Some theoretical notes on agrobiodiversity: spatial heterogeneity and population interactions
Diego Griffon & Maria-Josefina Hernandez
Ecological interactions are fundamental in ecological pest management,
and these interactions form networks. The properties of these networks,
where interactions of all possible nature (positive, neutral, negative)
coexist, are key for management, but little is known about them. The
main reasons for this lack of knowledge are the difficulties in
obtaining empirical evidence. These problems may be partially bypassed
using a theoretical approach. Here, by means of mathematical models that
represent networks of ecological interactions in agroecosystems, we
characterize some architectural features that promote the
self-regulation of population densities in these networks. The results
show that the key features are: spatial heterogeneity and a high
proportion of positive interactions.
Biodiversity
and spatial heterogeneity strongly benefit agricultural landscapes.
Among others, these benefits are related to population regulation of
organisms that feed on cultivated plants (Duflot et al. 2015; Fahrig et
al. 2011; Letourneau et al. 2011; Rusch et al. 2016, 2010; Tscharntke et
al. 2002, 2012; Vandermeer 1989). However, there is a need for further
theoretical development to help us understand the processes behind these
empirical observations, particularly from a mechanistic point of view.
In a very general sense, in the agroecological literature it is proposed
that increasing agricultural biodiversity involves an increase in the
number of trophic interactions of the ecological community, which in
turn promotes the stability of the whole system (Altieri 1983; Altieri
and Nicholls 2000, 2004; Nicholls and Altieri 2002). On the other hand,
we acknowledge that the ecological evidence concerning the relationship
between the number of species (richness) and the number of trophic
interactions in natural ecosystems is ambiguous (Hall and Raffaelli
1997) and, that from a theoretical point of view, the relationship
between complexity and stability is an issue far from resolved (Allesina
and Tang 2015; Bersier 2007; Ings et al. 2009; Namba 2015). However,
when it comes to contrasting a monoculture with a multidiverse
agroecosystem, these topics may have clearer answers (Griffon and
Hernández 2014; Griffon and Rodríguez 2017; Rusch et al. 2016, 2010;
Tscharntke et al. 2012).
In
a conventional monoculture, the system is explicitly designed and
managed to reduce as much as possible the unplanned associated
biodiversity(typically by using insecticides, herbicides, etc.).
Paradoxically this may contribute(among other things) to the long term
establishment of phytophagous organisms in the system, eradicating at
the same time their biological controllers (Jonsson et al. 2015; Landis,
Wratten, and Gurr 2000; Levins and Vandermeer 1990). In this type of
farming system most species are related directly to one (the
monoculture) by a victim-exploiter relationship (i.e., predation,
parasitism, parasitoidism and herbivory), where the monoculture species
(the crop) typically plays the role of the victim. So, the system has a
star-like architecture (i.e., many nodes connected to a central hub)
with the monoculture in the center (Griffon and Torres-Alruiz 2008),
which is a structure that favours the occurrence of pest situations and
crops losses (Griffon and Hernández 2014; Griffon and Rodríguez 2017).
Alternatives
to the ecological oversimplification of monocultures are companion
crops. One of the aims of this cropping strategy is the population
regulation of phytophagous and phytopathogenic species by means of
ecological interactions (Altieri and Nicholls 2004). The success of this
approach not only depends on the occurrence of a more complex trophic
web, but also on the occurrence of other ecological interactions
(competition, mutualism, amensalism and commensalism) that together make
up the ecological network system, i.e., a network consisting of all
types of ecological interactions (Ings et al. 2009). We have very little
information on the structure of ecological networks (Pocock et al.
2012) and we also lack knowledge on how ecological networks promote the
regulation of phytophagous population densities. Given the need of
information, coupled with the difficulty and effort involved in
achieving it in the field, this paper addresses the issue from a
theoretical perspective. In order to do this, we build and numerically
evaluate mathematical models that simulate networks of hypothetical
ecological interactions associated with agricultural ecosystems. This is
done with the objective of finding patterns that can provide guidelines
on architectural features associated with self-regulation in
populations.
Another
related topic must be considered. There is abundant field information
that shows the positive effects of spatial heterogeneity (Batáry et al.
2011; Duflot et al. 2015; Fahrig 2013; Fahrig et al. 2011, 2015; Jonsson
et al. 2015; Landis, Wratten, and Gurr 2000; Rusch et al. 2016;
Tscharntke et al. 2012; Tuck et al. 2014) on the maintenance of
associated agricultural biodiversity (sensu Vandermeer and Perfecto
1995; Altieri et al. 2005). In some cases, space heterogeneity may even
play a more important role than intra-farm diversity in the regulation
of phytophagous population densities (Fahrig et al. 2011, 2015). But
surely the two components (intra and inter farm diversity) relate
synergistically.
For
the spatial heterogeneity to have a positive effect on the internal
dynamics of agroecosystems it is necessary, on the one hand, an
insidefarm design that attracts biological controllers (e.g., flower
strips or beetle banks) (Altieri, Ponti, and Nicholls 2005; Nicholls and
Altieri 2002) and on the other, the existence of nearby sources of
organisms with enough internal complexity to provide the necessary
control agents (Rusch et al. 2016, 2010; Tscharntke et al. 2012).
So,
metapopulation and metacommunity dynamics seem to be crucial for the
long term survival of species in heterogeneous environments (Alfonzo et
al. 2009; Aberg et al. 1995; Cantrell, Cosner, and Fagan 1998; Delin and
Andren 1999; Griffon, Alfonzo, and Hernandez 2010; Griffon and
Hernández 2014; Gustafson and Gardner 1996; Perfecto, Vandermeer, and
Wright 2009; Sisk, Haddad, and Ehrlich 1997; Tejat et al. 2002;
Vandermeer and Carvajal 2001; Vandermeer and Perfecto 2007). In general
terms, the spatial structure of populations, along with processes of
dispersal, migration and colonization, allows the emergence of dynamics
that make possible the persistence and coexistence of species (Hanski
1994, 1998; Hanski and Gilpin 1997; Hanski et al. 1996; Leibold et al.
2004). Thus, spatial heterogeneity may enhance the configuration of the
complex ecological networks needed for a successful ecological pest
management program (Batáry et al. 2011; Fahrig et al. 2011; Rusch et al.
2016, 2010; Tejat et al. 2002; Tscharntke et al. 2002, 2012). For this
reason, in the mathematical approach used here we also include the
effect of spatial heterogeneity on population dynamics.
In
short, the objective of this work is to find architectural features
that promote the self-regulation of population densities in ecological
networks associated to agroecosystems. To do this, we built mathematical
models that represent ecological networks, both for a single community
and for metacommunitarian systems. We must make clear that whenever we
say ‘ecological network’, we are considering the potential presence of
‘all’ types of interactions, i.e., competition, mutualism,
victim-exploiter, amensalism and commensalism.
Species
survivals. Initial richness, for six different initial conditions
(defined in Fig 3). Blue: survival percentage in one community (no
spatial heterogeneity). Grey: survival percentage of one community in a
metacommunitarian background. The curves are averages of 40 simulations
for each initial condition.
Uniform
perturbation (example). A persistent network obtained under the
20:10:10:60 initial condition is perturbed by a little increase in the
densities of each population. Left: dynamics after the perturbation. Two
populations reach very low (but non zero) densities. Right: network
before and after the perturbation. Notice that both networks are the
same.
The
more relevant results of the models evaluated and discussed in this
article can be summarized as follows: (i) The conditions under which
persistent networks are obtained after the iterative process are very
restricted. However, when persistent networks are obtained, they are
fundamentally resilient to perturbations. (ii) Mutualistic (and positive
in general) interactions have an important and extensive effect under
certain (very specific) conditions. (iii) Spatial heterogeneity
increases the possibility of persistence in hypothetical communities.
(iv) Ecological interactions that somehow have been neglected in the
past (commensalism and amensalism, the forgotten sisters), may be: 1-
More frequent than generally thought, and 2- Important for the
persistence of communities.
.
