Transcript Poster - Biology Department | UNC Chapel Hill

Edge Width of Forest Fragments in North Carolina Agroecosystems
Amanda R. Senft, Jason D. Fridley and Robert K. Peet
Results
2) Species Turnover
We did not find a significant change in species richness from forest edge to interior at
any scale. Thus, we could not determine edge width based on species richness.
N.S.
Upper Confid. Limit
Lower Confid. Limit
N.S.
Large
Med
X
1 m2
100 m2
10 m2
-4
Distance to Edge (m)
N.S.
-1
Our method of quantifying edge width was unable to
detect a significant change in species turnover from
edge to interior. However, there is an interesting
explanation for this. The pattern of turnover from edge
to interior is triphasic: that is, there is a significant
change in species from 0 to 50 meters into the forest,
there is a period of relative stability, and then another
change in species composition. An opposite trend is
shown at the smallest, 1m2 scale.
Edge Width Based on Species Richness
40
35
30
25
20
15
10
5
0
Pattern of Species turnover based on Jaccard Similarity
-2
1) Species richness:
-3
The long term sustainability of a habitat patch in a fragmented landscape is dependent on factors related to
its size, shape and context. Important determinants of patch size and viability are the width, sharpness and
influence of its edge. However, the choice of the variable measured can have a profound effect on
perceived edge width. We set out to quantify edge width by using a multi-metric, multi-scalar approach. We
measured change in species turnover, species richness, richness of invasive exotics, and floristic
composition along 25 forest-farm field edges all located in piedmont agricultural landscapes of the North
Carolina Piedmont. We surveyed vascular plant richness and percent cover in a series of nested quadrats
varying in scale from 1 to 100 m2 and arrayed in a series of transects. Transects extended from the edge of
the agricultural field, across the field-forest boundary, and up to 200 meters into the forest interior. All
measurements of edge width were variable and scale dependent. In general, edge width is most clearly
defined floristic turnover. Our results highlight the need for multiple characterizations of edge width and
habitat in the assessment of how species distributions are influenced by habitat fragmentation.
Log(Jaccard's Similarity)
Abstract
Small
Scale
0
50
100
150
200
250
Distance to Edge
Introduction
Edge Width Based on Exotic Species
40
30
20
10
0
Med
20
10
N.S.
5
Large
100 m2
10
5
Large
Med
Edge Width Based on Trees
30
25
20
N.S.
15
N.S.
10
5
N.S.
60
N.S.
50
40
30
20
10
0
0
1. Edge width is relatively consistent for the variables tested. We found a significant
edge effect for exotics, field species, graminoids and trees. The edge width for all of
these variables ranged only 10 meters for the largest scale and 30 meters for the
smallest scale.
2. Edge width is dependent on scale. There was as much variation between scales as
there was between variables, indicating that scale is an important consideration in any
study of edge width. We also found a random pattern of composition at the smaller two
scales, resulting in non-significant edge effects. This suggests that scales smaller than
100 m2 are not as useful for studying compositional change.
3. Edge width determined in this study is lower than currently accepted values in the
literature. The variables that we chose, most based on vegetation composition, suggest
that edge effects do not penetrate much further than 20 meters into the forest. This
contradicts previous studies of edge effects, which report average edge widths of 50
meters and maximums of hundreds of meters.
Small
Scale
Edge Width Based on Graminoids
Discussion and Conclusions
10 m2
15
0
Small
Med
Scale
Field Species were defined using Indicator Species Analysis (Dufrene and Legendre 1997) in PC-ORD (McCune and Mefford 1999). The 100m2
plots were classified by distance from the edge; those located completely within the field were given negative distances. We sorted the species
according to the distance group for which they had the highest indicator value. To test for statistical significance, indicator values are compared
to those of species randomly assigned to groups. This Monte Carlo procedure was run 1000 times to derive a p-value indicating whether a
species occurs more frequently in a group than it would by chance given its relative frequency among all plots. All species with a negative
distance (field plots) that had a significant p-value were labeled as field species.
1 m2
Med
Large
Before data analysis, we used Kartesz and Meacham (1999) to classify species as trees, shrubs, forbs, herbs, or graminoids, and native or
exotic. Nitrogen Fixers were identified using the USDA Plants Database (USDA 2004) as well as primary literature (Allen and Allen 1981).
Species turnover was quantified as the log of Jaccard’s similarity for each plot from the first plot in its transect.
20
Scale
4) Physiognomic turnover:
We arranged plots along transects reaching into the mixed hardwood forests in the Northern Piedmont region of North Carolina. We set up 17
transects directly adjacent and perpendicular to the edge between these forests and agricultural fields and ran them up to 300 meters into the
forest interior. At each transect, we surveyed vegetation using a modified version of the Carolina Vegetation Survey protocol (Peet et al 1998), a
nested sampling design which tallies species identity and richness at multiple scales. Each plot was surveyed at three spatial scales:
100m2,10m2, and 1m2, known in the rest of this text as large,medium and small (Figure 1).
25
0
Small
There is a significant edge width of 20 meters determined at the large scale for both
trees (an increase in trees from edge to interior) and graminoids (a decrease in
graminoides from edge to interior). No signifcant trend was determined for either the
medium or small scales for these traits. Also, no significant trend was determined for
proportion of forbs, vines or shrubs.
N.S.
N.S.
15
Scale
Methods
Small
We quantified edge width using breakpoint regression. A sharp bend in a loess regression line using
aggregated data can yield poor estimates if individual transects are heterogeneous. A superior method is to
Med
use hierarchical breakpoint regression, an alternative to the commonly used "Helmert Procedure". In this
technique, if there is an edge effect, the regression line between distance and the trait of interest will have a
nonzero slope through the edge and will level off in the forest interior. Edge width is the point at which we
switch between these two pieces of the regression. Each individual transect regression coefficient is treated
Large
as a random deviation from an underlying population value in a nonlinear mixed-effects model. Point
estimates and 95% profile-likelihood based confidence intervals of the population edge width are reported.
Figure 1: Scales of
All statistical analyses were performed in S+, Version 6.1 and R, Version 1.9.1.
Sampling
50
30
25
Distance to Edge (m)
60
Large
Edge Width Based on Field Species
Edge Width Based on Nitrogen Fixers
Distance from edge (m)
We hypothesize that edge width will vary with scale and with the
following variables:
1. Species richness: Species richness should be higher at the edge
than in the interior, as there is both a blending of two communities as
well as unique habitat for early successional and edge preferring
species.
2. Species turnover: Species turnover is predicted to be higher at the
edge because of the rapid change in environmental variables.
3. Proportion of exotics, field species and nitrogen fixers: There
should be a higher abundance of nitrogen fixers and early successional
field species closer to the edge, where there is available light. In
addition, there should be more exotics near the edge, where propagule
pressure and light availability are higher.
4. Physiognomic turnover: When moving from the edge to interior,
there should be a higher proportion of trees, a lower proportion of forbs
and grasses and a lower proportion of vines.
Distance to Edge (m)
Edge effects have been assessed in many different
ways, at many scales of observation and using many
different variables. The purpose of this study is to
examine several of these variables at multiple
scales and to quantify the change in perceived
edge width with each perspective. An additional
purpose of this study is to propose a novel, statistically
robust method of quantifying edge width.
There is a significant, consistent edge width based on the proportion of exotics. Both the large and medium scales have distances around 15 meters
while a slightly longer distance of 40 meters is determined from the smallest scale. There is also a significant, consistent edge width based on
proportion of field species. A short distance of 24, 12 and 8 meters is determined from the data taken at large, medium and small scales,
respectively. We were unable to determine edge width from the proportion of nitrogen fixers. All scales indicate an insignificant trend of nitrogen
fixers from edge to interior.
Distance from edge (m)
The Piedmont region of North Carolina has been
farmed intensively since European settlement,
partitioning the once contiguous mixed evergreenhardwood forests into small fragments. Forest
fragments in agricultural landscapes have been
relatively understudied. However, in order to conserve
and manage the remaining forest fragments, it is
essential to understand the effects that exterior forces
exert on the boundary of these fragments. The
changes in vegetation pattern resulting from these
exterior conditions are termed edge effects.
3) Proportion of exotics, field species and nitrogen fixers:
HYPOTHESIS
Distance from Edge (m)
BACKGROUND
Small
Large
Med
Small
Scale
Acknowledgements
Field Crew: Alenna Clements, Joel Gramling, Todd Jobe,
Annika Dollander, Matt Kudla, Jenny Mayer, Eric
Fridley, Christina Zakas, and Jessica Long.
Financial Support: UNC Department of Biology,
Syngenta Corporation
Statistical Advice: Jack Weiss
Literature Cited
Allen, O.N. and E.K.Allen. 1981. The Leguminosae. The
University of Wisconsin Press. Madison, WI.
Dufrene, M. and P. Legendre. 1997. Species assemblages and
indicator species: the need for a flexible asymmetrical approach.
Ecological Monographs 67:345-366.
Kartesz, J.T. and C.A. Meacham. 1999. Syntehsis of the North
American Florida. Version 1.0. North Carolina Botanical Garden.
Chapel Hill, NC.
McCune, B. and M.J. Mefford. 1999. PC-ORD. Multivariate
Analysis of Ecological Data Version 4.27. MjM Software Design,
Gleneden Beach, Oregon.
Peet, R.K., Wentworth, T.R. and P.S. White (1998). A Flexible,
Multipurpose Method for Recording Vegetation Composition and
Structure. Castanea 63(3): 262-274.
USDA, NRCS. 2004. The PLANTS Database, Version 3.5
(http://plants.usda.gov). National Plant Data Center, Baton
Rouge, LA 70874-4490 USA.