Transcript Probability metric using null model

Spatial Turnover and Beta Diversity:
What causes species turnover and why
is this important from a conservation
standpoint?
Amanda Senft
Bio 255
Beta Diversity (species turnover
through space)
Previously, we discussed how to measure beta
diversity…. but what processes are we
measuring? In other words, what causes beta
diversity?
When we conserve areas of high beta diversity
(biogeographic crossroads), are we conserving:

Areas of high species turnover?
OR

The processes causing this high species turnover?
Whittaker (1960)
NON - GRADIENT
β=γ/α
GRADIENT
Half changes
Similarity v. distance
Harrison
Cowling
Wilson & Schmida
Rey Benayas
Cowling
Porembski
Willig & Sandlin
Harrison
Blackburn & Gaston
Nekola & White
Oliver
Cody
Young
Syuzo
Pharo…..
Givnish…..
Slide stolen from Meghan
with an H
What causes beta diversity?
From Duivenvoorden et al 2002
Dispersal limitation
I= # seeds per area
P= probability that a seed predator won’t get the seed
PRC=population recruitment curve: distance from adult that its offspring is
likely to appear
From Janzen 1970
Climatic and Edaphic factors?
How did Condit et al consider beta
diversity?
Similarity in species composition as a
function of distance.
All pairs of plots considered together.
Spatial heterogenity (e.g. ecotones) not
considered in null model
Tested Dispersal Limitation as a major
driving force
How did Spector consider beta
diversity?
Caused by environmental factors:
geological, climatic, geometric features
Preserved by evolutionary factors:

Ecotones are “evolutionarily active
zones”…see Enserink 1992
and by “underlying processes”

Fire or another ecological gradient
But…considered only a single scale
Today’s papers…
Condit, R., Pitman, N., Leigh, E. G., Chave, J.,
Terborgh, J., Foster, R. B., Nunez, P., Aguilar, S.,
Valencia, R., Villa, G., Muller-Landau, H. C., Losos,
E., and S.P Hubbell 2002. Beta-Diversity in
Tropical Forest Trees. Science 295(5555): 666669.
Spector, S. 2002. Biogeographic Crossroads as
Priority Areas for Biodiversity Conservation.
Conservation Biology. 16(6): 1480-1487.
How do you measure ecological
distance between 2 sites?
Distance metrics: Similarity between plots
A
w
B
Sorenson 2w / A+B : Shared abundance over total
abundance
Jaccard w / A+B-w: Proportion of combined abundance
not shared
Other distance measures they could have used:
Euclidean, Correlation distance, Mahalenobis, Chisquare
Probability metric using null model:
Probability: F=Σfi1 f j2 The sum of the abundance of a spp
at site 1 times abundance at site 2 for all species.
Probability that two trees chosen randomly from 2 plots
are of the same species. Takes into account relative
abundance of species.
Condit et al. Figure 1
In Panama, plots within 1 km very similar, but
very different from plots 50 km away
In Peru and Ecuador, plots within 1 km nearly
as similar as plots 50 km away
Panama: Species turnover due to
environmental factors: climate, geology, soil
Peru and Ecuador: Species turnover due to
dispersal limitation
Null model based on a Bessel
function of decay:
Null hypothesis: Only dispersal and
speciation affect spp distributions.
Used Hubbell’s null model: all spp
identical in death, reproduction,
dispersal, speciation=extinction, etc.
Probability that 2 individuals r
kilometers apart are the same species
is a function of distance, speciation,
dispersal distance and tree density.
Condit et al Figure 2
Graphs showing
experimental (points) vs.
theoretical decay of species
similarity with distance
- Variance is higher for
Panama
-Decay in similarity from 0100 m not well simulated by
model
Condit et al Figure 2
Graphs showing
experimental (points) vs.
theoretical decay of species
similarity with distance
- Variance is higher for
Panama
-Decay in similarity from 0100 m not well simulated by
model
Condit et al Figure 2
Graphs showing
experimental (points) vs.
theoretical decay of species
similarity with distance
- Variance is higher for
Panama
-Decay in similarity from 0100 m not well simulated by
model
Conclusions from Condit paper:
Species more aggregated than dispersal
theory predicts
Abundant species not well modeled with
dispersal theory (they’re just everywhere,
regardless of distance)
Null model of dispersal works well for
distances .2-50 km, but other factors
influence beta diversity at different scales
Duivenvoorden et al: Variance
partitioning
Using Condit’s data, found that (linear) distance,
elevation, precipitation, stand age and bedrock
type were all significant predictors of species
similarity.
Still, distance and environment together
explained left 59% of floristic variation
unexplained
A re-reanalysis using presence-absence data
instead of abundances, and log distances, was
able to leave only 41% of data unexplained
(Ruoklainen and Tuomisto, 2002).
Spector paper
We should conserve
“biogeographic crossroads” where
there is high species turnover:


Areas meet goals of
representativeness and
complementarity
Conserve evolutionary processes:
speciation and coevolution
Congruence assumed to be high:

Environmental factors act on most
taxa to create zones of high turnover
Scarab beetle
How have others considered
biogeographic crossroads?
Clements 1905: Tension zones where
principle species from adjacent
communities meet
Delcourts (numerous): shifting ecoclines
Sky Islands (stacked biotic
communities)…
Marshall 1957 cited by Warshall 2003
Spector’s Caveats
Ecotones harbor species at the
edges of their ranges, which are
often more fragmented, more
extinction prone (is this true or
not?)
Park size should equal the size of
the ecological gradient and
processes maintaining the ecotone.
How does one quantify this?
Discussion questions…
Do “biogeographic crossroads” exist? At what scale?
Given that ecotones often mark the edge of a given
species’ range, should we mark areas of high species
turnover for conservation? Consider also that ecotones
are predicted to be highly sensitive to climate change
(Turner, Gardner, and O’Neill 1991).
What caused the differences in diversity patterns
between Panama and the two South American sites?
Which variables predicted beta diversity in the Tropics?
How transferable are those results?
Are these “biogeographic
crossroads?”