ECOLOGY
The effects of defaunation on plantsÕcapacity to
track climate change
Evan C. Fricke1,2*, Alejandro Ordonez^3 , Haldre S. Rogers^4 , Jens-Christian Svenning^3
Half of all plant species rely on animals to disperse their seeds. Seed dispersal interactions lost
through defaunation and gained during novel community assembly influence whether plants can adapt to
climate change through migration. We develop trait-based models to predict pairwise interactions and
dispersal function for fleshy-fruited plants globally. Using interactions with introduced species as an
observable proxy for interactions in future novel seed dispersal networks, we find strong potential to
forecast their assembly and functioning. We conservatively estimate that mammal and bird defaunation
has already reduced the capacity of plants to track climate change by 60% globally. This strong
reduction in the ability of plants to adapt to climate change through range shifts shows a synergy
between defaunation and climate change that undermines vegetation resilience.
S
eed dispersal influences global vegetation
dynamics by supporting plant regener-
ation and enabling species ranges to shift
in response to environmental changes
( 1 – 3 ). Roughly half of plant species are
dispersed by animals, and seed dispersal is the
most widespread mutualistic function provided
by vertebrates ( 1 , 4 ). Therefore, disruption of
the seed dispersal mutualism is a key ecological
consequence of defaunation ( 5 ). The impacts of
defaunation on plants are exemplified by the
extinction of mutualistic interactions involv-
ing Pleistocene megafauna. The extinction of
megafauna—such as the elephant-like gom-
photheres in South America—not only severed
coevolutionary relationships supporting plant
regeneration ( 6 , 7 ) but also caused large
declines in long-distance dispersal, which
is critical for population spread, according
to dispersal reconstructions based on body
mass allometries ( 8 ).
Plants today are experiencing mutualist loss
as a result of past and ongoing defaunation
( 9 ) as well as climate change ( 10 ), with many
populations needing to shift hundreds of
meters to tens of kilometers per year to track
their climatic niche ( 11 ). Further, novel com-
munities are assembling as species introduc-
tions and shifting ranges result in cooccurrence
of species that do not share coevolutionary
history ( 12 ). The mutualistic interaction net-
works that assemble in these communities will
likely influence whether certain plant species
persist and spread ( 13 ). Developing the ability
to predict how novel interactions and interac-
tion extinctions affect seed dispersal function
at macroecological scales is key for moni-
toring global human impacts on ecosystem
functioning and forecasting future vegeta-
tion dynamics.
Predicting species interactions and quan-
tifying how they affect ecosystem functioning
are pressing goals for ecologists, and a grow-
ing consensus supports doing so by using a
trait-based approach ( 14 , 15 ). Specifically, re-
searchers can predict interactions by model-
ing the links between observed interactions
and species’traits such as body mass in
predator-prey interactions ( 16 ) or bill and corolla
length in hummingbird-plant interactions ( 17 ).
Complementary data, often also related to
traits, can establish how interactions translate
to ecosystem functioning ( 18 , 19 ). For example,
traits may be used to predict whether a flower-
animal interaction results in effective pollina-
tion or nectar robbing ( 20 ). The increasingavailability of data on species’traits and inter-
actions brings global predictions within reach,
and machine learning tools now offer suitable
methods to capture the complex relationships
that shape these interactions ( 17 ). We leverage
these data and models to apply this trait-based
approach to (i) make quantitative predictions
of pairwise species interactions and resulting
ecosystem function that can be applied at
the global scale, (ii) assess the predictability of
novel species interactions and their potential
to constrain range shifts, (iii) estimate how
altered species composition has affected seed
dispersal function globally, and (iv) quantify
the resulting impact on the ability of plants to
track climate change.
Our first aim was to build predictive models
for seed dispersal interactions involving
mammals and birds, and to estimate two key
functional outcomes of these interactions for
plants: seed dispersal distance and the impact
of gut passage on germination. By extending
standard methods based on trait-matching and
body mass allometries ( 2 , 8 ), we developed trait-
based models for discrete components of the
seed dispersal process for birds and mammals
(fig. S1) with boosted regression trees, a syn-
thesis of data on the seed dispersal process,
and databases of plant and animal traits ( 21 ).
Analyses of the seed dispersal process involved
a database of ~18,000 distinct interactions
from 406 local networks where seed dispersal
interactions were recorded at a given location
and time period ( 22 ), data on disperser move-
ment and seed retention time involving 302
animal species ( 23 , 24 ), and the results of 2215
experiments that tested germination of seeds
that were consumed and passed by frugivores
versus germination of seeds untouched by210 14 JANUARY 2022•VOL 375 ISSUE 6577 science.orgSCIENCE
(^1) National Socio-Environmental Synthesis Center, University of
Maryland, Annapolis, MD 21401, USA.^2 Department of
BioSciences, Rice University, Houston, TX 77005, USA.
(^3) Center for Biodiversity Dynamics in a Changing World
(BIOCHANGE), Department of Biology, Aarhus University, Ny
Munkegade 114, DK-8000 Aarhus C, Denmark.^4 Department
of Ecology and Evolutionary Biology, Iowa State University,
Ames, IA 50011, USA.
*Corresponding author. Email: [email protected]
Fig. 1. Functional outcomes of plant-
frugivore interactions predicted on the
basis of plant and animal traits.(A) Mean
seed dispersal distances and (B) germination
effects (log odds ratio; values >0 indicate
improvement in germination relative to
baseline germination for seeds untouched
by frugivores) predicted on the basis of
multiple traits and plotted against jittered
animal body mass. Each point represents
an observed plant-frugivore interaction in the
network database. [Photos left to right via
Wikimedia: T. Castro, K. Schneeberger, G. Stolz,
M. Thyssen, J. Boone, C. Sharp, V. Panzirsch]
RESEARCH | REPORTS