Deborah E. Goldberg, Tara Rajaniemi, Jessica Gurevitch, and Allan Stewart-Oaten.
1999. Empirical approaches to quantifying interaction intensity: competition
and facilitation along productivity gradients. Ecology 80:1118-1131.
Supplements
Supplement 1: Data on interaction
intensity among plants for a meta-analyses of patterns in competition intensity
along productivity gradients.
Ecological Archives E080-006-S1.
Author
File list
Description
Reference list meta-analysis
Citations
Copyright
Author
Deborah Goldberg
Department of Biology
University of Michigan
Ann Arbor, MI 48109-1048
TEL: 734 764-1490
FAX: 734 647-0884
(degold@umich.edu)
File list
Viewable Supplement
Data files
metadata.txt
is a text file containing these metadata.
varcodes.txt
is a text file describing the variables and codes for the data file.
metacomp.csv
is the data file (comma-delimited ASCII)
citation.txt
is a text file with the full citations that correspond to the reference
numbers listed in. metacomp.csv
E080006.zip
is a zip file containing all supplemental files.
Description
Abstract: This supplement contains data on interaction intensity extracted
from the primary literature for a meta-analyses of patterns in competition intensity
along productivity gradients in plants. Goldberg et al. (1999) report on analyses
using a subset of these data and of the possible relationships among variables.
Research period and location: This work was conducted as part of the
Meta-analysis Working Group (1996-1997, Meta-analysis, interaction strength
and effect size: application of biological models to the synthesis of experimental
data) supported by the National Center for Ecological Analysis and Synthesis,
a Center funded by NSF (DEB-94-21535), the University of California - Santa
Barbara, and the State of California.
Explanation of variables and codes in the supplement (download
as varcodes.txt)
Note that variable names are written in bold and possible code values the variable
may take are written in italics.
General terminology: Targets are the plants whose response to interactions
are being measured, neighbors are those plants that are assumed to be affecting
the targets and are manipulated.
Criteria for inclusion in the database: The studies were chosen from
those cited in four previous reviews of plant competition: Goldberg and Barton
(1992), Gurevitch et al. (1992), Goldberg (1996), and Goldberg and Novoplansky
(1997) according to the following criteria. All experiments included in the
database were conducted in naturally-occurring vegetation in the field and have
at least the following two treatments to quantify interaction intensity: a control
in which no neighbors were removed and a removal in which roots and/or shoots
of all naturally-occurring neighbors within some distance were removed or assumed
to be dead. All also have either a direct measure of productivity or at least
one environmental measure that should correlate with productivity (standing
crop, rainfall, total N in soil). Standing crop measures usually includes both
living material and litter; although many papers did not distinguish or specify
which was measured. Thus, there may be considerable variation in standing crop
which does not reflect productivity.
Methods: Most values in the database were garnered from descriptions
in the text of the original source. The quantitative data for productivity and
its estimators (Var21-23), for the means and variances in removal and
control treatments (Var28-29, 37-39), and for the indices and variances
of interaction intensity (Var30-32, 39) were usually copies from tables
or scanned from figures (using SigmaScan, Jandel Scientific Software); see specific
variable descriptions for further details.
Goldberg et al. 1999 analysis: To recreate their data set, select the
following values for these variables:
- Parts.removed=all
- Stand.crop not missing
- Response.var in resp.var.units=
- biomass in g/ind
- OR expgrowth in g/g/d
- OR surv in proportion
- Target.form ne tree OR shrub
- Neighbor.form ne tree OR shrub
- In addition, because only results at the end of the first growing season
were included (when multiple measurements on the same plants were taken),
- for Ref=13, retain Census.day=95 (exclude census.day=34,
63)
- for Ref=15, retain Census.day=120 (exclude census.day=60,
240, 365, 485, 730)
- This should yield 296 cases in 14 studies.
All regressions were performed using RCI or lnRR as the dependent
variables and stand.crop as the independent variable, either over all
cases with the same response variable (Fig. 1 in Goldberg et al. 1999) or all
cases for a given taxon within a study (Fig. 2 in Goldberg et al. 1999)
CITATION
- Var1. Ref: reference code (matches with full citation in citation.txt)
REPLICATION
- Var2. Replsite: names of distinct sites in which the entire experiment
was repeated. This does NOT include
- different steps along a productivity gradient—these are considered
the same site. If the experiment was only done in a single site, the value
is ‘none’.
- Var3. Repltime: years in which the entire experiment was repeated.
This does NOT include information
- on how long an experiment was monitored.
- None=experiment initiated only once
- yr1=results from the first time an experiment was initiated
(if initiated more than once), regardless of how long the experiment
was monitored
- yr2=results from the second time an experiment was initiated
(none were repeated more than two years in a row)
EXPERIMENTAL DESIGN--ENVIRONMENTAL VARIATION
- Var4. Gradient: type of productivity gradient within study
- None=no productivity gradient within the study
- Natural=natural productivity gradient within the study
- Exptl=added resources to some plots
- Var5. Envt.factor: specific environmental factor varying along natural
or
- experimental productivity gradient (if identified,
- otherwise ‘none’). In some cases different levels
of the factor are identified with different experimental treatments
(e.g., nitrogen [added], water [added], ctrl (control),
in some cases discrete habitats or microhabitats are identified (e.g.
slope, ridge or different species of neighbors) and in some cases
a factor that varies continuously with productivity or standing crop
is identified (e.g., SOM [soil organic matter] or soil depth.
In the latter case, the same factor is listed for all values of productivity
in that experiment listed in the data base.
- Var6. Amt-added: amount of resource added if an experimental productivity
gradient and reported
-
-
- EXPERIMENTAL DESIGN—QUANTIFYING INTERACTIONS
- Var7. Parts.removed: component of vegetation removed experimentally
- all=roots+shoots removed
- roots=roots removed, shoots left
- shoots=shoot removed, roots left
- canopy=canopy trees removed, understory left
- Var8. Pulse.press: whether or not the experimental removals were
continuous through the experiment
- press=continual removal of neighbors throughout the experiment
- once=single (pulse) removal of neighbors at the beginning of
the experiment
- Var9. removal.method: how vegetation components were removed
- pull=pulling from ground without soil disturbance (used for
seedlings)
- clip=clipping at ground level
- herb=herbicide
- trench=trenching
- tie=tieback of aboveground parts to eliminate shading
- gird=girdling tree trunks to kill trees
- herbclip=herbicide and clip
- cliptren=clip and trench
- clippull=clip and pull
- clipsoil=clip and surface soil disturbance (to remove roots)
- h-tr-p=herbicide, trench, and pull
- h-tr-c-s=herbicide, trench, clip, and soil disturbance (raking)
- Var10. Herb.excluded: types of herbivores excluded from the experiment
- none=full complement of natural herbivores present
- large=large (usually mammalian) herbivores excluded
- small=small herbivores excluded
- insect=insect herbivores excluded
- below=belowground herbivores excluded
- Var11. Plot.size: area (m2) from which neighbors were
removed in removal treatments
-
-
- SPECIES DESCRIPTIONS
- Var12. Genus of target taxon:
- mean=results reported for several species averaged together
- Var13. Species of target taxon
- mean=results reported for several species averaged together
- Var14. Neighbor.form: neighbor growth form
- annual
- shrub
- tree
- rosette=herbaceous with most of leaves in basal rosette
- erect=herbaceous, erect
- gram=herbaceous, graminoid
- herb=herbaceous (mixed or unspecified)
- vine
- Var15. Target.form: target growth form, defined as for neighbor.form
- Var16. Neighbor.stage: neighbor life history stage
- seed
- sdlg=seedling
- juvenile
- adult
- none=none specified
- Var17. Target.stage: target life history stage, defined as for neighbor.stage
- Var18. Target.source: whether target individuals were naturally-occuring
individuals or transplanted to
- the experimental plots
- natural=naturally-occurring individuals
- transpl=transplanted to experimental plots
- Var19. Num.targets: number of target plants per experimental plot
(1000 indicates a large and unknown number)
-
- ENVIRONMENT
Var20. Comm.type: community type
- desert=deserts
- grass=grasslands
- shr-sav=shrub or savanna
- wet=wetlands of various types
- forest=closed canopy tree-dominated communities
- weed=early successional, usually annual dominated
- lowherb=persistent, short herbaceous vegetation such as low
tundra
- Var21. Productivity: estimate of productivity (g/m2/yr)
- Var22. Stand.crop: standing crop (g/m2, usually at end
of season but not always and usually
- including both live biomass and litter (not always distinguished in
the source)
- Var23. TotalN: total nitrogen (mg/kg)
- Var24. Rainfall: mean annual rainfall (mm/yr)
-
RESPONSE VARIABLE
Var25. Census.day: days from the beginning of the experiment to the time
of measurement of the dependent variable.
- This is most often the end of the experiment but some studies have series
of repeated measurements which are recorded as separate cases in the database.
- Var26. Response.var: variable measured on target plants
- biomass=biomass/plant at the census day reported
- expgrowth=exponential growth rate (change in some size measure
per unit time, assuming the plant has grown exponentially from the beginning
of the experiment to the census day reported
- emerg=emergence
- lingrowth=linear growth rate (change in some size measure per
unit time, assuming the plant has grown linearly from the beginning
of the experiment to the census day reported
- seed=seed production
- surv=survival
- xpp=xylem pressure potential
- tiller=tiller number
- Var27. Resp.var.units: units for response variable measurements.
- Var28. Mean.ctrl: mean of the response variable in the control treatment
(missing if only an interaction
- intensity index was reported)
- Var29. Mean.rem: mean of the response variable in the removal treatment
(missing if only an interaction
- intensity index was reported)
-
-
- INDICES OF INTERACTION INTENSITY
Three measures of interaction intensity are included in the database:
- Relative competition intensity (RCI=rem - ctrl / rem)
- Absolute competition intensity (ACI=rem - ctrl)
- Log response ratio (lnRR=ln(rem / ctrl))
- Where rem and ctrl indicate the mean value of the response variable in removal
and control treatment plots, respectively. Where mean treatment values were
reported in the original paper, all three indices were calculated from these
means. The single exception is in cases of zero or negative growth, where
ln(RR) could not be calculated. In many cases, only RCI values were reported
rather than mean treatment values; this was especially likely when a continuous
gradient of standing crop was used rather than two or three discrete levels
with multiple replicates at a level. In this case, lnRR was calculated as
lnRR=-ln(1-RCI) and ACI could not be calculated at all. In a few studies,
only ACI was reported, in which case neither of the other indices could be
calculated.
-
- Var30. RCI: reported or calculated value as above
- Var31. ACI: reported or calculated value as above
- Var32. lnRR: calculated value as above
-
SAMPLE SIZES
Sample sizes are reported for the means of control and treatment treatments
and for RCI and ACI if reported in the original paper (no sample sizes are
given for lnRR since this index was not reported in any of the original papers).
If the interaction intensity indices were calculated by Goldberg et al. (1999)
from treatment means reported in the original paper, sample size for the index
is reported as missing in the database. For some studies, the values of an
interaction intensity index were taken from scatterplots in which each data
point represented a single pair of control and removal plots—for these
the sample size of the index at each value of standing crop or productivity
is reported in the database as 1.
-
- Var33. n.ctrl: the number of control treatment plots
- Var34. n.rem: the number of removal treatment plots
- Var35. n.rci: the number of paired control-removal plots for calculation
of RCI
- Var36. n.aci: number of paired control-removal plots for calculation
of ACI
-
VARIABILITY MEASURES
The values in the datebase are for a variety of measures of variability (identified
in Var36), as reported in the original papers. These measures of variability
are NOT necessarily the appropriate ones for weighting in meta-analysis; the
potential user of these data should see Hedges et al. (1999) and references
therein.
-
- Var36. Var.meas: the measure of variability reported in the original
paper and in var37, var38, var39 in this database
- sd=standard deviation
- se=standard error
- ci=95% confidence interval
- lsd=least significant difference
- Var37. Var.ctrl: value of the variability measure defined in Var36
for the control treatment, as reported in the
- original paper
- Var38. Var.rem: value of the variability measure defined in Var36
for the removal treatment, as reported
- in the original paper
- Var39. Var.RCI: value of the variability measure defined in Var36
for RCI as reported in the original paper
- (no variability measures were given for any other interaction intensity
indices in the original papers)
-
-
- USEFUL VARIABLES FOR TAKING SUBSETS OF THE CASES
If any of the first three of these variables are greater than 1, the study
is useful for comparisons of interaction intensity within studies (i.e., over
time, among dependent variables, or among taxa) but care must be taken in
meta-analyses because the data suffer from varying degrees of nonindependence.
- Var40. n.times: the number of times measurements were taken on the
same plants in the same expt
- Var41. n.depvar: the number of different response variables taken
on the same plants
- Var42. n.taxa: number of different species studied in the same experiment
- Var43. Use: whether (‘yes’) or not (‘no’)
this case was used in the analyses reported in Goldberg et al. (1999).
CITATIONS
-
- Goldberg, D.E. 1996. Simplifying the study of competition at the individual
plant level: the consequences of distinguishing between effect and response
for forest vegetation management. New Zealand Journal of Forestry Science
26: 19-38.
-
- Goldberg, D. E., and A. M. Barton. 1992. Patterns and consequences of
interspecific competition in natural communities: a review of field experiments
with plants. American Naturalist 139: 771-801.
-
- Goldberg, D. E., and A. Novoplansky. 1997. On the
relative importance of competition in unproductive environments.
Journal of Ecology 85: 409-418.
-
- Goldberg, D.E., T. Rajaniemi,
J. Gurevitch, and A. Stewart-Oaten. 1999. Empirical approaches to quantifying
interaction intensity: competition
and facilitation along productivity gradients. Ecology 80: 1118-1131.
-
- Gurevitch, J., L. L. Morrow, A. Wallace, and J. S.
Walsh. 1992. A meta-analysis of competition in field experiments. American
Naturalist 140: 539-572.
-
- Hedges, L. V., J. Gurevitch, and P. S. Curtis. 1999.
The meta-analysis of response ratio experiments in experimental ecology.
Ecology 80: 1150-1156.
ESA
Publications | Ecological Archives
| Permissions | Contacts