Transcript Macroinvertebrate Communities in Ephemeral Ponds and the Effects

Macroinvertebrate
Communities in Ephemeral
Ponds: Effects of Competition,
Landscape, and Hydroperiod
on Species Richness
Edmund Hart
University of Vermont
ESA 2007
What is an Ephemeral Pond?
• A landscape depression regularly filling in either the fall
or the spring.
• Rarely hold water for more than 4 - 5 months after
spring ice out, but can be flooded for multiple years.
• Range in size from 68 – 2941 m2
• Regular drying and no inflow or outflow prevents the
establishment of fish populations.
Ponds Undergo Regular Drying
Common Ephemeral Pond Taxa
Community assembly
Communities reform each year with some
taxa overwintering, and others colonizing in
the early spring.
• Only certain taxa can exploit ephemeral habitats.(Wiggins et al 1980,
Williams 1997).
• Have resting eggs (cladocerans, Branchiopods) or
dessicant resistant life stages.
• Rapid development (mosquitoes, other Diptera).
• Early colonizers from nearby permanent habitat
(Hemiptera, Dytiscidae).
• Wissinger and Gallagher (1999) found between 63 and 71% of
post-drought insect taxa were from dessicant resistant stages.
Schematic of community
structure.
Hydroperiod
.2
.3
.2
.6
.5
.4
.2
.9
.1
.4
.8
.7
.3
.6
.1
.1
.8 .3
.9
.5
Model drawn after Schnieder and Frost (1996)
.1
.2
= Taxa that can
utilize ephemeral
habitat
= Taxa already present
In habitat
= Potential colonizer
Study question
Do competitive interactions with salamander larva
Or abiotic habitat variables determine invertebrate
species richness?
Study Location
Delaware
Water Gap
National
Recreation
Area
•
67,000 acres in New Jersey and Pennsylvania
bordering 40 miles of the Delaware river.
•
Includes two large ridges on either side of the river
valley and numerous tributaries.
Experimental Design
I. Two independent factors crossed, third added post-hoc
a)
Open or closed canopy
b)
Ambystoma spp. larva present or absent
c)
Third factor added, late or early drying
d)
One site for each of the initial factors, 4 total sites
Experimental Design
II. Sites sampled every other week from 3/25/04 to
6/25/04 or until dry
a)
Three 1-Meter dipnet sweeps taken and then pooled, and
picked for 10 minutes to make 1 composite sample
b)
Three composite samples taken per visit.
c)
All three samples pooled into a single species richness count
per sampling date
Model Development and Analysis
Data was analyzed using the lmer function for mixed
models in the lme4 package (Bates and Sarkar 2007) of R
(R Development Core Team 2007).
a)
Calculated Aikake’s Information for small
sample sizes (AICc) (Burnam and Anderson 2002)
b)
Calculated Aikake weights,
c)
Calculated model selection frequencies and
95% confidence intervals using 10,000
bootstrap replicates
Candidate Models
Model
Interpretation
Null
No biological interactions at work
Non-forested sites have greater abundance or species
richness
Sites with longer hydroperiod have greater abundance or
species richness
Competition with salamander larva structures species
richness or abundance
Ponds with open habitat and long drying period will have
the highest abundance or species richness
Salamander comp etition and habitat together determi ne
species richness
Salamander comp etition and hydroperiod together
determine species richness
Habit at
Hydroperiod
Competition
Habit at and
Hydroperiod
Habit at and
Competition
Competition and
Hydroperiod
Since design was not fully crossed, a saturating likelihood
results when too many factors are added into the model.
Therefore interaction terms could not be considered, only
additive effects of two parameters.
Results
Species Richness by Treatment
Results
Individual based rarefaction curves generated by
EstimateS (Colwell 2005). Open sites have higher
species diversity.
Model Results
Model
Null
Habit at
Hydroperiod
Competition
Habitat and
Hydroperiod
Habit at and
Competition
Competition
and
Hydroperiod
Log-Likelihood
Lower
95% CI
AICc
Upper
95% CI
Delta
AICc
AICc
Weights
Bootsrap
Selection
Frequencies
-21.68
-19.31
-21.25
-21.46
47.9
45.76
49.64
50.06
33.34
29.59
34.89
35.56
53.46
52.06
55.17
55.46
3.1
0.96
4.84
5.26
0.091
0.265
0.038
0.031
0.013
0.153
0
0
-17.4
44.8
26.24
51.79
0
0.429
0.68
-18.57
47.14
30.84
53.82
2.34
0.133
0.154
-20.96
51.92
37.33
57.21
7.12
0.012
0
The model with habitat and hydroperiod predictors
best fits the data, also having the highest AICc
weight and bootstrap frequency.
Conclusions
• The strongest predictor of species richness was abiotic
habitat variables.
• Open habitat was a strong predictor on its own.
•Open habitats could be better quality habitat (Tarr
et al 2005)
•Open habitats are more detectable by aerial
colonizers (My own wild rank speculation)
• Competitive interactions with Ambystoma larva have little
impact on macroinvertebrate species richness.
•Agrees with Corti et al’s (1997) prediction of low
effects of predation on highly disturbed systems
•Also larval densities probably didn’t reach high
enough levels
Questions?
Worms sure are
tastier than bugs
Thanks DEWA staff,
Rich Evans,
USGS BRD
Brian Beckage,
Nick Gotelli