Synthesis
Shaping the Latitudinal Diversity Gradient: New Perspecti ves
from a Synthesis of Paleobiology and Biogeog raphy
David Jablonski,
1,
* Shan Huang,
2
Kaustuv Roy,
3
and Ja mes W. Va lentine
4
1. Department of Geophysical Sciences, University of Ch icago, Chicago, Illinois 60 637; 2. Senckenberg Biodiversity and Climate Res earch
Center, Senckenberganlage 25, D-60325 Frankfurt (Main), Germany; 3. Section of Ecology, Behavior, and Evolution, Division of Biolo gical
Sciences, University of California, San Diego, La Jolla, California 92093; 4. Department of Integrative Biology, University of California,
Berkeley, California 94720
Submitted July 13, 2016; Accepted Septemb er 16, 2016; Electronically published December 2, 2016
Online enhancements: supplemental tables. Dryad data: http://dx.doi.org/10.5061/dryad.qd53c.
abstract: An impediment to understanding the origin and dynam-
ics of the latitudinal diversity gradient (LDG)— the most pervasive
large-scale biotic pattern on Earth—has been the tendency to focus
narrowly on a single causal factor when a more synthet ic, integrative
approach is needed. Using marine bivalves as a model system an d
draw ing on other systems where possible, w e review paleobiologic
and biogeographic support for two supposedly opposing vie ws, that
the LDG is shaped primarily by (a) local environmental factors that
determine the number of species and higher taxa at a given latitude
(in situ hypoth eses) or ( b) the entry of lineages a rising elsewhere
into a focal region (spatial dynamics hypotheses). Support for in situ
hypotheses includes the fit of present-d ay diversity trends in many
clades to such environmen tal factors as temperature and the correla-
tion of extinction inten sities in Pliocene bivalve faunas with net re-
gional temp erature changes. Suppor t for spatial dynamics hypoth-
eses includes the age-frequency distribution of bivalve genera across
latit udes, whic h is consistent with an out-of-the-tropics dyna mic, as
are the higher species diversities in temperate southeastern Australia
and southe astern Japan than in the tro pical Caribbean. Th us, both in
situ and spati al dynamics processes must shape the bivalve LDG and
are likely to operate in other group s as well. The relative strengths of
the two processes may differ among groups showing similar LDGs,
but dissecting their effec ts wil l require improved methods of in tegrat-
ing fossil data with molecular phylogenies. We highlight several po-
tential research directions and argue that many of the most dramatic
biotic pat terns, past and present, are likely to have been generated by
diverse, mutually reinforcing drivers.
Keywords: extinction, speciation, geographic range dynamics, phy-
loge ny, h istorical biogeography, paleobiology.
Introduction
The latitudinal diversity gradient (LDG), the most pervasive
large-scale biological pattern on Earth, has been documented
and debated for more than two centuries. Multiple reviews
have attested to the increase in diversity of species and higher
taxa from poles to tropics and cataloged the many hypoth-
eses on underlying mechanisms (e.g., Pianka 1966; Rohde
1992; Gaston 2000; Hillebrand 2004; Mittelbach et al. 2007;
Krug et al. 2009b; Donoghue and Edwards 2014; Fine 2015;
Schluter 2016), leading Brown (2014, p. 9) to state that “even
as the patterns have become clearer . . . the explanations have
remained elusive and controversial.” However, others have
suggested that the many proposed explanations are begin-
ning to coalesce (e.g., Fine 2015). In this essay, we go one step
further to argue that a number of supposedly opposing hy-
potheses about the LDG actually represent different facets
that should be integrated to gain a fuller understanding of
the origin and maintenance of this biodiversity pattern.
One such artificial d ichotomy, the chief focus of this es-
say, involves hypotheses that th e equator-t o-pole profile of
the LDG is essentially shaped by local environmental fac-
tors that determine the number of species and higher taxa
at a given la titude (for brevity, we will call these in situ hy-
potheses), as opposed to hypothese s that center on the long-
term spatial dyn amics of clades across latitudes. The crux
of this dich otomy i s the explanatory power of past and pres-
ent conditions within the re gion where diversity is being
measured, a s opposed to processes that closely link div er-
sity in the focal region to events outside it. We argue that,
in fact, in situ processes and spatial dynamics have likely
shaped the modern LDG in concert, as illustrated by the
simple conceptual model in figure 1. Consider two ad jacent
regions, a and b, which can represent two adjace nt latitu-
dinal bands or biogeographic provinces anywhere along
the LDG. ( Note that the focus of this essay is on the mech-
anisms underlying the shape of the full LDG rather than
* Corresponding author; e-mail: djablons@uchicago.edu.
ORCIDs: Huang, http://orci d.org/0000-0002-5055-1308.
Am. Nat. 2017. Vol. 189, pp. 1–12. q 2016 by The University of Chicago.
0003-0147/2017/18901-57110$15.00. All rights rese rved.
DOI: 10.1 086/689739
vol. 189, no. 1 the american naturalist january 2017
This content downloaded from 141.002.253.017 on August 01, 2018 03:50:47 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
those resp onsible for a binary, tropical-vs.-extratropical
contrast in rich ness; the latter would mostly fall under an
in situ heading, aside from those hypotheses holding that
tropical diversity i s promoted by the influx of taxa from
higher latitudes.) The richness of each region at a given time
is partly a function of (i) the environmental and habitat
characteristics of each regi on (T
a
and T
b
), (ii) the specia-
tion rates in each region (s
a
and s
b
), and (iii) their respec-
tive extinctio n rates (x
a
and x
b
), which together constitute
the in situ component because they op erate within each re-
gion. In addition, the richness of each region is also a func-
tion o f dispersal into each area (d
a
and d
b
), a spatial dy-
namic that involves regional processes elsewhere that affect
the focal area. Clearly, all of these processes should be eval-
uated together for a comprehensive understanding of the
LDG. Yet, as discussed below, traditionally they have been
treated separately, and relatively little is known about how
they interact.
We will support this integrative view by a combination
of present-day and fossil data that provide evidence for each
type of dynamic and, therefore, for their simultaneous op-
eration. We frame our discussion using the marine Bival-
via, which hav e become a model system for large-scale spa-
tial and temporal analyses for at leas t three reasons. First,
their well-documented biogeography shows a strong LDG
(fig. 2) that mirrors that of other groups on land and sea
(Hillebrand 2004; Tomasovych et al. 2016). Second, they
exhi bit relatively high taxonomic, p henotypic, and func-
tional diversity, including a wide range of trophic group s
rang ing from suspension fe eding through chem o- and pho-
tosy mbiosi s to predation and parasitism. And third, their
rich fossil record allows robust est imates of origination, re-
gional and global extinction, and geographic rang e dynam-
ics through time (Crame 2000a, 2000b, 200 2; Jablonski et al.
2006, 2013; Bieler et al. 2 013; Berke et al. 2 014; Huang et al.
2015; Mondal and Harries 2016). The ideas presented here
should apply broadly across different groups of organisms,
and where possible we draw on examples from other groups
as well. We will also briefly touch on a related ( false) dichot-
omy, th at is, whether the tropics are a cr adle or museum of
biodiversity.
Evidence for In Situ Controls
Many analyses, both marine and terrestri al, have found that
pres ent-day temperature (annual mean and, to some ex-
tent, variance) along with a handful of other physical fac-
tors, such as precipitation on land and productivity in the
sea, are powerful first-order predictors of the profile of the
LDG on global and continental scales (Fine 2015 and refer-
ences therein) and of the boundaries between biogeographic
units, such as provinces (Olson et al. 2001; Lomolino et al.
2010; Belanger et al. 2012; Briggs and Bowen 2012). The con-
sensus, reasonably, has been that such correlations provide
strong evidence for in situ controls and that species and
higher-taxon richness within latitudinal bands or provinces
track climate almost instantaneously, in that diversity often fits
well with present-day factors despite the enormous changes
in climate at most locations over the past 18,000 years. How-
ever, the species-rich bathyal (deep-sea) benthos suggests that
tropical temperatures are not necessary to generate and main-
tain high levels of diversity (Rex and Etter 2010; Valentine
and Jablonski 2015). One hypothesis is that high bathyal
diversity indicates a role for damped seasonality in, among
other potential factors, temperature and organic carbon flux
(and thus trophic stability), in contrast to the cold but more
Figu re 1: Simple mode l for diversity dynamics. The taxonomic di-
versity in r egions a and b a re functions of in situ factors— environ-
mental and habitat characteristics (T
a
and T
b
), w ithin-region rates of
speciation (s
a
and s
b
), and extinction rates (x
a
and x
b
)—and dispersal
into each area (d
a
and d
b
), that is, the larger-scale spatial dynamic
(for further details, see Goldberg et al. 2005; Jablonski et al. 2006;
Roy and Goldberg 2007).
0
200
400
600
800
1000
1200
-90 -60 -30 0 30 60 90
Number of Species
Latitude
W Pacific
E Pacific
W Atlantic
Figure 2: Bivalve latitudinal diversity gradients (species and genus
level) at shelf depths (0–200 m) along three well-sampled coastlines.
Compiled from the primary literature and m useum collections (data
from Tomasovych et al. 2016).
2 The American Naturalist
This content downloaded from 141.002.253.017 on August 01, 2018 03:50:47 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
seasonal and depauperate polar sea (Valentine and Jablonski
2015). More generally, while a strong correlation between
latitudinal trends in mean annual temperature and regional
species richness is frequently emphasized (including for
bivalves; Roy et al. 1998), many other environmental vari-
ables covary along this gradient, as do such historical effects
as changes in past climate and habitats (e.g., those driven
by Pleistocene glaciations). This multitude of potential driv-
ers makes it difficult to evaluate the role played by different
present-day variables and to separate present-day and his-
torical factors (e.g., Jetz et al. 2004; Ricklefs 2004; Erwin
2009; Field et al. 2009; Svenning et al. 2015; Valentine and
Jablonski 2015).
Large-scale natural experiments can shed light on the
role played by in situ climate changes in shaping the LDG,
and inferences about the variability of past climatic and other
envi ronmental conditions hav e been used to explain re-
gional diversity patterns. The bivalve fossil record shows
that regional extinction is st rongly correlated with net tem-
perature change since the mid-Pliocene and regional ex-
tinction in Pliocene bivalve faunas (fig. 3). In particular,
the striking correspondence between modest net tempera-
ture changes and regional losses in the warm-temper ate Pli-
ocene faunas is consistent with the view that such historical
factors as past climatic changes and extinctions have played
a major role in shaping the modern LDG (e.g., Mannion
et al. 2014). The temperature differences between modern
and full-glacial Pleistoc ene conditions were greater than
those in the Pliocene, albeit with ro ughly parallel interre-
gional patterns (see Annan and Hargreaves 2013; Precht
and Aronson 2016), and so could have contributed to, or
reinforced, the observed trend. In any case, bivalves show
differential regional extinctions with latitude that are posi-
tively related to net re gional cooling over the past 3 million
years and, probably, to the distribution of te mperature min-
ima in the Pleistocene. Most of these extinction s were re-
gional, with individ ual clades contracting equatorward in
the face of climate cooling (Valentine et al. 2013), progres-
sively sharpening the LDG as th e poles shi fted to a refriger-
ated state (albeit with smaller, repeated distri butional shi fts
duri ng glacial/interglacial cycles). Detailed interregional
analy ses comparing the effect s of time-integrated tempera-
ture change or maximum regional excursion s against net
biotic change woul d be valuable.
The pattern in figure 3 also supports the more specific
view that re gions that experienced lesser climate change
have harbored, and perhaps have g enerated, more diversity
than regions that experienced greater changes (e.g., Jan sson
and Dynesius 2002; Pyron and Wiens 2013; Mannion et al.
2014; Fine 2015; Claramunt and Cracraft 2015; Pulido-
Santacruz and Weir 2016). The effe cts of such variation
can be manifest at a variety of scales and locations. For
example, with in the tropics, present-day local diversity in
marine fishes is inversely related to distance from histori-
cally stable reef areas (Pellissier et al. 201 4). In the temper-
ate zones, the high plant diversity in the South African
Cape Floristi c Region has been a ttribu ted to the apparently
damped climate fluctuat ions for its latitude (Dynesius and
Jans son 2000; Schnitz ler et al. 2011; for additional exam-
ples, see also Jansson 2003; Svenning et al. 2015), and climate
velocity—defined as the rate of displacement of climatic
conditions across a region— has been shown to influence
geographic patterns of endemism in multiple vertebrate
groups (Sandel et al. 2011), which in turn affects their LDGs.
In the deep-sea benthos, more direct analyses using esti-
mates of paleotempera ture and paleoproductivity show
that regional species richness of deep-sea invertebrates re-
sponded predictably to past changes in climate (Hunt et al.
2005; Yasuhara et al. 2012).
Taken together, figure 3, the extensive literature show-
ing a close relation today between both marine and terres-
trial diversity and regional clima te, and the smaller litera-
ture relating fossil divers ity to climate shifts (Erwin 20 09;
Mannion et al. 2014) strongly sugg est that regional diver-
sity can change substantially wh en th e environmental pa-
Figu re 3: Regional extinction in Pliocene faunas versus net temper-
ature change from the Pliocene to the present day. Regional temper-
ature changes are estimated using a spatially explicit model of Pliocene
paleotemperatures based on stable iso topes in microfossils (Dowsett
et al. 2010, 2013) and thus are independent of molluscan distribu-
tional data; bivalve data are from Valentine et al. (2013) and Jablonski
et al. (2013). CA p California; Indo p Indonesia; Ice p Iceland;
Med p Med iterranean; Mid-Atl p M id-Atlantic states, United
States; N Sea p North Sea region; NZ p New Zealand; Venez p
Venezuela.
Shap ing the Latitudinal Diversity Gradient 3
This content downloaded from 141.002.253.017 on August 01, 2018 03:50:47 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
rameters shift strongly. For bivalves, such c limate-driven
steepening of the LDG does not appear to be phylogenet-
ically patterned, at least not at the family level, where com-
prehensive phylogenetic data are available (Bieler et al.
2014). Comparing diversity changes along each of the tem-
perate coasts of North America from the Pliocene to the
present day, a given family may show different extinction
rates and/or dispersal dynamics on different coasts (Huang
et al. 2015). A better predictor of clade behavior in the face
of cooling climates is the geographic range size of species
at the Pliocene starting point, with the narrowest-ranging
species most prone to disappear from the focal region
(Huang et al. 2015; see also Saupe et al. [2015], who found
that a measure of niche breadth lacks the predictive power
shown by geographic range per se). This result is not surpris-
ing in that those species most able to tolerate and disperse
among a range of latitudes and environments might be
expected to survive longest, but it usefully demonstrates that
this tolerance effect can override family-level clade member-
ship. Studies of present-day terrestrial birds and mammals,
using more indirect methods, have also concluded that cli-
mate changes “can force multiple groups into similar diver-
sity patterns even when evolutionary trajectories differ”
(Hawkins et al. 2012, p. 825). These findings drive home
the message that a given clade can attain similar diversities
in two regions in the absence of similar diversification dy-
namics and that different diversities for a given clade in
two regions today need not signal a long divergent history.
Evidence for Spatial Dynamics
Changing climates, and thus tracking of in situ environ-
mental chan ge with or without lags, are not the only im-
petus for large-scale biogeographic shifts. Our initial work,
and that of many others, on the dynamics underlying the
LDG was motivated by Stebbins’s (1974) famous question:
are the tr opics a cradle or a museum of diversity? Stebbins
used the cradle metaphor for regions that are diversity gen-
erators, presumably with hi gh origination rates, while mu-
seums are areas that accumulate diversity, presumably with
low extinction rates. Although such a binary characteriza-
tion of the evolutionary drivers of the LDG has led to con-
siderable empirical work, this dichotomy is increasingly
proving to be a false one. The fossil record shows that, over
the past 12 million years, bivalve genera preferentially orig-
inated in the tropics and expanded their ranges poleward
over time—even in the Pleistocene, during times of glaci-
ated poles and steep thermal gradients—while maintaining
their tropical presence (Jablonski et al. 2006, 2013 [the lat-
ter with substantially improved data based on further inte-
gration of literature and museum records]). The pattern is
particularly striking given that the sampling is strongly bi-
ased to the temperate zones (Valentine et al. 2013), so that
some of the clades that originated in the tropics would not
have entered into the fossil record until they entered the
midlatitudes, thereby underestimating the strength of the
dynamic.
Percent (%)
0 100 200 300 400
0
5
10
15
20
25
ARCTIC
N = 62 genera
(+3 not known fossil)
Median age = 54.0 Myr
Percent (%)
0 100 200 300 400
0
10
20
30
40
TROPICAL
N = 772 genera
(+125 not known fossil)
Median age = 25.6 Myr
Percent (%)
0 100 200 300 400
0
5
10
15
20
25
30
35
ANTARCTIC
N = 31 genera
(+2 not known fossil)
Median age = 42.9 Myr
Age (Myr)
Figure 4: Geologic ages of bivalve genera at high and low latitudes,
showing a deficiency of young taxa near both poles a nd a long tail
of older taxa in all three regions. Frequency distributions for the
Arctic and Antarctic differ significantly from that of the tropics
(Komogorov-Smirnov test; tropics vs. Arctic, P ! .001; tropics vs.
Antarctic, P p .004) but not from each other (P p .78). Genus ages
are given in table S1 in the Dryad Digital Repository: http://dx
.doi.org/10.5061/dryad.qd53c (Jablonski et al. 2017).
4 The American Naturalist
This content downloaded from 141.002.253.017 on August 01, 2018 03:50:47 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Because nearly all bivalve clades that expand out of the
tropics (OTT) maintain their presence in the tropics even
as t hey push poleward and because the highest latitudes ev-
idently have low origination rates, the polar regions harbor
very few young genera while the tropics have many, with a
median age roughly half that of the polar fauna (e.g., Gold-
berg et al. 2005). However, tropical genera also have a long
tail to their age distribution, which includes ∼90% of the
taxa that constitute the polar tails (fig. 4). Thus, the tropics
are both a cradle and a museum of diversity in that they
harbor both the oldest and the youngest genera. The poles
are mainly a museum of diversity, as they harbor mostly
older genera. Of course, present-day age distributions in a
given region are uninformative about when taxa actually ar-
rived, but the fossil data can determine the polarity of such
movement.
These latitudinal dy namics could still fit into an in situ
paradigm if the OTT genera were sim ply filling gaps left
by regional extinctions to keep the regional biota in equi-
librium with key environmental variables. Tests on, for ex-
ample, ecological equivalency between extinctions an d suc-
cessful OTT range expansions would therefore be valuable.
However, comparison of West Pacific and West Atlantic
diversity patterns falsify any simple in situ hypothesis.
Southeastern Japan and southeastern Australia, each firmly
lodged today in the temperate zone, contain 30% and 14%
more species, respectively, than the Caribbean (table 1; see
also Tittensor et al. 2010, in which diversity maps incorpo-
rating a variety of marine groups show similar or greater
coastal diversity in one or both of our two temperate re-
gions relative to the tropical Caribbean, aside from two
strongly tropical clades [corals and mangroves]). In situ
models would predict the opposite: the Caribbean has an
order of magnitude more shelf area than either of the tem-
perate provinces; its mean annual sea surface temperature
is 57–67C warmer and significantly less seasonal; it is topo-
logically and ecologically more complex, with coral reefs
and a scatter of islands of various sizes in contrast to the
essentially linear temperate provinces; and it almost cer-
tainly hosts a greater variety and intensity of biotic inter-
actions, another factor posited to promote taxonomic di-
versity (e.g., Mittelbach et al. 2007; Schemske et al. 2009).
Diversity is evidently spilling over from the massively di-
verse west Pacific tropical region into the two adjacent tem-
perate provinces, to an extent that overwhelms the simple
environment-diversity relationships that are often seen when
examining a single coastline or cumulative diversity in global
latitudinal bands.
The very low diversity near the poles relative to diver-
sity maxima in all tropical regions also has analytical con-
sequences: a global model focusing exclusively on the
equator-pole gradient will tend to downplay longitudinal
interocean differences, which, though little studied, have
the potential to be highly informative about the interac-
tion of processes that set large-scale diversity patterns. For
example, linear models linking diversity to in situ envi-
ronmental factors will tend to derive strong support from
extreme polar and tropical values, which then drive the
conclusions. Yet the deviations from model predictions at
midlatitudes and elsewhere, often dismissed as noise, actu-
ally represent biologically informative variance. To be sure,
some interocean differences arise from historical events, the
consequences of which are chiefly restricted to a single ocean
basin. The uplift of the Panamanian Isthmus and the ocean-
ographic response was such an event, potentially account-
ing in part for the surprisingly low modern Caribbean diver-
sity. Although spatially heterogeneous within the Caribbean,
this turnover event involved a 14% extinction of bivalve gen-
era and subgenera, by one estimate, but is followed by a 20%
diversity rebound on a per-taxon basis (Todd et al. 2002)
and so is unlikely to account by itself for the observed dis-
cordance between environment and diversity seen in these
tropical-temperate contrasts.
Synthesis an d Future Directions
Large-scale diversity patterns are set by the spatial and tem-
poral dynamics of origination, extinct ion, and geog raphic
range shifts. These dynamics must operate within an envi-
ronme ntal framework so that constraints on diversity im-
posed by the physical and biotic environment are also im-
porta nt. The role played by envir onmental controls has
been corroborat ed in both paleontological and neontolog-
ical data sets via the fit of regional diver sity variations—
Table 1: Comparisons of bivalve faunas of the Caribbean,
southeastern Australia, and southeastern Japan
Region
No.
species
Shelf
area
(km
2
)
Latitudinal
extent (7)
Mean annual
sea-surfac e
temperature
(7)
SE Japan 769 71,200 7 21.4
Caribbean 537 912,100 27 27.0
SE Australia 612 61,750 14 22.3
Note: The Caribbean Province is taken to exten d from southern Florida and
the Yucatan Peninsula to the mouth of the Amazon, the southeastern Japan
Province is taken to extend fr om the southern tip of Kyushu to the Boso Pen-
insula, and the southeaster n Australian (Peronian) Province is taken to encom-
pass southern Queensland and New South Wales (for coastal marine biogeo-
graphic compartmen ts, see Spalding et al. 2007; Belanger et al. 2012; Briggs
and Bowen 2012). Caribbean data are available in datafile S1 of Jablonski
et al. (2013). Northeastern Australia and southeastern Japan species lists are
provided in table S2 in the Dryad Digital Repository: http://dx.doi.org/10
.5061/drya d.qd53c (Jablonski et al. 2017). Shelf area and latitudin al extent were
estimated using ArcGIS. Mean annual sea-surface temperature was ca lculated
from MODIS (http://modis .gsfc.nasa.gov/).
Shap ing the Latitudinal Diversity Gradient 5
This content downloaded from 141.002.253.017 on August 01, 2018 03:50:47 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
