Download Functional Ecology draft manuscript April 16 2008

yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Big Exciting Title of some kind
Authors – no order: Steve, Inder, Scott, Eran, Niv, Laurie, Kazik, Hongyu, Alana, Emily,
Lead authors: self-selected group committed to turning the draft into a polished manuscript.
The pack: everyone else who contributed to the work (defined generously) in alphabetical
At the end in “head of lab” position = Steve.
Running head:
Key words:
A recent debate amongst ecologists is whether spatial patterns of biodiversity are
determined primarily by dispersal (neutral processes) or by environmental conditions (niche
processes). One pattern that has been extensively investigated is the decreasing similarity of
communities as geographical distance increases (Nekola and White 1999, Soininen et al.
2007). This pattern may be due to a variety of factors including decreasing similarity among
habitat features (niche processes or species sorting, Soinenen et al 2007; Nekola and White
1999; Tuomisto et al. 2003, Gilbert and Lechowicz 2004), the dispersal limitations dictated
by the spatial configuration of the landscape (Nekola and White 1999, Garcillan and Ezcurra
2003) or the neutral expectation that community similarity will decrease with distance
without regard to heterogeneity in the environment – simply due to organisms’ limited
dispersal abilities (Hubbell 2001 OTHER CITATIONS?). Most of the studies about distance
decay of plant community similarity to date have been conducted in terrestrial ecosystems
(Nekola and White 1999, Condit et al. 2002, Tuomisto et al. 2003, Gilbert and Lechowicz
2004, Qian et al. 2005), and generally these studies showed the important role of
environmental determinism but provided a weak support for neutrality.
The types of patterns that are evident, and the forces that create them, may vary
depending on the spatial scale examined. Across large spatial scales and for very diverse
habitats, niche-based processes such as adaptation to larger scale regional environmental
conditions have been predicted to be strongest (Gilbert and Lechowicz 2004) while at
intermediate and fine spatial scales CAN WE BE MORE PRECISE ABOUT WHAT
HAVE VERY DIFFERENT INTERPRETATIONS the interactions of both environmental
change and distance decay has made these effects more difficult to separate.
Patterns in the distribution of species over space may also vary depending on whether
we choose to consider the unique identities of all species or to consider different phylogenetic
or functional group arrangements of these species.
Community patterns may not be clear when looking at individual species, but evident
when group characteristics are considered. If two species are functionally redundant, one
might replace the other across space through a random process, and yet the composition of
the community (from a functional perspective) would remain identical and could be highly
determined by the environment. Groups can be identified as functional (based on traits) or
taxonomic (if phylogeny constrains traits). A trait is a well-defined, measurable property of
an organism, that varies more between than within species (McGill, et al. 2006). A functional
trait strongly influences performance (McGill, et al. 2006).
The characteristics of plant community similarity decay with geographical distance
have never been explicitly examined in salt marshes. Salt marshes are widely distributed at
geographical scales (Chapman 1960; Reimold 1977; Odum 1988; Mitsch and Gosselink
1993), contain a relatively simple community of species, and are known to be strongly
structured by deterministic abiotic gradients at local scales (CITATIONS). These
characteristics make salt marshes strong candidate systems in which to examine community
The seashore-upland direction, which provides suitable condition to test the different
rates of similarity decay with distance between direction along strong environmental
gradients (seashore-upland direction) and direction with weaker environmental gradients
(direction parallel with seashore), therefore address the underlying mechanisms controlling
plant community structure in salt marsh. And we examined the distance decay of plant
community similarity both in Georgia (east coast) and Texas (Gulf coast) and this enables us
to have an idea of how the pattern of similarity decay change at region scale. Meanwhile, we
also explored patterns of community similarity decay using functional groups based on plant
traits and taxonomic groups based on phylogeny instead of the traditional species-based
Study sites and dataset
We analyzed a dataset of salt marsh plant community composition from 49 sites in
Texas and 59 sites in Georgia (Kunza and Pennings in review, datasets XXX and YYY at
GCE website). Data were collected in April and May 2005 (TX) and June and July 2005
(GA). Henceforth we will refer to TX and GA as geographic “regions”.
Within each region, sites were chosen to include a range of mainland, barrier island,
and back-barrier island (hammock) locations. Sites were limited to those dominated by salt
marsh vegetation (tidal brackish and tidal fresh marshes were not sampled). Plant community
composition was documented along a single transect at each site. Each transect began at the
lower elevational limit of vegetation and continued perpendicular to the water's edge to the
lower edge of the shrub community at the upper marsh border (operationally defined by the
presence of a non-stunted shrub). Transects in TX (range X to Y, average 85+6 m) were
somewhat shorter than those in GA (range X to Y, average 122+14 m).
A 0.5 x 0.5 m quadrat was located at 1 m intervals along the transect, and the percent
cover of each plant species was scored in each quadrat on a 6 point scale: absent, present but
less than 5%, between 5 and 12%, between 13 and 25%, between 26 and 50%, and greater
than 50%. Scores were later converted to percent cover values using the midpoint of each
cover range. Plants were identified following Duncan and Duncan (1987).
Zonation patterns in each region
We analyzed plant zonation patterns in both geographical regions. We normalized the
data on species coverage on each transect by dividing transects into 25 intervals of equal
length spanning the entire length of the transects, and binning the plots into these intervals.
Data were averaged within each bin and then averaged for each bin across all the sites within
each region. We normalized the data on functional group and taxonomic group coverage by
summing the species coverage within each functional or taxonomic group within each plot.
Following that we averaged the data within each bin and averaged the data within region as
Plant functional groups
We collected data on 13 plant traits from the USDA Plants web site (
and local floras (table of traits and sources is available upon request/in appendix). Three
traits were continuous (maximum height, seed mass, USDA regional wetland indicator status)
and two were categorical (growth form (Cornelissen et al.,), longevity (annual, biennial or
perennial)) and eight were binomial (C3 vs. C4 photosynthetic pathway, monocot vs. dicot,
tall vs. short, and clonal, herbaceous, succulent, parasitic, and scramble or not). These traits
were chosen because they have commonly been used to define functional groups and data
were available for all species.
We used the list of plant traits to calculate the dissimilarity between each pair of
species. Let Tki be the value of trait k in species i. Let dki,j be the dissimilarity between
species i and j in trait k. For categorical and binomial traits, dki,j = 1 if Tki = Tkj and 0 if Tki !=
Tkj. For continuous traits, dki,j = abs(Tki - Tkj) / (max Tk – min Tk). Let Di,j = Average(dki,j), be
the average dissimilarity between species i and j. We used the UPGMA (Unweighted Pair
Group Method with Arithmetic mean) clustering method to construct a dendrogram of species
relationship based on functional traits (Figure 1). Results of the clustering analysis were
robust against the exclusion of particular traits.
Based on the dendrogram and our knowledge of the plant species, we defined ten
functional groups: 1) parasitic plants (a single species, Cuscuta indecora), 2) perennial
succulents, 3) annual succulents, 4) annual dicots, 5) rushes and sedges, 6) tall grasses, 7)
species, Cynanchum angustifolium), 9) perennial dicots, and 10) shrubs. The number of
species in each functional group ranged from 1 to 5.
Plant taxonomic groups
Using the online interface Phylomatic (Webb & Donoghue 2005; accessed March,
2008) and the Davies et al. (2004) angiosperm supertree, we constructed a phylogeny for all
38 (36?) species (Figure 1). Because phylogenies constructed using the “maximally resolved
seed plant tree” and the “conservative seed plant tree” yielded similar results as the supertree
in Davies et al. (2004), we used data from the latter. To create the same number of taxonomic
as functional groups, we used the dendrogram and our knowledge of the plant species to
define 10 taxonomic groups: 1) Lamiales , 2) Solanales, 3) Asclepidaceae + Gentianaceae, 4)
Asteraceae + Lilaeopsis, 5) Bataceae, 6) Plumbaginaceae, 7) Cactaceae + Aizoaceae, 8)
Amaranthaceae, 9) Poaceae, 10) Cyperaceae + Juncaceae. The number of species in each
taxonomic group ranged from 1-9. COMPARE THE TWO TREES.
Community similarity versus distance
We analyzed community composition data based on species composition, functional
group composition, and taxonomic group composition to examine community similarity at
three spatial scales.
First, to examine community similarity within sites, we compared the Euclidean
distance of all pairs of plots within a site versus their physical distance in meters, with data
from all sites within a region pooled. We compared analyses based on species, functional
groups and taxonomic groups by examining the slope and R2 of each relationship.
Second, to examine community similarity within regions, we compared the Euclidean
distance of all pairs of sites (based on the average of all plots within each site) versus the
distance between sites in meters (calculated from GPS values using the Haversine great circle
distance formula). We compared analyses based on species, functional groups and
taxonomic groups by examining the slope and R2 of each relationship.
Third, to examine community similarity between regions, we compared the Euclidean
distance of all pairs of sites within Georgia, all pairs of sites within Texas, and all
between-region pairs of sites using ANOVA.
Zonation patterns in each region
Plant zonation patterns differed between marshes in Georgia and Texas (Figure 2).
Georgia marshes contained three dominant zones. The shoreward and largest zone was a
monospecific stand of the tall grass, Spartina alterniflora; at some sites this zone covered
over 90% of the transect. Further inland, a transition zone began consisting of a mixture of
S. alterniflora and perennial succulents (Salicornia virginica, Batis maritima). At some
sites these succulents formed distinct monospecific stands that covered over 30% of the
transect; at other sites they were mixed with Distichlis spicata. The succulent zone was often
bordered inland by a dense stand of the rush Juncus roemerianus, sometimes accompanied by
the shrub Borrichia frutescens.
Texas marshes were less structured and more diverse (Kunza & Pennings, in review).
Although Spartina alterniflora also created monospecific stands on the shoreline, this zone
did not extend as far inland as in Georgia marshes. A mixed zone of succulents occupied a
wider range of the mid-marsh. This succulent zone consisted of a number of species
including Batis maritima, Salicornia virginica and Salicornia bigelovii. A number of other
species were commonly present, including Limonium carolinianum, Monanthochloe littoralis
and Lycium carolinianum. Sometimes the succulents were accompanied or replaced by
Distichlis spicata. Occasionally Scirpus robustus was present throughout the marsh. Juncus
roemerianus was present in Texas but occurred less frequently than in Georgia, and in mixed
species patches rather than large monospecific stands.
Community similarity versus distance
Within sites. Figure 3.
Within regions. Figure 4.
Between regions. Figure 5.
We thank the Environmental Institute of Houston for funding data collection and the
Georgia Coastal Ecosystem LTER program (NSF grant number) for logistical support and
archiving data sets. This work was done as part of a Distributed Graduate Seminar funded
and facilitated by the National Center for Ecological Analysis and Synthesis. We thank
Jason Kreitler for assistance.
Literature cited
Cornelissen et al.
Davies TJ, Barraclough TG, Chase MW, Soltis PS, Soltis DE and Savolainen V (2004)
Darwin's abominable mystery: Insights from a supertree of the angiosperms. Proceedings of
the National Academy of Sciences of the United States of America 101: 1904-1909
Duncan, W.H. and M.B. Duncan. 1987. Seaside plants of the Gulf and Atlantic Coasts.
Smithsonian Institution Press, Washington D.C.
Webb CO and Donoghue MJ (2005) Phylomatic: tree assembly for applied phylogenetics.
Molecular Ecology Notes 5: 181-183.
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Zonation patterns.
Regressions within sites.
Regressions within regions.
Within versus between regions bar graph.