morphological transformation into bacteroids. Rhizo-
bia can fix nitrogen only in this transformed state,
which is reached through a terminal development
event during which bacteroids may lose the ability
to reproduce (Zhouet al. 1985). Loss of reproductive
viability is generally prevalent in the nodules of the
indeterminate types, such as nodules found on peas
and alfalfa. In such cases it is believed that the rhizo-
sphere populations are replenished by the undiffer-
entiated infection threads (Denison 2000). In
determinate nodules, such as in soybean, reproduc-
tive viability of the transformed bacteroids is high
(Sutton and Paterson 1980), and it is thought that
bacteroids remain viable after the nodule senesces.
Exceptions to these generalizations exist, and more
research is needed to understand the processes
that control reproductive viability of rhizobia in no-
dules.
One of the most intriguing aspects of the symbi-
osis is how the relationship has persisted for
millions of years. Rhizobia expend an incredible
amount of their energy in fixing nitrogen and sup-
plying it to their host (Gutschick 1981). This is
thought to incur considerable costs in terms of re-
productive fitness for the rhizobia (Denison and
Kiers 2004a). It is true that, by supplying its host
with nitrogen, an individual rhizobium enhances
host photosynthesis, thereby potentially increasing
the rhizobium’s own access to photosynthate. How-
ever, we also know that plants are typically infected
by more than one strain of rhizobia (Hagen and
Hamerick 1996; Westet al. 2002b and references
within). This means that rhizobia that supply their
host with nitrogen may indirectly benefit compet-
ing strains of rhizobia infecting the same individual
plant. As the number of strains per plant increases,
evolutionary theory predicts a rise in symbiont par-
asitism (Smith and Szathmary 1995). The situation
is analogous to the classic tragedy of the commons
problem from human economics (Hardin 1968).
The tragedy is that ‘free-rider’ rhizobia, those that
cheat by extracting carbohydrates from the host
while fixing little to no nitrogen, are predicted to
spread at the expense of efficient nitrogen-fixing
strains (Denison 2000; Kierset al. 2002; Denison
et al. 2003). The problem of hosting multiple part-
ners is a recurring theme in rhizosphere mutual-
isms (Kiers and Denison 2008). The question is,
why do rhizobia expend resources on fixing nitro-
gen for the benefit of their host plant (perhaps also
indirectly benefiting competing rhizobial strains)
when they could use those resources for their own
reproduction (Denison 2000; Westet al. 2002a, b)?
Bever and Simms (2000) proposed a model in
which nitrogen-fixing bacteroids in the nodules
would convey reproductive benefits to their free-
living kin in the rhizosphere. They hypothesized
that, because rhizobia reproduce largely by asexual
fission, genetically identical kin of the bacteroids
inside the nodule could be found directly outside
the nodule, when there was little mixing in the soil.
Thus, although a bacteroid may have sacrificed its
own reproduction in the process of fixing nitrogen,
the transformation could be evolutionarily advan-
tageous if the benefits from enhanced exudates of
the host could increase the fitness of its related kin
in the rhizosphere. This was hypothesized to offset
the loss in reproductive capacity of the bacteroid
(Bever and Simms 2000).
There are three flaws with this reasoning. First,
the bacteroid form is not a reproductive dead-end
for all rhizobia, as explained above. Their hypothe-
sis would apply only to those rhizobia in indeter-
minate nodules. Second, this hypothesis is
inconsistent with the way natural selection operates
in nature. The rhizosphere surrounding the host
plant root may be colonized by kin, but it is also
colonized by rhizobia of many different lineages,
some of which may not even be fixing nitrogen
(Denison and Kiers 2004b). Yet these ‘cheaters’,
even though unrelated to those in the bacteroid,
will also benefit by the increases in exudates from
the altruistic sacrifice of the transformed bacterium.
Third, even if strong spatial structuring in the soil
(no mixing) increased the chances that the exudates
would be directed to kin, increased spatial structure
also increases competition between relatives,
making competition more local. These contrasting
effects tend to balance each other out (Westet al.
2002b). Therefore, spatial structuring of soil popu-
lations will not necessarily favour greater mutual-
ism (see Denison and Kiers 2004a).
In contrast, preferentially allocating resources di-
rectly to cooperative rhizobial strains represents a
stable evolutionary strategy (Westet al. 2002a, b). If
host plants are able to discriminate among nodules,186 FUTURE DIRECTIONS