Describing Communities

Species Diversity Concepts

There are two important
Species Richness descriptors of a community:
Species-Area Curves
Rarefaction 1) its physiognomy
Diversity Indices (physical structure), as
- Simpson's Index described in the previous
- Shannon-Weiner Index lecture, and
- Brillouin Index
Species Abundance Models 2) the number of species
present and their relative
abundances (species
richness and diversity).

Species Richness Species Richness

The simplest way to describe a community is to list the While simply finding and listing the species is useful, this
species in it. method has many limitations.

Species richness (S) is the number of species on that list, and If we wish to compare two or more communities, we need
is most often used as the first pass estimate of diversity for a comparable samples, otherwise we might just find a
community. difference because one was sampled more intensively than
the other.
How would one generate such a list? A simple and widely
used method is to define the boundaries of the community This begs the question, how much sampling should we do
and then walk through it seasonally, noting all the species in order to be confident that we have found most of the
you encounter. This is what we call a flora. species in each community?

Species-Area Curve Species-Area Curve

One way to make this 30 After many years of study, we now realize that the
interpretation is through the
25 classical perception of the shape of a "typical" species-
use of a species-area curve.
No. of Species (S)

20 area curve is an artifact of the type of communities in

which the relationship was first described.
A graph of the total number 15

of species found as the 10

Moreover, it is now apparent, that habitat grain, patch
number of quadrats 5
size, and equitability will have a substantial influence on
increases explains the 0
the shape of the curve and must be evaluated in that
1

relationship. We know that

11

13

15

No. of 1m 2 quadrats context.

we have sampled
sufficiently when the curve Conclusion: 10-12 quadrats would
begins to "plateau". be sufficient to describe the
vegetation in this community.

1
 Species-Area Curve Species-Area Curve
Fine-grained; high equitability Fine-grained; low equitability
(classical condition)

Species-Area Curve Species-Area Curve

Coarse-grained; equal patch size Coarse-grained; unequal patch size

Grain, Patch Size, Equitability Species Richness

While many studies include S as a descriptive factor
associated with the community, it is largely
uninformative in as much as it does not reflect relative
abundance.

Example: suppose two communities (1 & 2) each contain

100 individuals distributed among five species (A-E):

A B C D E

Comm-1 20 20 20 20 20

Comm-2 96 1 1 1 1

Are these two communities equivalent?

2
 Species Richness Rarefaction

There are two ways to overcome this problem: One of the fundamental tenants of diversity is that the
number of species found in a given sample is strongly
1) incorporate both species richness and abundance dependent upon the size of that sample.
information in to one diversity index.
This makes good intuitive sense in that the more
2) rely on species richness but control for the effects of quadrats one samples in a plant community, the more
sample size by a procedure called rarefaction. likely you are to pick up more and more rare species.

We will examine both alternatives as they are widely used One method of avoiding incompatibility of
in ecology. measurements resulting from samples of different sizes
is called rarefaction.

Rarefaction Rarefaction
S
  N - Ni   N   
E(S) = ∑ 1 -   /   
The term  N  is a "combination" that is calculated as:
n
i=1   n   n     
N N!
 n  = n! N n !
Where E(S) is the expected number of species in the  
rarefied sample, n is the standardized sample, N is the where N! is a "factorial", e.g., 5! = 5×4×3×2×1 = 120
total number of in the sample to be rarefied, and Ni is
the number of individuals in the ith species in the This combination is important, because it allows us to
sample to be rarefied, summed over all species calculate all the possible numbers of unique species
counted. combinations...

Rarefaction Rarefaction

For example, if we have four species, A, B, C, D, Let's look at a fully worked real-world example (taken
then we have six species pairs: AB, AC, AD, BC, from Magurran 1988).
BD, CD.
Imagine two moth traps that have been set out in a
Using the combinatorial equation: forest to monitor moth diversity. Trap-B was
inadvertently left out for only about half the time as
Trap-A.
 4 4! 24
 2  = 2! 4 2 ! = 4 = 6
  We know this will be a problem because Trap-A
Thus  N  is the number of unique combinations of N "sampled" the environment more (longer period of
  time) and is likely to pick up more species. To compare
taken nnat a time; i.e., the number of different ways of
S between traps would be misleading and inappropriate.
picking species pairs from four different species.

3
 No. of Individuals
Rarefaction Species Trap-A Trap-B
Moth light traps are
1 9 1
typically suspended
2 3 0
above the vegetation As expected, Trap-A
3 0 1
and contain a battery has more species. The 4 4 0
powered light. The best way to correct for 5 2 0
trap is set to operate the difference in 6 1 0
only during the sampling time is to ask, 7 1 1
evening hours. Moths 8 0 2
are drawn to the light How many species 9 1 0

and become would we expect to find 10 0 5

entrapped in the in Trap-A if it too 11 1 3

contained 13 12 1 0
canister below.
S 9 6
individuals?
N 23 13

Rarefaction Rarefaction
Ni Expected
Continue this same set of 9 1.00
First, take the number of individuals of each species
calculations for each species (to 3 0.93
from Trap-A and insert them into the formula.
determine the expected number) 4 0.98
and then sum the values (as per 2 0.82
For species in Trap-A: N = 23, n = 13, Ni = 9, N-Ni = 14
the Σ in the equation). 1 0.57
1 0.57
N 23!
1 0.57
 n  = 13! 23 - 13 ! Zero values need not be
  ( ) included as they have no 1 0.57
 N-N i  14! influence on the estimate. 1 0.57
 n  = 13! 14 - 13 !
  ( ) E(S) 6.58
therefore:
  14!   23!    Conclusion: If Trap-A contained 13 individuals, we would
1 -  13! × 1!  /  13! × 10!    = {1 − [14 /1144066]} = 1 − 0.00 = 1.00 expect it to contain 6.58 species--about the same as Trap-B.
     

Diversity Indices Simpson's Index

As already alluded to, the diversity of a community needs Simpson's Index is considered a dominance index because it
(in most instances) to account for both species richness weights towards the abundance of the most common
and the evenness with which individuals are distributed species.
among species.
Simpson's Index gives the probability of any two
One way to do this is through the use of a proportional individuals drawn at random from an infinitely large
abundance index. There are two major forms of these community belonging to different species.
indices: dominance indices and information indices.
For example, the probability of two trees, picked at random
While more than 60 indices have been described, we will from a tropical rainforest being of the same species would
look at the three most widely used in the ecological be relatively low, whereas in boreal forest in Canada it
literature: Simpson's, Shannon-Weiner, and Brillouin. would be relatively high.

4
 Simpson's Index Simpson's Index

A worked Tree No.

The bias corrected form of Simpson's Index is: example for
spp. Individuals
A 100
201 trees of 5

DS =∑
(n (n − 1))
S
i i species B 50

( N ( N − 1))
i =1
assessed in
several
C
D
30
20
quadrats: E 1

where ni is the number of individuals in the ith species. Total 201

Since Ds and diversity are negatively related, Simpson's  100 × 99   50 × 49   1× 0 

DS =   +  + ...   = 0.338
index is usually expressed as the reciprocal (1-D) so that  201× 200   201× 200   201× 200 
as the index goes up, so does diversity. Then 1/D = 1/0.388 = 2.96

Shannon-Weiner Index Shannon-Weiner Index

The Shannon-Weiner Index belongs to a subset of indices The Shannon Index assumes that all species are
that maintain that diversity can be measured much like the represented in a sample and that the sample was obtained
information contained in a code or message (hence the randomly:
name information index). S
H' = -∑ pi ln pi
i=1
The rationale is that if we know a letter in a message, we
can know the uncertainty of the next letter in a coded where pi is the proportion of individuals found in the ith
message (i.e., the next species to be found in a species and ln is the natural logarithm.
community).

The uncertainty is measured as H', the Shannon Index. A

message coded bbbbbb has low uncertainty (H' = 0).

Shannon-Weiner Index Shannon-Weiner Index

A worked example from a community containing 100
trees distributed among 5 species: The most important source of error in this index is
failing to include all species from the community in the
Species Abund pi pi ln pi
sample.
A 50 0.5 -0.347
B 30 0.3 -0.361 This makes a species-area curve assessment very
C 10 0.1 -0.230
important at the beginning of a study.
D 9 0.09 -0.217
Values of the Shannon diversity index for real
E 1 0.01 -0.046 communities typically fall between 1.5 and 3.5.
Total 5 100 1.00 -1.201

H' = 1.201

5
 Shannon-Weiner Index Evenness

We can compare the actual diversity value to the maximum

The Shannon index is affected by both the number of possible diversity by using a measure called evenness.
species and their equitability, or evenness.
The evenness of the sample is obtained from the formula:
A greater number of species and a more even distribution
BOTH increase diversity as measured by H'. Evenness = H'/Hmax = H'/lnS

The maximum diversity (Hmax) of a sample is found when By definition, E is constrained between 0 and 1.0. As with
all species are equally abundant. Hmax = ln S, where S is the H', evenness assumes that all species are represented within
total number of species. the sample.

Brillouin Index Brillouin Index

When the randomness of a sample cannot be guaranteed, Species No. Individuals ln ni !

the Brillouin Index HB is preferable to the H': 1 5 4.79
2 5 4.79
ln N! - ∑ ln n i ! 3 5 4.79
HB = 4 5 4.79
N
5 5 4.79
where N is the total number of individuals and ni is the S=5 N = 25 Σ = 23.95
number of individuals in the ith species.

A worked example follows...

HB =
ln N! - ∑ ln n ! =
i ln 25! - 23.95
= 1.362
N 25

Evenness Diversity Indices

As you have probably figured out, the choice of a

Evenness for the Brillouin Index is estimated as:
particular index is chosen with respect to the goals of the
study (emphasis on abundant vs rare species) and to what
HB extent sampling can be assured to be random.
E=
H
where HBmax represents the maximum possible Brillouin There are other factors that come in to play, but these are
diversity, that is, a completely equitable distribution of the 3 most widely used measures of diversity that
individuals between species. incorporate both richness and evenness into the
determination.
In our example, we had complete equitability, therefore,
HBmax = HB = 1.0. Note: There is generally NO relationship between one
index and another.

6
 Species Abundance Models Species Abundance Models

One of the earliest observations made by plant ecologists A species abundance model is generated by graphing the
was that species are not equally common in a given abundance of each species against its rank order abundance
community. Some were very abundant, other were from 1 = highest to N = lowest.
uncommon.
One of four distributions usually arise:
A graphical way was sought to describe this pattern, and so
arose species abundance models. Log normal distribution
Geometric series
These models are strongly advocated among some ecologists Logarithmic series
because they emphasize abundance while utilizing species McArthur's broken stick model
richness information and therefore provide the most
complete mathematical description of the data.

Species Abundance Models Species Abundance Models

(Changes through succession - Bazzaz 1975)