RESEARCH ARTICLE

The Value of Long-Term Assessment of Restoration:

Support from a Seagrass Investigation
Susan S. Bell,1,2 Michael L. Middlebrooks,1 and Margaret O. Hall3

Abstract levels of seagrass cover initially remained poor perform-

ers, 4–7 years post-planting. By 2008, substantial seagrass
The importance of judging success of restoration studies
spillover, contiguous with over 75% of plots, was recorded.
over extended time periods has been repeatedly voiced
When both within-plot coverage and spillover were consid-
but convincing information to justify increased monitor-
ered, seagrass restoration success was attained 6 years after
ing is generally unavailable. Building on Bell et al. (2008),
initiation. Our findings provide an example of compara-
we investigated the development of areal coverage of the
tively longer-term monitoring of a restoration effort lead-
seagrass, Halodule wrightii , as a metric for assessing the
ing to reversal of an earlier evaluation of project success.
outcome of a restoration effort conducted near Tampa
Moreover, unique information on H. wrightii temporal
Bay, Florida, U.S.A., over 7 years, thereby expanding the
dynamics emerged from the 7 year study, further illus-
timescale over which a subtropical seagrass restoration
trating the value of long-term assessment of restoration.
project was evaluated for success. In each of 12 plots,
Extending the duration of post-planting surveys of sea-
500 planting units of H. wrightii were introduced in 2002,
grass coverage may address multiple needs as it advances
and the seagrass cover level documented annually through
the field of seagrass restoration.
2009. Although only low-moderate levels of H. wrightii
cover were recorded after 3 years, a rapid increase to Key words: dynamics, Florida, Halodule wrightii , monitor-
high coverage levels was evident in many plots after 2006 ing, physiological integration, recruitment, rhizome, sub-
and sustained through 2009. Plots that supported only low tropical.

Introduction and persistence after restoration in temperate settings is not

Studies to assess the effectiveness of seagrass restoration
have largely been based on efforts conducted in temperate
coastal settings and focused upon species within the genera
of Zostera or Posidonia. Numerous examples of establishing
new seagrass beds are available (Paling et al. 2003; Fishman
et al. 2004; Li et al. 2010) but questions still remain about
wide-scale restoration as a management strategy (Thom et al.
2012). Typically, most of these seagrass restoration efforts
have been monitored over a short duration (2–3 years) (Li
et al. 2010) although a few studies have examined the fate of
restoration plots over longer time (Thom et al. 2012). Among
the longer-term studies, some discuss plantings that were
not continuously monitored over most years post-introduction
(McGlathery et al. 2012) or consider data only from plots
within which seagrass cover successfully developed (Evans &
Short 2005). Currently, an extended view of seagrass dynamics
1 Department of Integrative Biology, University of South Florida, 4202 E. Fowler
Avenue, Tampa, FL 33620-5200, U.S.A.
2 Address correspondence to S. S. Bell, email sbell@usf.edu
3 Florida Fish and Wildlife Conservation Commission, Florida Fish and Wildlife
Research Institute, 100 Eighth Avenue S.E., St. Petersburg, FL 33701, U.S.A.
© 2014 Society for Ecological Restoration
doi: 10.1111/rec.12087
commonly available.
Seagrass restoration efforts in subtropical settings have
embraced many of the approaches utilized in temperate studies,
although much less attention has been directed to subtropical,
compared to temperate, investigations. As in many temperate
studies, seagrass restoration in subtropical/tropical areas has
relied mainly on introduction of planting units to initiate sea-
grass bed development (Fonseca et al. 1996; Bell et al. 2008)
sometimes in conjunction with re-engineering of blowout areas
from boating activities (Fonseca et al. 2004). Monitoring sea-
grass development in subtropical/tropical sites has revealed
low to moderate levels of shoot cover early after initiation of
restoration (Sheridan et al. 1998), with bioturbation (Townsend
& Fonseca 1998) and/or high turbidity (Kaldy et al. 2004)
suggested as possible factors curtailing cover establishment.
Yet, while subtropical and temperate seagrass studies may
share similar challenges associated with establishing vegeta-
tion in coastal waters, some marked differences between the
two regions with respect to seagrass bed development after
introducing vegetation may be expected. Specifically, seagrass
in subtropical waters may display temporal dynamics distinct
from those in temperate sites given that both seagrass species
composition and environmental conditions (sediments, water
quality, and temperature range) differ. But, as for temperate

304 Restoration Ecology Vol. 22, No. 3, pp. 304–310 MAY 2014

locations, most studies in subtropical/tropical locales which
document planting success often encompass 2–3 years post-
planting (Bell et al. 1993; Fonseca et al. 1996; Sheridan et al.
1998; Kaldy et al. 2004).
The importance of judging success of marine restoration
over extended time periods has been repeatedly voiced across
numerous coastal habitats (Lirman & Miller 2003; Cunha et al.
2012). In fact, upon review of rhizome elongation rates, some
investigations have noted that the time required for seagrass
beds to achieve coverage to “natural” levels likely surpasses
the duration of typical restoration monitoring (Fonseca et al.
2004; Irving et al. 2010). If monitoring of restoration efforts
were lengthened, then empirical data gathered from sequential
surveys of seagrass cover might offer an increasing possibility
of detecting successful establishment. A comparatively longer
time course over which restoration success is evaluated may
also allow detection of potential scenarios of seagrass dynam-
ics not gleaned from shorter-term studies. For example, if
conditions which limit development of seagrass coverage were
of restricted duration, then the probability of detecting delayed,
but increasing, levels of seagrass cover would increase as mon-
itoring time is extended. Additionally, increasing monitoring
time may allow opportunities to observe not only increases,
but also possible declines, of seagrass coverage. Lengthened
surveys may also inform models of spatial patterns of expan-
sion, generating greater confidence in model predictions of
longer-term growth patterns (Renton et al. 2011). Finally, over
expanded time intervals, seagrass coverage may extend beyond
initial plot demarcations especially if rhizome growth under-
lies space occupation (Jensen & Bell 2001). If spillover of
seagrass footprints beyond plot dimensions represents produc-
tion of new seagrass from the restoration effort (Fonseca et al.
1996; Uhrin et al. 2008), it thereby merits inclusion in met-
rics of newly created habitat. Thus, information documenting
any of the aforementioned scenarios contributes to improved
assessment of restoration success.
We followed development of the seagrass, Halodule
wrightii , in a restoration effort, by documenting coverage
of aboveground shoots over 7 years, thereby expanding
the timescale over which a subtropical seagrass restoration
is evaluated for success. Success is operationally defined
as attaining seagrass coverage similar to that of existing
“reference” conditions and at greater than or equal to 70%
of plot areas. Previously, Bell et al. (2008) followed devel-
opment of H. wrightii seagrass near Tampa Bay, Florida,
U.S.A. and, while finding some evidence for establishment
of seagrass at the plot level, seagrass cover scores were
low-moderate over 3 years post-planting. Additionally some
plots performed poorly. Using data from restoration plots and
literature values of seagrass extension rates Bell et al. (2008)
predicted that experimental plots would require 8–30 years
to attain coverage approximating that of reference sites. This
suggestion was aligned with previous reports that development
of seagrass cover over areas of 10–100 s m2 may require
decades (Fonseca et al. 2004) given that new shoots arise
largely from rhizome elongation for H. wrightii in Tampa
Bay, Florida, U.S.A. Flowering is rare in this region and seeds
MAY 2014 Restoration Ecology
Long-Term Seagrass Restoration
have not been recorded in over 1,200 sediment cores from
H. wrightii beds from 1985 to 2011 (S. Bell, unpublished
data). Thus, a longer-term evaluation of bed development of
this particular restoration effort offers a unique opportunity to
not only evaluate dynamics of seagrass coverage through 7
years post-planting, in comparison to predictions based upon
earlier results, but also determine if any of the three scenarios
of spatial change/pattern outlined above (i.e. delayed increase,
decrease, and spillover) can be discerned.
Herein, building upon Bell et al. (2008), we ask:
(1) What are the magnitude and direction of change in
seagrass coverage during years 4–7 of the seagrass
restoration.
(2) Do plots that performed comparatively poorly or well
with respect to post-planting coverage in years 1–3, do
so through later stages (4–7 years).
(3) Does seagrass growth extend beyond the areas of orig-
inally planted plots and, if so, does this spillover effect
modify the evaluation of restoration success?
Methods
Study Site
We conducted our study at Shell Key (27◦ 40 09 N, Lat;
82.23◦ 44 14 W, Long) near St. Petersburg, Florida, from 2006
to 2009 using the same restoration plots included in Bell et al.
(2008). Details of the restoration protocol are included in Bell
et al. (2008) but briefly, 15 plots were established in June
2002 spanning approximately −0.2 to −0.4 m (MLW) water
depths at time of planting (Fig. 1). A natural bed of Halodule
wrightii , the taxon planted at the start of the study, was
located approximately 50–75 m from the experimental plots
(# 1 and 10) (Fig. 1) and at a greater distance parallel to the
row containing plots # 11 and 12. Sediments were well-sorted
sands, and often rippled. Over the study duration salinity and
water temperature ranged from 31 to 34◦ C and 15 to 30◦ C,
respectively.
A total of 6,000 planting units consisting of intact shoots,
meristem, rhizomes and roots were collected from H. wrightii
beds adjacent to the restoration area. All planting units were
introduced into plots in 2002. Additionally, a set of three
“control” (unplanted) plots were erected to detect any evidence
of seagrass recruitment, although not as a result of planting
efforts.
As in Bell et al. (2008), we documented the presence and
level of seagrass cover at each of 500 points uniformly dis-
tributed within each plot (per date total = 6,000 points in
experimental plots; 1,500 points in control plots). Visual sur-
veys were conducted using the Braun-Blanquet (BB) tech-
nique (Braun-Blanquet 1972) to estimate cover categories (0,
no seagrass; 1, <5%; 2, 5–25%; 3, 26–50%; 4, 51–75%;
5, 76–100%). Previous work established that seagrass shoot
number, aboveground biomass and belowground biomass were
all well-estimated by BB scores (Bell et al. 2008). Plots
were surveyed yearly in summer months from 2006 to 2009;
305
Long-Term Seagrass Restoration

extension could sometimes exceed this threshold. Any sea-

grass extension into the interior areas between longest sides
of plots was determined from the edges of only one of the
adjacent
 plots. Area of seagrass extension was approximated
as [ (linear distance (m) seagrass extension from plot mar-
gin at point of measurement) × (distance (m) between adjacent
measurement points)].

Figure 1. Approximate arrangement and position of the 12 experimental
restoration and unplanted “control” (U1-3) plots. Not to scale.
however, because of logistical limitations, only three of the
planted plots and controls were examined in 2006.
Spatial patterns of seagrass cover within each plot were
determined from annual surveys of the planted and control
plots. For each plot, spatial representations of cover were
prepared using two different protocols. The first protocol
utilized the BB score for the presence of seagrass exhibiting
any cover level (BB: 1–5) while the second utilized only those
BB scores ≥ 3. A BB score of 3 represents the mean cover
score for H. wrightii in natural beds across many sites in
the Tampa Bay region including natural seagrass beds near
the planting area (Bell et al. 2008). For each plot and date,
percent seagrass cover was tabulated as either: (1) number of
points with BB (1–5)/500 points or (2) number of points with
BB ≥3/500 points. Accordingly, a plot with 100% cover at the
plot level would reflect the presence of seagrass at each of 500
points examined, although the cover value at any point could
range from 1 to 5. Using the second, more restrictive criterion,
a plot would be considered to have 100% cover only when all
cover scores were equal to, or exceeded, the mean value of
natural sites (i.e. BB ≥ 3) at each of 500 points.
In 2008, area of seagrass cover that extended beyond the
original plot dimensions (Fig. 1) was recorded. At selected
locations along each plot perimeter, we quantified the spatial
extension of seagrass perpendicular to an original plot bound-
ary. Measurement locations were at: (1) every plot corner; (2)
1.8 m intervals along the short sides of a plot; and (3) 4.6 m
intervals along the longer side of plots. Only seagrass that
continued across, or was located within 1 m of the plot edge,
was included. Measurements were performed to an arbitrary
maximum of 8 m from the outside plot edge, although seagrass
Statistical Analyses
The total number of points with seagrass covers assigned BB
scores = 1–5 and BB ≥ 3 were determined for each plot by
year. To determine if the percentage of plot locations with
seagrass cover across the 12 plots differed by year after
planting, a Friedman Repeated Measures Analysis of Variance
(ANOVA) on Rank was conducted; plots were considered
replicates (n = 12 per year). Both new data from 2007 to
2009 and data from 2002 to 2005 (Bell et al. 2008) were
included in this set of comparisons. The Friedman Repeated
Measures ANOVA was conducted separately for each of the
two BB scoring protocols. If differences in percent cover
between years were detected, then a Tukey post hoc test was
used to conduct pairwise tests to identify which years were
significantly different from each other. Coverage values from
2006 were excluded from this analysis because data were not
available for all plots.
Results
Seagrass cover generally increased over the 4–7 years post-
planting interval (Tables 1 & 2; Fig. S1, Supporting Informa-
tion) and this was visually evident for many of experimental
plots (Fig. S2). Friedman Repeated Measures ANOVA on
Ranks indicated significant differences between median val-
ues of percent seagrass cover among years for both scoring
protocols (BB scores = 1–5: χ 2 = 41.764, p = <0.001; BB
scores ≥ 3: χ 2 = 50.359, p = <0.001). When seagrass cover
reflecting BB scores = 1–5 was further examined, Tukey post
hoc analysis of pairwise multiple comparisons indicated signif-
icantly lower cover in plots from 2003 to 2005 than in plots
in either 2007, 2008, or 2009; cover values for 2004 were
lower but not significantly different than those in 2007–2009
(Table S1). When Tukey post hoc tests were repeated for the
protocol using BB scores ≥ 3, seagrass cover in plots from

Table 1. Percentage of seagrass cover [(# locations with seagrass of BB score 1–5 × 500−1 ) × 100] by plot from 2003 to 2009.

Year Plot 1 Plot 2 Plot 3 Plot 4 Plot 5 Plot 6 Plot 7 Plot 8 Plot 9 Plot 10 Plot 11 Plot 12 Mean

2003 27.6 31.2 27.6 39.4 18.8 7.8 8 8.8 2.6 32 48.8 6.2 21.6
2004 76.4 82.4 58 76 54.6 5.8 24.2 4.6 5.2 85.6 96.6 18.4 48.9
2005 49.2 61.6 32.8 29 14 2.4 6.2 3.4 1 83.8 88 11 31.7
2006 79.2 89 — — — — — — — 99.6 — — —
2007 92.4 94.6 74.6 79.8 72.8 15.6 21 11.2 9.2 100 100 32.2 58.6
2008 92.2 94.8 75.6 81.6 78.2 20 16 10 3.2 100 100 36 58.9
2009 97.8 95.6 83 94.4 79.4 10.4 15 10 0 100 100 36.8 60.2
Data from 2003 to 2005 are from Bell et al. (2008) containing any seagrass cover by year. No seagrass was recorded in unplanted control plots.

306 Restoration Ecology MAY 2014

 Long-Term Seagrass Restoration

Table 2. Percentage of seagrass cover [(# locations with seagrass of BB score ≥ 3 × 500−1 ) × 100] by plot from 2003 to 2009.

Year Plot 1 Plot 2 Plot 3 Plot 4 Plot 5 Plot 6 Plot 7 Plot 8 Plot 9 Plot 10 Plot 11 Plot 12 Mean

2003 0 0 0 0.4 0 0 0 0 0 0.6 1.4 0 0.2

2004 9 10.8 0 1.6 0 0 0 0 0 2.4 4.2 0 2.3
2005 7.2 14.6 0 2 0 0 0.4 0.2 0 15.6 27.8 0 5.7
2006 15 26.2 — — — — — — — 65.4 — —
2007 83 83.2 67.2 72.2 64.4 7.2 14.8 7.2 1.8 99.8 99.8 28.6 52.4
2008 89.2 93 62.6 76.8 70.8 6.2 15.6 6.8 0.8 98.4 98.8 29.2 54.0
2009 95.8 94.2 80.2 93.6 76.2 7.2 14.8 9.4 0 98.6 100 36.8 58.9
Data from 2003 to 2005 are from Bell et al. (2008) containing any seagrass cover by year. No seagrass was recorded in unplanted control plots.

Table 3. Area of seagrass cover (m2 ) using both BB scoring protocols inside of experimental plots (total area =5019.6 m2 ) in 2008 (see Tables 1 and 2)
and areal extension (m) of seagrass found adjacent to each experimental plot (= spillover) in summer 2008.

Plot 1 Plot 2 Plot 3 Plot 4 Plot 5 Plot 6 Plot 7 Plot 8 Plot 9 Plot 10 Plot 11 Plot 12 Total (m2 )

Plot area: BB = 1–5 385.7 396.5 316.2 341.3 327.1 83.7 66.9 41.8 13.4 418.3 418.3 150.6 2960
Plot area: BB ≥ 3 373.1 389.0 261.9 321.3 296.2 25.9 65.3 28.4 3.3 411.6 413.3 122.1 2711.4
Spillover (m2 ) 279.4 384.5 326.5 233.4 353.6 104.8 42.7 10.3 0.0 441.7 371.8 109.0 2657.7

2003 to 2005 was significantly lower than that recorded in

years 2007 through 2009 (Table S1).
The largest increases in mean percent seagrass cover across
all experimental plots occurred between 2005 and 2007 (Tables
S2 & S3). In plots for which data in 2006 were available,
a 21–23% increase in mean seagrass cover was charted in
2006 compared to 2005 (Tables S2 & S3). Mean seagrass
cover (BB score 1–5) in 2007 reflected the establishment
of approximately 119 m2 of new seagrass × plot−1 , compared
to 2005 levels (Table 1). When cover levels in 2007 were
examined using BB scores ≥ 3, a mean of 207 m2 × plot−1
of new seagrass was established compared to 2005 levels
(Table 2). For plots monitored in both 2005 and 2006, a rapid
increase in seagrass cover was recorded within 1 year (Figs. S1
& S2). In contrast, all plots decreased in seagrass cover by a
mean 17.1% between 2004 and 2005 (Tables 1 & 2). Changes
in seagrass cover after 2007 were of relatively small magnitude
and no significant differences in mean seagrass cover per plot
were detected between 2007 and 2009 (Fig. S3). Note however
that during 2007–2009 some plots were near, or at, 100%
cover as determined using either of the BB scoring protocols
(Tables 1 & 2). In contrast, plots that performed very poorly in
2002–2005 (plots # 7, 8, and 9) remained so 4–7 years post-
planting. Plots # 6, 7, 8, and 9 never attained cover exceeding
25% of plot area, with 16% the highest cover recorded in 2009.
One plot (# 9) contained no seagrass after 7 years. Seagrass
was absent from control plots from 2003 to 2009.
In 2008, 2657 m2 of seagrass was recorded in areas outside
of, but contiguous with, plots containing seagrass (Table 3;
Fig. 2). Seagrass extending from plots occupied areas both
surrounding and between plots. Combined, seagrass coverage
established outside and within plots in 2008 for BB scores ≥ 3
totaled 5368.4 m2 representing 106% of total planted area.
Likewise, the combined seagrass cover within and outside of
plots for BB scores = 1–5 was 5619.4 m2 or 112% of total
planted area.
Figure 2. Location of seagrass spillover across the experimental plots in
2008. Diamonds indicate that location along a side of a plot at which
spillover was observed. Blackened areas indicate that seagrass filled the
space between plots. Magnitude of spillover is presented in Table 3.
Discussion
Information from a limited number of studies on seagrass
dynamics has begun to reveal the temporal scales over which
expansion and contraction of seagrasses in natural landscapes
occur (Robbins & Bell 2000; Kendrick et al. 1999). Assem-
bling results from restoration studies has potential to add to
empirical information, but, to date, descriptions of seagrass
cover dynamics post-planting have generally been restricted
to short time intervals (≤3 years) across both temperate and
subtropical settings. In our 7-year study, contrasting onto-
genetic patterns of seagrass development were noted among
plots: some supported only low levels of seagrass coverage
while many, after a few years of low to moderate seagrass
coverage, displayed rapid increases to high levels of cover-
age. Additionally, coincident with these high levels, seagrass
spillover of substantial magnitude beyond original plot dimen-
sions was detectable after 5 years Importantly, restoration
success (defined previously) assessed using seagrass coverage
was not evident 3 years post-planting but was supported by

MAY 2014 Restoration Ecology 307

Long-Term Seagrass Restoration
overall coverage levels at the site 6 years after initiation of the
restoration effort.
Evidence of contrasting differences in the rate of increase
in areal cover in plots into which seagrass was introduced
was first detected 3–5 years post-planting, even when coverage
values were low. However, in many experimental plots, a syn-
chronous and large increase in seagrass cover was evident from
2005 to 2007. These findings are supported by statistically sig-
nificant differences between cover levels across plots in early
(2002–2005) and later (2007–2009) years. Such trends match
one of the possible scenarios presented earlier (i.e. a delayed,
but increasing level of seagrass cover), but only when the tem-
poral survey exceeded 3 years. Such response demonstrates
the ability of Halodule wrightii to expand relatively quickly
under favorable field conditions. The relatively rapid increase
in seagrass expansion 2005–2007 in many plots suggests
changing water column conditions might underlie the response.
We could not identify any noticeable variation in patterns of
water temperature over this time interval although marked
increases in chlorophyll a concentration in late summer-early
fall in 2002–2004 were noticeably absent in 2005–2009
across nearby stations (see http://optics.marine.usf.edu/cgi-
bin/vb?area=TB&station=03,04,05). Of special interest was
the persistence of high levels of seagrass abundance inside of
plots from 2007 to 2009, the last sampling date.
Rapid occupation of bare sediments has been reported pre-
viously for H. wrightii via rhizome extension (Jensen & Bell
2001; Fonseca et al. 2004), although this appeared infrequently
during early stages of bed development at our site. Con-
ditions promoting rapid expansion were likely unfavorable
for most plots during early years as evidenced by overall
declines in seagrass coverage between 2004 and 2005 (BB
scores 1–5) and low percent coverage (BB scores ≥ 3). When
results from early coverage were used to predict the time
course required for seagrass to attain even moderate cover-
age across plots (Bell et al. 2008), a considerably longer time
range was forecast (8–30 years) than the time range here (i.e.
5–7 years) over which such levels were actually observed in
many plots. This direct test of rates of predicted seagrass
coverage hints that development of seagrass landscapes origi-
nating from planting units may increase quickly after reaching
some threshold cover level, possibly with positive feedbacks
maintaining space occupation (van der Heide et al. 2007). The
long-term data from our restoration study quantifying vari-
ability in expansion rates of seagrass over a suite of temporal
scales reveal a prevailing limitation of forecasting coverage
patterns when factors generating patch expansion and contrac-
tion of this taxon remain understudied.
While many of the plots displayed development of high
levels of seagrass cover, plots that performed comparatively
poorly with respect to coverage in early stages (1–3 years) of
the restoration displayed similar below-optimal performance
over 7 years. For example, seagrass cover did not develop
beyond low levels in plots # 7, 8, and 9 through 2009. How-
ever, a poor-performing plot (# 12) (based upon cover in 2005),
displayed moderate levels of seagrass cover by 2008/2009 as
well as some evidence of seagrass spillover by 2008. The
308
underlying explanation for successful seagrass establishment
over some, but not all, experimental plots has yet to be promul-
gated but results suggest that conditions promoting seagrass
growth likely operate differentially over small spatial scales
(10 m). Other studies have suggested that even small changes
in water depth and light conditions might explain variable per-
formance of seagrass over the scale of meters (Sheridan et al.
1998; Kaldy et al. 2004); this may have been the case here.
Horn et al. (2009) showed that exposure of seagrass (Posi-
donia sinuosa) planting units to air during collection resulted
in negative effects on photosynthesis. Whether such physio-
logical stress impacted establishment of H. wrightii in plots
our study is unknown. Clearly, environmental conditions in
2007–2008 supported development of seagrass in plots # 6
and 12 to cover levels higher than in earlier years. These
plots were adjacent to plots # 5 and 11 and near inter-plot
locations—all of which contained substantial amounts of sea-
grass cover. The encroachment of seagrass originating from
adjacent plots may have resulted in new seagrass in plots # 6
or 12. Successful expansion of seagrass into “poor performing”
plots via spillover may be explained by physiologically inte-
grated ramets extending from well- to poor-performing areas.
Physiological integration of ramets has been documented
for other seagrasses from coastal Florida (Syringodium fili-
forme: Schwarzschild & Zieman 2008; Thalassia testudinum:
Tomasko & Dawes 1989) and other tropical seagrasses (Marbà
et al. 2002), although not currently for H . wrightii . Note that
asynchronous development of seagrass coverage across plots is
likely if plots only support seagrass growth after encroachment
from adjacent locations.
In contrast to earlier years, many plots not only devel-
oped substantial levels of seagrass cover (>70%) after 2005,
but also sustained these levels through 2009. The lack of
seagrass recruitment into control plots from 2003 to 2009
provides compelling evidence that development of seagrass
cover within experimental plots is most likely linked to the
fate of introduced planting units and their resultant spread.
The lack of recruitment into control plots also supports the
above suggestion that new records of seagrass in some poor-
performing plots may be attributed to spillover. Fonseca et al.
(1996) reported no seagrass recruitment into planted plots in
Tampa Bay after 3 years, and Sheridan et al. (1998) failed
to detect new seagrass in controls over two years following
a H. wrightii restoration effort in Texas, U.S.A. In contrast,
both of these aforementioned investigations reported that sea-
grass appeared in previously bare areas adjacent to planted
plots. These findings, along with experimental data of Jensen
and Bell (2001), agree that the primary mechanism underly-
ing the expansion of H. wrightii in this coastal system reflects
seagrass growth patterns generated from rhizomes which are
often highly branched (Pangallo & Bell 1988). More broadly,
at our site, unassisted development of seagrass beds appears
to be constrained by the lack of recruitment of the targeted
seagrass.
Spillover effects of expanding seagrass cover have received
little attention previously, although greater than 100% cover at
a restoration site was noted by Fonseca et al. (1996) (see also
Restoration Ecology MAY 2014

Uhrin et al. 2008). Only by following seagrass development
over a comparatively long time were spillover effects revealed
however. While seagrass had not appeared outside of plots
through 2005 (S. Bell, personal observation), spillover was
commonly encountered by 2008 and maintained through 2009.
Recording spatial patterns of spillover in our study was
especially fortuitous as it possibly explained the appearance
of seagrass in plots # 6 and 12 on the last two sampling dates.
Across the site, the areal expansion of seagrass outside of plot
boundaries was equal to or exceeded the areal dimensions of
plots in which seagrass establishment was low or absent. When
areal coverage of seagrass in plots is combined with that from
outside plot expansion, the restoration effort achieved greater
than 100% of seagrass cover in 2008.
Extending the duration of post-planting surveys of seagrass
coverage offers an opportunity to address multiple needs as it
advances the field of seagrass restoration. For example, tem-
poral sequences of patch dynamics can be gleaned from long-
term restoration studies and assessment of progress toward
project goals (Kendrick et al. 2005; Frederiksen et al. 2004).
New information on the rate of recovery of ecosystem func-
tion linked to seagrass development, a topic typically discussed
over limited durations (Bell et al. 1993; but see Simenstad &
Thom 1996) can be compiled. Data collected for extended
time after seagrass restoration is initiated would be especially
useful for identifying the stage of development from which
project outcomes can be predicted from early performance.
Renton et al. (2011) argued that long-term patterns of cover
development might be predicted using models of seagrass
growth and initial seagrass planting designs, and long-term
data from restoration projects could contribute to validating
these models.
Early reviews of seagrass restoration in the United States
found that approximately half of seagrass restoration projects
failed to meet success criteria (Fonseca et al. 1998) and con-
cerns about a global strategy of seagrass restoration have been
expressed (Cunha et al. 2012). However, it remains poorly
known how the overall success rate might be modified if
seagrass restoration was evaluated over longer time periods.
Our study provides much needed data supporting the call
to expand the time scale of evaluating restoration success.
While this suggestion has appeared before, convincing infor-
mation to justify increased monitoring is scarce. Our findings
demonstrate that data from “longer-term” studies of seagrass
restoration may, in some cases, lead to reversal of an ear-
lier assessment of project success. In turn, a modified per-
spective of the efficacy of restoration efforts may emerge.
Ultimately, it would be instructive to determine the required
timeline for monitoring restoration efforts in order to predict
success.
Implications for Practice
• We show that increasing the typical duration of post-
restoration seagrass monitoring led to a reversal of an
MAY 2014 Restoration Ecology
Long-Term Seagrass Restoration
earlier evaluation of project success. Our understanding
of the efficacy of restoration may be incomplete because
of limited duration of monitoring programs.
• Results from long-term studies that follow temporal
dynamics of seagrass can improve the design of strategic
or efficient sampling schedules for restoration projects
and inform the planning of future restoration efforts.
• At restoration sites, measurement of seagrass extension
beyond planted areas should be considered if taxa with
high rhizome extension rates are of interest.
Acknowledgments
Funded by grants from the National Fish and Wildlife Founda-
tion (# 2003-0092-010 and # 10024) to S.S.B. R. Thom offered
insightful comments on an earlier version of the manuscript.
We acknowledge the assistance of: M. Fonseca, W. Ellis,
and A. Tewfik. Numerous USF students and FWC personnel
assisted with field/laboratory tasks.
LITERATURE CITED
Bell, S. S., L. A. J. Clements, and J. Kurdziel. 1993. Production in natural
and restored seagrasses – a case-study of a macrobenthic polychaete.
Ecological Applications 3:610–621.
Bell, S. S., A. Tewfik, M. O. Hall, and M. S. Fonseca. 2008. Evaluation of
seagrass planting and monitoring techniques: implications for assessing
recovery times and habitat equivalency analysis. Restoration Ecology
16:407–416.
Braun-Blanquet, J. 1972. Plant sociology: the study of plant communities.
Hafner, New York.
Cunha, A. H., N. N. Marbá, M. M. VanKatwijk, C. Pickerell, M. Henriques,
G. Bernard, M. A. Ferreira, S. Garcia, J. M. Garmendia, and P. Manent.
2012. Changing paradigms in seagrass restoration. Restoration Ecology
20:427–430.
Evans, N. T., and F. T. Short. 2005. Functional trajectory models for assessment
of transplanted eelgrass, Zostera marina L., in the Great Bay Estuary,
New Hampshire. Estuaries 28:936–947.
Fishman, J. R., R. J. Orth, S. Marion, and J. Bieri. 2004. A comparative test
of mechanized and manual transplanting of eelgrass, Zostera marina in
Chesapeake Bay. Restoration Ecology 12:214–219.
Fonseca, M. S., W. J. Kenworthy, and F. X. Courtney. 1996. Development of
planted seagrass beds in Tampa Bay, Florida, USA. I. Plant components.
Marine Ecology: Progress Series 132:127–139.
Fonseca, M. S., W. J. Kenworthy, and G.W. Thayer. 1998. Guidelines for
the conservation and restoration of seagrasses in the United States and
adjacent waters. NOAA Coastal Ocean Program Decision Analysis Series
No. 12, National Oceanic and Atmospheric Administration Coastal Ocean
Office, Silver Spring, Maryland.
Fonseca, M. S., P. E. Whitfield, W. J. Kenworthy, D. R. Colby, and B. E.
Julius. 2004. Use of two spatially explicit models to determine the effect
of injury geometry on natural resource recovery. Aquatic Conservation
in Marine and Freshwater Ecosystems 14:281–298.
Frederiksen, M., D. Krause-Jensen, M. Holmer, and J. S. Laursen. 2004. Spatial
and temporal variation in eelgrass (Zostera marina) landscapes: influence
of physical setting. Aquatic Botany 78:147–165.
Horn, L. E., E. I. Paling, and M. VanKeulen. 2009. Photosynthetic recovery
of transplanted Posidonia sinuosa, Western Australia. Aquatic Botany
90:149–156.
Irving, A. D., J. E. Tanner, S. Seddon, D. Miller, G. J. Collings, R. J. Wear, S.
L. Hoare, and M. J. Theil. 2010. Testing alternate ecological approaches
309
Long-Term Seagrass Restoration
to seagrass rehabilitation: links to life-history traits. Journal of Applied
Ecology 47:1119–1127.
Jensen, S. L., and S. S. Bell. 2001. Seagrass growth and patch dynamics:
cross-scale morphological plasticity. Plant Ecology 155:201–217.
Kaldy, J. E., K. H. Dunton, J. L. Kowalski, and K. S. Lee. 2004. Factors
controlling seagrass revegetation on dredged material deposits: a case
study in Lower Laguna Madre, Texas. Journal of Coastal Research
20:292–300.
Kendrick, G. A., J. Eckersley, and D. I. Walker. 1999. Landscape-scale changes
in seagrass distribution over time: a case study from Success Bank,
Western Australia. Aquatic Botany 65:293–309.
Kendrick, G. A., C. M. Duarte, and N. Marba. 2005. Clonality in seagrasses,
emergent properties and seagrass landscapes. Marine Ecology: Progress
Series 290:291–296.
Li, W. T., J. Kim, J. Park, and K. Lee. 2010. Assessing establishment success of
Zostera marina transplants through measurements of shoot morphology
and growth. Estuarine, Coastal and Shelf Science 88:377–384.
Lirman, D., and M. W. Miller. 2003. Modeling and monitoring tools to assess
recovery status and convergence rates between restored and undisturbed
coral reef habitats. Restoration Ecology 11:448–456.
Marbà, N., M. A. Hemminga, M. A. Mateo, C. M. Duarte, Y. E. M. Mass, J.
Terrados, and E. Garcia. 2002. Carbon and nitrogen translocation between
seagrass ramets. Marine Ecology: Progress Series 226:287–300.
McGlathery, K. J., L. K. Reynolds, L. W. Cole, R. J. Orth, S. R. Marion, and
A. Schwarzschild. 2012. Recovery trajectories during state change from
bare sediment to eelgrass dominance. Marine Ecology: Progress Series
448:209–221.
Paling, E. I., M. Van Keulen, K. D. Wheeler, and J. Phillips. 2003. Influence
of spacing on mechanically transplanted seagrass survival in a high wave
energy regime. Restoration Ecology 11:56–61.
Pangallo, R., and S. S. Bell. 1988. Dynamics of the aboveground and below
ground structure of the seagrass Halodule wrightii (Asherson) Asherson.
Marine Ecology: Progress Series 43:297–301.
Renton, M., M. Airey, M. L. Cambridge, and G. A. Kendrick. 2011. Modeling
seagrass growth and development to evaluate transplanting strategies for
restoration. Annals of Botany 108:1213–1223.
Robbins, B. D., and S. S. Bell. 2000. Dynamics of a subtidal seagrass
landscape: seasonal and annual change in relation to water depth. Ecology
81:1193–1205.
Schwarzschild, A. C., and J. C. Zieman. 2008. Effects of physiological
integration on the survival and growth of ramets and clonal fragments
310
in the seagrass, Syringodium filiforme. Marine Ecology: Progress Series
372:97–104.
Sheridan, P., G. McMahan, K. Hammerstrom, and W. Pulich Jr. 1998. Factors
affecting restoration of Halodule wrightii to Galveston Bay, Texas.
Restoration Ecology 6:144–158.
Simenstad, C. A., and R. M. Thom. 1996. Functional equivalency trajectories
of the restored Gog-Le-Hi-Te estuarine wetland. Ecological Applications
6:38–56.
Thom, R. M., H. L. Diefenderfer, J. Vavrinec III. , and A. B. Borde.
2012. Restoring resiliency: case studies from Pacific Northwest estuarine
eelgrass (Zostera marina L.) ecosystems. Estuaries and Coasts 35:78–91.
Tomasko, D. A., and C. J. Dawes. 1989. Evidence for physiological integration
between shaded and unshaded short shoots of Thalassia testudinum.
Marine Ecology: Progress Series 54:299–305.
Uhrin, A., M. O. Hall, M. F. Merello, and M. S. Fonseca. 2008. Survival
and expansion of mechanically transplanted seagrass sods. Restoration
Ecology 17:359–368.
Van der Heide, T., E. H. VanNes, G. W. Geerling, A. J. P. Smolders, T. J.
Bouma, and M. M. VanKatwij. 2007. Positive feedbacks in seagrass
ecosystems: implications for success in conservation and restoration.
Ecosystems 10:1311–1322.
Supporting Information
Additional Supporting Information may be found in the online version of this
article:
Figure S1. Box plot of percent cover of seagrass (# locations × 500–1) from
experimental restoration plots by year (except 2006) as determined using: (a) BB
score 1–5 and (b) BB score ≥ 3. The bottom of the box represents the lower quartile
(Q1), the middle line represents the median (Q2) and the top of the box represents
the upper quartile (Q3).
Figure S2. Example spatial representations of seagrass cover for three experi-
mental restoration plots for 4 years between 2003 and 2009. BB score is indicated by
color scale: 5 = darkest green; 1 = lightest green; yellow areas represent bare areas
(BB score = 0).
Table S1. Statistical analyses of seagrass cover between years for both cover
protocols.
Table S2. Change in cover of seagrass (values from Table 1) from the previous
annual value.
Table S3. Change in cover of seagrass (values from Table 2) from the previous
annual value.
Restoration Ecology MAY 2014