Assessment of habitat specialization of Southeastern trees using largeextent cooccurrence data

1
Assessment of habitat specialization of
Southeastern trees using large-extent
co-occurrence data

• David B. Vandermast, Jason D. Fridley, Dane
  Kuppinger, and Robert K. Peet
• University of North Carolina at Chapel Hill

2
How do we characterize habitat generalists and
specialists?

• In ecological literature species tendencies are
  commonly characterized by range of habitat use,
  particularly with respect to gradients

habitat specialist
habitat generalist
Abundance
Environmental gradient
3
Habitat generalism vs. specialization without
reference to gradients?

• A species can be a generalist along one gradient
  but specialist along another
• In theory, impossible or not feasible to measure
  all relevant gradients
• Why not let patterns of species co-occurrence
  reveal habitat generalists and specialists?

4
New approach Use large-extent species
co-occurrence data as a biological assay of
habitat specialization

• Specialists occur in few habitats. Therefore
• Compositional turnover within plots containing a
  specialist should be low.
• Generalists occur in many habitats. Therefore
• Compositional turnover within plots containing a
  generalist should be high.
• How to quantify species-centered compositional
  turnover?

5
How to calculate species-centered turnover

• Whittakers additive partition of diversity ß is
  species turnover among plots
• ? ?-?(?)
• where
• ? the cumulative species among plots, and
• ?(?) mean plot species richness

6
What the Species ß metric does

• Take all plots of a focal species
• Generalists should occur with more species over
  their range.
• All else equal, generalists should have higher
  gamma and beta values.
• BUT there are two ways this could happen NOT
  associated with turnover among plots
• 1. If a species occurs in a particularly
  species-rich habitat (a high associated alpha
  diversity). Removed by partition.
• 2. If a species is well sampled relative to its
  overall abundance in the region (a high plot
  frequency). Removed by selecting a constant
  number of plots (50) for each species, taking the
  mean of 1000 replicates.

7
Large extent co-occurrence data of the Carolinas

• Carolina Vegetation Survey (CVS) database of
  plots throughout North and South Carolina, and
  Georgia
• Search limited to trees 10 cm
• 2500 plots containing 112 tree species
• Use of large woody flora
• trees used as habitat indicators
• fewer species and co-occurrences and more life
  history data than herbs

8
CVS Plot locations
9
Results were consistent with predictions

• Common wide-ranging understory trees (American
  holly, ironwood) have among the highest ß values
• Species restricted to few habitat types (longleaf
  pine, pond cypress) have among the lowest ß values

10
5 highest and 5 lowest ß values
11
Meaning of species-centered ß
In a random sample of 50 plots containing
American holly, it co-occurs with (on average)
108 species, after subtracting for mean plot
richness (14 species).
12
Meaning of species-centered ß
In a random sample of 50 plots containing
American holly, it co-occurs with (on average)
108 species, after subtracting for mean plot
richness (14 species).
Compare specialist like pond cypress...
13
ß is correlated with plot frequency
14
ß is not correlated with µ(?) after a low
threshold
15
Is Species ß robust?

• If we resample using a smaller number of plots,
  or a geographic subset of the data, will species
  be similarly distributed along the
  generalist-specialists gradient?

16
Random subsets of data generate same relative ß
values
17
Regional data subsets are similar but yield
interesting exceptions
18
Is ß correlated with species life-history traits
and environmental ranges?

• Using primary sources
• Climate change website (Iverson and Prasad)
• USDA PLANTS database
• Radford, Ahles, and Bell 1968
• We regressed species ß against many variables
  describing life history traits and environmental
  tolerances

19
Habitat generalists were strongly associated with

• Life history traits
• Deciduousness
• Shade-tolerance
• Short lifespans
• Bird-dispersed seed
• p
• p
• p

• Environmental range
• Annual temperature
• Potential evapotranspiration
• Soil pH
• soil organic matter

20
Summary

• Use of additive partition of species richness and
  co-occurrence data from a large-extent database
  appears to be a robust method for placing species
  along a continuum of habitat generalism vs.
  specialization
• Our data indicate certain life history traits and
  environmental ranges are strongly correlated with
  species generalism

21
Acknowledgments

• 600 CVS participants since 1988
• CVS supported by NSF
• UNC Plant Ecology Lab
• Peter White, UNC
• Tom Wentworth, NCSU