Global biogeography of Albian ammonoids: A network-based
approach
Alexis Rojas1, 2, Pedro Patarroyo3, Liang Mao4, Peter Bengtson5, and Micha Kowalewski1
1
Division of Invertebrate Paleontology, Florida Museum of Natural History, University of Florida, Gainesville, Florida 32611, USA
2
Department of Geological Sciences, University of Florida, Gainesville, Florida 32611, USA
3
Departamento de Geociencias, Universidad Nacional de Colombia, Bogot, Colombia
4
Department of Geography, University of Florida, Gainesville, Florida 32611, USA
5
Institut fr Geowissenschaften, Universitt Heidelberg, Im Neuenheimer Feld 234, 69120 Heidelberg, Germany
ABSTRACT
Large-scale biogeographic provinces of Cretaceous ammonoids,
as currently defined in the literature, were delimited using qualita-
tive assessments of taxonomic inventories. Using aggregated species
occurrences in the Paleobiology Database, we generated a geographic
network to quantify connectivity of Albian epicontinental basins and
used the flow-based Infomap algorithm to delineate bioprovinces.
Despite taxonomic, stratigraphic, and geographic limitations of the
data, the Infomap bioprovinces are largely concordant with the tra-
ditional, qualitatively derived biogeographic model, including the
Boreal-Pacific Subrealm, Arctic Subrealm, Tethyan Realm, and Aus-
tral Realm. An agglomerative hierarchical cluster analysis applied to
the same occurrence data failed to replicate the Infomap bioprovinces
or reproduce the traditional qualitative model. The observed asym-
metrical distribution of the Infomap bioprovinces is consistent with
the known hemispheric differences in paleogeographic and oceano-
graphic features of the Albian Earth. The geographic network derived
from ammonoid data is twice as dense as the one derived for Albian
benthic marine invertebrates and thus more effective in delineating
global biogeographic units. The network-based approach establishes
a reproducible quantitative framework for delineating geographic
boundaries of marine bioprovinces, tracking biogeographic changes
over evolutionary time scales, and identifying biotic and abiotic fac-
tors that influence global partitioning of marine biodiversity.
INTRODUCTION
Biogeographic studies of fossil organisms have contributed to our
current understanding of the relationships between plate tectonics and
evolution of life (Lieberman, 2005). However, the impact of those studies
extends beyond basic biogeographic questions to include, for example,
conservation biology and climate change (Perrin and Kiessling, 2012;
Garca-Molinos et al., 2015). The network theory has brought significant
advances to our understanding of biogeographic patterns in fossil organ-
isms (Brayard et al., 2007; Dera et al., 2011; Sidor et al., 2013; Vilhena
et al., 2013; Dunhill et al., 2016). One of the most relevant features of
network analysis is its ability to detect community structure, i.e., natural
partitioning of network nodes into densely connected subgroups (Newman
and Girvan, 2004). The network-based approach has been demonstrated to
be a powerful tool in modern biogeography (Vilhena and Antonelli, 2015;
Edler et al., 2017). Here we employ a network approach to examine the
biogeography of Albian (mid-Cretaceous) ammonoids using data from the
Paleobiology Database (https://paleobiodb.org), a major geoinformatics
initiative aimed at providing fossil occurrence data across all taxa retriev-
able from the geological record (Peters and McClennen, 2016).
Large-scale biogeographic regions for the Cretaceous have been
delimited by ammonoid experts using qualitative assessments of taxo-
nomic inventories (e.g., Kennedy and Cobban, 1976; Jagt-Yazykova, 2011).
GEOLOGY, July 2017; v. 45; no. 7; p. 659662 | Data Repository item 2017213 | doi:10.1130/G38944.1 | Published online 08 May 2017
GEOLOGY
| VolumeGold
2017 The Authors. 45 |Open
Number 7 | www.gsapubs.org
Access: This paper is published under the terms of the CC-BY license.
This approach in ammonoid research has deep historical roots and remains
unverified by quantitative assessments (Ifrim et al., 2015). Because of the
high dispersal potential of ammonoids, their biogeographic partitioning is
likely to reflect large-scale physical, climatic, and/or biotic environmen-
tal changes (Bengtson and Kakabadze, 1999). We focused on the Albian
because this stage is represented by the largest number of ammonoid records
in the Paleobiology Database compared to the other Cretaceous stages. We
explicitly test the qualitatively derived biogeographic model for the Albian
ammonoids as defined in the biogeographic synthesis of Lehmann et al.
(2015): the Boreal Realm (subdivided into Boreal-Atlantic, Boreal-Pacific,
and Arctic subrealms), Tethyan Realm, and Austral Realm. These bioprov-
inces have been consistently recognized in studies on the distribution of
ammonite taxa (e.g., Owen, 1973; Kennedy and Cobban, 1976; Page, 1996).
METHOD
Data
Occurrence data of Albian ammonoids were downloaded from the
Paleobiology Database on 20 January 2016. The search was restricted
to occurrences with species-level resolution, and those with uncertain
or provisional taxonomic identification (i.e., qualified by aff., cf., ex.gr.,
sensu lato, or quotation marks) were excluded. The following data fields
were downloaded: collection number, genus name, species name, geo-
logic formation, and present-day latitude and longitude. The reduced data
set contained a total of 1795 occurrences that met the filtering criteria.
Paleogeographic coordinates were determined using the PointTracker
Software (Scotese, 2010) and rotated occurrences were plotted on the
plate tectonic configuration from the PALEOMAP Project (Scotese, 2013).
The stage-level resolution used in this analysis reflects data limitations:
the number of species occurrences currently available is insufficient to
conduct a meaningful network analysis at finer stratigraphic resolution.
Network Construction and Partitioning
The network analysis implemented here is a four-step process (Fig.
DR1 in the GSA Data Repository1). First, we aggregate the rotated occur-
rence data into a geographical grid of 5 5 cells. This spatial resolution
is widely used in studies of mid-Cretaceous climate and paleogeography
(e.g., Fluteau et al., 2007). Second, aggregated data are used to generate
a bipartite network (G) between species (S) and grid cells (P) denoted G
= (V, E), where V is the node set of two disjoint subsets (S, P) (Fig. 1). E
P S is the edge set that links P and S subsets. The incidence matrix
(B) (Table DR1) representing the connections between S and P has the
elements Bij such that
1
GSA Data Repository item 2017213, Figures DR1DR6 and Tables DR1
DR5, is available online at http://www.geosociety.org/datarepository/2017/ or on
request from editing@geosociety.org.
659

species
grid cells
Figure 1. Bipartite occurrence network of Albian
ammonoids (G).
1, if taxon j belongs to grid cell i
Bij = . (1)
0, otherwise
Because no edges are possible within node subsets P and S, the pro-
portion of all possible edges that are actually present (i.e., density) in the
bipartite network G was calculated as |V| / (|P| |S|). Third, we performed
a weighted projection from the bipartite network G onto the node subset
P (Alzahrani and Horadam, 2016). The projection procedure generates a
geographic network GP = (P, EP) in which two grid cells k and l P are
linked together if they have at least one common taxon in S (Fig. 2A).
The adjacency matrix A representing the connections between grid cells
in the projected network GP has the elements Akl such that
1, if grid cells l and k have taxa in common
Akl = . (2)
0, otherwise
For a grid cell k P, let (k) denote the set of neighbors of k. Then, the
connection strength (CS) between distinct grid cells k and l in P is given by
( k ) (l )
CSkl = , k l, (3)
C ( k ) + C (l )
where C(k) and C(l) are the total number of collections recorded at grid
cells k and l. This common neighbors index, standardized using the num-
ber of collections, takes into account variations in sampling effort. We
also performed a projection from the bipartite network G onto the node
subset S and generated a network GS = (S, ES) in which two species are
linked if they occur together at least in one grid cell (Fig. DR2). Fourth,
we applied the Infomap clustering algorithm (Rosvall and Bergstrom
2008; Lambiotte and Rosvall, 2012) to partition the grid cells within the
projected network GP into bioprovinces (Edler et al., 2017). Infomap finds
the modular structure of the network with respect to flow by using random
walks. The algorithm calculates the theoretical limit of how concisely we
can describe the trajectory of a random walker on the network and selects
the partition that gives the shortest description length. The analyses were
performed using the R-package igraph 0.6 (Csardi and Nepusz, 2006).
RESULTS
The bipartite occurrence network of Albian ammonoids (G) comprises
a total of 540 nodes partitioned into 78 grid cells (P) and 462 taxa (S).
This network is fairly sparse, comprising 859 edges that represent only
2% of all possible connections (density 0.02). The number of edges per
node of the subset P is highly variable (mean degree 11.0 12.5 standard
660
90
45 A
Latitude (dd)
0
45
90
180 120 60 0 60 120 180
CS Longitude (dd)
0 0.5 1.0
90
ASr
BPSr
45 B
Latitude (dd)
TR
0 Curaao
45 AR
90
180 120 60 0 60 120 180
Longitude (dd)
Figure 2. A: Projected one-mode network, GP (dddecimal degrees).
Links are colored to indicate their connection strength (CS). B: Com-
munity structure in the projected network GP. Nodes are colored to
indicate the Infomap bioprovinces: BPSr Boreal-Pacific subrealm;
ASrArctic subrealm; TRTethyan Realm; ARAustral Realm. Black
squares are disconnected nodes.
deviation, SD) and correlated with the number of sampled formations per
grid cell in the data set (r = 0.7, n = 78, p < 0.01). Ammonoid species
(subset node S) are connected with ~2 grid cells on average (1.9 1.6 SD)
(Fig. 1). The emergent projection GP is a spatially explicit arrangement
of 78 nodes connected by 433 weighted edges. The GP projection covers
most of the Albian epicontinental basins and quantifies the strength of
the connectivity across most marine regions (Fig. 2A). This geographic
network is fairly sparse (density 0.14), and the number of edges per node
approximates a power law distribution (Fig. DR3). The maximum node
to node distance (diameter) of the network GP, as measured by number of
edges, equals 5. The numbers of mutual connections between GP network
nodes range from 1 to 43, with ~12 mutual connections on average. A
small fraction of nodes (~6%) are disconnected from all remaining nodes.
The Infomap algorithm applied to the geographic network GP parti-
tioned data into four non-overlapping bioprovinces. The number of nodes
per Infomap bioprovince ranged from 9 to 40. Disconnected nodes were
disregarded because they did not contain meaningful information about
the overall network structure. The Infomap bioprovinces match closely the
traditional qualitatively established biogeographic units: Boreal-Pacific
Subrealm, Arctic Subrealm, Tethyan Realm, and Austral Realm. The
modularity score of this division of the network GP, i.e., the fraction of
edges within the given Infomap bioprovince minus the fraction expected
if edges were distributed at random, is relatively low (Q = 0.24), but indi-
cates the presence of a significant community structure (Newman, 2006a).
To find out how stable this division of the network GP is, we compared
the results obtained by Infomap to six different partitioning procedures
using the Normalized Mutual Information similarity score (NMI) (Table
DR2). According to the NMI, the Label Propagation procedure (Raghavan
et al., 2007) produces a grouping most closely aligned with the Infomap
output. The increasingly less concordant (although still generally con-
sistent) groupings were derived by Walktrap (Pons and Latapy, 2006),
Multilevel (Blondel et al., 2008), Fastgreedy (Clauset et al., 2004), and
www.gsapubs.org | Volume 45 | Number 7 | GEOLOGY
 n Infomap LP WT ML FG LV EB Figure 3. Comparison of the
Infomap results with other AR

0.02 0.04
BPSr estimated grouping struc- ASr
tures (see text). LPLabel
ASr Propagation; WTWalktrap;

Density
TR
MLMultilevel optimization;
FGFastgreedy; LVLead- Curaao
ing Eigenvector; EBEdge
Betweenness. Disconnected
BPSr

0
TR nodes are represented as white
unfilled areas. Bioprovinces: 90 60 30 0 30 60 90
BPSrBoreal-Pacific sub- Latitude (dd)
realm; ASrArctic subrealm;
TRTethyan Realm; AR Figure 4. Latitudinal (dddecimal degrees) distribution
AR Austral Realm. Disconnected of nodes in the projected network GP grouped by Infomap
nodes (nodes with degree k = bioprovince. ARAustral Realm; TRTethyan Realm; ASr
(k=0) Arctic subrealm; BPSrBoreal-Pacific subrealm.
0) are colored in white.

Leading Eigenvector (Newman, 2006b) partitioning procedures. The Edge and endemism of the Great Artesian Basin (Wright, 1963; Henderson,
Betweenness partitioning algorithm (Girvan and Newman, 2002) produced 1990). The Tethyan Realm in our biogeographic model extends from the
numerous small-sized communities. Because two partitioning algorithms open waters of the Tethyan Ocean, the Central Atlantic, and the paleo-
can reach a similar level of performance (NMI) but produce different Caribbean to paleolatitude ~60S, and includes the northernmost Ant-
grouping structures (Orman et al., 2011), we implemented a qualitative arctic Peninsula region. This asymmetric pattern may be controlled by
comparison of the different partitions using the largest node overlap across the paleogeographic position of both South America and Africa, which
procedures. Despite some variations, the grouping structure of the network were still relatively close and largely located in the Southern Hemisphere.
GP was relatively consistent regardless of the algorithm and most nodes Such a geographic configuration appears to have provided a number of
were within the corresponding Infomap community. However, some of relatively well connected epicontinental basins suitable for ammonoids.
those algorithms failed to distinguish Boreal subrealms, whereas some In addition, southern latitudes were relatively warm and the northernmost
algorithms further subdivided the Tethyan Realm (Fig. 3). The Infomap Antarctic Peninsula region maintained relatively strong marine connec-
bioprovinces differ from those obtained by an agglomerative hierarchical tions with South America and Africa (Bice and Norris, 2002; Martin
clustering method (Unweighted Pair Group Method with Arithmetic Mean, and Hartnady, 1986). Overall, the spatial distribution of the Infomap
UPGMA) in which a few of the nodes clustered out of their geographic bioprovinces is reminiscent of the known large-scale paleogeographic
context (Fig. DR4). The network analysis was also implemented for Albian and oceanographic features of the Albian Earth: restricted Arctic Ocean,
benthic invertebrates in the Paleobiology Database (Table DR3). The con- relatively open Tethyan region, and partly isolated South Atlantic Ocean
struction procedure resulted in a geographic network with a similar size, (Sewall et al. 2007). The latitudinal distribution of the Infomap bioprov-
but less dense than that derived from ammonoids (Fig. DR5). The cluster- inces (Fig. 4) appears to support the hypothesis that latitude-related factors
ing procedure resulted in a network partition with a higher modularity for also played an important role in shaping the biogeographic partitioning of
the benthic invertebrates compared to the ammonoids (Table DR4), and the mid-Cretaceous ammonoids by reducing dispersal and constraining
failed to reproduce the ammonoid Infomap bioprovinces. bioprovinces in space (Reboulet, 2001; Ifrim et al., 2015). However, these
interpretations may partly depend on geographic and stratigraphic resolu-
DISCUSSION tions of analyzed data, and may thus be inapplicable for analyses carried
The partitioning of the geographic network GP into four groups using out at finer observational scales. Our results shed new light on the con-
the Infomap algorithm resulted in an analytical outcome that is largely troversial affinities of the Albian ammonoid faunas from the Netherlands
concordant with the traditional biogeographic model of Albian ammonoids. Antilles (see Owen and Mutterlose, 2006). For multiple network partition-
The most obvious difference between the qualitative and network-based ing procedures, the grid cell containing this tropical region clustered with
biogeographic model is that Infomap failed to delineate the Boreal-Atlantic the Arctic Subrealm (Fig. 2B). The spatial relationship among grid cells
Subrealm, either reflecting limitations of our data or inaccuracies of the of the Infomap bioprovince representing the Boreal Realm in our model
traditional biogeographic model. This Boreal-Atlantic Subrealm has been, suggests that a Boreal affinity for areas proximal to the paleoequator is
however, recognized in qualitative studies based on non-ammonoid taxa unlikely. This outcome is consistent with the Tethyan affinity suggested
(e.g., Iba et al., 2011). Our outcome may also reflect the resolution limits for this region by Owen and Mutterlose (2006). Centrality measurements
of the Infomap algorithm implemented in igraph, which uses standard on the projected network GS (Fig. DR2; Table DR5) allows us to identify
teleportation; a procedure used in random walk-based methods that allows important or highly connected species.
walkers to randomly teleport across the network on any node (Lambiotte Despite having a similar size, the geographic network derived from
and Rosvall, 2012). Because the primary goal of this study was the global- ammonoid data is twice as dense as the network derived for Albian ben-
scale test of major biogeographic bioprovinces, this limitation is not critical. thic marine invertebrates. The overall comparison indicates that benthic
The results presented here offer tentative evidence that the Boreal Realm invertebrates do not replicate the biogeographic patterns observed in the
was partitioned into the Arctic and Boreal-Pacific subrealms. The hierarchi- ammonoids, and thus confirms the premise that ammonoid biogeographic
cal cluster UPGMA failed to replicate the Infomap bioprovinces regardless partitioning is more likely to reflect large-scale environmental changes
of the level of similarity at which grid cells are joined (Fig. DR4), point- (Bengtson and Kakabadze, 1999). The approach utilized in this study
ing to the advantage of Infomap for detection of biogeographic structure. establishes an alternative, objective standard framework for studying
The location of the Austral-Tethyan boundary in our biogeographic the spatiotemporal dynamics of the marine paleobioprovinces over evo-
model differs substantially from that suggested by Lehmann et al. (2015). lutionary time scales (e.g., origination, extinctions, expansions, contrac-
In our model, the Austral Realm is restricted to the Australian epiconti- tions, and migrations). It is also a powerful methodological framework for
nental sea. However, a node representing the Northern Territory of Aus- comparative biogeographic studies within and across taxa. This approach
tralia clustered with the Tethyan Realm (Fig. 2B). This configuration is complements qualitative studies by providing a spatially unambiguous
driven by the occurrence of widespread forms in Australian borderlands and reproducible strategy for quantitative delineation of bioprovinces.

GEOLOGY | Volume 45 | Number 7 | www.gsapubs.org 661

ACKNOWLEDGMENTS
We thank the contributors to the Paleobiology Database (https://paleobiodb.org)
who collected Albian ammonoid data, especially Austin Hendy, Matthew Clapham,
and Loc Villier. We thank Martin Rosvall (Ume University, Sweden) for his
comments on the Infomap, Christopher Scotese (http://www.scotese.com) for
providing the PALEOMAP and PointTracker Software, and James E.T. Channell
(University of Florida, Gainesville) for his comments on an earlier version of the
manuscript. We thank Wolfgang Kiessling, Dieter Korn, an anonymous reviewer,
and editor Judith Totman Parrish for helpful comments and constructive criticism
that greatly improved this manuscript. Rojas is grateful to Elizabeth Sandoval for
her continuous support. Rojas was supported by a Jon L. and Beverly A. Thompson
Endowment Fund. Publication of this article was funded in part by the University
of Florida Open Access Publishing Fund.
REFERENCES CITED
Alzahrani, T., and Horadam, K.J., 2016, Community detection in bipartite networks:
Algorithms and case studies, in L, J., et al., eds., Complex systems and net-
works: Heidelberg, Springer, p. 2550, doi:10.1007/978-3-662-47824-0_2.
Bengtson, P., and Kakabadze, M.V., 1999, Biogeography of Cretaceous ammo-
nites: A review of procedures and problems: Neues Jahrbuch fr Geologie
und Palaontologie Abhandlungen, v.212, p.221239.
Blondel, V.D., Guillaume, J.-L., Lambiotte, R., and Lefebvre, E., 2008, Fast unfold-
ing of communities in large networks: Journal of Statistical Mechanics: Theory
and Experiment, v. 2008, P10008, doi:10.1088/1742-5468/2008/10/P10008.
Brayard, A., Escarguel, G., and Bucher, H., 2007, The biogeography of Early
Triassic ammonoid faunas: Clusters, gradients, and networks: Geobios, v.40,
p.749765, doi:10.1016/j.geobios.2007.06.002.
Bice, K.L., and Norris, R.D., 2002, Possible atmospheric CO2 extremes of the
middle Cretaceous (late AlbianTuronian): Paleoceanography, v. 17, p. 22-
122-17, doi:10.1029/2002PA000778.
Clauset, A., Newman, M.E.J., and Moore, C., 2004, Finding community structure
in very large networks: Physical Review E: Statistical, Nonlinear, and Soft
Matter Physics, v.70, 066111, doi:10.1103/PhysRevE.70.066111.
Csardi, G., and Nepusz, T., 2006, The igraph software package for complex net-
work research: International Journal of Complex Systems, v.1695, p.19.
Dera, G., Neige, P., Dommergues, J.-L., and Brayard, A., 2011, Ammonite pa-
leobiogeography during the PliensbachianToarcian crisis (Early Jurassic)
reflecting paleoclimate, eustasy, and extinctions: Global and Planetary Change,
v.78, p.92105, doi:10.1016/j.gloplacha.2011.05.009.
Dunhill, A.M., Bestwick, J., Narey, H., and Sciberras, J., 2016, Dinosaur biogeo-
graphical structure and Mesozoic continental fragmentation: A network-based
approach: Journal of Biogeography, v.43, p.16911704, doi:10.1111/jbi.12766.
Edler, D., Guedes, T., Zizka, A., Rosvall, M., and Antonelli, A., 2017, Infomap
bioregions: Interactive mapping of biogeographical regions from species dis-
tributions: Systematic Biology, v. 66, p. 197204, doi:10.1093/sysbio/syw087.
Fluteau, F., Ramstein, G., Besse, J., Guiraud, R., and Masse, J., 2007, Impacts
of palaeogeography and sea level changes on mid-Cretaceous climate: Pa-
laeogeography, Palaeoclimatology, Palaeoecology, v.247, p.357381, doi:
10.1016/j.palaeo.2006.11.016.
Garca Molinos, J., Halpern, B.S., Schoeman, D.S., Brown, C.J., Kiessling, W.,
Moore, P.J., Pandolfi, J.M., Poloczanska, E.S., Richardson, A.J., and Burrows,
M.T., 2015, Climate velocity and the future global redistribution of marine bio-
diversity: Nature Climate Change, v.6, p.8388, doi:10.1038/nclimate2769.
Girvan, M., and Newman, M.E.J., 2002, Community structure in social and bio-
logical networks: Proceedings of the National Academy of Sciences, v.99,
p.78217826, doi:10.1073/pnas.122653799.
Henderson, R.A., 1990, Late Albian ammonites from the Northern Territory, Aus-
tralia: Alcheringa: An Australasian Journal of Palaeontology, v.14, p.109148,
doi:10.1080/03115519008527815.
Iba, Y., Mutterlose, J., Tanabe, K., Sano, S., Misaki, A., and Terabe, K., 2011, Belem-
nite extinction and the origin of modern cephalopods 35 m.y. prior to the Cre-
taceous-Paleogene event: Geology, v.39, p.483486, doi:10.1130/G31724.1.
Ifrim, C., Lehmann, J., and Ward, P., 2015, Paleobiogeography of Late Cretaceous
ammonoids, in Klug, C., et al., eds., Ammonoid paleobiology: From macro-
evolution to paleogeography: Dordrecht, Springer, p. 259274, doi:10.1007
/978-94-017-9633-0_10.
Jagt-Yazykova, E., 2011, Palaeobiogeographical and palaeobiological aspects of
mid- and Late Cretaceous ammonite evolution and bio-events in the Russian
Pacific: Scripta Geologica, v.143, p.15121.
Kennedy, W.J., and Cobban, W.A., 1976, Aspects of ammonite biology, biogeog-
raphy, and biostratigraphy: Palaeontological Association of London Special
Papers in Palaeontology Number 17, 94 p.
Lambiotte, R., and Rosvall, M., 2012, Ranking and clustering of nodes in networks
with smart teleportation: Physical Review E: Statistical, Nonlinear, and Soft
Matter Physics, v.85,056107, doi:10.1103/PhysRevE.85.056107.
Lehmann, J., Ifrim, C., Bulot, L., and Frau, C., 2015, Paleobiogeography of Early
Cretaceous ammonoids, in Klug, C., et al., eds., Ammonoid paleobiology:
662
From macroevolution to paleogeography: Dordrecht, Springer, p. 229257,
doi:10.1007/978-94-017-9633-0_9.
Lieberman, B.S., 2005, Geobiology and paleobiogeography: Tracking the coevolu-
tion of the Earth and its biota: Palaeogeography, Palaeoclimatology, Palaeo-
ecology, v.219, p.2333, doi:10.1016/j.palaeo.2004.10.012.
Martin, A.K., and Hartnady, C.J.H., 1986, Plate tectonic development of the
south west Indian Ocean: A revised reconstruction of East Antarctica and
Africa: Journal of Geophysical Research, v.91, p.47674786, doi:10.1029
/JB091iB05p04767.
Newman, M.E.J., 2006a, Finding community structure in networks using the ei-
genvectors of matrices: Physical Review E: Statistical, Nonlinear, and Soft
Matter Physics, v.74, 036104, doi:10.1103/PhysRevE.74.036104.
Newman, M.E.J., 2006b, Modularity and community structure in networks: Pro-
ceedings of the National Academy of Sciences, v.103, p.85778582, doi:10
.1073/pnas.0601602103.
Newman, M.E.J., and Girvan, M., 2004, Finding and evaluating community struc-
ture in networks: Physical Review E: Statistical, Nonlinear, and Soft Matter
Physics, v.69,026113, doi:10.1103/PhysRevE.69.026113.
Orman, G.K., Labatut, V., and Cherifi, H., 2011, Qualitative comparison of com-
munity detection algorithms, in Cherifi, H., et al., eds., Digital information
and communication technology and its applications: Heidelberg, Springer, p.
265279, doi:10.1007/978-3-642-22027-2_23.
Owen, H.G., 1973, Ammonite faunal provinces in the middle and upper Albian
and their palaeogeographical significance, in Casey, R., and Rawson, P.F., eds.,
The Boreal Lower Cretaceous: Geological Journal Special Issue 5, p. 145154.
Owen, H.G., and Mutterlose, J., 2006, Late Albian ammonites from offshore Suri-
name: Implications for biostratigraphy and palaeobiogeography: Cretaceous
Research, v.27, p.717727, doi:10.1016/j.cretres.2005.12.001.
Page, K.N., 1996, Mesozoic ammonoids in space and time, in Landman, N.H., et al.,
eds., Ammonoid paleobiology: Boston, Massachusetts, Springer, p. 755794,
doi:10.1007/978-1-4757-9153-2_18.
Perrin, C., and Kiessling, W., 2012, Latitudinal trends in Cenozoic reef patterns and
their relationship to climate, in Mutti, M., et al., eds., Carbonate systems dur-
ing the Oligocene-Miocene climatic transition: Oxford, UK, Wiley-Blackwell,
p. 1733, doi:10.1002/9781118398364.ch2.
Peters, S.E., and McClennen, M., 2016, The Paleobiology Database application
programming interface: Paleobiology, v.42, p.17, doi:10.1017/pab.2015.39.
Pons, P., and Latapy, M., 2006, Computing communities in large networks using
random walks: Journal of Graph Algorithms and Applications, v.10, p.191
218, doi:10.7155/jgaa.00124.
Raghavan, U.N., Albert, R., and Kumara, S., 2007, Near linear time algorithm
to detect community structures in large-scale networks: Physical Review E:
Statistical, Nonlinear, and Soft Matter Physics, v.76,036106, doi:10.1103
/PhysRevE.76.036106.
Reboulet, S., 2001, Limiting factors on shell growth, mode of life and segregation
of Valanginian ammonoid populations: Evidence from adult-size variations:
Geobios, v.34, p.423435, doi:10.1016/S0016-6995(01)80006-1.
Rosvall, M., and Bergstrom, C.T., 2008, Maps of random walks on complex net-
works reveal community structure: Proceedings of the National Academy of
Sciences of the United States of America, v. 105, p. 11181123, doi:10.1073
/pnas.0706851105.
Scotese, C.R., 2010, PointTracker, Version 7h: Arlington, Texas.
Scotese, C.R., 2013, PALEOMAP Project: Evanston, Illinois, http://www.scotese.com.
Sewall, J.O., van de Wal, R.S.W., van der Zwan, K., van Oosterhout, C., Dijkstra,
H.A., and Scotese, C.R., 2007, Climate model boundary conditions for four
Cretaceous time slices: Climate of the Past, v.3, p.647657, doi:10.5194
/cp-3-647-2007.
Sidor, C.A., Vilhena, D.A., Angielczyk, K.D., Huttenlocker, A.K., Nesbitt, S.J.,
Peecook, B.R., Steyer, J.S., Smith, R.M.H., and Tsuji, L.A., 2013, Provin-
cialization of terrestrial faunas following the end-Permian mass extinction:
Proceedings of the National Academy of Sciences, v.110, p.81298133, doi:
10.1073/pnas.1302323110.
Vilhena, D.A., and Antonelli, A., 2015, A network approach for identifying and
delimiting biogeographical regions: Nature Communications, v.6, p.6848,
doi:10.1038/ncomms7848.
Vilhena, D.A., Harris, E.B., Bergstrom, C.T., Maliska, M.E., Ward, P.D., Sidor,
C.A., Strmberg, C.A.E., and Wilson, G.P., 2013, Bivalve network reveals
latitudinal selectivity gradient at the end-Cretaceous mass extinction: Scientific
Reports, v.3,1790, doi:10.1038/srep01790.
Wright, C.W., 1963, Cretaceous ammonites from Bathurst Island, Northern Aus-
tralia: Palaeontology, v.6, p.597614.
Manuscript received 8 January 2017
Revised manuscript received 22 March 2017
Manuscript accepted 23 March 2017
Printed in USA
www.gsapubs.org | Volume 45 | Number 7 | GEOLOGY