Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
303
2014 NORTHEASTERN NATURALIST 21(2):303–322
Natural Plant Establishment along an Urban Stream,
Onondaga Creek, New York
Catherine L. Landis1,* and Donald J. Leopold1
Abstract - Urbanization results in a suite of harmful effects to streams, including removal
or degradation of riparian vegetation. Many stream-restoration programs address this by
adding plants, with limited quantitative knowledge about vegetation dynamics already
occurring within the stream corridor. This project examined natural plant establishment
along an urbanized stream channel in Syracuse, NY. It had three objectives: first, to relate
plant establishment along an urban stream gradient to substrate condition; second, to
quantify seeds dormant in the soil at those same sites; and third, to indicate what passive
revegetation responses might occur to various treatments along a rural-to-urban gradient.
Three sites were selected along such a gradient on Onondaga Creek, near Syracuse, NY.
Vegetation plots were established at each site to assess plant germination and establishment
under substrate conditions designed to mimic restoration interventions. We also conducted
a seedbank study using soil cores collected from these sites. Plant communities were dominated
by grasses and forbs. Numbers of alien species increased from 34% at more rural
sites to 51% at more urban sites. Seedlings of native riparian trees nonetheless germinated
at all three sites along the gradient. Recruitment of native riparian trees (especially Populus
deltoides [Eastern Cottonwood], Fraxinus pennsylvanica [Green Ash], and Acer negundo
[Boxelder]) exceeded non-native and invasive ones. The riparian seedbank showed disproportionate
dominance by herbaceous plants (95.5% of individuals) at all locations surveyed,
and invasive species were common (about 25% of all germinants). This study shows some
potential for natural regeneration of native trees, but also found a significant source of invasive
plants in the soil seedbank that could reduce restoration success. Notably, the study
recorded the presence of 16 bryophyte taxa, and the common ones were those typically
associated with disturbances.
Introduction
Urban streams are among the most extensively disturbed and degraded aquatic
systems in North America (Hession et al. 2000). In particular, riparian deforestation
associated with urbanization reduces food availability for wildlife, affects
stream temperature, and disrupts sediment, nutrient, and toxin uptake from surface
runoff (Paul and Meyer 2001). Resulting changes in hydrology create “hydrologic
drought” by lowering water tables, which in turn alters soil, vegetation, and microbial
processes in urban riparian zones (Groffman et al. 2003).
There is growing interest in restoring function to urban waterways, including
that of the riparian vegetation that plays so many important roles along streams (Riley
1998). These efforts often involve adding plants to the site, but this step is costly
1Department of Environmental and Forest Biology, 241 Illick Hall, State University of
New York College of Environmental Science and Forestry, 1 Forestry Drive, Syracuse, NY
13210. *Corresponding author - cllandis@syr.edu.
Manuscript Editor: M.A. Leck
Northeastern Naturalist
304
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
and labor intensive. Passive restoration, which relies on propagules already in the
system to recover riparian vegetation, appeals due to its low cost and simplicity
(Bernhardt and Palmer 2007, White and Stromberg 2005). For example, streams
could transport seeds and plant parts from more upstream, natural sections to enhance
the flora of downstream, urbanized reaches of the stream. A second source of
seeds lies in the urban streamside soil seedbank. Few studies have examined these
sources—existing upstream propagules and the riparian seed bank—to assess their
role in urban stream restoration.
Many riparian plant species are disturbance tolerant, early successional, and
fast growing, traits that could allow them to compete well even in anthropogenic
settings. Moreover, in a system where upstream reaches of the watershed remain
relatively natural, those areas could provide a pool of propagules of native species
to replenish depauperate areas downstream. On the other hand, streams can
also facilitate dispersal of less-desirable species such as Polygonum cuspidatum
Siebold and Zucc. (Japanese Knotweed), Rhamnus cathartica (European Buckthorn),
and Phalaris arundinacea (Reed Canary Grass) (Tickner et al. 2001).
Thus, research on natural regeneration is important to inform stream-restoration
strategies, including the design of seeding and planting schemes (Gurnell et al.
2006). The purpose of this study was to investigate these existing pools of plant
material along an urban stream, in order to assess their potential contribution to
restoration efforts.
In some cases, especially where site conditions are harsh, added vegetation can
facilitate the arrival and establishment of other plants (Padilla and Pugnaire 2006).
Success of restoration projects in these instances can be enhanced by the addition
of such facilitators or “nurse plants”. Along streams, adding vegetation to barren
areas can directly add a propagule source. Also, by increasing bank roughness, these
added plants can increase seed capture from hydrochory. Streamside plants could
“nurse” any newly established seedlings by mitigating high-velocity erosive flows,
moderating soil moisture (via shading), and protecting against intense herbivory. In
this study, we added small trees (Ulmus americana L. [American Elm]) and Salix
nigra Marsh. (Black Willow) stakes (cuttings) to investigate facilitation effects.
We also conducted a seedbank study to assess the nature of this propagule resource,
with special focus on its potential role in restoration. Further, we collected
hydrological data to permit the assessment of stream stage relative to plot elevation
(e.g., recording frequency of inundation).
The first objective was to relate plant establishment along an urban stream to site
or substrate condition; here, the study focused especially on the establishment of
woody plants that could form a floodplain forest. The second objective was to quantify
seeds dormant in the soil. A third objective was to begin to relate plot treatment
as well as seedbank results to potential differences in vegetation restoration along
the urbanization gradient.
We hypothesized that plant colonization would be highest in the scarified plots
due to the removal of all existing vegetation, which resulted in reduced competition
for soil, light, and water resources in these areas. Colonization in the willow and
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
305
elm treatments was expected to follow, since these plots also experienced reduced
competition, in this case resulting from the herbicide treatment of existing vegetation
before the trees were added. We added four willow stakes, and one elm seedling
(≈50 cm), and expected differences in colonization here based on stem numbers and
plant architectural form. The “no mow” and “mow” plots were expected to have the
lowest levels of colonization because the established vegetation was not removed or
killed. We further hypothesized that numbers of native riparian species would decrease
with increasing urbanization, and that seedbanks in urban areas would harbor
fewer native plant seeds relative to rural areas.
Methods
Field-site description
Onondaga Creek, flowing through the heart of Syracuse, NY, represents an urban
stream for the final 14.5 km of its course (Fig. 1). From headwaters along the Valley
Heads Moraine near Tully, NY, to its mouth along the shores of Onondaga Lake,
Onondaga Creek flows about 43.5 km. Historically, Onondaga Creek meandered
through floodplain forests, cedar swamps and, near its outlet at Onondaga Lake, inland
salt marshes. Like many urban streams, Onondaga Creek was straightened and
deepened, while associated wetlands were drained and filled in f or development.
Study sites. Experimental units were set up at three sites along a rural-to-urban
gradient, from where stream channelization began in Nedrow, NY, to Franklin
Square in the city of Syracuse. Franklin Square is about 1 km south of the mouth
of Onondaga Creek in the Inner Harbor of Onondaga Lake (Fig. 2). The three sites
also represent various engineering treatments along Onondaga Creek. Site descriptions
are found in Table 1.
The study took place along a grass-lined channel that was mowed 1 to 3 times
per growing season, depending on the site. All three sites had a grass-lined “shelf”
or mini-floodplain along the stream where the study plots were located. The most
downstream site, Franklin, was unique in having concrete walls and hard rock
lining to the creek bottom in addition the vegetated streamside shelf. At all sites,
woody species had established in spite of the mowing regime. Due to the “coppiced”
growth form of these trees, seed production was minimal in the immediate
Table 1. Location and other data for study sites along Onondaga Creek. Sub-basin area and land-cover
figures are based on data prepared by M. Hall (SUNY ESF, Syracuse, NY) using NLCD data (USGS).
“Urban land cover” is described in the text.
Sites
Nedrow Seneca Franklin
Latitude 42.972188N 43.004248N 43.055651N
Longitude 76.150713W 76.149318W 76.158262W
Upstream contributing area (km2) 233.7 248.3 321.9
% upstream contributing area in urban land cover 6.84 32.7 60.5
Mowing regime 1x/year (June) 1–2x/year 2–3x/year
Channel lining Grass Grass Rock
Northeastern Naturalist
306
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
riparian area. Riparian soil texture at all three sites was almost uniformly a sandy
loam. Overflow frequency affects seed delivery via hydrochory, so we estimated the
number of times the channel overtopped the stream banks at each site. At least some
portion of the plots was inundated at Nedrow on 17 days, at Seneca on 11 days, and
at Franklin on 57 days for a total of 3.3%, 2.1%, and 11.0%, respectively, of the
520-day study period.
Figure 1. Onondaga Creek watershed. Map © Onondaga Environmental Institute (OEI).
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
307
At Nedrow, floodplain vegetation consisted of grasses, forbs, and woody plants
such as Fraxinus pennsylvanica (Green Ash) and Platanus occidentalis (American
Sycamore). At Seneca, Lolium arundinaceum (Schreb.) S.J. Darbyshire (Tall
Fescue) dominated the herbaceous layer. Cornus amomum (Silky Dogwood), Green
Ash, and Juglans nigra (Black Walnut) were among the woody plants present at
Figure 2. The three study sites (green triangles) along Onondaga Creek.
Northeastern Naturalist
308
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
this site. Tall Fescue, Japanese Knotweed, Salix fragilis L. (Crack Willow), and
Eastern Cottonwood all occurred near the stream at the Franklin site.
Urban-to-rural gradient. To quantify the rural-to-urban gradient, we estimated
percent urban cover. Land-cover analyses were conducted using ArcGIS for all
subwatersheds of the creek. We used these data to help meet study objective 3,
relating results to increasing urbanization. Urban cover values for areas included
in this study are shown in Table 1. Urban cover is defined according to the NLCD
1992/2001 Retrofit Change Product, using modified Anderson Level 1 class codes
and descriptions (MRLC 2011). Urban cover includes developed open spaces as
well as areas occupied by single-family housing units, multifamily housing units,
and retail, commercial, and industrial uses.
The study sites follow a gradient of urbanization. However, while we use “site”
effects as an indication of what might happen along an urbanization gradient, our
results cannot confirm an urbanization effect without true replication. We use our
“site” effects to highlight site differences, but caution that they indicate only a potential
pattern of differences along an urbanization gradient.
Plot sampling: treatment plots
Statistical design. The study was set up as split plots in a completely randomized
design (CRD), with 3 whole-plot treatments per site, each consisting of a collection
of 1-m2 subplot treatments (factorial treatment design). Intially 5 different substrate
treatments were applied, each one selected to mimic, on a smaller scale, conditions
that could occur in a restoration project. Treatments were as follows:
• Scarify soil. We removed turf to expose bare mineral soil. Preparation of
1-m2 scarified plots took place in mid-May 2006 mimicking the timing of
sand- and gravel-bar formation that follows recession of spring floods. A
second set of scarified plots was prepared in mid-May 2007.
• No mow. These plots were left unmowed for the duration of the study.
• Add plants: American Elm. A single elm sapling, approximately 50 cm
tall, was added to the center of the plot after application of herbicide.
• Add plants: Black Willow. Five 20” dormant Black Willow stakes (cuttings)
were added per plot after application of herbicide. Because greenwood
stakes did not root in 2006, dormant stakes were used to replace
them in 2007.
• Control: mow. The 1-m2 control plots were mowed according to the local
municipal maintenance schedule, but by the first author to avoid erroneous
removal of plant cover. Mowing was the default management treatment
and so was considered as the background or control for these sites.
Collectively, one of each type of the 1-m2 treatment plots were considered to make
a “whole plot” for the purpose of statistical design; the whole plots were replicated
three times at each site (Fig. 3). As a result, each site had 3 “whole plots” each made
up of 5 individual 1-m2 (small) treatment plots.
In the 1-m2 plots where trees (willow and elm) were added, existing vegetation
was cleared by application of Roundup herbicide (Monsanto Co., St. Louis, MO)
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
309
before planting. A new set of scarified 1-m2 plots was added at each site during
the second field season, raising the number of 1-m2 treatment plots from 5 to 6 per
whole plot in 2007.
To compare treatments, we used a plant species’ aerial cover as the response
variable, i.e., a modified Daubenmire cover-class system (Daubenmire 1959). Plant
names follow Mitchell and Tucker (1997). For woody plant recruitment, density of
seedlings of individual species became the response variable.
Seedbank study. The seedbank study used the same split-plot array as described
above, except that the collection of soil samples occurred at only 2 subplots, representing
the two elevations, high and low. Five soil cores, each 10 cm in diameter,
were collected from each elevation on either side of the stream per whole plot for
a total of 60 cores per site (Fig. 3). Where possible, soil was collected from two
elevations, high and low. Low elevations were those subject to periodic overflows,
as identified by driftlines of debris or actual observations of water levels in the
stream. High elevations were not subject to the same frequency of stream overflows
and occurred above driftlines in a drier zone.
Sampling depth was limited to the upper 5 cm of soil because this is the location
of most of the seedbank available for germination (Wilson et al. 1993). The resulting
volume per core was approximately 400 cm3 (392.75 cm3), and the volume per
sample (i.e., composite of all ten cores representing one elevation in a whole-plot)
totaled ≈3930 cm3. All soil sampling occurred 15–22 May 2007 in order to focus
Figure 3. Schematic of plot layout at a study site.
Northeastern Naturalist
310
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
collection on the persistent seedbank. Soil samples were kept refrigerated and moist
until 11 June 2007 when they could be placed in the greenhouse.
We used the seedling-emergence method to estimate numbers of viable seeds.
Each soil sample was sifted through 1.27-cm (0.5-inch) mesh to remove rocks, rhizomes,
and root pieces, then spread out to a depth of 1 cm on 4 cm of sterile planting
medium before being set out in the greenhouse.
Seedlings were counted for a 6-month period between 11 June and 11 December
2007. Soil in trays was kept moist by manual watering as needed. Greenhouse temperatures
varied according to season and time of day, with a range of approximately
16–30. °C, and nearly all light was solar (i.e., we did not add artificial grow lights).
Plants were removed upon identification in order to reduce competition in trays.
Unknown seedlings were potted separately and allowed to flower or otherwise
reach a stage of development at which they could be identified. Unknown species
were identified using Gleason and Cronquist (1991) with supplemental assistance
from Newcomb (1977) and Uva et al. (1997).
Two control trays containing only sterilized soil were used to detect contaminant
seeds. Parietaria pensylvanica Muhl. ex Willd. (Pennsylvania Pellitory),
Cardamine pensylvanica Muhl. ex Willd. (Pennsylvania Bittercress), Oxalis spp.
(Woodsorrel), Populus spp., and Salix spp. were identified as green house “weeds”
in this way and disqualified from data analyses.
Data analysis
To provide a measure of dominance and the relative contribution of a plant to the
structure of the community, we used relative cover as a metric. Further comparisons
between treatments, such as woody seedling density, were made using analysis of
variance (ANOVA) for split plots in a CRD. We analyzed these data using SAS software,
version 9.1.3 (SAS 2002–2003). Where necessary (as with much of the woody
seedling data), we square root transformed data as a remedy for non-normality and
heterogeneous variance (Kuehl 2000). We made post-hoc comparisons using least
square means test; alpha value for all statistical tests was 0.05. We used the restricted
maximum likelihood (REML) algorithm as the default variance-component–estimation
procedure used for mixed models in SAS (Littell et al. 1996).
We used species frequency, the number of plots in which a species appears
(Wilson et al. 1993), to indicate which plants were most common throughout the
areas sampled, in contrast to those species seldom encountered. We also recorded
community indicators for plants such as wetland-indicator status (based on USDA
2008), species origin (alien or native; USDA 2008), and whether or not the plant
was considered invasive (Invasive.org 2009).
Results
Treatment plots
Post-treatment, a total of 86 species was found in the 1-m2 field plots during the
two seasons of the study at the three sites along Onondaga Creek (Supplemental
Appendix 1, available online at https://www.eaglehill.us/NENAonline/suppl-files/
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
311
N21-2-983-Landis-s1, and, for BioOne subscribers, at http://dx.doi.org/10.1656/
N983.s1). Numbers of alien (non-native) species (45, or about 52%) slightly exceeded
those of natives (37, or 43%). Percentage of alien species increased from
34% to 40% to 51% of total species richness at the three sites along the rural–urban
gradient (Table 2).
Taken altogether, the three sites sampled were dominated by herbaceous plants,
especially grasses—not a surprising result for an area maintained as a grass-lined
channel. This result was true of the pre-treatment transects as well as the treated
plots. Woody plants, however, appeared in the treatment plots at frequencies—if not
abundances— comparable to the herbaceous ones. The eleven most common plants
are listed in Table 3. The list includes two native riparian trees, Green Ash and Acer
negundo (Boxelder).
Woody plant recruitment. The seedlings included in these counts were limited
to first-year germinants for either 2006 or 2007. For nearly all woody species,
the data were marked by considerable variability from year to year. Among the
most common woody seedlings were native riparian species: Green Ash = 113,
Eastern Cottonwood = 53, and Boxelder = 55 seedlings (Table 4). Eleven seedlings
of the nonnative Ailanthus altissima (Tree of Heaven) appeared but only in
plots at the most urban site, Franklin, over 2006–2007.
European Buckthorn, a calciphilic species that has invaded entire sections
of Onondaga Creek’s historic floodplain, did germinate in the streamside plots
Table 3. Plant species frequency for 2006–2007, with data combined for treatment plots along Onondaga
Creek. Total plots = 98. Native riparian trees are highlighted with *.
Plant Frequency Relative frequency (%)
Tall Fescue 57 58.16
Purple Loosestrife 57 58.16
Bedstraw 26 26.53
Narrowleaf Plantain 26 26.53
Green Ash* 25 25.51
Clasping-leaved Dogbane 24 24.49
Unknown grass 23 23.47
Queen Anne’s Lace 22 22.45
Scouring Rush 21 21.43
Yellow Foxtail 19 19.39
Boxelder* 16 16.33
Table 2. Community indicators for treatment plot study 2006–2007. Nedrow is most rural, Franklin
most urban. Wetland-indicator status based on USDA (2008). Obligate wetland species (OBL) almost
always occur in wetlands. Facultative wetland (FACW) species usually occur in wetlands but may
appear in non-wetlands. Species are categorized as invasive based on Swearingen (2008) and alien
based on USDA (2008).
Site Richness FACW/OBL Invasive Alien % Alien
Nedrow 47 18 3 16 34.0
Seneca 48 13 5 19 39.6
Franklin 51 9 3 26 51.0
Northeastern Naturalist
312
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
but at numbers (12 individuals total) well below most of the native woody
species. The only shrub represented was Cornus sp. (dogwood) with 6 new individuals.
No Salix sp., American Sycamore, or Acer saccharinum L. (Silver
Maple) appeared in the plots, although these species are present at least sporadically
throughout the stream corridor. For all woody species combined (Fig. 4a),
woody plant recruitment declined from the rural to urban sites. Mean seedling
density varied significantly across sites (P = 0.025), with the Franklin and Nedrow
sites, representing opposite ends of the urban–rural spectrum, accounting
for that difference (P = 0.021). Seedling densities at Seneca, the more centrally
located site, did not vary significantly from those at either Nedrow or Franklin
(Appendix 1). Mean seedling densities did vary significantly with treatment
(P = 0.021), a result probably driven by the fact that cottonwoods only germinated
on the scarified plots. No interaction effects between site and treatment
were detected (P = 0.221).
Eastern Cottonwood (Fig. 4b) appeared to germinate independently of site (P =
0.170), but was highly dependent on treatment as mentioned above, establishing
only on exposed soils. Results also revealed site x treatment interaction for cottonwood.
(P = 0.045). Treatment effects of scarification were magnified at the more
Table 4. Total number of woody seedlings germinating in treatment plots in 2006 and 2007. The same
set of plots was used in both years, with the addition of a second set of scarified plots in 2007. Plots
were observed near the end of the growing season (August) for each year, and new germinants only
were counted.
Species Nedrow Seneca Franklin
2006
Boxelder 2 0 5
Norway Maple 0 0 0
Tree of Heaven 0 0 8
Dogwood (Cornus spp.) 1 0 0
Green Ash 14 2 0
Eastern Cottonwood 19 14 6
Cherry (Prunus spp.) 0 0 0
European Buckthorn 3 0 0
Poison Ivy 0 0 0
Unknown woody seedling 0 0 0
Total 39 16 19
2007
Boxelder 30 18 0
Norway Maple 0 2 0
Tree of Heaven 0 0 3
Dogwood (Cornus spp.) 2 3 0
Green Ash 99 19 1
Eastern Cottonwood 11 3 0
Cherry (Prunus spp.) 0 2 0
European Buckthorn 7 2 0
Poison Ivy 3 0 0
Unknown woody seedling 0 1 0
Total 152 50 4
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
313
rural site. In contrast to cottonwood, mean seedling densities for Green Ash did not
vary with treatment.
Total woody seedling establishment in the plots did vary significantly by year
for the two field seasons 2006 and 2007 (P = 0.030). Numbers of cottonwood
seedlings showed no significant difference from year to year while mean seedling
density of ash did increase (P = 0.005) from 2006 to 2007. Values for Boxelder
(Table 3) also increased significantly from 2006 to 2007 ( P = 0.003).
Figure 4. Mean
seedling density
(± standard error)
of plants germinated
in treatment
plots, 2006 and
2007. (a) Results
for all species
combined.
(b) Eastern Cottonwood.
This
tree germinated
only on scarified
plots, regardless
of site. (c) Green
Ash. Ash showed
site effect, but no
treatment effect.
All data were analyzed
using Least
Square Means
test for mixed
models at α =
0.05.
Northeastern Naturalist
314
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
For all woody species combined, recruitment in the scarified plots was significantly
higher than the no-mow and mow treatments (Fig. 5). Remaining treatment
differences were not statistically significant. Mean woody seedling density was
actually highest in the elm plots at 5.55 seedlings/m2, but this result was due largely
to a single elm plot in which 74 elm seedlings were counted in September 2007.
Addition of an elm or a willow seemed to have no significant effect on plant colonization
in these plots.
Seedbank
A total of 2355 seedlings emerged during the 6-month duration of the seedbank
study. Mean seedling density was 199.9 seedlings/m2 across all sites, and ranged
from a low of 20.4 seedlings/m2 at an upper elevation plot at Seneca, to 863.3 seedlings/
m2 at an upper-elevation plot at Nedrow.
Over one hundred vascular plant species emerged from the soil samples
collected along Onondaga Creek (Supplemental Appendix 2, available online at
https://www.eaglehill.us/NENAonline/suppl-files/N21-2-983-Landis-s1, and, for
BioOne subscribers, at http://dx.doi.org/10.1656/N983.s1). Most were identified to
the species level, except one Juncus, one Cyperus, and seven of unknown identity.
Of the unknowns, most could at least be assigned to a family. Of the 102 species,
45 (44%) were native to North America, 55 (54%) were introduced species, while 2
were classified as both (NI; USDA 2008). The number of wetland-indicator species
(FACW and OBL combined) was 28. Seven woody species emerged.
When examined by site, the seedbank data revealed a pattern similar to that
observed in the field-treatment plots along the rural-to-urban gradient. Percentage
of alien species increased from 44.8 to 64.7, while numbers of wetland-indicator
species dropped substantially, from 22 at Nedrow to 3 at Franklin.
The ten most abundant species represented a mixture of wetland, oldfield, and
generalist species. Lythrum salicaria (Purple Loosestrife) outnumbered all other
Figure 5. Recruitment
(±
standard error)
by treatment
for all woody
plant germinants
across
all three sites
in the study.
Means having
the same letter
are not significantly
different.
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
315
species; in fact, its seedlings made up nearly one quarter (564/2355 = 23.95%) of
all germinants and was the most frequently encountered species, germinating in
25/30 soil samples. Of the 177 Juncus articulatus (Jointleaf Rush) that emerged,
151 were found in a single sample (from Nedrow, WP 1, low elevation).
Only 11 seedlings of 7 trees and shrubs species appeared. Of those 11 plants, 6
were Rubus occidentalis (Black Raspberry).
There were no significant differences detected between seedling density at low
versus high elevations (least significant means test: P = 0.846; Appendix 2). Significant
differences did appear, however, among seedling densities at sites, with
Nedrow being higher than both Franklin and Seneca (P = 0.015; Fig. 6).
In addition to vascular plants, a number of bryophytes appeared during the
course of this study (Supplemental Appendix 3, available online at http://www.
eaglehill.us/NENAonline/suppl-files/N21-2-983-Landis-s1, and, for BioOne subscribers,
at http://dx.doi.org/10.1656/N983.s1). Since none of these mosses appeared
in the control trays, these were probably not spore contaminants, but rather
from the soil samples collected along Onondaga Creek. The moss flora included
mainly disturbance-tolerant species such as Ephemerum crassinervum, Funaria
flavicans, and Physcomitrium pyriforme. Physcomitrella patens, a typical floodplain
bryophyte, also appeared. Some of these mosses could have emerged from an
actual persistent diaspore bank (Leyer 2006), and others from spores shed in early
spring the same year (2007) that soils were collected (soil samples were taken in
mid-May).
Discussion
Plant community: Treatment effects
We hypothesized that colonization by both herbaceous and woody species
would be highest in scarified plots, a projection borne out by the data. Remaining
Figure 6. Seedling
density (per
m2; (± standard
error) by site
and elevation for
all species that
germinated in
seed bank study.
Means having
the same letter
(a, b) are not significantly
different.
Northeastern Naturalist
316
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
treatment differences were not significantly different from the mowed control. Results
also supported the hypothesis that numbers of native riparian species decrease
with increasing urbanization, as other studies have shown (e.g., Burton et al. 2005,
Kowarik 1995).
Facilitation. Addition of an elm or a willow sapling or stake seemed to have no
significant effect on plant colonization in these plots. These non-significant results,
however, should be regarded as inconclusive rather than negative, for several reasons.
First, the Black Willow greenwood cuttings did not develop roots and grow
during the first year, and the second-year planting did not have sufficient time to influence
establishment of other species. Second, the herbicide used in the facilitation
treatment plots was applied in early spring before all herbaceous plants had emerged.
Roundup is only effective on actively growing plants, and not effective as a preemergence
herbicide. As a result, living vegetations was not eliminated from some
plots, making it difficult to assess effects of adding willows or an elm on a “new”
plant assemblage. Finally, facilitation effects resulting from addition of woody plants
could require a longer time period than the two field seasons available for this study.
Woody plant recruitment
In the course of this study, Eastern Cottonwood, Green Ash, and Boxelder germinated
in streamside plots under substrate conditions suitable to each species.
Recruitment occurred even under the current hydraulic conditions at all points
surveyed along Onondaga Creek. However although a critical first step, these seedlings
would take many years to form a mature riparian forest (Grubb 1977).
In this study, Green Ash readily colonized diverse riparian substrates except at
the most-urban site. Green Ash appears to be a versatile generalist colonizer in a
range of riparian zones from rural to urban, germinating in far greater numbers than
Silver Maple, for example, or American Sycamore in this system. Mortality as a
result of feeding by Agrilus planipennis Fairmaire (Emerald Ash Borer), however,
may soon eliminate Green Ash as a viable member of the floodplain forest (Herms
et al. 2004, Knight et al. 2007).
Woody plant recruitment and level of urbanization. In general, woody plant
recruitment decreased along the rural-to-urban gradient. This decline could result
from reduced seed sources in urban areas or it could be related to the more erosive
flows typical of urban stream hydrology. One exception to this pattern, Eastern
Cottonwood, appeared to recruit equally well regardless of site. In the case of cottonwood,
seed source was present within 50 m of the most-urban site, and these
trees are likely candidates for parentage of the cottonwood seedlings appearing in
the study plots. In addition, cottonwood seeds may disperse long distances by wind
or, especially, water (Van Haverbeke 2008).
Cottonwoods also produce copious amounts of seed, and good seed crops are the
rule. Estimates of annual seed production of a single open-grown tree have been as
high as 48 million seeds (Van Haverbeke 2008). Such fecundity and wind dispersal
of seeds can translate into seeds reaching more unoccupied gaps compared to other
species (Thompson et al. 2002).
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
317
Numbers of invasive woody plants recruited during the course of this study
were relatively few—11 individuals of Tree of Heaven, 2 Norway Maple, and 12
European Buckthorn—compared to native woody species. These low numbers are
in contrast to the herbaceous flora, of which invasives, or at least aliens, appear to
make up a greater proportion (based on both aerial cover and seedbank data). Few
individuals of European Buckthorn germinated in the treatment plots, despite the
fact that this invasive plant has colonized entire segments of the historic floodplain
along Onondaga Creek.
Seedbank
Few studies have examined seedbanks along an urban stream. It is doubtful that
we would find an “ancient” seedbank still present, previously deposited during the
accretion of the natural floodplain (Goodson et al. 2002), because channelization
has interrupted accretion and literally moved the stream to another position in the
valley. Channel construction along Onondaga Creek also involved the addition of
much fill transported from other locales (R. Peiffer, retired engineer who worked
on the channelization project, pers. comm.). Nonetheless these manipulated soils
may have accumulated seeds and spores that could be relevant to restoration and/
or different management schemes. Fluvial processes of deposition, transport, and
erosion continue even in a channelized stream. Overflows onto the instream shelf
built into flood-control channels such as Onondaga Creek deposited seeds that were
incorporated into the riparian seedbank.
Results revealed a seedbank heavily dominated by herbaceous species at urban
as well as more rural sites. These included important invasives such as Purple
Loosestrife. The presence of such weedy species suggests that the seedbank may
not be a reliable source of plants for restoration and may actually have adverse
economic or environmental impact (Pysek et al. 2004). Herbaceous weedy species
could interfere with survival of native trees. However, in the case of Purple Loosestrife,
Galerucella calmariensis L. (Loosestrife Leaf Beetle) was observed feeding
on it throughout the Onondaga Creek corridor, and could provide some check on
growth and reproduction of this prolific seedbank invasive.
Only six tree seedlings appeared among the >2000 plants that germinated in
the seedbank study. This small number may reflect an absence of viable tree and
shrub propagules in soils along Onondaga Creek. Because canopy trees along
these mowed reaches have been removed, the numbers of seeds may be reduced
compared to intact riparian forests. Also, due to the dynamic conditions of the
floodplain environment, the persistent-seedbank strategy might not be employed
by riparian plants. Silver Maple, cottonwood, and willows produce seeds with dispersal
and short-term viability that coincide with the recession of spring high flows
(Van Haverbeke 2008).
The seedbank did generate a remarkable diversity of plants, including wetland
indicators (18 FACW or OBL species appeared at Nedrow, the most rural site).
Unfortunately, few of these wetland species appeared to be incorporated into the
riparian seedbank at downstream, more-urban sites. Only 3 wetland indicator
Northeastern Naturalist
318
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
species—Eupatorium maculatum (Spotted Joe-pye Weed), Spergularia salina
(Saltmarsh Sand-spurrey), and Purple Loosestrife—were found at the most-urban
site. Due to their prolific output of tiny seeds (Grime et al. 1988) and because rivers
are known to move seeds over considerable distances (Goodson et al. 2002),
we might expect rushes (Juncus spp.), for example, that occur in the vegetation to
colonize riparian areas. However, although Jointleaf Rush appeared in 11/30 of the
seedbank samples, including urban ones, indicating transport, no Jointleaf Rush
plants were observed growing at the more downstream sites.
Low numbers of wetland plants found in the urban seedbank samples could
also reflect “flashy” flows experienced by city streams. Both the channel morphology
and vegetation of the riparian zone play important roles in controlling fluvial
deposition of seeds and sediment, and thus of seedbank characteristics. If these
conditions are not favorable to deposition of flood-borne debris, any viable seeds
could simply continue downstream with the flow.
In addition, variation in the seedling-emergence patterns could have been
influenced by experimental methods. Seedbanks may vary seasonally (Thompson
and Grime 1979), so results would have differed had we collected the soil
samples earlier to document the transient component, e.g., Impatiens capensis
(Jewelweed). The seedling-emergence method can underestimate seedbank composition
because germination requirements may not be met for all plant species
present (Brown 1992). The six-month emergence period may have been too short
for some plants, such as those requiring a double dormancy. Also, the seedbank
composition reflects the location sampled, and it would be interesting to sample a
lower elevation zone, even closer to the stream than the shelf chosen for the “low”
samples. During very low flows, Eleocharis (spikerush), Bidens (beggar-ticks),
Juncus (rushes), and other wetland herbs appeared at this basal elevation within
the grass-lined reaches of channel (C.L. Landis, pers. observ.). Soils at these elevations
may have greater interaction with fluvial dynamics and therefore different
seedbank composition.
Management implications
Plans are already underway to implement green infrastructure in the Onondaga
Creek urban watershed as a means of reducing stormwater runoff, thus reducing
combined sewer overflow (CSO) discharge to Onondaga Creek and Onondaga Lake
(Onondaga County 2001–2013; OEI 2009). In designing these projects, and in species
selection, data suggest species that could function as propagule sources for
local riparian restoration. Depending on site hydrology, these could include wetland
plants such as rushes, sedges, spikerushes, and Eupatorium spp., as well as selected
woody plants including Eastern Cottonwood, Green Ash, and Boxelder. Species
could be chosen not only for their fit to the site but also for aesthetics and for a
potential role in passive restoration of riparian vegetation both along the stream’s
main corridor as well as tributaries.
To replenish a depauperate urban seedbank, seed sources should be provided in
urban and suburban areas via plantings and gardens, with an eye to re-connecting
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
319
systems via tributaries and reducing dispersal distances along the main channel. As
much as possible, physical processes and morphologies that support plant community
development should be restored. This approach requires rethinking the form of
urban design to work with natural processes and the regenerative capacity of soils
and ecosystems (Ferguson et al. 2001).
Acknowledgments
Thanks to Ted Endreny, Karin Limburg, John Stella, Karen Missell, and Steve Stehman
for extended discussions of this research and for their review of this manuscript. Funding
was provided by the US EPA (to T. Endreny and D. Leopold) and The Edna Bailey Sussman
Foundation. We also thank the Onondaga Environmental Institute for assistance. The
detailed comments of independent reviewers much improved this paper, and to them we
extend our thanks. Keith Bowman helped identify bryophytes.
Literature Cited
Bernhardt, E.S., and M.A. Palmer. 2007. Restoring streams in an urbanizing world. Freshwater
Biology 52:738–751.
Brown, D. 1992. Estimating the composition of a forest seedbank: A comparison of
the seed-extraction and seedling-emergence methods. Canadian Journal of Botany
70:1602–1612.
Burton, M.L., L.J. Samuelson, and S. Pan. 2005. Riparian woody plant diversity and forest
structure along an urban–rural gradient. Urban Ecosystems 8:93–106.
Daubenmire, R. 1959. A canopy-coverage method of vegetational analyses. Northwest Science
33:43–64.
Ferguson, B., R. Pinkham, and T. Collins. 2001. Restorative redevelopment: The Nine Mile
Run model. Stormwater July/August 2001. Available online at www.stormh2o.com. Accessed
3 January 2007.
Gleason, H.A., and A. Cronquist. 1991. Manual of Vascular Plants of Northeastern United
States and Adjacent Canada. 2nd Edition. The New York Botanical Garden, Bronx,
NY. 910 pp.
Goodson, J.M., A.M. Gurnell, P.G. Angold, and I.P. Morrissey. 2002. Riparian seedbanks
along the lower River Dove, UK: Their structure and ecological implications. Geomorphology
47:45–60.
Groffman, P.M., D.J. Bain, L.E. Band, K.T. Belt, G.S. Brush, J.M. Grove, R.V. Pouyat,
I.C. Yesilonis, and W.C. Zipperer. 2003. Down by the riverside: Urban riparian ecology.
Frontiers in Ecology and the Environment 1:315–321.
Grime, J.P., J.G. Hodgson, and R. Hunt. 1988. Comparative Plant Ecology: A Functional
Approach to Common British Species. Unwin Hyman, London, UK.
Grubb, P.J. 1977. The maintenance of species-richness in plant communities: The importance
of the regeneration niche. Biological Reviews 52:107–145.
Gurnell, A.M., A.J. Boitsidis, K. Thompson, and N.J. Clifford. 2006. Seedbank, seed dispersal,
and vegetation cover: Colonization along a newly created river channel. Journal
of Vegetation Science 17:665–674.
Herms, D.A., A.K. Stone, and J.A. Chatfield. 2004. Emerald Ash Borer: The beginning of
the end of ash in North America? Special Circular No. 193. Ohio Agricultural Research
and Development Center,Wooster, OH. Pp. 62–71.
Northeastern Naturalist
320
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
Hession, W.C., T.E. Johnson, D.F. Charles, D.D. Hart, R.J. Horwitz, D.A. Kreeger, J.E. Pizzuto,
D.J. Velinsky, J.D. Newbold, C. Cianfrani, T. Clason, A.M. Compton, N. Coulter,
L. Fuselier, B.D. Marshall, and J. Reed. 2000. Ecological benefits of riparian reforestation
in urban watersheds: Study design and preliminary results. Environmental Monitoring
and Assessment 63:211–222.
Invasive.org. 2009. WeedUS: Database of Plants Invading Natural Areas in the United
States. Center for Invasive Species and Ecosystem Health, University of Georgia.Available
online at http://www.invasive.org/species/list.cfm?id=76. Accessed 4 March 2007.
Knight, K.S., R.P. Long, A. Smith, K. Gandhi, and D.A. Herms. 2007. How fast will the
trees die? A transition matrix model of ash decline in forest stands infested by the Emerald
Ash Borer. Pp. 29–30, In V. Mastro, R. Reardon, and G. Parra (Eds.). Proceedings of
the Emerald Ash Borer Research and Technology Development Meeting; 23–24 October
2007; Pittsburgh, PA. FHTET 2008-07. USDA Forest Service, Forest Health Technology
Enterprise Team Morgantown, WV, .
Kowarik, I. 1995. On the role of alien species in urban flora and vegetation. Pp. 85–103,
In P. Pysek, K. Prach, M. Rejmanek, and P.M. Wade (Eds.). Plant Invasion—General
Aspects and Special Problems. SPB Academic, Amsterdam, Netherlands.
Kuehl, R.O. 2000. Design of Experiments: Statistical Principles of Research Design and
Analysis. Duxbury Press, Pacific Grove, CA.
Leyer, I. 2006. Dispersal, diversity, and distribution patterns in pioneer vegetation: The role
of river-floodplain connectivity. Journal of Vegetation Science 17:407–416.
Littell, R.C., G.A. Milliken, W.W. Stroup, and R.D. Wolfinger. 1996. SAS System for
Mixed Models. SAS Institute Inc., Cary, NC. 633 pp.
Mitchell, R.S., and G.C. Tucker. 1997. Revised checklist of New York State plants. New
York State Museum Bulletin No. 490. State Education Deptartment, Albany, NY.
Multi-Resolution Land Characteristics Consortium (MRLC). 2011. National Land Cover
Database (NLCD) 1992/2001 Retrofit Change Product, using modified Anderson Level
1 class codes and descriptions. Available online at http://www.mrlc.gov/faq.php. Accessed
31 October 2012.
Newcomb, L. 1977. Newcomb’s Wildflower Guide. Little, Brown, and Co., New York, NY.
Onondaga Environmental Institute (OEI). 2009. Onondaga Creek conceptual revitalization
plan draft report. Available online at http://www.oei2.org/OEIResources_OCRPDRAFT.
html. Accessed 13 January 2010.
Onondaga County. 2001–2013. Save the Rain program. Available online at http://www.
ongov.net/sustainability/water/str.html. Accessed 10 December 2010.
Padilla, F.M., and F.I. Pugnaire. 2006. The role of nurse plants in the restoration of degraded
environments. Frontiers in Ecology and the Environment 4:196–202.
Paul, M.J., and J.L. Meyer. 2001. Streams in the urban landscape. Annual Review of Ecology
and Systematics 32:333–365.
Pysek, P., D.M. Richardson, M. Rejmanek, G.L. Webster, M. Williamson, and J. Kirschner.
2004. Alien plants in checklists and floras: Towards better communication between taxonomists
and ecologists. Taxon 53:131–143.
Riley, A. 1998. Restoring Streams in Cities. Island Press, Washington, DC.
SAS Institute. 2002–2003. SAS version 9.1.3. SAS Institute, Cary, NC.
Thompson, K., and J.P. Grime. 1979. Seasonal variation in the seedbanks of herbaceous
species in ten contrasting habitats. Journal of Ecology 67:893–921.
Thompson, K., L.C. Rickard, D.J. Hodkinson, and M. Rees. 2002. Seed dispersal: The
search for trade-offs. Pp. 155–172, In J.M. Bullock, R.E. Kenward, and R.S. Hails
(Eds.). Dispersal Ecology. Blackwell, Oxford, UK.
Northeastern Naturalist Vol. 21, No. 2
C.L. Landis and D.J. Leopold
2014
321
Tickner, D.P., P.G. Angold, A.M. Gurnell, and J.O. Mountford. 2001. Riparian plant invasions:
Hydrogeomorphological control and ecological impacts. Progress in Physical
Geography 25:22–52.
US Department of Agriculture (USDA). 2008. Plants database. Available online at http://
plants.usda.gov. Accessed numerous times.
Uva, R.H., J.C. Neal, and J.M. DiTomaso. 1997. Weeds of the Northeast. Cornell University
Press, Ithaca, NY.
Van Haverbeke, D.F. 2008. Populus deltoides Bartr. ex Marsh., Eastern Cottonwood. Silvics
of North America, volume 2. Available online at http://www.na.fs.fed.us/spfo/pubs/
silvics_manual/volume_2/populus/deltoides.htm. Accessed 25 January 2010.
White, J., and J.C. Stromberg. 2005. Opportunities for passive restoration of the Salt River
riparian corridor. University of Arizona Water Resources Research Center publications.
Available online at http://www.ag.arizona.edu/AZWATER/research/2005/Stromberg.
pdf. Accessed 11 March 2007.
Wilson, S.D., D.R.J. Moore, and P.A. Keddy. 1993. Relationships of marsh seedbanks to
vegetation patterns along environmental gradients. Freshwater Biology 29:361–370.
Northeastern Naturalist
322
C.L. Landis and D.J. Leopold
2014 Vol. 21, No. 2
Appendix 1. ANOVA results for split plots, field treatment plots (woody plant recruitment)
study: all woody seedlings (10 species) combined, Eastern Cottonwood, and Green Ash.
Dependent variable = seedling density.
Covariance
parameter
Effect Num df Den df F value Pr > F estimates
All woody seedlings (10 species)
Site 2 6 7.28 0.0248 0.0502
Treatment 5 75 2.84 0.0212 1.2110
Site*treatment 10 75 1.35 0.2211 1.2110
Eastern Cottonwood
Site 2 6 2.41 0.1702 0.0000
Treatment 5 75 15.00 less than 0.0001 0.2588
Site*treatment 10 75 2.00 0.0451 0.2588
Green Ash
Site 2 6 4.60 0.0616 0.1052
Treatment 5 75 0.64 0.6666 0.7441
Site*treatment 10 75 1.70 0.0972 0.7441
Appendix 2. ANOVA results for seedbank densities examining site, elevation and interactions.
Dependent variable = seedling density.
Covariance
Effect Num df Den df F value Pr > F parameter estimates
Site 2 6 9.27 0.0146 0
Elevation 1 15 0.04 0.8459 20705
Site*elevation 1 15 1.50 0.2398 20705