How does habitat diversity affect species richness? Perhaps intuition
suggests that habitat diversity increases
species richness by facilitating niche or resource partitioning among
species. But, for a fixed
area, as habitat heterogeneity increases, the area that can be allocated to each
habitat type decreases. In a recent paper,
Allouche and colleagues (2012)
provide a theoretical and empirical treatment of the habitat
area-heterogeneity trade-off’s consequences for species richness. Both
treatments of the subject indicated that the relationship between
habitat heterogeneity and species richness may be unimodal, rather than
Conceptually, this is expected to occur when on the left side of the curve,
increasing habitat heterogeneity opens up new regions in niche space,
facilitating colonization by new species. However, as heterogeneity
continues to increase, each species has fewer habitat patches to utilize,
population sizes decrease, and local extinction risk increases
due to demographic stochasticity. To explore this idea theoretically,
Allouche et al. (2012) developed an individually based model using
a continuous time Markov process. The details of their modeling approach
can be found in
the supplementary material
to their article,
which I recommend. In this post, I’ll
demonstrate how to implement a discrete time version of their model in R.
Thanks to the agent-based modeling
working group at the University of Colorado for providing motivation to
code up model in R.
This model is spatially implicit, with A equally connected sites.
Each site falls on an environmental condition axis, receiving
some value E that characterizes local conditions. The environmental
conditions for each site are uniformly distributed between two values that
dictate the range of environmental conditions in a focal area. The
local range of environmental conditions is
a subset of some global range. There are N species
in the regional pool that can colonize habitat patches. Each species has
some environmental optimum , and some niche width ,
which together define a Gaussian function for the probability of
establishment given a colonization attempt and a habitat patch
environmental condition E.
The image above illustrates the probability of establishing for five
species across the global range of environmental conditions possible.
For any focal area, the realized
range of environmental conditions is some subset of this global range.
It is assumed that all individuals that occupy a patch have the same
of death and reproduction. If an individual reproduces, the number of
offspring it produces is a Poisson distributed random variable, and each
individual offspring attempts to colonize one randomly selected site.
At each time-step, every site has an equal probability of a colonization
attempt by an individual from each species in the regional pool. Every
habitat patch holds only one individual.
Offspring and immigrants from
the regional pool do not displace individuals from habitat patches when
they attempt to colonize. In empty sites, offspring receive
with regional colonization occurring after breeding. When multiple
offspring or immigrants from the regional pool could establish in
an empty site, one successful individual is randomly chosen to establish
regardless of species identity.
The following parameters are supplied to the function
A = number of sites;
N = number of species in the regional pool;
ERmin = global environmental conditions minimum;
ERmax = global environmental conditions maximum;
Emin = local environmental minimum;
Emax = local environmental maximum;
sig = niche width standard deviation for all species;
pM = per timestep probability of mortality;
pR = per timestep probability of reproduction;
R = per capita expected number of offspring; and
I = per timestep probability of attempted colonization by an immigrant
from the regional pool for each patch.
Implementation in R
alloucheIBM() does the majority of work for this model:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
The function returns a list containing a vector of species richness at
each timestep, the proportion of sites occupied at each timestep,
a state array containing all occupancy information for each patch,
species, and timestep, and lastly
a dataframe containing information on the niches of each species in
the regional pool.
Using this function we can begin to explore the dynamics of the model
Repeating this process a few times, we can get a picture of the
expected species richness and it’s variance for some set of parameters.
Finally, we can address the issue of habitat heterogeneity and its
effect on species richness. There are many ways to approach this issue,
and many parameter combinations to consider. Allouche et al. (2012)
provides a thorough treatment of the subject; I’ll
demonstrate just one result: that under certain conditions, species
peaks at intermediate levels of habitat heterogeneity.
To construct a range of habitat heterogeneity values, let’s construct
an interval and take subsequently narrower intervals centered around the
middle of the original interval.
1 2 3 4 5 6 7
Here are the intervals:
Now, for each interval, we can iteratively run the model and track
species richness. Because species richness tends to vary through time,
let’s take the mean of the final 100 timesteps as a measure of species
richness for each model run, and record the standard deviation to track
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Of course, the shape of this relationship is sensitive to the parameters.
As an example, changing niche width to increase or decrease niche overlap
will mediate the strength of interspecific competition for space.
Also, increasing reproductive rates may buffer each species from
stochastic extinction so that the relationship between environmental
heterogeneity and richness is monotonically increasing. Furthermore,
here I centered all intervals around the same value, but the exact
position of the environmental heterogeneity interval will affect the net
establishment probability for each site, depending on how the interval
relates to species niches. The parameter space is yours to explore.
These types of stochastic simulation models are fairly
straightforward to implement in R. Indeed there’s a package
dedicated to facilitating the implementation of such models:
simecol. There’s even a book:
A Practical Guide to Ecological Modelling: Using R as a Simulation Platform.
Allouche O, Kalyuzhny M, Moreno-Rueda G, Pizarro M, & Kadmon R. (2012)
Area-heterogeneity tradeoff and the diversity of ecological communities.
Proceedings of the National Academy of Sciences of the United States of
America, 109 (43): 17495-17500.