Oikos 122: 306–312, 2013

doi: 10.1111/j.1600-0706.2012.20716.x
© 2013 The Author. Oikos © 2013 Nordic Society Oikos
Subject Editor: Jotaro Urabe. Accepted 18 July 2012

Comparing aquatic and terrestrial grazing ecosystems:

Deron E. Burkepile
D. E. Burkepile (deron.burkepile@fiu.edu), Dept of Biological Sciences, Florida International Univ., North Miami, FL 33181, USA.

‘Grazing ecosystem’ is typically used to describe terrestrial ecosystems with high densities of mammalian herbivores
such as the Serengeti in East Africa or the Greater Yellowstone Ecosystem in North America. These abundant, large
herbivores determine plant community dynamics and ecosystem processes. The general concepts that define grazing eco-
systems also aptly describe many aquatic ecosystems, including coral reefs, seagrass beds, and lakes, where herbivores such
as parrotfishes, turtles, and zooplankton have strong impacts on ecosystem processes. Here, I compare the ecology of
grazing ecosystems in search of common concepts that transcend the terrestrial-aquatic boundary. Specifically, I evaluate:
1) the feedbacks between herbivory and primary production, 2) the roles of herbivore richness and facilitation, 3) how
predators and diet quality shape patterns of herbivory, and 4) how altering herbivory mediates alternative states.

Grazing ecosystems and the role of herbivores McNaughton 1985, Carpenter 1986). Here, I compare
the ecological processes shaping grazing ecosystems across
In terrestrial ecosystems, the term ‘grazing’ typically refers terrestrial, marine, and freshwater habitats to identify areas
to the consumption of grasses while ‘browsing’ refers to where cross-fertilization of ideas could be productive in
the consumption of forbs or trees and the more general facilitating the study of these herbivore-dominated systems.
term ‘herbivory’ refers to consumption of plants in general There is a long history of comparing general ecologi-
(McNaughton and Georgiadis 1986). The term ‘grazing cal processes across wet and dry ecosystems (Strong 1992,
ecosystem’ is often used to describe terrestrial ecosystems Pace et  al. 1999), and recent syntheses have identified
that are dominated by large herds of migratory herbivores broad patterns in plant–herbivore interactions across
such as wildebeest and gazelles in the Serengeti and bison marine, freshwater, and terrestrial systems (Cebrian and
and caribou in North America (Bell 1971, Frank et  al. Lartigue 2004, Hillebrand et al. 2007, Gruner et al. 2008).
1998). The high biomass of herbivores in these ecosystems, While these syntheses are important for understanding
typically the highest among terrestrial systems, results in common patterns in ecological processes across ecosystems,
herbivory being an important regulator of ecosystem pro- limiting comparisons to systems with similar characteris-
cesses with  50% of primary production removed as com- tics may allow a deeper investigation of the mechanisms
pared to ~ 9% removal, on average, for other terrestrial that underlie relationships between plants and herbivores.
ecosystems (Frank et al. 1998). Here, I focus on comparing the ecological processes across
These characteristics also apply to many aquatic eco- aquatic and terrestrial ecosystems (i.e. grazing ecosystems)
systems where ‘grazing’ is often used interchangeably with that share some fundamental characteristics such as the
‘herbivory’ and refers to the consumption of fast-growing relative high abundance of herbivores, their relatively long
primary producers such as phytoplankton, filamentous generation times relative to primary producers, and their
algae or sea grasses. In marine systems, for example, large relatively large body size as compared to their food source.
herbivores such as fishes, urchins, turtles and dugongs Specifically, I evaluate: 1) the feedbacks between herbivory
regulate primary production and community structure of and primary production, 2) the roles of herbivore rich-
coral reefs, seagrass beds and kelp forests (Burkepile and ness and interspecific facilitation, 3) how predators and
Hay 2006). Likewise, in freshwater ecosystems, waterfowl, forage quality shape patterns of herbivory, and 4) how her-
zooplankton and benthic invertebrates impact the rates of bivores mediate regime shifts. My goal is not to compre-
primary production and nutrient regeneration (Lamberti hensively review herbivory (Schmitz 2008) but to highlight
and Resh 1983, Carpenter et  al. 1987). Across these graz- key concepts and mechanisms where further comparative
ing ecosystems, 50–90% of primary production gets passed research across different wet and dry ecosystems could
up the food chain via herbivory (Cargill and Jefferies 1984, be fruitful.

306
Herbivores and the facilitation of primary facilitates the flux of nutrients from the water column to
production the producers thereby decreasing nutrient limitation
(Carpenter and Williams 2007). Herbivores may also
Although herbivory is often viewed as a negative interaction facilitate primary production by altering primary producer
from the perspective of a plant, theoretical and empirical species composition such as on corals reefs where her-
studies in grasslands suggest that individual plants, as well bivores prevent colonization of larger macroalgae which
as ecosystem-level primary productivity, may respond in a have slower growth rates than filamentous algae (Bruno
neutral or even a positive manner to grazing (McNaughton et al. 2006). This mechanism is likely at work in terrestrial
1979, Frank et al 1998). This ‘grazing optimization’ hypo­ systems as well since removal of grazers leads to encroach-
thesis suggests that plants may compensate for biomass lost ment of woody vegetation (Bond 2008) that is likely less
to herbivory by increasing primary production following productive on an annual basis. Smaller crustaceans and
grazing (McNaughton 1979). The hypothesized mecha- gastropods across a range of benthic freshwater and marine
nisms whereby herbivores facilitate plant production in ecosystems may also facilitate production on grazing
these systems are that: 1) herbivores reduce self-shading in lawns of microalgae (Lamberti and Resh 1983, Jaschinski
grasses, particularly by preventing the accumulation of and Sommer 2010) by preventing self-shading, facilitat-
dead plant biomass and 2) herbivores efficiently recycle ing nutrient flux to the algae, and/or recycling limiting
limiting nutrients to back to the soil via urine and dung nutrients given their close spatial relationships with their
(McNaughton 1979, 1983). Although the hypothesis is con- prey base (Bracken and Nielsen 2004).
troversial (Briske 1993), empirical work in North American Although herbivores can facilitate primary produc-
and African grasslands shows that grazed grass communi- tion across a variety of ecosystems, many studies in both
ties can produce more biomass over the course of a growing terrestrial and aquatic systems document suppression by
season as compared to ungrazed areas (McNaughton 1979, or no effect of herbivory on primary productivity rather
1985, Frank et al. 2002). than facilitation (Zieman et al. 1984, Williams 1988, Briske
In pelagic aquatic systems, herbivory is unlikely to 1993, Knapp et  al. 2012). Overall, facilitation appears
release single-celled phytoplankton from self shading, strongly dependent upon grazing intensity and resource
but herbivorous zooplankton are important recyclers of availability. Herbivores can facilitate primary production at
limiting nutrients that stimulate primary production low to medium grazing intensity but suppress production
(Sterner et  al. 1992). The stoichiometry of the nutrients at higher intensities (McNaughton 1983, Hik and Jefferies
(nitrogen:phosphorus) recycled by zooplankton can deter- 1990). Further, resource availability appears to drive much
mine the impact of primary production (Elser and Urabe of the context-dependency in facilitation of primary pro-
1999, Sterner and Elser 2002). The stoichiometry of duction. In terrestrial systems, herbivores are most likely
zooplankton-derived nutrients may even control primary to facilitate primary production when water is not limit-
production on a global scale in pelagic marine ecosystems. ing (Augustine and McNaughton 2006) but either light or
A modeling study suggested that herbivorous zooplankton soil nutrients are limiting such that herbivores either
with high nitrogen demand recycle low N:P ratio wastes remove plant biomass to increase light availability or
which favors phosphorus-limited, N2-fixing phytoplank- recycle nutrients to the soil to remove nutrient limitation
ton that, in turn, facilitate nitrogen-limited phytoplank- (Milchunas and Lauenroth 1993, de Mazancourt et  al.
ton (Nugraha et  al. 2010). This stoichiometric focus is 1998). Likewise in seagrass ecosystems, facilitation may
missing from many terrestrial studies of herbivore nutrient be resource-dependant as urchins increased seagrass pro-
subsidies as the focus is typically on nitrogen input with duction under high light intensity (shallow areas or during
little emphasis on how this impacts eventual phosphorus summer) but not under low light intensity (deeper seagrass
limitation. beds or during spring) (Valentine et  al. 2000). Thus, the
For many subtidal marine systems, as in grass-dominated potential positive feedbacks between grazing and primary
systems, intense herbivory keeps primary producers in production are likely spatially discrete and depend on the
short, fast-growing growth forms where there is minimal context of abiotic factors that influence the potential rate of
self-shading and competition for light. Seagrass beds may primary production (Fig. 1).
be the best analogs to terrestrial grazing ecosystems where Work in terrestrial and pelagic aquatic systems has led
dugongs, turtles, fishes and urchins can graze up to 90% the way in fleshing out these context-dependent factors, but
of primary production (Heck and Valentine 2006). These more emphasis is needed to understand under what condi-
large grazers can increase primary production by up to tions herbivores suppress or facilitate production in benthic
40% over ungrazed areas by removing heavily-epiphytized, aquatic systems. For example, given that turbulence strongly
seagrass blades which relieves light limitation and facili- influences nutrient limitation in benthic marine produc-
tates the production of new, fast growing shoots. (Valentine ers (Carpenter and Williams 2007), grazing may facilitate
et  al. 1997, 2000, Moran and Bjorndal 2005). Likewise, primary production in areas of moderate turbulence where
grazing by herbivorous urchins in coral reef systems can grazing-induced alterations to the producer canopy maxi-
increase algal production by 3–4 times over ungrazed areas mizes nutrient delivery to the benthos while suppressing
(Carpenter 1986). Wave motion and currents would make production in areas of lower flow. Thus, facilitation may
it unlikely that herbivores could fertilize most benthic be more prevalent on the edges of seagrass beds rather than
primary producers directly via their excretions. Yet, her- the interior where the surrounding seagrass canopy retards
bivores may indirectly fertilize benthic primary producers water motion or on turbulent shallow forereefs on coral reefs
as the short primary producer canopy created by grazing rather than calmer deep forereefs and lagoonal reefs.

307
 across aquatic systems (Stachowicz et  al. 2007), and
research in these wet ecosystems has helped drive the field
forward in understanding the role of consumer diversity in
ecosystems given that similar experimental manipulations
are intractable in many terrestrial systems.
In species rich grazing systems, the diversity of body sizes,
adaptations for herbivory, and foraging scales (Cromsigt
and Olff 2006), may lead to important facilitative effects
among herbivore species in addition to significant effects
on primary producer communities. The most common
mechanism of interspecific facilitation in grazing ecosys-
tems appears to be herbivore-mediated increases in forage
quality. In terrestrial systems, large, ecosystem-engineering
megaherbivores such as white rhinoceros and hippo-
potamus can create and maintain grazing lawns of short,
nutritious grasses that facilitate nutrient acquisition by
smaller herbivore species that preferentially feed in these
lawns (Verweij et  al. 2006, Waldram et  al. 2008). Dense
concentrations of smaller herbivores can also increase
forage quality, likely increasing the carrying capacity of
grasslands (McNaughton et  al. 1997). However, these
Figure 1. Conceptual graph of the relationship between productiv- facilitative effects may only be evident during the growing
ity potential of the primary producer community and the impact season as herbivores would be more likely to compete
of herbivores on primary production in grazing ecosystems. with, rather than facilitate, one another during the dry
Herbivores will negatively impact primary production when abiotic
season when plants are not growing and forage quality and
factors controlling production such as rainfall in terrestrial ecosys-
tems or turbulence or light in aquatic ecosystems are low. As quantity are limiting (Arsenault and Owen-Smith 2002,
the potential for productivity increases due to higher rainfall Odadi et al. 2011).
or increased turbulence, the impact of herbivores on primary pro- Although facilitative interactions among herbivores
duction transitions from being net negative to net positive as her- in aquatic systems have received much less attention, the
bivory maintains primary producers in short, fast-growing growth mechanism of increased food quality via intense grazing and
forms where they can take advantage of readily available resources. nutrient recycling may be similar. In both salt marshes
and seagrass beds, large vertebrate grazers such as geese
and turtles repeatedly graze plants, which mobilizes nitro-
Herbivore richness, competition, and facilitation gen from root stores and increases the nitrogen content of
plant regrowth (Cargill and Jefferies 1984, Zieman et  al.
The range in species diversity, body size, and adaptations 1984, Moran and Bjorndal 2005, Aragones et  al. 2006).
to herbivory across many grazing ecosystems (e.g. 10–20 These patches of high quality forage could, in turn, facilitate
species of dominant herbivores in African savannas and nutrient acquisition in smaller herbivores such as hares in
30 on some coral reefs) sets the stage for complex inter- salt marshes and urchins and parrotfishes in seagrass beds.
actions among herbivore species (du Toit 2003, Bellwood Changes to the nutrient stoichiometry of primary pro-
et  al. 2004). Although low diversity herbivore guilds can ducers via consumer-driven nutrient cycling may also be
strongly impact plant communities, herbivore richness an important mechanism of facilitation. In mesocosm
appears to facilitate rates of herbivory via complementary experiments, Urabe et  al. (2002) demonstrated that her-
feeding in areas of high herbivore diversity. For example, bivorous zooplankton showed positive density-dependent
in the Serengeti, mixed-species herds of zebra, wildebeest, growth as they recycled phosphorus back to P-limited
gazelle, and buffalo consumed 82% of plant biomass phytoplankton, which led to increased diet quality and
while single-species herds consumed an average of 54% somatic growth of the zooplankters. Benthic freshwater
(ranging from 31–74% depending on herbivore species) grazers also show complex impacts on the stoichiometry
(McNaughton 1985). Feeding preferences for different of microalgae (Hillebrand et al. 2008) setting the table for
plant species and plant heights likely facilitate efficient complex facilitative or competitive effects among herbivores.
biomass removal. Similarly, Burkepile and Hay (2008) The focus on the stoichiometric relationships between
used in situ mesocosms to manipulate richness of herbiv- herbivores and primary producers in freshwater grazing
orous fishes on a coral reef in the Caribbean and showed ecosystems has produced fundamental insights into the
that increasing species richness allowed for complementary dynamic nature of the feedbacks among grazing, nutri-
feeding on different species of algae and a more efficient ents, and primary production (Sterner and Elser 2002).
removal of algal biomass, which led to higher survivorship However, for some benthic marine systems and many
and growth of corals. Likewise, species rich assemblages terrestrial systems the focus is often on how grazers impact
of zooplankton increased grazing rates and phytoplank- single nutrients (often nitrogen). Given that terrestrial
ton removal as compared to single species assemblages herbivores may be more nutrient limited than their aquatic
(Sommer et al. 2004). Significant direct and indirect effects counterparts (Cebrian et al. 2009), a stoichiometric frame-
of consumer diversity on ecosystem function are common work for understanding feedbacks between herbivores and

308
producers could be especially useful. Recent work showing systems with minimal human impact have inverted trophic
that grasshoppers exert complex effects on plant C:N:P pyramids with top-predators such as sharks, jacks and
stoichiometry (Zhang et  al. 2011) further emphasizes the grouper representing up to 85% of the fish biomass
need for a more explicit stoichiometric framework for (Sandin et al. 2008) – the equivalent of ~ 5 lions per wilde-
understanding plant–herbivore and herbivore–herbivore beest in the Serengeti and clearly an unsustainable trophic
interactions. pyramid in a terrestrial system. The longer food chains,
higher predator biomass, and stronger top–down control
Top–down vs bottom–up control of grazing in aquatic systems may ultimately stem from higher quality
ecosystems primary producers (Cebrian et  al. 2009). However, herbi-
vores in aquatic grazing systems may ultimately be nutrient
Although the impact of herbivores on plants may be simi- limited (Hillebrand et  al. 2009) which may put an upper
lar across these different grazing ecosystems, the relative limit on the amount of biomass that can be removed from
effects of top–down and bottom–up forces on herbivore these systems and may determine how and where these
populations shows some fundamental differences between herbivores forage within ecosystems. For example, limit-
aquatic and terrestrial grazing ecosystems (Cebrian and ing nutrients, particularly nitrogen, can influence foraging
Lartigue 2004, Shurin et  al. 2006). Trophic cascades in decisions in herbivorous fishes, as in ungulates, as fishes
many systems, especially aquatic ones, show that herbi- target areas of higher quality forage within the landscape
vores in grazing systems can be limited by predators (Shurin (Burkepile and Hay 2009).
et  al. 2002). However, the strength of the effect of preda- The interaction between predators and nutrient limi-
tors depends on a variety of factors such as the metabolism tation in driving herbivore foraging decisions will be a
and allometry of the herbivore and predator (Borer et  al. fruitful area for comparing terrestrial and aquatic grazing
2005, Shurin and Seabloom 2005). The fact that trophic systems. Ungulates often select habitats to decrease preda-
cascades tend to become stronger as the herbivore:primary tor encounter or increase the likelihood of escape once a
producer size ratio increases (Shurin and Seabloom 2005) predator is encountered (Ripple and Beschta 2004, Riginos
suggests that most grazing ecosystems should exhibit strong and Grace 2008). This predator-influenced foraging can
trophic cascades. However, the nutrient content of primary have significant impacts on the spatial impacts of herbivores
producers can limit the amount of herbivory in ecosystems on primary producers and nutrient cycling (Frank 2008).
(Cebrian and Lartigue 2004, Cebrian et  al. 2009). Thus, Similar patterns are well known in marine grazing eco-
the relatively nutrient poor primary producers in terrestrial systems, such as seagrass beds of Western Australia. Here,
grazing ecosystems and the subsequent nutrient imbal- herbivorous dugongs readily forage in deeper, less produc-
ance in terrestrial herbivores may mean they are more often tive areas where they are safer from their tiger shark pred-
limited from the bottom–up than would be marine or fresh- ators as opposed to foraging in shallower areas that have
water herbivores (Elser et al. 2000, Hillebrand et al. 2009). better quality forage but also a higher risk of encountering
The classic grazing ecosystems of African savannas do a shark (Wirsing et al. 2007). However, in the same seagrass
not exhibit a trophic cascade dynamic but show strong ecosystem, green sea turtles that are in poor body condi-
bottom–up regulation. Predators such as wild dogs, leopard, tion are more likely to make foraging decisions that are
and lion often account for 100% of mortalities in smaller riskier than those their more healthy conspecifics make in
herbivores such as gazelle or impala but negligible mortality order to gain higher quality forage (Heithaus et al. 2007).
for megaherbivores such as rhinoceros (Sinclair et al. 2003, Thus, foraging decisions and acceptable levels of risk are
Owen-Smith and Mills 2008). For medium to large-sized dependant on herbivore body condition and the need for
ungulates like buffalo and wildebeest, bottom–up pro- quality forage. Since terrestrial herbivores are more likely
cesses such as rainfall and forage availability are often the to be nutrient limited and require high quality forage, ter-
main drivers of population dynamics, especially for migra- restrial herbivores may have to make riskier foraging deci-
tory populations (Mduma et  al. 1999). The quest for sions to acquire the necessary limiting nutrients as compared
limiting nutrients or the highest quality forage is a strong to their marine counterparts. This state-dependant frame-
driver of herbivore movement and habitat selection with work for foraging decisions should serve as a nice structure
herbivores targeting areas high in sodium, potassium, to clarify understanding of the foraging choices that herbi-
nitrogen and digestibility (McNaughton 1988, Fryxell vores make.
et  al. 2004). Because these herbivores are major prey for
the dominant predators such as lions, these systems likely Alternative states in grazing ecosystems
have strong bottom–up drivers that cascade through the
ecosystem (Sinclair et  al. 2010). The bottom–up controls The strong interactions and positive feedbacks between
on large herbivores in terrestrial systems are more evident plants and herbivores in grazing ecosystems suggest that
on a global scale where rainfall and soil nutrients interact to disrupting these connections has severe consequences for
create hotspots of high quality forage and high diversity of ecosystem integrity. In fact, some of the classic examples of
large herbivores (Olff et al. 2002). alternative states, or regime shifts, come from ecosystems
Aquatic systems are much more likely than terrestrial where both primary production and herbivory are typically
systems to have inverse biomass pyramids with a large high (Scheffer et  al. 2001). A classic example is from the
standing stock of predators and a high turnover of primary strong trophic cascades in freshwater lakes where the pres-
producer biomass (Shurin et al. 2006, Sandin et al. 2008). ence of top predators, piscivorous fishes, facilitate abun-
Recent expeditions to remote coral reefs have shown that dant zooplankton and intense grazing on phytoplankton

309
(Carpenter et  al. 1987). When these top predators are woodlands to open savannas (Dublin et  al. 1990). Thus,
removed, mid-level carnivores increase, prey on zooplank- the forces that maintain savannas in open grass-dominated
ton, and release phytoplankton from top–down control. The state (i.e. herbivory) may not be sufficient to reverse the
reduction in herbivory has been linked, in part, to the tran- altered woodland state. Likewise, in reef systems, the grazing
sition from clear-water lakes with high levels of submerged pressure required to reverse a reef from being algal-
vegetation to turbid lakes with little to no submerged vege­ dominated to being coral-dominated is much higher
tation and abundant phytoplankton (Jeppesen et al. 1997). than the grazing pressure needed to maintain the coral-
Trophic cascades can also lead to an overabundance of dominated state (Mumby et al. 2007). Understanding what
herbivores in some grazing systems and the overgrazing processes control the nonlinear transitions within these eco-
of the primary producers. One classic example is the systems will facilitate their conservation and restoration.
strong trophic cascade in North Pacific kelp beds where
reductions in populations of predatory sea otters leads to Conclusions
population explosions of herbivorous urchins and the con-
version of kelp forests to crustose algal barrens (Estes et al. Recent syntheses point to key differences among wet
1998). Similarly, overfishing of invertebrate-eating fishes and dry ecosystems such as higher quality forage in aquatic
on coral reefs in the Indian Ocean leads to outbreaks ecosystems (Cebrian et  al. 2009) and opposite effects
of herbivorous urchins which overgraze the algal commu- of nutrient loading on primary producer richness across
nity and eventually feed on corals and erode the reef frame- aquatic and terrestrial ecosystems (Hillebrand et al. 2007).
work (McClanahan and Shafir 1990). Decreasing fishing Yet, there are also important similarities across these sys-
pressure on predators of urchins can help reverse this tems such as similar effects of herbivores and nutrient load-
trajectory. Overgrazing is common in terrestrial grazing ing on primary producer biomass (Gruner et al. 2008) and
ecosystems but is rarely linked to predator removal (Sinclair similar patterns in nutrient limitation (Elser et  al. 2007).
et al. 2010) and typically only happens where natural eco- In addition to identifying broad divergence or convergence
systems are converted to livestock grazing with extremely in patterns and processes of plant–herbivore interactions
high biomass of ungulates (Milchunas and Lauenroth 1993). across ecosystems, I suggest that focusing on wet and dry
The integrity of grazing ecosystems also suffers when ecosystems with similar inherent properties (e.g. size ratio
rates of herbivory decline due to overexploitation or disease. of herbivores:primary producers, trophic structure, primary
On many coral reefs, overfishing of herbivorous fishes com- producer growth forms) may allow for a better comparison
promises the ability of corals to recover after disturbances. of the mechanisms that may underlie larger patterns.
When grazing is reduced, larger, less productive macroalgae A comparison of aquatic and terrestrial grazing ecosys-
often become abundant, and they can hinder coral growth, tems is useful as different ecosystems seem to be leading
survivorship and recruitment (Hughes et al. 2007, Burkepile the way in addressing different mechanisms driving plant–
and Hay 2008). Thus, the removal of the herbivore from animal and animal–animal interactions. For example,
this strong herbivore–producer connection hampers the terrestrial ecologists have established a solid theoretical
recovery of the coral-dominated state and facilitates a shift framework for examining how herbivores impact the rate
to macroalgal-dominated reefs (Mumby 2009). Similarly of primary production across different abiotic conditions
in terrestrial systems, the loss of large herbivores via disease (Briske 1993, de Mazancourt et  al. 1998) that could be
or poaching leads to increased recruitment of woody used productively by aquatic ecologists for examining the
vegetation (Dublin et  al. 1990, Prins and Vanderjeugd interaction of herbivores and abiotic forcing on produc-
1993) which can ultimately convert open grassland to tion. However, like Gruner et  al. (2008), I suggest that
closed woodland if fire, the other major driver of woody research in marine and terrestrial ecosystems follow the lead
plant abundance, is infrequent (Bond 2008). of freshwater ecologists (Elser and Urabe 1999, Urabe et al.
In many of these ecosystems, the trajectory between 2002) by using a more explicit stoichiometric framework
alternative states does not follow the same trajectory forward for assessing how herbivores facilitate (or suppress) pri-
and back along the same path, a property called hysteresis mary production and each other. Although recent synthesis
(Scheffer et  al. 2001). Thus, shifting from one alternative suggests that predators control both herbivore and primary
state to another in a grazing ecosystem may take more producer abundance across disparate aquatic and terres-
than just reestablishing herbivore populations (or control- trial ecosystems (Borer et  al. 2006), in grazing ecosystems
ling overgrazing) depending on how substantial the shift in there may be fundamental differences in top–down control
the ecosystem has become. The roles of ungulates, ele- with stronger bottom–up forcing of vertebrate herbivore
phants, and fire on the transition between open savanna to populations in terrestrial ecosystems (Mduma et  al. 1999,
closed woodland emphasizes this point. Large herbivores Sinclair et  al. 2010). Despite this difference, a behavioral
and frequent fires tend to facilitate a grass-dominated framework integrating predation risk and resource acquisi-
system and prevent encroachment of woody vegetation tion has been useful across ecosystems for understanding
(Bond 2008), but removal of these herbivores can facili- herbivore foraging decisions (Wirsing et al. 2007, Schmitz
tate the conversion of open savannas to woodlands (Prins 2008, Riginos and Grace 2008). The next challenge across
and Vanderjeugd 1993). Once these woodlands become systems ranging from savannas to coral reefs is to link these
established, even large herbivores such as elephants, which foraging decisions to spatial differences in impacts on plant
are uniquely destructive to large trees, seem unable to con- communities. Ultimately, cross-fertilization of ideas among
vert established woodland back to open savanna. Only an ecosystems could lead to a more unified theory of the pro-
increase in fire frequency appears able to return established cesses and mechanisms that structure these ecosystems.

310
Acknowledgements – I thank M. D. Smith, C. E. Burns, S. L. de Mazancourt, C. et  al. 1998. Grazing optimization and
Collins, R. W. S. Fynn, N. Govender, K. P. Kirkman, A. K. Knapp, nutrient cycling: when do herbivores enhance plant produc-
C. J. Tambling, D. I. Thompson, A. Zinn and the rest of the tion? – Ecology 79: 2242–2252.
Savanna Convergence Experiment for introducing this marine du Toit, J. T. 2003. Large herbivores and savannah heterogeneity.
biologist to the grazing ecosystems of Africa and North America. – In: du Toit, J. T. et al. (eds), The Kruger experience: ecology
S. L. Collins, R. W. S. Fynn, D. S. Gruner, J. Idjadi, K. A. Mooney, and management of savanna heterogeneity. Island Press,
R. Newman, R. D. Rotjan and D. I. Thompson provided useful pp. 292–309.
comments on the manuscript. Dublin, H. T. et al. 1990. Elephants and fire as causes of multiple
stable states in the Serengeti Mara woodlands. – J. Anim.
Ecol. 59: 1147–1164.
References Elser, J. J. and Urabe, J. 1999. The stoichiometry of consumer-
driven nutrient recycling: theory, observations and conse-
Aragones, L. V. et al. 2006. Dugong grazing and turtle cropping: quences. – Ecology 80: 735–751.
grazing optimization in tropical seagrass systems? – Oecologia Elser, J. J. et  al. 2000. Nutritional constraints in terrestrial
149: 635–647. and freshwater food webs. – Nature 408: 578–580.
Arsenault, R. and Owen-Smith, N. 2002. Facilitation versus compe- Elser, J. J. et al. 2007. Global analysis of nitrogen and phosphorus
tition in grazing herbivore assemblages. – Oikos 97: 313–318. limitation of primary producers in freshwater, marine and
Augustine, D. J. and McNaughton, S. J. 2006. Interactive effects terrestrial ecosystems. – Ecol. Lett. 10: 1135–1142.
of ungulate herbivores, soil fertility, and variable rainfall on Estes, J. A. et al. 1998. Killer whale predation on sea otters linking
ecosystem processes in a semi-arid savanna. – Ecosystems 9: oceanic and nearshore ecosystems. – Science 282: 473–476.
1242–1256. Frank, D. A. 2008. Evidence for top predator control of a grazing
Bell, R. H. V. 1971. A grazing ecosystem in the Serengti. – Sci. ecosystem. – Oikos 117: 1718–1724.
Am. 225: 86–94. Frank, D. A. et  al. 1998. The ecology of the Earth’s grazing
Bellwood, D. R. et  al. 2004. Confronting the coral reef crisis. ecosystems. – Bioscience 48: 513–521.
– Nature 429: 827–833. Frank, D. A. et  al. 2002. Consumer control of grassland plant
Bond, W. J. 2008. What limits trees in C-4 grasslands and production. – Ecology 83: 602–606.
savannas? – Annu. Rev. Ecol. Evol. Syst. 39: 641–659. Fryxell, J. M. et  al. 2004. Predictive models of movement by
Borer, E. T. et al. 2005. What determines the strength of a trophic Serengeti grazers. – Ecology 85: 2429–2435.
cascade? – Ecology 86: 528–537. Gruner, D. S. et  al. 2008. A cross-system synthesis of consumer
Borer, E. T. et  al. 2006. Asymmetry in community regulation: and nutrient resource control on producer biomass. – Ecol.
effects of predators and productivity. – Ecology 87: 2813–2820. Lett. 11: 740–755.
Bracken, M. E. S. and Nielsen, K. J. 2004. Diversity of intertidal Heck, K. L. and Valentine, J. F. 2006. Plant–herbivore interactions
macroalgae increases with nitrogen loading by invertebrates. in seagrass meadows. – J. Exp. Mar. Biol. Ecol. 330:
– Ecology 85: 2828–2836. 420–436.
Briske, D. D. 1993. Grazing optimization – a plea for a balanced Heithaus, M. R. et al. 2007. State-dependent risk-taking by green
perspective. – Ecol. Appl. 3: 24–26. sea turtles mediates top–down effects of tiger shark intimi­
Bruno, J. F. et  al. 2006. Partitioning the effects of algal species dation in a marine ecosystem. – J. Anim. Ecol. 76: 837–844.
identity and richness on benthic marine primary production. Hik, D. S. and Jefferies, R. L. 1990. Increases inthe net above-
– Oikos 115: 170–178. ground primary production of a salt marsh forage grass: a test
Burkepile, D. E. and Hay, M. E. 2006. Herbivore vs nutrient of the predictions of the herbivore optimization model.
control of marine primary producers: context-dependent – J. Ecol. 78: 180–195.
effects. – Ecology 87: 3128–3139. Hillebrand, H. et  al. 2007. Consumer versus resource control of
Burkepile, D. E. and Hay, M. E. 2008. Herbivore species richness producer diversity depends on ecosystem type and producer
and feeding complementarity affect community structure and community structure. – Proc. Natl Acad. Sci. USA 104:
function on a coral reef. – Proc. Natl Acad. Sci. USA 105: 10904–10909.
16201–16206. Hillebrand, H. et  al. 2008. Ecological stoichiometry of indirect
Burkepile, D. E. and Hay, M. E. 2009. Nutrient versus herbivore grazer effects on periphyton nutrient content. – Oecologia
control of macroalgal community development and coral 155: 619–630.
growth on a Caribbean reef. – Mar. Ecol. Prog. Ser. 389: Hillebrand, H. et  al. 2009. Herbivore metabolism and stoichio­
71–84. metry each constrain herbivory at different organizational
Cargill, S. M. and Jefferies, R. L. 1984. The effects of grazing by scales across ecosystems. – Ecol. Lett. 12: 516–527.
lesser snow geese on the vegetaion of a sub-Arctic salt marsh. Hughes, T. P. et al. 2007. Phase shifts, herbivory, and the resilience
– J. Appl. Ecol. 21: 669–686. of coral reefs to climate change. – Curr. Biol. 17: 360–365.
Carpenter, R. C. 1986. Partitioning herbivory and its effects on Jaschinski, S. and Sommer, U. 2010. Positive effects of mesograzers
coral reef algal communities. – Ecol. Monogr. 56: 345–363. on epiphytes in an eelgrass system. – Mar. Ecol.-Prog. Ser.
Carpenter, R. C. and Williams, S. L. 2007. Mass transfer limitation 401: 77–85.
of photosynthesis of coral reef algal turfs. – Mar. Biol. 151: Jeppesen, E. et  al. 1997. Top–down control in freshwater lakes:
435–450. the role of nutrient state, submerged macrophytes and water
Carpenter, S. R. et al. 1987. Regulation of lake primary producti­ depth. – Hydrobiologia 342: 151–164.
vity by food web structure. – Ecology 68: 1863–1876. Knapp, A. K. et al. 2012. A test of two mechanisms proposed to
Cebrian, J. and Lartigue, J. 2004. Patterns of herbivory and decom- optimize grassland aboveground primary productivity in
position in aquatic and terrestrial ecosystems. – Ecol. Monogr. response to grazing. – J. Plant Ecol. doi: 10.1093/jpe/rts020
74: 237–259. Lamberti, G. A. and Resh, V. H. 1983. Stream periphyton and
Cebrian, J. et  al. 2009. Producer nutritional quality controls insect herbivores: an experimental study of grazing by a
ecosystem trophic structure. – PLoS One 4. caddisfly population. – Ecology 64: 1124–1135.
Cromsigt, J. and Olff, H. 2006. Resource partitioning among McClanahan, T. R. and Shafir, S. H. 1990. Causes and conse-
savanna grazers mediated by local heterogeneity: an experi- quences of sea urchin abundance and diversity in Kenyan coral
mental approach. – Ecology 87: 1532–1541. reef lagoons. – Oecologia 83: 362–370.

311
McNaughton, S. J. 1979. Grazing as an optimization process – Shurin, J. B. and Seabloom, E. W. 2005. The strength of trophic
grass ungulate relationships in the Serengeti. – Am. Nat. 113: cascades across ecosystems: predictions from allometry and
691–703. energetics. – J. Anim. Ecol. 74: 1029–1038.
McNaughton, S. J. 1983. Compensatory plant growth as a response Shurin, J. B. et  al. 2002. A cross-ecosystem comparison of the
to herbivory. – Oikos 40: 329–336. strength of trophic cascades. – Ecol. Lett. 5: 785–791.
McNaughton, S. J. 1985. Ecology of a grazing ecosystem: the Shurin, J. B. et  al. 2006. All wet or dried up? Real differences
Serengeti. – Ecol. Monogr. 55: 259–294. between aquatic and terrestrial food webs. – Proc. R. Soc. B
McNaughton, S. J. 1988. Mineral nutrition and spatial concen­ 273: 1–9.
trations of African ungulates. – Nature 334: 343–345. Sinclair, A. R. E. et  al. 2003. Patterns of predation in a diverse
McNaughton, S. J. and Georgiadis, N. J. 1986. Ecology of African predator–prey system. – Nature 425: 288–290.
grazing and browsing mammals. – Annu. Rev. Ecol. Syst. 17: Sinclair, A. R. E. et al. 2010. Trophic cascades in African savanna:
39–65. Serengeti as a case study. – In: Terborgh, J. and Estes, J. A.
McNaughton, S. J. et  al. 1997. Promotion of the cycling of (eds), Trophic cascades: predators, prey and the changing
diet-enhancing nutrients by African grazers. – Science 278: dynamics of nature. Island Press, pp. 255–274.
1798–1800. Sommer, U. et al. 2004. Complementary impact of copepods and
Mduma, S. A. R. et al. 1999. Food regulates the Serengeti wilde- cladocerans on phytoplankton. – Ecol. Lett. 4: 545–550.
beest: a 40-year record. – J. Anim. Ecol. 68: 1101–1122. Stachowicz, J. J. et al. 2007. Understanding the effects of marine
Milchunas, D. G. and Lauenroth, W. K. 1993. Quantitative biodiversity on communities and ecosystems. – Annu. Rev.
effects of grazing on vegetation and soils over a global range Ecol. Evol. Syst. 38: 739–766.
of environments. – Ecol. Monogr. 63: 327–366. Sterner, R. W. and Elser, J. J. 2002. Ecological stoichiometry.
Moran, K. L. and Bjorndal, K. A. 2005. Simulated green turtle – Princeton Univ. Press.
grazing affects structure and productivity of seagrass pastures. Sterner, R. W. et  al. 1992. Stoichiometric relationships among
– Mar. Ecol. Prog. Ser. 305: 235–247. producers, consumers and nutrient cycling in pelagic ecosys-
Mumby, P. J. 2009. Phase shifts and the stability of macroalgal tems. – Biogeochemistry 17: 49–67.
communities on Caribbean coral reefs. – Coral Reefs 28: Strong, D. R. 1992. Are trophic cascades all wet – differentiation
761–773. and donor- control in speciose ecosystems. – Ecology 73:
Mumby, P. J. et  al. 2007. Thresholds and the resilience of 747–754.
Caribbean coral reefs. – Nature 450: 98–101. Urabe, J. et al. 2002. Herbivorous animals can mitigate unfavour-
Nugraha, A. et  al. 2010. Influence of consumer-driven nutrient able ratios of energy and material supplies by enhancing
recycling on primary production and the distribution of nutrient recycling. – Ecol. Lett. 5: 177–185.
N and P in the ocean. – Biogeosciences 7: 1285–1305. Valentine, J. F. et al. 1997. Experimental evidence that herbivory
Odadi, W. O. et  al. 2011. African wild ungulates compete with increases shoot density and productivity in a subtropical
or facilitate cattle depending on season. – Science 333: turtlegrass (Thalassia testudinum) meadow. – Oecologia 112:
1753–1755. 193–200.
Olff, H. et  al. 2002. Global environmental controls of diversity Valentine, J. F. et al. 2000. Role of sea urchin Lytechinus variegatus
in large herbivores. – Nature 415: 901–904. grazing in regulating subtropical turtlegrass Thalassia
Owen-Smith, N. and Mills, M. G. L. 2008. Predator–prey size testudinum meadows in the Florida Keys (USA). – Mar. Ecol.
relationships in an African large-mammal food web. – J. Anim. Prog. Ser. 200: 213–228.
Ecol. 77: 173–183. Verweij, R. J. T. et al. 2006. Grazing lawns contribute to the sub-
Pace, M. L. et  al. 1999. Trophic cascades revealed in diverse sistence of mesoherbivores on dystrophic savannas. – Oikos
ecosystems. – Trends Ecol. Evol. 14: 483–488. 114: 108–116.
Prins, H. H. T. and Vanderjeugd, H. P. 1993. Herbivore population Waldram, M. S. et  al. 2008. Ecological engineering by a mega-
crashes and woodland structure in east Africa. – J. Ecol. 81: grazer: white rhino impacts on a South African savanna.
305–314. – Ecosystems 11: 101–112.
Riginos, C. and Grace, J. B. 2008. Savanna tree density, herbivores Williams, S. L. 1988. Thalassia testudinum productivity and grazing
and the herbaceous community: bottom–up vs top–down by green turtles in a highly disturbed seagrass bed. – Mar. Biol.
effects. – Ecology 89: 2228–2238. 98: 447–455.
Ripple, W. J. and Beschta, R. L. 2004. Wolves and the ecology of Wirsing, A. J. et al. 2007. Fear factor: do dugongs (Dugong dugon)
fear: can predation risk structure ecosystems? – Bioscience trade food for safety from tiger sharks (Galeocerdo cuvier)?
54: 755–766. – Oecologia 153: 1031–1040.
Sandin, S. A. et al. 2008. Baselines and degradation of coral reefs Zhang, G. M. et  al. 2011. Rapid top–down regulation of plant
in the Northern Line Islands. – PLoS One 3: e1548. C:N:P stoichiometry by grasshoppers in an Inner Mongolia
Scheffer, M. et al. 2001. Catastrophic shifts in ecosystems. – Nature grassland ecosystem. – Oecologia 166: 253–264.
413: 591–596. Zieman, J. C. et  al. 1984. Herbivory effects on Thallassia
Schmitz, O. J. 2008. Herbivory from individuals to ecosystems. testudinum leaf growth and nitrogen content. – Mar. Ecol.
– Annu. Rev. Ecol. Evol. Syst. 39: 133–152. Prog. Ser. 15: 151–158.

312