North America after
the K/Pg extinction event, with the floras
from the NGP being relatively even and homo-
geneous on the landscape, but low in species
richness, while the floras from the SJB, and
potentially other areas in southern North
America, were uneven but species rich and
very heterogeneous. These differences may
have been driven by climate in the earliest
Paleocene, in which the warm temperatures in
southern North America allowed more rapid
speciation following the K/Pg extinction
event. Additionally, the latitudinal gradient in
floral species richness in the early Paleocene
documented here is consistent with a relatively
steep mammalian latitudinal diversity gradient
in the early Paleocene (Marcot et al. 2016),
suggesting that they may be related.
Interestingly, early Paleocene fossil floras
from Patagonia (Iglesias et al. 2007; Comer
et al. 2015) had species richness similar to the
SJB floras (rarefied richness using only dicot
angiosperm leaves downsampled to 2000
specimens: SJB = 28.28 ± 1.18 morphotypes;
Salamanca Flora = 32.47 ± 0.67 morphotypes).
The high species richness in Patagonia has
been linked to lower rates of extinction than
in North America and dispersal of refugia
taxa following the K/Pg boundary (Donovan
et al. 2016). Similar to the SJB flora, the early
Paleocene megaflora from the Salamanca
Formation in Argentina (64.67–63.49 Ma) had
>50% more species than comparable North
American localities (Iglesias et al. 2007). This
suggests that the pattern in macrofloral diver-
sity found in the SJB could be more similar to
South American floras than to floras from the
NGP in North America, but further work will
need to be done to test this hypothesis.
Regional Patterns of Paleoclimate and Paleoecol-
ogy.—Paleoclimate estimates for the SJB, DB,
and WB indicate that climate in the SJB was
markedly warmer and somewhat wetter than
contemporaneous basins in western North
America (Table 5,Fig. 8). MAT in the SJB is
reconstructed to be >6°C warmer than in the
DB and >11°C warmer than in the WB (Table 5,
Fig. 8), and the paleolatitudes of the SJB, DB,
and WB were ∼42°N, ∼44°N, and ∼51°N,
respectively (reconstructed based on Torsvik
et al. [2008]). This difference in MAT and the
latitudinal differences between the sites implies
that continental latitudinal temperature gradi-
ents across the Western Interior of North Amer-
ica were at least as steep in the early Paleocene
as in the modern (e.g., North et al. 1981).
Further, when compared with modern biomes
(Whittaker 1975), the DB and WB climate esti-
mates indicate a temperate deciduous forest
biome, while the SJB is reconstructed as a trop-
ical seasonal forest (Fig. 8). Today, tropical sea-
sonal forests are primarily restricted to north
and south latitudes <30° (Murphy and Lugo
1986; Mooney et al. 1995; Dupuy et al. 1999).
Thus, the climate and ecosystem reconstruc-
tions for the SJB, DB, and WB indicate greater
latitudinal expansion of tropical biomes and
the transition between tropical and temperate
forests in the earliest Paleocene compared
with the present.
In modern ecosystems, there is a strong lati-
tudinal diversity gradient, and tropical forests
have higher α- and β-diversity than temperate
forests (e.g., Givnish 1999; Phillips and Miller
2002; Leigh et al. 2004; Kraft et al. 2011;
Brown