megafauna, because they would have had large
ecosystem impacts in relation to their abundance
(49, 50). The megafaunal concept must, however, be viewed as context-dependent, because
in island ecosystems, the largest native frugi-vore may be an order of magnitude lighter than
those in continental systems, yet loss of large
island frugivores can result in more sizable
cascading effects owing to the lower functional
redundancy on islands (51). The most important
application of ecological replacements to date
has been in the restoration of herbivory and
seed dispersal functions in island ecosystems.
Extinction of large frugivores can disrupt seed
dispersal, interrupt recruitment, and reduce
genetic variation of large-seeded fleshy-fruited
plants (52); it can also drive rapid evolutionary
reduction in seed size, affecting seed surviv-
al (45). There is evidence of the ecosystem-
engineering role of giant tortoises, as tortoises
are important dispersers of large-seeded plants,
and their grazing and trampling is critical for
creating and maintaining some vegetation com-
munities (53). To restore grazing functions and
the seed dispersal of native large-seeded plants,
exotic Aldabra giant tortoises (Aldabrachelys
gigantea) have been introduced to Mauritian
offshore islands to replace the extinct Mauri-
tian Cylindraspis species (54, 55) (Fig. 3). Not
only has seed dispersal resumed, but passage
through the tortoise gut also improves seed
germination success (55). Further projects are
planned or under way to use ecologically sim-
ilar species of giant tortoise to reinstate pro-
cesses lost with the extinction of endemic giant
tortoises in the islands of Madagascar, the
Galapagos, the Mascarenes, the Seychelles, and
the Caribbean (56).
The future challenge is the identification of suitable replacements to perform the desired ecosystem functions within a given system. The
longer the time since the extinction of the original form, the greater the uncertainty about the
best substitute. The best replacements might
not be closely related taxa. If and when risk and
uncertainty are adequately evaluated, radical substitutions could be considered, such as the use
of tortoises as replacements for moa-nalo, a
group of extinct gooselike ducks, in Hawaii (57).
The focus must be more on reinstatement of
functions and processes to restore degraded ecosystems (58) and to enhance ecosystem resilience,
rather than on restoration to some arbitrary historical state. For any conservation introduction,
the risk of unintended effects must be evaluated
and weighed against the expected benefits (7).
The greatest progress will come from carefully
designed experimental substitutions using spe-
cies that can be readily monitored and managed
(58) and easily removed should the manifesta-
tion of unwanted effects reach some predeter-
mined threshold.
Rewilding
In 1998, the concept of “rewilding” was proposed
as a “fourth current in the modern conservation
movement” that would complement the protection of representative biotic elements (59). The
original concept of rewilding was built around
the keystone role played by wide-ranging, large
animals—particularly carnivores—able to maintain ecosystem structure, resilience, and diversity
through top-down trophic interactions (46, 59).
Rewilding would entail restoration of “big wilderness” through the creation and management
of large, strict, core protected areas, enhanced
connectivity between core reserves, and critically, the restoration of keystone species (59). The
term rewilding is going through a surge in popularity in the media, but its original meaning is
often misinterpreted or lost. Rewilding has been
widely and variously misused to mean: (i) the
reintroduction of any recently extirpated species;
(ii) the rehabilitation of ecosystems through reintroductions; (iii) the return of an ecosystem
to a prehuman state; or (iv) the release of non-native, rather than native, species. The increased
410 25 JULY 2014 • VOL 345 ISSUE 6195
sciencemag.org SCIENCE
Fig. 3. Rebuilding ecosystems by removing
invasive species and introducing ecological
replacements. The extinction (–) of keystone
ecosystem engineers, such as the Mauritian
giant tortoises (Cylindraspis species), and the
addition (+) of non-native mammalian herbivores
and invasive plants degraded (gray arrow) Round
Island’s ecosystem. The restoration phase (green
arrow) first entailed the eradication of goats and
rabbits. Without vertebrate herbivory, exotic vegetation flourished, suppressing native plants adapted
to tortoises’ grazing pressure. Restoration efforts
then focused on weeding invasive flora and rebuilding the native plant community, although
weeding was costly and limited in spatial area. A
long-term, more cost-effective solution sought
to restore the grazing and seed dispersal functions once performed by the giant tortoises. In
2007, a small population of Aldabra giant tortoises was introduced as part of a reversible
experiment to restore and increase ecosystem
resilience (68). Tortoises are preferentially grazing
the fast-growing exotic plants and avoiding much
of the native vegetation believed to have evolved
to withstand the high density of Mauritian giant
tortoises. [Image credits: Giant tortoise 1600s
(J. P. Hulme), giant tortoises today (Z. Ahamud),
1990s (C. Griffiths), 1972 (C. Jouanin)]
+
+
-
Degraded
ecosystem with
non-native
mammalian
herbivores
Original ecosystem
with native tortoises
Ecological replacement
tortoises
2007
+
Control of exotic fora
Reintroduction of
native fora
2002
+
Rabbits
1986
-
Goats
1978
-
Native tortoises
1844
Exotic fora
1870
Exotic rabbits and goats
1800s
Ecosystem state relative to pre-human condition
Intact
Fully
functional
Nonfunctional
Degraded
Level of
ecosystem
function
1972
1600s
1990s
Today
Less degraded ecosystem
with introduced tortoises
Degraded
ecosystem without
vertebrate herbivores
