Onuf 96
MARINE ECOLOGY PROGRESS SERIES
Vol. 138: 219-231.1996 Published July 25
Mar Ecol Prog Ser
Seagrass responses to long-term light reduction
by brown tide in upper Laguna Madre, Texas:
distribution and biomass patterns
Christopher P. Onuf*
National Biological Service, Southern Science Center, 6300 Ocean Drive, Corpus Christi, Texas 78412, USA
ABSTRACT- A brown tide caused by a very dense bloom of an as yet undescribed species of the new
class Pelagophyceae was first reported in upper Laguna Madre, Texas, USA, in June 1990 and has
been there continuously through December 1995. No change in response to reduced light was evident
in the distribution of the seagrass Halodule wrightii along transects sampled before the brown tide in
1988 and resampled after initiation of the brown tide in 1991 and 1992; however, in winter 1993-94
losses were documented over 2.6 km2 of bottom and by winter 1994-95 the area of vegetation lost had
more than tripled to 9.4 km2 Changes in biomass presaged the changes in distribution. Decreases in
biomass at depths >1.4 m were evident 2 yr before bare areas were detected. Reductions in biomass
were more pronounced toward the south, in keeping with a gradient of increasing light attenuation
from north to south. Support of a hminishing number of new shoots by reclamation of nutrients and
stored reserves from senescing shoots and rhizomes may allow H. wrightii to persist under conditions
of insufficient light for periods greatly in excess of the life span of any one shoot. This postulated capa-
bility would account for the pattern of diminishing biomass over time where the seagrass persists in
deeper areas and the long lag between light reduction and change in distribution where the seagrass
succumbed.
KEY WORDS: Distribution . Biomass . Seagrass . Halodule wrightii . Light . Irradiance . PAR . Brown
tide . Subtrop~cal Texas
INTRODUCTION tions, has occurred gradually over an extended period,
has been accompanied by many structural and hydro-
Clear water is increasingly recognized as a key req- logic changes of the coastal environment, and has
uisite for the development and maintenance of healthy occurred with little or no documentation of conditions
seagrass meadows, and conversely, reductions in before losses were noticed. Consequently, the connec-
water clarity have been implicated in large-scale tion between seagrass loss and water clarity can be dif-
losses of seagrass (Kenworthy & Haunert 1991). The ficult to document.
causes of reduced water clarity range from nutrient In contrast to a gradual, anthropogenically driven
enrichment (Costa 1988, Lewis 1989, Pulich & White change, the advent of a brown tide in upper Laguna
1991, Batiuk et al. 1992, Dennison et al. 1993), to Madre, Texas, USA, in June 1990 resulted in almost
increased suspended loads resulting from hydrological instantaneous light reduction that has persisted to
alteration (Giesen et al. 1990) and frequent resuspen- December 1995. Also, unlike many locations suffering
sion of dredge deposits (Onuf 1994).In most cases, the diminished water clarity and seagrass loss, antecedent
process of water clarity loss and seagrass decline has conditions were well documented. Seagrass distribu-
been the result of anthropogenic inputs and modifica- tion and biomass had been assessed in 1988 (Quam-
men & Onuf 1993),and a multidisciplinary study of the
'Present address: National Biological Service, Midwest Sci-
upper lagoon had been under way for a year when the
ence Center. 6300 Ocean Drive, Corpus Christi, Texas brown tide began, and was continued for 3 yr after
78412, USA. E-mail: chris-onuf@nbs.gov (Stockwell et al. 1993, Dunton 1994). These unusual
Q lnter-Research 1996
Resale o f full article not permitted
Mar Ecol Prog Ser 138: 219-231, 1996
circumstances lend themselves to a particularly strong
assessment of effects of reduced light on seagrass
meadows. The abrupt and persistent reduction in light
by the phytoplankton bloom is more akin to the manip-
ulative field experiments of Backman & Barilotti (1976)
and Dennison & Alberte (1982) than to the inferred re-
constructions necessary in most trend analyses. How-
ever, the change of the light regime is on the scale of
the whole ecosystem, not a tiny fragment.
The effect of the brown tide on the light regime of La-
guna Madre was abrupt and large and of unprece-
dented persistence [Stockwell et al. 1993). Prior to the
initiation of the brown tide, upper Laguna Madre was
renowned for the clarity of its waters (Pulich 1980). In
the 13 mo leading up to the beginning of the brown tide
in June 1990, chlorophyll concentrations never
reached as high as 20 pg I-'. In the next 11 mo, chloro-
phyll concentrations reached 90 pg 1-' and seldom fell
as low as 20 pg I-' (Stockwell et al. 1993, their Fig. 2).
The brown tide alga, an undescribed species in the
newly recognized class Pelagophyceae (DeYoe & Suttle
1994) was not seen before June 1990 but exceeded 106
cells ml-' thereafter. Light transmission through the top
1 m of the water column dropped from a mean of 47 %
the year before the brown tide to 19% the first year of
the brown tide (calculated from Stockwell et al. 1993,
their Fig. 2). Continued sampling at one site documents
that the influence of the brown tide persists. Annual to-
tal irradiance reaching the seagrass canopy in the 5 yr
since the start of the brown tide has ranged from 33 to Fig. 1 Map of south Texas coastal region showing location of
study area in the northern part of upper Laguna Madre (USA)
54 % of what it had been the year before the brown tide
began (Dunton 1994, pers. comm.).That the brown tide
is largely responsible for the light reduction is indicated Laguna Madre of Texas north of Baffin Bay (Fig. l ) ,the
by a highly significant ( p < 0.001, r2 = 0.60) relationship part of the lagoon in which the brown tide was most
between monthly measures of diffuse attenuation coef- prevalent. A survey used to determine the status of sea-
ficient and chlorophyll concentration (Dunton 1994). grasses in Laguna Madre in 1988 (Quammen & Onuf
Here, I report an assessment of the spatial pattern 1993, Onuf 1996) provided an assessment of seagrass
and time course of effects of light attenuation by the distribution and biomass-depth relations before the
brown tide on seagrass distribution and biomass brown tide. Sampling transects generally ran perpen-
afforded by resampling seagrasses in 1991, 1992, 1993 dicular from shore either west or east to the Gulf Intra-
and 1994 at the same time of year along the same tran- coastal Waterway (Fig. 2). Dredge deposits on one side
sects as were sampled in 1988. In addition, I predict the of the waterway or the other prevented continuous
ultimate extent of seagrass loss owing to light reduc- transects across the whole lagoon. Intervals between
tion by brown tide by integrating reports of minimum transects averaged 5 km. Sampling stations were l 0 0 to
light requirements for Halodule wrightii (Kenworthy et 400 m apart, depending on location and proximity to a
al. 1991, Onuf 1991, Dunton 1994, this study) with boundary, along 17 transects. Positions were deter-
measurements of the underwater light regime and mined by LORAN C or by dead reckoning in 1988, by
depth in different parts of the lagoon. LORAN C in 1991 and by Global Positioning System
(GPS) in 1992, 1993, and 1994. Two transects were
added to the sampling array in 1991 to provide more
METHODS detailed coverage of deeper parts of the lagoon, where
effects of llght reduction by brown tide were expected
Distribution. Observations to detect effects of light to be most strongly expressed. All sampling was carried
attenuation by brown tide on a seagrass meadow dom- out in October and November, except that the 1992
inated by Halodule wrightii were made in the upper sampling was not completed until 8 January 1993.
Onuf. Seagrass loss in response to llght reduction 221
Four 80 cm2 by 15 cm deep cores with plants were middle of the study area (Fig 3). The difference
collected at each station 2.5 m apart along the port between the mean depth of all stations on a transect
side of the boat. The time and depth were recorded at in 1988, before installation of the tide gauge, and the
each, and depths for 1991, 1992, 1993 and 1994 sur- mean of the depths corrected to MSL for the same
veys were standardized by adjusting measured depth LORAN coordinates in 1991 were used to refer 1988
by the deviation from mean sea level (MSL) at sam- depths to MSL.
pling time for the Texas Coastal Ocean Observation
Network tide gauge at Bird Island, located near the
SEAGRASS LOST1993 WLAND
- TRANSECTS SAMPLED 1991-94
IIII~II~~
TRANSECTS SAMPLED ALL YEARS
m SEAGRASS LOST 1994
W
---- GULF !NTRACOASTAL WATERWAY R g 3. Map s h o w ~ n g location of light monitors, tide gauge and
areas of seagrass cover lost durlng the brown tlde Light mon-
itors are designated as in the text and Fig 6 Areas a r e num-
Fig 2 Map of study area showlng location of transects sam- bered ( d l 4 4 ) as in the text and in Table I Dashed line indi-
pled ~n different years cates the course of the Gulf Intracoastal Waterway
222 Mar Ecol Prog Ser 138: 219-231, 1996
Cores were washed on 1 mm screens, the retained 1991 vs 1993 and 1992 vs 1993 for each depth class to
plant material identified, and the dominant and other determine whether there was a cumulative effect over
contributors to cover recorded. Occurrence data were the whole period of the brown tide and whether effects
plotted on maps to depict the distribution of sea- could be discriminated over shorter periods, up to the
grasses. last year for which biomass determinations were avail-
In 1993 and 1994, the edges of bare areas were able. The expectation was that if light attenuation
defined to <0.01' of longitude or latitude by repeatedly resulting from brown tide is affecting biomass, it will
halving the interval between the last station where be manifested in reduced biomass in later years in the
seagrass was encountered and the first station along deeper depth classes. Other differences might suggest
the same transect that was bare (4 bare core samples other causes.
and no seagrass rooted material pulled up on the Light. Underwater photosynthetically active radia-
anchor at that location). When vegetated and bare tion (PAR, 400 to 700 nm) was measured at 3 locations
samples were 0.01' or less apart, a stake was driven at (Fig. 3): Stn L1,just inside the outer boundary of a
the outermost vegetated station. These locations were meadow toward the south end of the study area
also marked with a plastic line-float attached by a (1.66 m MSL, 27" 21.80' N, 97" 22.10' W); Stn L2, 200 m
stainless steel cable secured to a screw-in anchor in shoreward of the boundary location (1.52 m MSL, 27"
some cases, because stakes sometimes disappeared 21.80' N, 97" 21.96' W); and Stn L3, in a deep, continu-
between visits. Similar sampling was conducted be- ously vegetated area near the middle of the study area
tween the regularly sampled transects and along the (1.91 m MSL, 27" 30.93' N, 97" 19.53' W). Underwater
long axis of the bare areas to delineate the bare areas PAR measurements were made at 1 min intervals and
more closely. Differences in the location of the outer integrated every hour on a continuous basis using an
boundary between observations were determined by LI-193SA spherical quantum sensor inputting to an LI-
plotting on maps or were measured with a fiberglass 1000 datalogger (LI-COR, Lincoln, NE, USA). The sen-
tape. In January and February 1995, the locations of all sor was positioned at canopy level, the top of the sen-
surviving boundary markers were determined using sor 25 cm above the bottom. The sensor was wrapped
differential GPS. In most cases, there was no ambigu- in transparent plastic wrap that was replaced at 1 or
ity as to what was meadow and what was not. All 4 wk intervals, depending on season, to minimize foul-
samples at a station were either vegetated or bare. ing. The periods of record were May 1993 to August
Patchy stations were mostly limited to a 20 m band at 1994 for Stns L1 and L2 and October 1993 to August
the outer edge. 1994 for Stn L3. Daily total fluxes of underwater PAR
Biomass. The retained material from 2 of the 4 cores were referred to daily total fluxes of incident PAR as
from each station, randomly chosen by coin toss, was measured with a LI-19OSA 2rc sensor mounted on top of
placed in plastic bags and returned to the laboratory a building at Port Aransas, Texas (27" 52' N, 97" 03' W)
on ice, where the samples were frozen until processed. by K. Dunton, University of Texas.
Processing consisted of thawing and separating into The spatial pattern of light attenuation in the lagoon
live (turgid green and white to beige structures) and was evaluated at approximately monthly intervals
dead (flaccid brown to washed-out maroon) fractions. from March 1992 to November 1994 with a 20 cm
The live fraction was sorted further according to spe- diameter Secchi disc. Measurements were made and
cies and into aboveground (green portions of shoots) depth recorded at stations -100 m from the eastern and
and belowground (root, rhizome, and unpigmented western shores of the lagoon and at the midpoint on
portions of shoots) fractions. The sample fractions were east-west transects at 27" 204,22', 26', 30', 34', 38', and
dried to constant weight at 60°C (72 h), weighed, 40'. All Secchi depth determinations were made by the
ashed at 530°C for 3 h, and weighed again. Dry weight same observer outside the shadow of the boat and
and ash-free dry weight were calculated for all plant were the means of the depths of disappearance of the
parts and dead material and expressed on a per m2 disk on descent and reappearance on ascent. Secchi
basis. depths were related to PAR at depth by simultaneously
Biomass samples were separated into depth classes measuring Secchi depths and determining the percent
to determine whether light attenuation resulting from of surface irradiance (SI) reaching Secchi depth with
the brown tide affected biomass. Depth classes were an LI-193SA spherical quantum sensor mounted on a
defined to yield a sample size of at least 10 in any year lowering frame inputting to an LI-1000 datalogger on
for statistical comparisons between years: <85, 85-1 15, 22 June, 12, 13, 31 October, and 2, 3, 4, 7, 8 November
115-130, 130-140, and 2140 cm depths. Variances 1994. Hourly wind data were obtained from Blucher
were not equal between years regardless of transfor- Institute, Texas A&M University, Corpus Christi,
mation. Consequently, Mann-Whitney U tests were Texas, for the Bird Island station of the Texas Coas-
performed on pair-wise comparisons for 1988 vs 1993, tal Ocean Observation Network (Fig. 3) to interpret
Onuf: Seagrass loss in response to light reduction 223
large differences in light attenuation for small Secchi depths were measured was then conlputed as the long-
depths. Conditions were classified as calm for winds term measure of available light at different depths in
c24 km h-' (15 miles h-') at the time of light and Secchi different parts of the lagoon. An estimate of 15% of
measurements and for protected locations (within surface light reaching canopy level, near the midpoint
200 m of shore on the lee side with respect to prevail- of the published range for the minimum light require-
ing wind) regardless of wind speed. Conditions were ment of Halodule wrjghtii (Kenworthy et al. 1991, Onuf
classified as rough at exposed locations when wind 1991, Dunton 1994), was then used to predict the ulti-
speed was >24 km h-' mate extent of loss of seagrass likely to result from the
The frequency of hourly observations of winds prolonged occurrence of the brown tide.
>24 km h-' from 1 October 1993 to 17 October 1995
was assessed to determine whether high light attenua-
tion associated with rough water was likely to differ RESULTS
between the November to April period of frequent
frontal passage and the May to October period of pre- Distribution
vailing southeasterly breezes. Winds >24 km h-' were
further categorized by direction according to 90" sec- No change in seagrass distribution was evident
tors centered on the long axis of upper Laguna Madre between surveys of 1988 and 1991 or 1992, despite
or at right angles to the long axis. The fetch of the large reductions in light reaching the bottom in upper
winds aligned with the long axis was greater than for Laguna Madre after initiation of the brown tide in
winds oriented cross-wise and presumably promoted June 1990 (Stockwell et al. 1993). These conditions
the development of rougher seas. have persisted at least through December 1995 (Dun-
These relations of % S1 reaching Secchi depth were ton 1994, pers. comm.). The first evidence of distribu-
then used to compute diffuse attenuation coefficients tional change came from installation of a light moni-
corresponding with Secchi depths. The attenuation tor just inside the boundary of the seagrass meadow
coefficients were applied to bathymetric profiles at near the south end of the study area in May 1993. In
intervals of 0.1' longitude along transects at intervals of November 1993, the outer boundary was 30 m land-
1' latitude to generate estimates of the proportion of ward of the monitor (Fig. 3: Area #3). Elsewhere, bare
surface light reaching the bottom according to the areas were encountered in November 1993 on 3 tran-
Beer's law relation: sects at stations that had been vegetated in 1992
(Fig. 3: Areas # l and #2).In aggregate, 2.6 km2 of sea-
Iz/Io = e-kz
grass meadow had been lost between observations
where I, is photon flux density (pm01 photons m-' S-') (Table 1).
at depth z, I. is photon flux density at the surface, k is Boundary locations marked in November and
diffuse attenuation coefficient (In m-'), and z is depth December 1993 were reexamined in December 1994
(m).The mean of these values for all dates that Secchi and January-February 1995. Boundaries had receded
from 50 to 800 m . In January 1995,
Table 1. Statistics on extent of loss of seagrass cover in upper Laguna Madre Area in the the lagoon
(Texas, USA) since the advent of the brown tide. Area numbers ( # l 4 4 ) are as was 8 km long and on average 0.8 km
in the text and Fig. 3 wide (Fig. 3). On the other side of the
Gulf Intracoastal Waterway, Area #2
Area Width of meadow lost ( m ) Length Area Number of extended 5 km, constricted by dredge
Mean M~nimum Maximum (m) (km2) determinations of disposal areas. The outer boundaries
outer boundary
for Areas #3 and #4 In the south had
November 1992 or May 1993 to November 1993 retreated 140 m on average (Fig. 3).
#l 330 260 550 5100 1.7 11 Over the whole period, 9.4 km2 of sea-
#2 560 - - 1450 0.8 1 grass has been lost, more than a 3-fold
#3 30 30 30 3000 01 2 increase since winter 1993 (Table 1).
#4 not surveyed
Total 2.6
November 1993 to January 1995
Biomass
#l 770 220 1340 8100 6.2
#2 430 220 800 5400 2.3
#3 180 150 210 2700 05 Tiends in biomass over time differed
#4 140 80 220 2900 0.4 according to depth (Fig. 4, Table 2). In
Total 9.4 the 2 shallowest depth classes, there
was no change over time. In the 115-
224 Mar Ecol Prog Ser 138: 219-231. 1996
130 cm depth class, the main difference
was that biomass was >30% higher In
1991 than other years. The significant
reduction in biomass between 1992 and
1993 is consistent with a brown tide ef-
fect; however, the lack of a significant
difference between the pre-brown tide
year and 1993 argues against this inter-
pretation.
In the 130-140 cm depth class, bio-
mass was elevated in 1991 also (Fig. 4).
However, in this case, there was a sig-
nificant reduction in biomass between
the pre-brown tide year and 1993, the
last year for which biomass data are
available. The reduction from 1992 to
85 85-115 115-130 130- 140 > 140
1993 was not significant (Table 2). DEPTH CLASS (CM) WITH RESPECTTO MEAN SEA LEVEL
The strongest indication of a cumula-
tive effect of brown-tide-caused light
m
1988 1991 m 1992 0 1993
attenuation on seagrass biomass is the Fig. 4. Mean biomass in 5 depth classes over 4 yr. 1988, before the advent of
monotonic decrease of biomass over the brown tide; and 1991, 1992, 1993, under the influence of the brown tide.
See Table 2 for statistical comparisons
time in the 2140 cm depth class (Fig. 4 ) .
Over the whole period, biomass diminished by >60%.
The reductions were significant from 1988 to 1993 and
from 1991 to 1993 but not from 1992 to 1993 (Table 2).
For the 1140 cm depth class, spatial differentiation
also was evident (Fig. 5). Biomass in the southern half
of the study area was less than in the north in 1991 and
1992 but not in 1993 (Table 3).
Table 2. Mann-Whitney U tests for reductions In biomass in
different depth classes for 3 time periods: 1988 vs 1993, 1991
vs 1993 and 1992 vs 1993. U test statistic; z: standardized
normal deviate; N: sample size earlier year, later year;
p: probability
l
1991 19m 1993
Depth class Time period U z N P YEAR
, north +south
< 85 1988 VS 1993 164 -0.18 20, 17 0.43 Fig. 5. Biomass In the >140 cm depth class In northern vs
1991 vs 1993 148 -0.43 19, 17 0.33 southern parts of the study area from 1991 to 1993. See Table 3
1992 vs 1993 209 -0.68 28, 17 0 25 for statistical comparisons
Table 3. Mann-Whltney U tests for differences ~n biomass in
the > l 4 0 cm depth class in the northern (less affected by
brown tide) and southern (more affected by brown tide) parts
of the study area In different years. U: test statistic; z: stan-
dardized normal deviate; N: sample slze north, sample
size south; p: probability
I year U z N P
Onuf: Seagrass loss in response to light reduction 225
Light
In May 1993, continuously recording
underwater light monitors were in-
stalled at canopy height at the outer
boundary of a seagrass meadow (Stn
L1) and 200 m into the meadow (Stn L2)
to document with precis~onn ~ i n i m u n ~
and sufficient levels of PAR to sustain
established seagrass meadows. In Oc-
tober 1993 a third monitor was installed
17 km to the north in a deep site with
continuous seagrass cover (Stn L3),
with the intent of documenting the light
environment at a clearer location. By
December 1993, the outer boundary I
May-93 Aug-93 Nov-93 Mar-94 Jun-94 Sep-94
had receded 30 m landward from Stn DATE
L1 (Fig. 3). By January 1995, the outer ,site L1 ++ site L2 -site L3
boundary was within 30 m of Stn L2
and had receded past Stn L3 (Fig. 3). Fig. 6. Weekly mean percentage incident PAR reaching canopy level at 3 sta-
Over the period with records from all 3 tions in upper Laguna Madre. See Fig. 2 for locations of monitors In relation to
statlons (10 October 1993 to 21 August shifts in seagrass boundaries
1994), 9 % of incident PAR reached
canopy level at Stn L1, 14% at Stn L2, and 16% at (28 vs 18%),but the relationship was reversed the fol-
Stn L3; however, the temporal distribution of percent lowing spring and summer (6% of incident PAR reach-
of surface PAR reaching seagrass canopy level was ing canopy level at Stn L3 vs 12% at Stn L 2 ) . By No-
highly variable and differed between locations (Fig. 6). vember 1995, seagrass had disappeared at Stn L2 as
A higher percentage of incident PAR reached the sea- well.
grasses in winter (November 1993 to March 1994) than Periodic Secchi disk measurements at stations
in the spring-summer period of active growth (April to throughout the study area revealed a north-south gra-
August 1994) at all stations. Much more light pene- dient in light attenuation (Fig. 7). Based on pooled data
trated to canopy level at Stn L3 than at Stn L2 in winter for 20 sampling dates and 1 to 3 stations on a latitude
transect of sufficient depth to measure
Secchi depths, the percent frequency of
Secchi depths <60 cm (relatively high
light attenuation) increased from 15 to 55
from the north end of the study area to the
south, while the percent frequency of
Secchi readings > 100 cm (relatively clear
water) decreased from 50 to 15. This in-
crease in light attenuation from north to
south is consistent with the higher biomass
seen in the 2140 cm depth range in the
northern half of the study area than in the
southern half in 1991 and 1992 (Fig. 5).
In order to apply this north-south gra-
dient in Secchi depths to predictions of
the area of seagrass likely to be lost
in different parts of the study area as a
40&38 34 30 26 22&20 result of brown tide shading, a corre-
MINUTES NORTH OF 27 DEGREES NORTH
spondence had to be established be-
CM a 6 0 - S O C M 6 0 - 1 0 0 ~~ ~> ~ O O C M tween Secchi depth and light attenua-
Fig. 7 Percentage frequency of observations in different Secchi depth cate-
tion. The % S1 reaching Secchi depth
gories along a latitudinal gradient in upper Laguna Madre, April 1992 to proved to be extremely variable, ranging
September 1994 from 15 to 40 (Fig. 8). For Secchi depths
226 Mar Ecol Prog Ser 138: 219-231, 1996
>60,% S1 reaching Secchi depth tended to decrease as equation for the 'calm' array of points in Flg. 8 ( A and +
Secchi depth increased. For Secchi depths <60 cm,% symbols). Short Secchi depths in November to March
S1 tended to be lower when winds exceeded 24 km h-' were assumed to be the result of episodes of sediment
at exposed locations ( X = 24, SE = 2.1) than for all loca- resuspension associated with winter storms. Conse-
tlons when winds were <24 km h-' (57 = 30, SE = 1.3) quently, the mean % S1 for the 'rough' array in Fig. 8
and for protected locations when winds exceeded 24 was used for Secchi values <60 cm in November to
km h-' ( X = 31, SE = 1.8). April. The relationship for the calm array was used for
The difference in % S1 for Secchi depths <60 cm Secchi depths >60 cm.
between rough water conditions (stiff breeze, exposed
locations) and calrn (moderate breeze or protected
location) was significant ( t = 3.14, df = 22). Therefore,
the mean of the 9 observations for exposed locations
with winds >24 km h-' was taken as representative for
rough conditions (24 % S1 reaching Secchi depth). For
calm conditions, estimates of % S1 were generated
from the linear regression equation for all other points
in Fig. 8: % SIsD= 35.1 - 0.0698 SD (r2= 0.37, df = 39),
where SD is Secchi depth.
Winds exceeded 24 km h-' 39% of the time in the
November to April period of frequent frontal passage
compared to 15 % of the time in May to October, based
on the 2 yr of records analyzed for the monitoring sta-
tion near Bird Island (Fig. 3). The difference was more
pronounced for winds with long fetch, oriented along
the long axis of the lagoon. They prevailed 17 % of the
time in November to April compared to 3 % of the time
in May to October.
The geographic survey of light attenuation by Secchi
depth, the light versus Secchi depth relations and a
bathymetric survey (Fig. 9) were then integrated to
depict % S1 at canopy height over the study area.
Because of the low frequency of winds r 2 4 km h-'
from May to October, especially oriented along the
long axis of the lagoon, estimates of light at depth for
that period were derived from the linear regression
101 I
0 50 100 150 200 S O
SECCHI DEPTH (CM)
A calm X windy, exposed + windy, protected
Fig. 8. Relationship between Secchi depth and percentage of Fig. 9. Bathyrnetric map of study area Dashed line marks the
incident PAR reaching Secchi depth under calm and rough course of the Gulf Intracoastal Waterway Small channels and
conditions emergent or shoal areas are not depicted
Onuf. Seagrass loss in response to light reduction 227
BARE VEGETATED UPLAND
Fig. 10.Bare areas in upper Laguna Madre (a) predicted based on measured hght attenuation, depth and assumption that c15%
of incident PAR reaching canopy level is limiting to meadow development. (b) determined by survey October-November 1988,
before advent of the brown tide; (c) determined by survey October-November 1994, after 4 growing seasons of brown tide
Dashed line indicates the course of the Gulf Intracoastal Waterway
Applying the criterion that long-term mean light ity of light is governing distribution in this system, a
reaching canopy level must exceed 15% of incident seagrass meadow should survive there, regardless of
PAR to sustain Halodule wrightiimeadows (Kenworthy how long the brown tide persists. To the south, increas-
et al. 1991, Onuf 1991, this study) to the Secchi depth ing light attenuation and deeper water yield predic-
and light data of this study, the predicted maximum tions of larger expanses of bare bottom (Fig. 10a). In
depth of seagrasses varied from 180 cm at 27' 38-41' all, on the basls of measured light attenuation, 24 % of
N to 170 cm at 27" 34-37' N, 160 cm at 27" 30-33' N, the study area is predicted to be too deep to support
150 cm at 27" 26-29' N, and 140 cm at 27" 20-25' N. seagrass meadow. In contrast, only 6 5 of the study site
%
The northern part of the lagoon is mostly shallow (82 % was bare before the advent of the brown tide
< 1 m deep north of 27" 37' N compared to 22 % < 1 m (Fig. l o b ) . The greatest change in cover between pre-
deep south of 27" 30' N; Fig. 9 ) .Therefore, i availabil-
f brown tide conditions and that predicted based on
228 Mar Ecol Prog Ser 138: 219-231. 1996
Table 4. Final % of study area predicted to go bare if brown tion. First, much of the lagoon is so shallow that even
tide persists long enough for distribution of seagrasses to drastic light reduction would have no effect on distrib-
reach steady state with the light regime for different assump-
tions about minimum light requirements and relations
ution or biomass over much of the area. Almost 50% of
between light attenuation and Secchi depth the study area is <1.0 m deep, while the Halodule
wrightii meadow extended at least to 1.8 m before the
% S1 at Secchi depth Minimum l ~ g h requirement
t
brown tide began. Second, there was a d~stinct gradi-
for Halodule wrightii ent in light attenuation increasing from north to south.
15% S1 18% S1 Since the long-term light record for the upper lagoon
(Dunton 1994) is from the south, much of the lagoon
Varies with season and wind 18 24
was not subject to such severe light deprivation as
Does not vary systematically 20 27
(use mean for all conditions)
measured there. Third, the resolution of the initial
sampl~ng program in 1988 was so coarse that substan-
tial changes in distribution could have occurred with-
out being detected. The original sampling design was
available llght with brown tide is in the middle of the patterned after an earlier survey of the lagoon
study area (Fig. 10a, b). Documented losses also are (Merkord 1978) to maximize the power of the analysis
concentrated in the middle of the study area (Fig. 10c); for trends between studies (Quammen & Onuf 1993).
however, as yet, the area pred.icted to go bare from This meant that relatively little effort was expended in
measurements of the underwater light regime is con- the most critical areas for detecting possible effects of
siderably larger than what is bare (Fig. 10a, c). light limitation: deep areas with vegetation and the
Because the assumptions about the seasonality of zone where vegetation terminated in deep water.
light attenuation versus Secchi depth and the mini- Fourth, the morphology of the lagoon is such that the
mum light requirement for Halodule wrightii are sub- middle region is relatively flat, while the slope is rela-
ject to question, losses were estimated for 2 other tively steep toward the edge. In the one part of the
assumptions. In one case, no systematic variation of % lagoon where the seagrass meadow terminated in
S1 with respect to Secchi depth was assumed, and the deep, bare bottom, the outer boundary lay in this
mean for all 50 observations (27.8% S1 reaching Secchi region of steeper slope. Therefore, any lateral shift in
depth) was used to estimate '10 S1 reaching canopy outer boundary corresponding to a decrease in the
level (25 cm above bottom) under all conditions. In the compensation depth for the seagrasses was relatively
other case, 18% S1 was assumed to be the minimum small.
necessary to support a seagrass meadow as reported Regardless of the characteristics of the sampling
by Dunton (1994), rather than the 15% suggested in program diminishing its sensitivity to detect change
this study. Both alternative assumptions lead to some- resulting from light reduction, the area of bottom expe-
what higher estimates of bare bottom at steady state riencing light reduction below that required to sustain
with the brown-tide-influenced light regime than as a seagrass meadow was far in excess of the detect~on
described above (Table 4). limits of the sampling program. One possible explana-
tion for the persistence of an established seagrass
meadow under conditions of limiting light is that sur-
DISCUSSION viving shoots cannibalize the resources of adjacent
shoots as they succumb to light limitation. According to
The depth limit of seagrass meadows is commonly this conceptual model, an established meadow might
assumed to be set by light attenuation underwater be able to survive much longer under insufficient light
(Bulthuis 1983, Orth & Moore 1983, Iverson & Bittaker than the life span of individual shoots would suggest
1986, Dennison 1987, Duarte 1991, Dennison et al. by translocation of reserves from dying parts to rhi-
1993).In this study, >50% reduction of light caused by zomes, analogous to the process of reclamation of
a dense phytoplankton bloom compared to pre-bloom nutrients and presumably other materials from senes-
conditions had no discernible effect on distribution cent leaves to support new growth in Zostera manna
until after the fourth growing season under reduced (Pederson & Borum 1993) These resources might then
light. However, an effect on biomass was seen sooner. subsidize survival of existing shoots and even develop-
Biomass at depths > l 4 0 cm was less in the first sam- ment of new shoots. This progressive pooling of
pling after the initiation of the brown tide, after 2 resources from a formerly larger population to support
growing seasons of reduced light, than before. new growth in a progressively smaller population
Four factors contribute to the apparent discordance operating at a deficit would seem necessary to account
between the magnitude of the environmental change for the persistence over years of a species in which the
and the expression of an effect on seagrass distribu- mean life span of individual shoots is 110 d (Gallegos et
Onuf: Seagrass loss in response to light reduction 229
al. 1994). Alternatively, the life span of individual to occur under calmer conditions than previously
shoots may be longer under conditions of light limita- (Ward et al. 1984) a n d more frequently. An indirect
tion than under the conditions assessed by Gallegos effect may be the efflux of nutrients from the sedi-
et al. (1994). ments a s the recently dead seagrasses a r e remineral-
No data a r e available to evaluate the responses of ized, enhancing the growth of the brown tide organ-
populations of shoots; however, the gradual decrease ism.
in biomass in d e e p areas where Halodule wrightii Obviously, there is a limit to how long this holding
persists is consistent with this proposed mechanism. If action by seagrasses can go on, if indeed light has been
reclamation of nutrients and stored reserves is respon- reduced below the ecological compensation point
sible for the persistence of seagrasses in Laguna (sensu Kenworthy et al. 1991). That limit was sur-
Madre under conditions of light limitation caused by passed in some parts of the lagoon in 1993. Approxi-
the brown tide, then a reduction in seagrass biomass mately 3 km2 of seagrass meadow went bare between
over time should be evident in deeper parts of the distributional surveys made in fall 1992 a n d fall 1993,
lagoon where light is now limiting, compared to shal- and the process of loss of seagrass cover has acceler-
lower areas where light remains sufficient. This trend ated through fall 1994 (Fig. 3).
was apparent. While there was no downward trend Predictions of how much further the losses might
over time in seagrass biomass in depths up to 115 cm proceed a r e difficult to make because the light regime
and equivocal changes between 115 a n d 140 cm, bio- of the lagoon is not uniform and no determination of
mass diminished by >60% in the 2140 cm depth range minimum light requirements was made before the
by the end of the 1993 growing season (Fig. 4 ) . The fact advent of the brown tide. Therefore, it is necessary to
that the reduction from 1992 to 1993 was not signifi- use determinations made elsewhere of minimum light
cant may signal that shoot density in deep areas has requirements to support development of continuous
equilibrated with the low light conditions of the brown meadows of Halodule wrightii and evaluate them in
tide and that no more losses will occur. An alternative terms of the measurements made in this study, when
explanation is that the process of biomass loss contin- the outer boundaries were not fixed. Kenworthy et al.
ues, but the mean for the depth class has not changed, (1991), working in Hobe Sound, Florida, USA, and
because some of the lowest biomass sites of 1992 had Onuf (1991), in lower Laguna Madre, calculated from
gone bare by 1993 and are not represented in the 1993 long-term monitoring of light along transects crossing
determination of mean biomass for vegetated samples. the outer boundary of meadows at many locations that
Detailed observations by Dunton (1996) at one site the ecological compensation point for the species was
where he had also collected continuous data on light where 12 to 20% of surface light reached canopy level.
reaching the bottom are consistent with this interpreta- In this study, sites where 9 a n d 16% of incident PAR
tion of persistence accomplished by reclamation of reached to the depth of the seagrass canopy over the
nutrients and other resources from dying parts. Peak 10 mo period of record had gone bare by December
biomass of rhizomes and roots based on quarterly 1994. The remaining site where 1 4 % of surface light
determinations had diminished every year since the reached canopy level had gone bare by November
beginning of the brown tide, from >500 g mW2 1989 in 1995. Two other studies from Texas closely bracket the
to <200 g m-2 in 1993. minimum light requirement for H. wrightii. In a shad-
The difference in biomass in d e e p water in 1991 and ing experiment in eastern Corpus Christi Bay, Texas,
1992 between the northern and southern parts of the H. wnghtii was eliminated within 9 mo when receiving
study area (Fig. 5) is consistent with the gradient seen 16% S1 (Czerny & Dunton 1995). In upper Laguna
in water clarity (Fig 7). If, as hypothesized, the light Madre, H. wrightii persists at a site receiving a n aver-
regime of the Laguna Madre is driving the changes in a g e of 18 % S1 since the advent of the brown tide (Dun-
biomass and distribution observed over the last 6 yr, ton 1994). These results indicate that there is a region
then the diminution of the north-south contrast in bio- of uncertainty in the determination of minimum trans-
mass in 1993 suggests that the latitudinal gradient in mission of light required to sustain a seagrass meadow,
light has broken down. This is suggested by the con- but that 15% of incident PAR reaching canopy level is
tinuous monitoring records for light at Stns L1 and L2 in the middle of that range. The difference in the sea-
near the south e n d of the study area a n d Stn L3 near sonal patterns of light availability between Stns L2 a n d
the middle of the study area (Fig. 6). The probable L3 (Fig. 6) may explain the anomaly of longer persis-
cause for the deterioration of water clarity at Stn L3 is tence of seagrass at the site with less light. Perfor-
that the bottom has gone bare around it (Fig. 3). With- mance of the seagrasses is likely to be governed more
out an intact seagrass meadow to buffer the action of strongly by light availability in the growing season
waves and the fabric of rhizomes and roots to bind the than by availability over the whole year. Between April
sediments, resuspension of bottom sediments is likely and August 1994, half as much light reached canopy
230 Mar Ecol Prog Ser 138: 219-231, 1996
level at Stn L3 as at Stn L2, even though more llght got ditlonal North American wintering areas such as
to canopy level at Stn L3 over the whole period of Chesapeake Bay, Pamlico Sound and Galveston Bay
record from October 1993 to August 1994. have been abandoned already (midwinter waterfowl
These findings generally validate the estimate of survey compilations, Office of Migratory Bird Manage-
minimum light necessary to maintain the seagrass ment Files, U.S. Fish and Wildlife Service, Laurel, MD)
meadow that was used in this study to project the ulti- and major modifications are proposed for the Laguna
mate extent of seagrass loss caused by the persistent Madre de Tamaulipas, just across the delta of the Rio
brown tide. Therefore, the difference between the Grande in Mexico, the second most important redhead
extent of bare bottom measured so far (Fig. 10c) and wintering area.
that expected on the basis of where llght at canopy This study sheds no light on the critical question of
level drops below 15% of what reaches the surface of how long the brown tide may last. Stockwell et al.
the water (Fig. 10a) suggests that considerably more (1993) reasonably speculated that regional drought
seagrass will be lost before seagrass meadows of upper and hypersalinity established conditions that favored
Laguna Madre come to equilibrium with the brown- the brown tide organism. Then a freeze-caused fish
tide-influenced light regime. This interpretation is kill conceivably provided the limiting organic or inor-
least reliable for the northern part of the study area. ganic compounds necessary to promote rapid growth.
where continuous records of underwater light are not Limited water exchange and reduced grazing pressure
available and where incursions of the brown tide are undoubtedly contribute to unprecedented persistence
shorter and more sporadic than to the south. Here, the of the brown tide (Stockwell et al. 1993); however,
1 to 2 mo intervals between sampling in the Secchi disc some additional source of nutrients also seems neces-
survey may not have been sufficient to represent the sary (T.Whitledge pers. comm.),and ammonium is the
light regime, or there may have been episodes of clear most likely candidate, since the brown tide alga is not
water during the growing season sufficient to sustain able to use nitrate (DeYoe & Suttle 1994).One possible
seagrasses, despite low light for the year as a whole. source could be ammonium regenerated from dying
The weakest assumptions used in predicting the ulti- seagrasses. If this is the case, the brown tide might be
mate extent of seagrass cover to be lost as a result of expected to persist as long as the seagrass meadows
brown tide shading are the choice of 15% of incident continue to recede. However, possible contributions
PAR reaching canopy level as the minimum light from watershed sources cannot be discounted.
required by Halodule wrightii and the attribution of
different Secchi depth versus PAR attenuation rela- Acknowledgements. J . J Ingold assisted in all phases of the
tionships (Fig. 8) to winter and summer periods based study. L. J. Hyde assisted in the laboratory. K. H. Dunton
on their different wind regimes. Results using Dunton's kindly supplied data on incident PAR. Data on water level and
wind speed and direction were kindly supplied by the Texas
(1994)estimate of 18 % S1 as the minimum requirement Coastal Ocean Observation Network, operated by the Conrad
of H. wrightii and attenuation based on the mean of all Blucher Institute for Surveying and Science at Texas A&M
observations of % S1 yielded somewhat higher esti- University, Corpus Christi. D. W. Hicks and B. B. Hardegree
mates of area of seagrass lost (Table 4 ) . Since H. prepared final versions of maps The manuscript has bene-
fited from the comments of E. J Buskey, K H. Dunton, J. J .
wrightii still survives at Dunton's (1994) site and has Ingold and 2 anonymous reviewers.
succumbed at the 3 stations with continuous light mon-
itoring records in this study, the actual limit may lie
between. I have no independent evidence to evaluate LITERATURE CITED
the relative merits of the alternative approaches for
estimating light at canopy level from Secchi measure- Backman TW, Barilotti DC (1976)Irradiance reduction: effects
ments. on standing crop of the eelgrass Zostera manna in a
coastal lagoon. Mar Biol34:33-40
Whether losses extend to 18 or 27 % of the study Batiuk RA, 01th RJ, Moore KA, Dennison WC, Stevenson JC,
area, they are of considerable concern, because they Staver L W , Carter V, Rybich NB, Hickman RE, Kollar S,
follow major losses of seagrass cover and displacement Bieber S, Heasly P (1992) Chesapeake Bay submerged
of Halodule wnghtii by other species in lower Laguna aquatic vegetation habitat requirements and restoration
targets: a technical synthesis. US Environmental Protection
Madre (Quammen & Onuf 1993). In aggregate, the
Agency, Chesapeake Bay Program, Report CBP/TRS 83/92
area of H. wnghtii in the Laguna Madre has dimin- Bulthuis DA (1983) Effects of in situ light reduction on density
ished >30% between the mid 1960s and 1994. Laguna and growth of the seagrass Heterozostera tasmanica,
Madre is the primary wintering area of redhead ducks (Martens ex Ascher.) in Western Port, Victoria, Australia.
Aythya americana, and, while in residence, the red- J Exp Mar Biol Ecol6?:91-103
Cornelius SE (1977) Food resource utilization by wintering
heads feed almost exclusively on H. wrightii (Cor- redheads on lower Laguna Madre. J Wildl Mgmt 4 1 .
nelius 1977). Degradation of the forage stock in the 374-385
lagoon is a particular concern, because some other tra- Costa J E (1988) Eelgrass in Buzzards Bay: distribution, pro-
Onuf: Seagrass loss in response to light reduction 23 1
duction, and historical changes in abundance. US Envi- ing programs to protect seagrasses. NOAA Tech Mem
ronmental Protection Agency, Office of Marine and Estu- NMFS-SEFC-287, NOAA
arine Protection Report No. EPA 503/4-88-002 Lewis RR I11 (1989)Biology and eutrophication of Tainpa Bay.
Czerny AB, Dunton KH (1995) The effects of in situ light In: Estevez ED (ed) Tampa and Sarasota Bays: issues,
reduction on the growth of two subtropical seagrasses, resources, status, and management. National Oceanic and
Thalassia testudinum and Halodule wrightii. Estuaries 18: Atmospheric Administration, Estuarine Programs Office,
418-427 NOAA Estuary-of-the-Month Seminar Series No. 11,
Dennison WC (1987) Effects of light on seagrass photosynthe- Washington, p 89-1 12
sis, growth and depth distribution. Aquat Bot 27:3-14 Merkord GW (1978) The d~stributionand abundance of sea-
Dennison WC, Alberte RS (1982) Photosynthetic response of grasses in Laguna Madre of Texas. MS thesis, Texas A&I
Zostera marina L (eelgrass) to in situ manipulations of University, Kingsville
light intensity. Oecologia 55:137-144 Onuf CP (1991) Light requirements of Halodule wrightii,
Dennison WC, Orth RJ, Moore KA. Stevenson JC, Carter V. Syringodium filiforme, and Halophila engelmanni in a
Kollar S, Bergstrom PW, Batiuk RA (1993) Assessing water heterogeneous and variable environment inferred from
quality with submersed aquatic vegetation: habitat long-term monitoring. In: Kenworthy WJ, Haunert D (eds)
requirements as barometers of Chesapeake Bay health The light requirements of seagrasses: proceedings of a
BioSci 43:86-94 workshop to examine the capability of water quality crite-
DeYoe HR, Suttle CA (1994)The inability of the Texas 'brown n a , standards and monitoring programs to protect sea-
tide' alga to use nitrate and the role of nitrogen in the ini- grasses. US Department of Commerce, National Oceanic
tiation of a persistent bloom of this organism. J Phycol 30: and Atmospheric Administration, National Marine
800-806 Fisheries Service, NOAA Tech Mem NMFS-SEFC-287,
Duarte CM (1991) Seagrass depth limits. Aquat Bot 40: p 95-105
363-377 Onuf CP (1994) Seagrasses, dredging, and light in Laguna
Dunton KH (1994) Seasonal growth and biomass of the sub- Madre, Texas, USA. Estuar Coast Shelf Sci 39.75-91
tropical seagrass Halodule wrighth in relation to continu- Onuf CP (1996) Biomass patterns in seagrass meadows of the
ous measurements of underwater irradiance. Mar Biol Laguna Madre, Texas. Bull Mar Sci 58:404-420
120:479-489 Orth RJ. Moore KA (1983) Chesapeake Bay: a n unprece-
Dunton KH (1996) Photosynthetic production and biomass of dented decline in submerged aquatic vegetation. Science
the subtropical seagrass Halodule wrightii along an estu- 2225-53
arine gradient. Estuaries 19436-447 Pedersen MF, Borum J (1993) An annual nitrogen budget for
Gallegos ME, Merino M, Rodriguez A, Marba N. Duarte CM a seagrass Zostera marina population. Mar Ecol Prog Ser
(1994) Growth patterns and demography of pioneer 101:169-177
Caribbean seagrasses Halodule wnghtli and Syringodlum Pulich W Jr (1980) The ecology of a hypersaline lagoon: the
filiforme. Mar Ecol Prog Ser 109:99-104 Laguna Madre. In: Fore PL, Peterson RD (eds) Proceed-
Giesen WBJT, van Katwijk MM, den Hartog C (1990) Eel- ings of the Gulf of Mexico coastal ecosystems workshop.
grass condition and turbidity in the Dutch Wadden Sea. US Fish and Wildlife Service Biological Report FWS/OBS-
Aquat Bot 37:71-85 80/30, Albuquerque, NM
lverson RL, Bittaker HF (1986) Seagrass distribution and Pulich WM, White WA (1991) Decline of submerged vegeta-
abundance in eastern Gulf of Mexico waters. Estuar Coast tion in the Galveston Bay system: chronology and relation-
Shelf Sci 22577-602 ships to physical processes. J Coast Res ?:909-926
Kenworthy WJ, Fonseca MS, DiPiero SJ (1991) Defining the Quammen ML, Onuf CP (1993) Laguna Madre seagrass
ecological light con~pensationpoint for seagrasses Halod- changes continue decades after salinity reduction. Estuar-
ule wrightij and Syringodium fil~formefrom long-term ies 16:303-311
submarine light regime monitoring in the southern Indian Stockwell DA, Buskey EJ. Whitledge TE (1993) Studies on
River. In: Kenworthy WJ, Haunert DE (eds) The light conditions conducive to the development and mainte-
requirements of seagrasses: proceedings of a workshop to nance of a persistent 'brown tide' in Laguna Madre,
examine the capability of water quality criteria, standards Texas. In- Smayda TJ, Shimizu Y (eds) Toxic phytoplank-
and monitoring programs to protect seagrasses. NOAA ton blooms in the sea. Elsevier Science Publishers, New
Tech Mem NMFS-SEFC-287, NOAA, p 106-113 York, p 693-698
Kenworthy WJ, Haunert DE (1991) The light requirements of Ward LG, Kemp WM, Boynton WR (1984) The influence of
seagrasses. proceedings of a workshop to examlne the waves and seagrass communities on suspended particu-
capability of water quality criteria, standards and monitor- lates in a n estuarine embayment. Mar Geol 59:85-103
This article was presented by G. W. Thayer (Senior Editorial Manuscript first received: A p r ~ 25, 1995
l
Advisor), Beaufort, North Carolina, USA Revised version accepted: January 24, 1996
Vol. 138: 219-231.1996 Published July 25
Mar Ecol Prog Ser
Seagrass responses to long-term light reduction
by brown tide in upper Laguna Madre, Texas:
distribution and biomass patterns
Christopher P. Onuf*
National Biological Service, Southern Science Center, 6300 Ocean Drive, Corpus Christi, Texas 78412, USA
ABSTRACT- A brown tide caused by a very dense bloom of an as yet undescribed species of the new
class Pelagophyceae was first reported in upper Laguna Madre, Texas, USA, in June 1990 and has
been there continuously through December 1995. No change in response to reduced light was evident
in the distribution of the seagrass Halodule wrightii along transects sampled before the brown tide in
1988 and resampled after initiation of the brown tide in 1991 and 1992; however, in winter 1993-94
losses were documented over 2.6 km2 of bottom and by winter 1994-95 the area of vegetation lost had
more than tripled to 9.4 km2 Changes in biomass presaged the changes in distribution. Decreases in
biomass at depths >1.4 m were evident 2 yr before bare areas were detected. Reductions in biomass
were more pronounced toward the south, in keeping with a gradient of increasing light attenuation
from north to south. Support of a hminishing number of new shoots by reclamation of nutrients and
stored reserves from senescing shoots and rhizomes may allow H. wrightii to persist under conditions
of insufficient light for periods greatly in excess of the life span of any one shoot. This postulated capa-
bility would account for the pattern of diminishing biomass over time where the seagrass persists in
deeper areas and the long lag between light reduction and change in distribution where the seagrass
succumbed.
KEY WORDS: Distribution . Biomass . Seagrass . Halodule wrightii . Light . Irradiance . PAR . Brown
tide . Subtrop~cal Texas
INTRODUCTION tions, has occurred gradually over an extended period,
has been accompanied by many structural and hydro-
Clear water is increasingly recognized as a key req- logic changes of the coastal environment, and has
uisite for the development and maintenance of healthy occurred with little or no documentation of conditions
seagrass meadows, and conversely, reductions in before losses were noticed. Consequently, the connec-
water clarity have been implicated in large-scale tion between seagrass loss and water clarity can be dif-
losses of seagrass (Kenworthy & Haunert 1991). The ficult to document.
causes of reduced water clarity range from nutrient In contrast to a gradual, anthropogenically driven
enrichment (Costa 1988, Lewis 1989, Pulich & White change, the advent of a brown tide in upper Laguna
1991, Batiuk et al. 1992, Dennison et al. 1993), to Madre, Texas, USA, in June 1990 resulted in almost
increased suspended loads resulting from hydrological instantaneous light reduction that has persisted to
alteration (Giesen et al. 1990) and frequent resuspen- December 1995. Also, unlike many locations suffering
sion of dredge deposits (Onuf 1994).In most cases, the diminished water clarity and seagrass loss, antecedent
process of water clarity loss and seagrass decline has conditions were well documented. Seagrass distribu-
been the result of anthropogenic inputs and modifica- tion and biomass had been assessed in 1988 (Quam-
men & Onuf 1993),and a multidisciplinary study of the
'Present address: National Biological Service, Midwest Sci-
upper lagoon had been under way for a year when the
ence Center. 6300 Ocean Drive, Corpus Christi, Texas brown tide began, and was continued for 3 yr after
78412, USA. E-mail: chris-onuf@nbs.gov (Stockwell et al. 1993, Dunton 1994). These unusual
Q lnter-Research 1996
Resale o f full article not permitted
Mar Ecol Prog Ser 138: 219-231, 1996
circumstances lend themselves to a particularly strong
assessment of effects of reduced light on seagrass
meadows. The abrupt and persistent reduction in light
by the phytoplankton bloom is more akin to the manip-
ulative field experiments of Backman & Barilotti (1976)
and Dennison & Alberte (1982) than to the inferred re-
constructions necessary in most trend analyses. How-
ever, the change of the light regime is on the scale of
the whole ecosystem, not a tiny fragment.
The effect of the brown tide on the light regime of La-
guna Madre was abrupt and large and of unprece-
dented persistence [Stockwell et al. 1993). Prior to the
initiation of the brown tide, upper Laguna Madre was
renowned for the clarity of its waters (Pulich 1980). In
the 13 mo leading up to the beginning of the brown tide
in June 1990, chlorophyll concentrations never
reached as high as 20 pg I-'. In the next 11 mo, chloro-
phyll concentrations reached 90 pg 1-' and seldom fell
as low as 20 pg I-' (Stockwell et al. 1993, their Fig. 2).
The brown tide alga, an undescribed species in the
newly recognized class Pelagophyceae (DeYoe & Suttle
1994) was not seen before June 1990 but exceeded 106
cells ml-' thereafter. Light transmission through the top
1 m of the water column dropped from a mean of 47 %
the year before the brown tide to 19% the first year of
the brown tide (calculated from Stockwell et al. 1993,
their Fig. 2). Continued sampling at one site documents
that the influence of the brown tide persists. Annual to-
tal irradiance reaching the seagrass canopy in the 5 yr
since the start of the brown tide has ranged from 33 to Fig. 1 Map of south Texas coastal region showing location of
study area in the northern part of upper Laguna Madre (USA)
54 % of what it had been the year before the brown tide
began (Dunton 1994, pers. comm.).That the brown tide
is largely responsible for the light reduction is indicated Laguna Madre of Texas north of Baffin Bay (Fig. l ) ,the
by a highly significant ( p < 0.001, r2 = 0.60) relationship part of the lagoon in which the brown tide was most
between monthly measures of diffuse attenuation coef- prevalent. A survey used to determine the status of sea-
ficient and chlorophyll concentration (Dunton 1994). grasses in Laguna Madre in 1988 (Quammen & Onuf
Here, I report an assessment of the spatial pattern 1993, Onuf 1996) provided an assessment of seagrass
and time course of effects of light attenuation by the distribution and biomass-depth relations before the
brown tide on seagrass distribution and biomass brown tide. Sampling transects generally ran perpen-
afforded by resampling seagrasses in 1991, 1992, 1993 dicular from shore either west or east to the Gulf Intra-
and 1994 at the same time of year along the same tran- coastal Waterway (Fig. 2). Dredge deposits on one side
sects as were sampled in 1988. In addition, I predict the of the waterway or the other prevented continuous
ultimate extent of seagrass loss owing to light reduc- transects across the whole lagoon. Intervals between
tion by brown tide by integrating reports of minimum transects averaged 5 km. Sampling stations were l 0 0 to
light requirements for Halodule wrightii (Kenworthy et 400 m apart, depending on location and proximity to a
al. 1991, Onuf 1991, Dunton 1994, this study) with boundary, along 17 transects. Positions were deter-
measurements of the underwater light regime and mined by LORAN C or by dead reckoning in 1988, by
depth in different parts of the lagoon. LORAN C in 1991 and by Global Positioning System
(GPS) in 1992, 1993, and 1994. Two transects were
added to the sampling array in 1991 to provide more
METHODS detailed coverage of deeper parts of the lagoon, where
effects of llght reduction by brown tide were expected
Distribution. Observations to detect effects of light to be most strongly expressed. All sampling was carried
attenuation by brown tide on a seagrass meadow dom- out in October and November, except that the 1992
inated by Halodule wrightii were made in the upper sampling was not completed until 8 January 1993.
Onuf. Seagrass loss in response to llght reduction 221
Four 80 cm2 by 15 cm deep cores with plants were middle of the study area (Fig 3). The difference
collected at each station 2.5 m apart along the port between the mean depth of all stations on a transect
side of the boat. The time and depth were recorded at in 1988, before installation of the tide gauge, and the
each, and depths for 1991, 1992, 1993 and 1994 sur- mean of the depths corrected to MSL for the same
veys were standardized by adjusting measured depth LORAN coordinates in 1991 were used to refer 1988
by the deviation from mean sea level (MSL) at sam- depths to MSL.
pling time for the Texas Coastal Ocean Observation
Network tide gauge at Bird Island, located near the
SEAGRASS LOST1993 WLAND
- TRANSECTS SAMPLED 1991-94
IIII~II~~
TRANSECTS SAMPLED ALL YEARS
m SEAGRASS LOST 1994
W
---- GULF !NTRACOASTAL WATERWAY R g 3. Map s h o w ~ n g location of light monitors, tide gauge and
areas of seagrass cover lost durlng the brown tlde Light mon-
itors are designated as in the text and Fig 6 Areas a r e num-
Fig 2 Map of study area showlng location of transects sam- bered ( d l 4 4 ) as in the text and in Table I Dashed line indi-
pled ~n different years cates the course of the Gulf Intracoastal Waterway
222 Mar Ecol Prog Ser 138: 219-231, 1996
Cores were washed on 1 mm screens, the retained 1991 vs 1993 and 1992 vs 1993 for each depth class to
plant material identified, and the dominant and other determine whether there was a cumulative effect over
contributors to cover recorded. Occurrence data were the whole period of the brown tide and whether effects
plotted on maps to depict the distribution of sea- could be discriminated over shorter periods, up to the
grasses. last year for which biomass determinations were avail-
In 1993 and 1994, the edges of bare areas were able. The expectation was that if light attenuation
defined to <0.01' of longitude or latitude by repeatedly resulting from brown tide is affecting biomass, it will
halving the interval between the last station where be manifested in reduced biomass in later years in the
seagrass was encountered and the first station along deeper depth classes. Other differences might suggest
the same transect that was bare (4 bare core samples other causes.
and no seagrass rooted material pulled up on the Light. Underwater photosynthetically active radia-
anchor at that location). When vegetated and bare tion (PAR, 400 to 700 nm) was measured at 3 locations
samples were 0.01' or less apart, a stake was driven at (Fig. 3): Stn L1,just inside the outer boundary of a
the outermost vegetated station. These locations were meadow toward the south end of the study area
also marked with a plastic line-float attached by a (1.66 m MSL, 27" 21.80' N, 97" 22.10' W); Stn L2, 200 m
stainless steel cable secured to a screw-in anchor in shoreward of the boundary location (1.52 m MSL, 27"
some cases, because stakes sometimes disappeared 21.80' N, 97" 21.96' W); and Stn L3, in a deep, continu-
between visits. Similar sampling was conducted be- ously vegetated area near the middle of the study area
tween the regularly sampled transects and along the (1.91 m MSL, 27" 30.93' N, 97" 19.53' W). Underwater
long axis of the bare areas to delineate the bare areas PAR measurements were made at 1 min intervals and
more closely. Differences in the location of the outer integrated every hour on a continuous basis using an
boundary between observations were determined by LI-193SA spherical quantum sensor inputting to an LI-
plotting on maps or were measured with a fiberglass 1000 datalogger (LI-COR, Lincoln, NE, USA). The sen-
tape. In January and February 1995, the locations of all sor was positioned at canopy level, the top of the sen-
surviving boundary markers were determined using sor 25 cm above the bottom. The sensor was wrapped
differential GPS. In most cases, there was no ambigu- in transparent plastic wrap that was replaced at 1 or
ity as to what was meadow and what was not. All 4 wk intervals, depending on season, to minimize foul-
samples at a station were either vegetated or bare. ing. The periods of record were May 1993 to August
Patchy stations were mostly limited to a 20 m band at 1994 for Stns L1 and L2 and October 1993 to August
the outer edge. 1994 for Stn L3. Daily total fluxes of underwater PAR
Biomass. The retained material from 2 of the 4 cores were referred to daily total fluxes of incident PAR as
from each station, randomly chosen by coin toss, was measured with a LI-19OSA 2rc sensor mounted on top of
placed in plastic bags and returned to the laboratory a building at Port Aransas, Texas (27" 52' N, 97" 03' W)
on ice, where the samples were frozen until processed. by K. Dunton, University of Texas.
Processing consisted of thawing and separating into The spatial pattern of light attenuation in the lagoon
live (turgid green and white to beige structures) and was evaluated at approximately monthly intervals
dead (flaccid brown to washed-out maroon) fractions. from March 1992 to November 1994 with a 20 cm
The live fraction was sorted further according to spe- diameter Secchi disc. Measurements were made and
cies and into aboveground (green portions of shoots) depth recorded at stations -100 m from the eastern and
and belowground (root, rhizome, and unpigmented western shores of the lagoon and at the midpoint on
portions of shoots) fractions. The sample fractions were east-west transects at 27" 204,22', 26', 30', 34', 38', and
dried to constant weight at 60°C (72 h), weighed, 40'. All Secchi depth determinations were made by the
ashed at 530°C for 3 h, and weighed again. Dry weight same observer outside the shadow of the boat and
and ash-free dry weight were calculated for all plant were the means of the depths of disappearance of the
parts and dead material and expressed on a per m2 disk on descent and reappearance on ascent. Secchi
basis. depths were related to PAR at depth by simultaneously
Biomass samples were separated into depth classes measuring Secchi depths and determining the percent
to determine whether light attenuation resulting from of surface irradiance (SI) reaching Secchi depth with
the brown tide affected biomass. Depth classes were an LI-193SA spherical quantum sensor mounted on a
defined to yield a sample size of at least 10 in any year lowering frame inputting to an LI-1000 datalogger on
for statistical comparisons between years: <85, 85-1 15, 22 June, 12, 13, 31 October, and 2, 3, 4, 7, 8 November
115-130, 130-140, and 2140 cm depths. Variances 1994. Hourly wind data were obtained from Blucher
were not equal between years regardless of transfor- Institute, Texas A&M University, Corpus Christi,
mation. Consequently, Mann-Whitney U tests were Texas, for the Bird Island station of the Texas Coas-
performed on pair-wise comparisons for 1988 vs 1993, tal Ocean Observation Network (Fig. 3) to interpret
Onuf: Seagrass loss in response to light reduction 223
large differences in light attenuation for small Secchi depths were measured was then conlputed as the long-
depths. Conditions were classified as calm for winds term measure of available light at different depths in
c24 km h-' (15 miles h-') at the time of light and Secchi different parts of the lagoon. An estimate of 15% of
measurements and for protected locations (within surface light reaching canopy level, near the midpoint
200 m of shore on the lee side with respect to prevail- of the published range for the minimum light require-
ing wind) regardless of wind speed. Conditions were ment of Halodule wrjghtii (Kenworthy et al. 1991, Onuf
classified as rough at exposed locations when wind 1991, Dunton 1994), was then used to predict the ulti-
speed was >24 km h-' mate extent of loss of seagrass likely to result from the
The frequency of hourly observations of winds prolonged occurrence of the brown tide.
>24 km h-' from 1 October 1993 to 17 October 1995
was assessed to determine whether high light attenua-
tion associated with rough water was likely to differ RESULTS
between the November to April period of frequent
frontal passage and the May to October period of pre- Distribution
vailing southeasterly breezes. Winds >24 km h-' were
further categorized by direction according to 90" sec- No change in seagrass distribution was evident
tors centered on the long axis of upper Laguna Madre between surveys of 1988 and 1991 or 1992, despite
or at right angles to the long axis. The fetch of the large reductions in light reaching the bottom in upper
winds aligned with the long axis was greater than for Laguna Madre after initiation of the brown tide in
winds oriented cross-wise and presumably promoted June 1990 (Stockwell et al. 1993). These conditions
the development of rougher seas. have persisted at least through December 1995 (Dun-
These relations of % S1 reaching Secchi depth were ton 1994, pers. comm.). The first evidence of distribu-
then used to compute diffuse attenuation coefficients tional change came from installation of a light moni-
corresponding with Secchi depths. The attenuation tor just inside the boundary of the seagrass meadow
coefficients were applied to bathymetric profiles at near the south end of the study area in May 1993. In
intervals of 0.1' longitude along transects at intervals of November 1993, the outer boundary was 30 m land-
1' latitude to generate estimates of the proportion of ward of the monitor (Fig. 3: Area #3). Elsewhere, bare
surface light reaching the bottom according to the areas were encountered in November 1993 on 3 tran-
Beer's law relation: sects at stations that had been vegetated in 1992
(Fig. 3: Areas # l and #2).In aggregate, 2.6 km2 of sea-
Iz/Io = e-kz
grass meadow had been lost between observations
where I, is photon flux density (pm01 photons m-' S-') (Table 1).
at depth z, I. is photon flux density at the surface, k is Boundary locations marked in November and
diffuse attenuation coefficient (In m-'), and z is depth December 1993 were reexamined in December 1994
(m).The mean of these values for all dates that Secchi and January-February 1995. Boundaries had receded
from 50 to 800 m . In January 1995,
Table 1. Statistics on extent of loss of seagrass cover in upper Laguna Madre Area in the the lagoon
(Texas, USA) since the advent of the brown tide. Area numbers ( # l 4 4 ) are as was 8 km long and on average 0.8 km
in the text and Fig. 3 wide (Fig. 3). On the other side of the
Gulf Intracoastal Waterway, Area #2
Area Width of meadow lost ( m ) Length Area Number of extended 5 km, constricted by dredge
Mean M~nimum Maximum (m) (km2) determinations of disposal areas. The outer boundaries
outer boundary
for Areas #3 and #4 In the south had
November 1992 or May 1993 to November 1993 retreated 140 m on average (Fig. 3).
#l 330 260 550 5100 1.7 11 Over the whole period, 9.4 km2 of sea-
#2 560 - - 1450 0.8 1 grass has been lost, more than a 3-fold
#3 30 30 30 3000 01 2 increase since winter 1993 (Table 1).
#4 not surveyed
Total 2.6
November 1993 to January 1995
Biomass
#l 770 220 1340 8100 6.2
#2 430 220 800 5400 2.3
#3 180 150 210 2700 05 Tiends in biomass over time differed
#4 140 80 220 2900 0.4 according to depth (Fig. 4, Table 2). In
Total 9.4 the 2 shallowest depth classes, there
was no change over time. In the 115-
224 Mar Ecol Prog Ser 138: 219-231. 1996
130 cm depth class, the main difference
was that biomass was >30% higher In
1991 than other years. The significant
reduction in biomass between 1992 and
1993 is consistent with a brown tide ef-
fect; however, the lack of a significant
difference between the pre-brown tide
year and 1993 argues against this inter-
pretation.
In the 130-140 cm depth class, bio-
mass was elevated in 1991 also (Fig. 4).
However, in this case, there was a sig-
nificant reduction in biomass between
the pre-brown tide year and 1993, the
last year for which biomass data are
available. The reduction from 1992 to
85 85-115 115-130 130- 140 > 140
1993 was not significant (Table 2). DEPTH CLASS (CM) WITH RESPECTTO MEAN SEA LEVEL
The strongest indication of a cumula-
tive effect of brown-tide-caused light
m
1988 1991 m 1992 0 1993
attenuation on seagrass biomass is the Fig. 4. Mean biomass in 5 depth classes over 4 yr. 1988, before the advent of
monotonic decrease of biomass over the brown tide; and 1991, 1992, 1993, under the influence of the brown tide.
See Table 2 for statistical comparisons
time in the 2140 cm depth class (Fig. 4 ) .
Over the whole period, biomass diminished by >60%.
The reductions were significant from 1988 to 1993 and
from 1991 to 1993 but not from 1992 to 1993 (Table 2).
For the 1140 cm depth class, spatial differentiation
also was evident (Fig. 5). Biomass in the southern half
of the study area was less than in the north in 1991 and
1992 but not in 1993 (Table 3).
Table 2. Mann-Whitney U tests for reductions In biomass in
different depth classes for 3 time periods: 1988 vs 1993, 1991
vs 1993 and 1992 vs 1993. U test statistic; z: standardized
normal deviate; N: sample size earlier year, later year;
p: probability
l
1991 19m 1993
Depth class Time period U z N P YEAR
, north +south
< 85 1988 VS 1993 164 -0.18 20, 17 0.43 Fig. 5. Biomass In the >140 cm depth class In northern vs
1991 vs 1993 148 -0.43 19, 17 0.33 southern parts of the study area from 1991 to 1993. See Table 3
1992 vs 1993 209 -0.68 28, 17 0 25 for statistical comparisons
Table 3. Mann-Whltney U tests for differences ~n biomass in
the > l 4 0 cm depth class in the northern (less affected by
brown tide) and southern (more affected by brown tide) parts
of the study area In different years. U: test statistic; z: stan-
dardized normal deviate; N: sample slze north, sample
size south; p: probability
I year U z N P
Onuf: Seagrass loss in response to light reduction 225
Light
In May 1993, continuously recording
underwater light monitors were in-
stalled at canopy height at the outer
boundary of a seagrass meadow (Stn
L1) and 200 m into the meadow (Stn L2)
to document with precis~onn ~ i n i m u n ~
and sufficient levels of PAR to sustain
established seagrass meadows. In Oc-
tober 1993 a third monitor was installed
17 km to the north in a deep site with
continuous seagrass cover (Stn L3),
with the intent of documenting the light
environment at a clearer location. By
December 1993, the outer boundary I
May-93 Aug-93 Nov-93 Mar-94 Jun-94 Sep-94
had receded 30 m landward from Stn DATE
L1 (Fig. 3). By January 1995, the outer ,site L1 ++ site L2 -site L3
boundary was within 30 m of Stn L2
and had receded past Stn L3 (Fig. 3). Fig. 6. Weekly mean percentage incident PAR reaching canopy level at 3 sta-
Over the period with records from all 3 tions in upper Laguna Madre. See Fig. 2 for locations of monitors In relation to
statlons (10 October 1993 to 21 August shifts in seagrass boundaries
1994), 9 % of incident PAR reached
canopy level at Stn L1, 14% at Stn L2, and 16% at (28 vs 18%),but the relationship was reversed the fol-
Stn L3; however, the temporal distribution of percent lowing spring and summer (6% of incident PAR reach-
of surface PAR reaching seagrass canopy level was ing canopy level at Stn L3 vs 12% at Stn L 2 ) . By No-
highly variable and differed between locations (Fig. 6). vember 1995, seagrass had disappeared at Stn L2 as
A higher percentage of incident PAR reached the sea- well.
grasses in winter (November 1993 to March 1994) than Periodic Secchi disk measurements at stations
in the spring-summer period of active growth (April to throughout the study area revealed a north-south gra-
August 1994) at all stations. Much more light pene- dient in light attenuation (Fig. 7). Based on pooled data
trated to canopy level at Stn L3 than at Stn L2 in winter for 20 sampling dates and 1 to 3 stations on a latitude
transect of sufficient depth to measure
Secchi depths, the percent frequency of
Secchi depths <60 cm (relatively high
light attenuation) increased from 15 to 55
from the north end of the study area to the
south, while the percent frequency of
Secchi readings > 100 cm (relatively clear
water) decreased from 50 to 15. This in-
crease in light attenuation from north to
south is consistent with the higher biomass
seen in the 2140 cm depth range in the
northern half of the study area than in the
southern half in 1991 and 1992 (Fig. 5).
In order to apply this north-south gra-
dient in Secchi depths to predictions of
the area of seagrass likely to be lost
in different parts of the study area as a
40&38 34 30 26 22&20 result of brown tide shading, a corre-
MINUTES NORTH OF 27 DEGREES NORTH
spondence had to be established be-
CM a 6 0 - S O C M 6 0 - 1 0 0 ~~ ~> ~ O O C M tween Secchi depth and light attenua-
Fig. 7 Percentage frequency of observations in different Secchi depth cate-
tion. The % S1 reaching Secchi depth
gories along a latitudinal gradient in upper Laguna Madre, April 1992 to proved to be extremely variable, ranging
September 1994 from 15 to 40 (Fig. 8). For Secchi depths
226 Mar Ecol Prog Ser 138: 219-231, 1996
>60,% S1 reaching Secchi depth tended to decrease as equation for the 'calm' array of points in Flg. 8 ( A and +
Secchi depth increased. For Secchi depths <60 cm,% symbols). Short Secchi depths in November to March
S1 tended to be lower when winds exceeded 24 km h-' were assumed to be the result of episodes of sediment
at exposed locations ( X = 24, SE = 2.1) than for all loca- resuspension associated with winter storms. Conse-
tlons when winds were <24 km h-' (57 = 30, SE = 1.3) quently, the mean % S1 for the 'rough' array in Fig. 8
and for protected locations when winds exceeded 24 was used for Secchi values <60 cm in November to
km h-' ( X = 31, SE = 1.8). April. The relationship for the calm array was used for
The difference in % S1 for Secchi depths <60 cm Secchi depths >60 cm.
between rough water conditions (stiff breeze, exposed
locations) and calrn (moderate breeze or protected
location) was significant ( t = 3.14, df = 22). Therefore,
the mean of the 9 observations for exposed locations
with winds >24 km h-' was taken as representative for
rough conditions (24 % S1 reaching Secchi depth). For
calm conditions, estimates of % S1 were generated
from the linear regression equation for all other points
in Fig. 8: % SIsD= 35.1 - 0.0698 SD (r2= 0.37, df = 39),
where SD is Secchi depth.
Winds exceeded 24 km h-' 39% of the time in the
November to April period of frequent frontal passage
compared to 15 % of the time in May to October, based
on the 2 yr of records analyzed for the monitoring sta-
tion near Bird Island (Fig. 3). The difference was more
pronounced for winds with long fetch, oriented along
the long axis of the lagoon. They prevailed 17 % of the
time in November to April compared to 3 % of the time
in May to October.
The geographic survey of light attenuation by Secchi
depth, the light versus Secchi depth relations and a
bathymetric survey (Fig. 9) were then integrated to
depict % S1 at canopy height over the study area.
Because of the low frequency of winds r 2 4 km h-'
from May to October, especially oriented along the
long axis of the lagoon, estimates of light at depth for
that period were derived from the linear regression
101 I
0 50 100 150 200 S O
SECCHI DEPTH (CM)
A calm X windy, exposed + windy, protected
Fig. 8. Relationship between Secchi depth and percentage of Fig. 9. Bathyrnetric map of study area Dashed line marks the
incident PAR reaching Secchi depth under calm and rough course of the Gulf Intracoastal Waterway Small channels and
conditions emergent or shoal areas are not depicted
Onuf. Seagrass loss in response to light reduction 227
BARE VEGETATED UPLAND
Fig. 10.Bare areas in upper Laguna Madre (a) predicted based on measured hght attenuation, depth and assumption that c15%
of incident PAR reaching canopy level is limiting to meadow development. (b) determined by survey October-November 1988,
before advent of the brown tide; (c) determined by survey October-November 1994, after 4 growing seasons of brown tide
Dashed line indicates the course of the Gulf Intracoastal Waterway
Applying the criterion that long-term mean light ity of light is governing distribution in this system, a
reaching canopy level must exceed 15% of incident seagrass meadow should survive there, regardless of
PAR to sustain Halodule wrightiimeadows (Kenworthy how long the brown tide persists. To the south, increas-
et al. 1991, Onuf 1991, this study) to the Secchi depth ing light attenuation and deeper water yield predic-
and light data of this study, the predicted maximum tions of larger expanses of bare bottom (Fig. 10a). In
depth of seagrasses varied from 180 cm at 27' 38-41' all, on the basls of measured light attenuation, 24 % of
N to 170 cm at 27" 34-37' N, 160 cm at 27" 30-33' N, the study area is predicted to be too deep to support
150 cm at 27" 26-29' N, and 140 cm at 27" 20-25' N. seagrass meadow. In contrast, only 6 5 of the study site
%
The northern part of the lagoon is mostly shallow (82 % was bare before the advent of the brown tide
< 1 m deep north of 27" 37' N compared to 22 % < 1 m (Fig. l o b ) . The greatest change in cover between pre-
deep south of 27" 30' N; Fig. 9 ) .Therefore, i availabil-
f brown tide conditions and that predicted based on
228 Mar Ecol Prog Ser 138: 219-231. 1996
Table 4. Final % of study area predicted to go bare if brown tion. First, much of the lagoon is so shallow that even
tide persists long enough for distribution of seagrasses to drastic light reduction would have no effect on distrib-
reach steady state with the light regime for different assump-
tions about minimum light requirements and relations
ution or biomass over much of the area. Almost 50% of
between light attenuation and Secchi depth the study area is <1.0 m deep, while the Halodule
wrightii meadow extended at least to 1.8 m before the
% S1 at Secchi depth Minimum l ~ g h requirement
t
brown tide began. Second, there was a d~stinct gradi-
for Halodule wrightii ent in light attenuation increasing from north to south.
15% S1 18% S1 Since the long-term light record for the upper lagoon
(Dunton 1994) is from the south, much of the lagoon
Varies with season and wind 18 24
was not subject to such severe light deprivation as
Does not vary systematically 20 27
(use mean for all conditions)
measured there. Third, the resolution of the initial
sampl~ng program in 1988 was so coarse that substan-
tial changes in distribution could have occurred with-
out being detected. The original sampling design was
available llght with brown tide is in the middle of the patterned after an earlier survey of the lagoon
study area (Fig. 10a, b). Documented losses also are (Merkord 1978) to maximize the power of the analysis
concentrated in the middle of the study area (Fig. 10c); for trends between studies (Quammen & Onuf 1993).
however, as yet, the area pred.icted to go bare from This meant that relatively little effort was expended in
measurements of the underwater light regime is con- the most critical areas for detecting possible effects of
siderably larger than what is bare (Fig. 10a, c). light limitation: deep areas with vegetation and the
Because the assumptions about the seasonality of zone where vegetation terminated in deep water.
light attenuation versus Secchi depth and the mini- Fourth, the morphology of the lagoon is such that the
mum light requirement for Halodule wrightii are sub- middle region is relatively flat, while the slope is rela-
ject to question, losses were estimated for 2 other tively steep toward the edge. In the one part of the
assumptions. In one case, no systematic variation of % lagoon where the seagrass meadow terminated in
S1 with respect to Secchi depth was assumed, and the deep, bare bottom, the outer boundary lay in this
mean for all 50 observations (27.8% S1 reaching Secchi region of steeper slope. Therefore, any lateral shift in
depth) was used to estimate '10 S1 reaching canopy outer boundary corresponding to a decrease in the
level (25 cm above bottom) under all conditions. In the compensation depth for the seagrasses was relatively
other case, 18% S1 was assumed to be the minimum small.
necessary to support a seagrass meadow as reported Regardless of the characteristics of the sampling
by Dunton (1994), rather than the 15% suggested in program diminishing its sensitivity to detect change
this study. Both alternative assumptions lead to some- resulting from light reduction, the area of bottom expe-
what higher estimates of bare bottom at steady state riencing light reduction below that required to sustain
with the brown-tide-influenced light regime than as a seagrass meadow was far in excess of the detect~on
described above (Table 4). limits of the sampling program. One possible explana-
tion for the persistence of an established seagrass
meadow under conditions of limiting light is that sur-
DISCUSSION viving shoots cannibalize the resources of adjacent
shoots as they succumb to light limitation. According to
The depth limit of seagrass meadows is commonly this conceptual model, an established meadow might
assumed to be set by light attenuation underwater be able to survive much longer under insufficient light
(Bulthuis 1983, Orth & Moore 1983, Iverson & Bittaker than the life span of individual shoots would suggest
1986, Dennison 1987, Duarte 1991, Dennison et al. by translocation of reserves from dying parts to rhi-
1993).In this study, >50% reduction of light caused by zomes, analogous to the process of reclamation of
a dense phytoplankton bloom compared to pre-bloom nutrients and presumably other materials from senes-
conditions had no discernible effect on distribution cent leaves to support new growth in Zostera manna
until after the fourth growing season under reduced (Pederson & Borum 1993) These resources might then
light. However, an effect on biomass was seen sooner. subsidize survival of existing shoots and even develop-
Biomass at depths > l 4 0 cm was less in the first sam- ment of new shoots. This progressive pooling of
pling after the initiation of the brown tide, after 2 resources from a formerly larger population to support
growing seasons of reduced light, than before. new growth in a progressively smaller population
Four factors contribute to the apparent discordance operating at a deficit would seem necessary to account
between the magnitude of the environmental change for the persistence over years of a species in which the
and the expression of an effect on seagrass distribu- mean life span of individual shoots is 110 d (Gallegos et
Onuf: Seagrass loss in response to light reduction 229
al. 1994). Alternatively, the life span of individual to occur under calmer conditions than previously
shoots may be longer under conditions of light limita- (Ward et al. 1984) a n d more frequently. An indirect
tion than under the conditions assessed by Gallegos effect may be the efflux of nutrients from the sedi-
et al. (1994). ments a s the recently dead seagrasses a r e remineral-
No data a r e available to evaluate the responses of ized, enhancing the growth of the brown tide organ-
populations of shoots; however, the gradual decrease ism.
in biomass in d e e p areas where Halodule wrightii Obviously, there is a limit to how long this holding
persists is consistent with this proposed mechanism. If action by seagrasses can go on, if indeed light has been
reclamation of nutrients and stored reserves is respon- reduced below the ecological compensation point
sible for the persistence of seagrasses in Laguna (sensu Kenworthy et al. 1991). That limit was sur-
Madre under conditions of light limitation caused by passed in some parts of the lagoon in 1993. Approxi-
the brown tide, then a reduction in seagrass biomass mately 3 km2 of seagrass meadow went bare between
over time should be evident in deeper parts of the distributional surveys made in fall 1992 a n d fall 1993,
lagoon where light is now limiting, compared to shal- and the process of loss of seagrass cover has acceler-
lower areas where light remains sufficient. This trend ated through fall 1994 (Fig. 3).
was apparent. While there was no downward trend Predictions of how much further the losses might
over time in seagrass biomass in depths up to 115 cm proceed a r e difficult to make because the light regime
and equivocal changes between 115 a n d 140 cm, bio- of the lagoon is not uniform and no determination of
mass diminished by >60% in the 2140 cm depth range minimum light requirements was made before the
by the end of the 1993 growing season (Fig. 4 ) . The fact advent of the brown tide. Therefore, it is necessary to
that the reduction from 1992 to 1993 was not signifi- use determinations made elsewhere of minimum light
cant may signal that shoot density in deep areas has requirements to support development of continuous
equilibrated with the low light conditions of the brown meadows of Halodule wrightii and evaluate them in
tide and that no more losses will occur. An alternative terms of the measurements made in this study, when
explanation is that the process of biomass loss contin- the outer boundaries were not fixed. Kenworthy et al.
ues, but the mean for the depth class has not changed, (1991), working in Hobe Sound, Florida, USA, and
because some of the lowest biomass sites of 1992 had Onuf (1991), in lower Laguna Madre, calculated from
gone bare by 1993 and are not represented in the 1993 long-term monitoring of light along transects crossing
determination of mean biomass for vegetated samples. the outer boundary of meadows at many locations that
Detailed observations by Dunton (1996) at one site the ecological compensation point for the species was
where he had also collected continuous data on light where 12 to 20% of surface light reached canopy level.
reaching the bottom are consistent with this interpreta- In this study, sites where 9 a n d 16% of incident PAR
tion of persistence accomplished by reclamation of reached to the depth of the seagrass canopy over the
nutrients and other resources from dying parts. Peak 10 mo period of record had gone bare by December
biomass of rhizomes and roots based on quarterly 1994. The remaining site where 1 4 % of surface light
determinations had diminished every year since the reached canopy level had gone bare by November
beginning of the brown tide, from >500 g mW2 1989 in 1995. Two other studies from Texas closely bracket the
to <200 g m-2 in 1993. minimum light requirement for H. wrightii. In a shad-
The difference in biomass in d e e p water in 1991 and ing experiment in eastern Corpus Christi Bay, Texas,
1992 between the northern and southern parts of the H. wnghtii was eliminated within 9 mo when receiving
study area (Fig. 5) is consistent with the gradient seen 16% S1 (Czerny & Dunton 1995). In upper Laguna
in water clarity (Fig 7). If, as hypothesized, the light Madre, H. wrightii persists at a site receiving a n aver-
regime of the Laguna Madre is driving the changes in a g e of 18 % S1 since the advent of the brown tide (Dun-
biomass and distribution observed over the last 6 yr, ton 1994). These results indicate that there is a region
then the diminution of the north-south contrast in bio- of uncertainty in the determination of minimum trans-
mass in 1993 suggests that the latitudinal gradient in mission of light required to sustain a seagrass meadow,
light has broken down. This is suggested by the con- but that 15% of incident PAR reaching canopy level is
tinuous monitoring records for light at Stns L1 and L2 in the middle of that range. The difference in the sea-
near the south e n d of the study area a n d Stn L3 near sonal patterns of light availability between Stns L2 a n d
the middle of the study area (Fig. 6). The probable L3 (Fig. 6) may explain the anomaly of longer persis-
cause for the deterioration of water clarity at Stn L3 is tence of seagrass at the site with less light. Perfor-
that the bottom has gone bare around it (Fig. 3). With- mance of the seagrasses is likely to be governed more
out an intact seagrass meadow to buffer the action of strongly by light availability in the growing season
waves and the fabric of rhizomes and roots to bind the than by availability over the whole year. Between April
sediments, resuspension of bottom sediments is likely and August 1994, half as much light reached canopy
230 Mar Ecol Prog Ser 138: 219-231, 1996
level at Stn L3 as at Stn L2, even though more llght got ditlonal North American wintering areas such as
to canopy level at Stn L3 over the whole period of Chesapeake Bay, Pamlico Sound and Galveston Bay
record from October 1993 to August 1994. have been abandoned already (midwinter waterfowl
These findings generally validate the estimate of survey compilations, Office of Migratory Bird Manage-
minimum light necessary to maintain the seagrass ment Files, U.S. Fish and Wildlife Service, Laurel, MD)
meadow that was used in this study to project the ulti- and major modifications are proposed for the Laguna
mate extent of seagrass loss caused by the persistent Madre de Tamaulipas, just across the delta of the Rio
brown tide. Therefore, the difference between the Grande in Mexico, the second most important redhead
extent of bare bottom measured so far (Fig. 10c) and wintering area.
that expected on the basis of where llght at canopy This study sheds no light on the critical question of
level drops below 15% of what reaches the surface of how long the brown tide may last. Stockwell et al.
the water (Fig. 10a) suggests that considerably more (1993) reasonably speculated that regional drought
seagrass will be lost before seagrass meadows of upper and hypersalinity established conditions that favored
Laguna Madre come to equilibrium with the brown- the brown tide organism. Then a freeze-caused fish
tide-influenced light regime. This interpretation is kill conceivably provided the limiting organic or inor-
least reliable for the northern part of the study area. ganic compounds necessary to promote rapid growth.
where continuous records of underwater light are not Limited water exchange and reduced grazing pressure
available and where incursions of the brown tide are undoubtedly contribute to unprecedented persistence
shorter and more sporadic than to the south. Here, the of the brown tide (Stockwell et al. 1993); however,
1 to 2 mo intervals between sampling in the Secchi disc some additional source of nutrients also seems neces-
survey may not have been sufficient to represent the sary (T.Whitledge pers. comm.),and ammonium is the
light regime, or there may have been episodes of clear most likely candidate, since the brown tide alga is not
water during the growing season sufficient to sustain able to use nitrate (DeYoe & Suttle 1994).One possible
seagrasses, despite low light for the year as a whole. source could be ammonium regenerated from dying
The weakest assumptions used in predicting the ulti- seagrasses. If this is the case, the brown tide might be
mate extent of seagrass cover to be lost as a result of expected to persist as long as the seagrass meadows
brown tide shading are the choice of 15% of incident continue to recede. However, possible contributions
PAR reaching canopy level as the minimum light from watershed sources cannot be discounted.
required by Halodule wrightii and the attribution of
different Secchi depth versus PAR attenuation rela- Acknowledgements. J . J Ingold assisted in all phases of the
tionships (Fig. 8) to winter and summer periods based study. L. J. Hyde assisted in the laboratory. K. H. Dunton
on their different wind regimes. Results using Dunton's kindly supplied data on incident PAR. Data on water level and
wind speed and direction were kindly supplied by the Texas
(1994)estimate of 18 % S1 as the minimum requirement Coastal Ocean Observation Network, operated by the Conrad
of H. wrightii and attenuation based on the mean of all Blucher Institute for Surveying and Science at Texas A&M
observations of % S1 yielded somewhat higher esti- University, Corpus Christi. D. W. Hicks and B. B. Hardegree
mates of area of seagrass lost (Table 4 ) . Since H. prepared final versions of maps The manuscript has bene-
fited from the comments of E. J Buskey, K H. Dunton, J. J .
wrightii still survives at Dunton's (1994) site and has Ingold and 2 anonymous reviewers.
succumbed at the 3 stations with continuous light mon-
itoring records in this study, the actual limit may lie
between. I have no independent evidence to evaluate LITERATURE CITED
the relative merits of the alternative approaches for
estimating light at canopy level from Secchi measure- Backman TW, Barilotti DC (1976)Irradiance reduction: effects
ments. on standing crop of the eelgrass Zostera manna in a
coastal lagoon. Mar Biol34:33-40
Whether losses extend to 18 or 27 % of the study Batiuk RA, 01th RJ, Moore KA, Dennison WC, Stevenson JC,
area, they are of considerable concern, because they Staver L W , Carter V, Rybich NB, Hickman RE, Kollar S,
follow major losses of seagrass cover and displacement Bieber S, Heasly P (1992) Chesapeake Bay submerged
of Halodule wnghtii by other species in lower Laguna aquatic vegetation habitat requirements and restoration
targets: a technical synthesis. US Environmental Protection
Madre (Quammen & Onuf 1993). In aggregate, the
Agency, Chesapeake Bay Program, Report CBP/TRS 83/92
area of H. wnghtii in the Laguna Madre has dimin- Bulthuis DA (1983) Effects of in situ light reduction on density
ished >30% between the mid 1960s and 1994. Laguna and growth of the seagrass Heterozostera tasmanica,
Madre is the primary wintering area of redhead ducks (Martens ex Ascher.) in Western Port, Victoria, Australia.
Aythya americana, and, while in residence, the red- J Exp Mar Biol Ecol6?:91-103
Cornelius SE (1977) Food resource utilization by wintering
heads feed almost exclusively on H. wrightii (Cor- redheads on lower Laguna Madre. J Wildl Mgmt 4 1 .
nelius 1977). Degradation of the forage stock in the 374-385
lagoon is a particular concern, because some other tra- Costa J E (1988) Eelgrass in Buzzards Bay: distribution, pro-
Onuf: Seagrass loss in response to light reduction 23 1
duction, and historical changes in abundance. US Envi- ing programs to protect seagrasses. NOAA Tech Mem
ronmental Protection Agency, Office of Marine and Estu- NMFS-SEFC-287, NOAA
arine Protection Report No. EPA 503/4-88-002 Lewis RR I11 (1989)Biology and eutrophication of Tainpa Bay.
Czerny AB, Dunton KH (1995) The effects of in situ light In: Estevez ED (ed) Tampa and Sarasota Bays: issues,
reduction on the growth of two subtropical seagrasses, resources, status, and management. National Oceanic and
Thalassia testudinum and Halodule wrightii. Estuaries 18: Atmospheric Administration, Estuarine Programs Office,
418-427 NOAA Estuary-of-the-Month Seminar Series No. 11,
Dennison WC (1987) Effects of light on seagrass photosynthe- Washington, p 89-1 12
sis, growth and depth distribution. Aquat Bot 27:3-14 Merkord GW (1978) The d~stributionand abundance of sea-
Dennison WC, Alberte RS (1982) Photosynthetic response of grasses in Laguna Madre of Texas. MS thesis, Texas A&I
Zostera marina L (eelgrass) to in situ manipulations of University, Kingsville
light intensity. Oecologia 55:137-144 Onuf CP (1991) Light requirements of Halodule wrightii,
Dennison WC, Orth RJ, Moore KA. Stevenson JC, Carter V. Syringodium filiforme, and Halophila engelmanni in a
Kollar S, Bergstrom PW, Batiuk RA (1993) Assessing water heterogeneous and variable environment inferred from
quality with submersed aquatic vegetation: habitat long-term monitoring. In: Kenworthy WJ, Haunert D (eds)
requirements as barometers of Chesapeake Bay health The light requirements of seagrasses: proceedings of a
BioSci 43:86-94 workshop to examine the capability of water quality crite-
DeYoe HR, Suttle CA (1994)The inability of the Texas 'brown n a , standards and monitoring programs to protect sea-
tide' alga to use nitrate and the role of nitrogen in the ini- grasses. US Department of Commerce, National Oceanic
tiation of a persistent bloom of this organism. J Phycol 30: and Atmospheric Administration, National Marine
800-806 Fisheries Service, NOAA Tech Mem NMFS-SEFC-287,
Duarte CM (1991) Seagrass depth limits. Aquat Bot 40: p 95-105
363-377 Onuf CP (1994) Seagrasses, dredging, and light in Laguna
Dunton KH (1994) Seasonal growth and biomass of the sub- Madre, Texas, USA. Estuar Coast Shelf Sci 39.75-91
tropical seagrass Halodule wrighth in relation to continu- Onuf CP (1996) Biomass patterns in seagrass meadows of the
ous measurements of underwater irradiance. Mar Biol Laguna Madre, Texas. Bull Mar Sci 58:404-420
120:479-489 Orth RJ. Moore KA (1983) Chesapeake Bay: a n unprece-
Dunton KH (1996) Photosynthetic production and biomass of dented decline in submerged aquatic vegetation. Science
the subtropical seagrass Halodule wrightii along an estu- 2225-53
arine gradient. Estuaries 19436-447 Pedersen MF, Borum J (1993) An annual nitrogen budget for
Gallegos ME, Merino M, Rodriguez A, Marba N. Duarte CM a seagrass Zostera marina population. Mar Ecol Prog Ser
(1994) Growth patterns and demography of pioneer 101:169-177
Caribbean seagrasses Halodule wnghtli and Syringodlum Pulich W Jr (1980) The ecology of a hypersaline lagoon: the
filiforme. Mar Ecol Prog Ser 109:99-104 Laguna Madre. In: Fore PL, Peterson RD (eds) Proceed-
Giesen WBJT, van Katwijk MM, den Hartog C (1990) Eel- ings of the Gulf of Mexico coastal ecosystems workshop.
grass condition and turbidity in the Dutch Wadden Sea. US Fish and Wildlife Service Biological Report FWS/OBS-
Aquat Bot 37:71-85 80/30, Albuquerque, NM
lverson RL, Bittaker HF (1986) Seagrass distribution and Pulich WM, White WA (1991) Decline of submerged vegeta-
abundance in eastern Gulf of Mexico waters. Estuar Coast tion in the Galveston Bay system: chronology and relation-
Shelf Sci 22577-602 ships to physical processes. J Coast Res ?:909-926
Kenworthy WJ, Fonseca MS, DiPiero SJ (1991) Defining the Quammen ML, Onuf CP (1993) Laguna Madre seagrass
ecological light con~pensationpoint for seagrasses Halod- changes continue decades after salinity reduction. Estuar-
ule wrightij and Syringodium fil~formefrom long-term ies 16:303-311
submarine light regime monitoring in the southern Indian Stockwell DA, Buskey EJ. Whitledge TE (1993) Studies on
River. In: Kenworthy WJ, Haunert DE (eds) The light conditions conducive to the development and mainte-
requirements of seagrasses: proceedings of a workshop to nance of a persistent 'brown tide' in Laguna Madre,
examine the capability of water quality criteria, standards Texas. In- Smayda TJ, Shimizu Y (eds) Toxic phytoplank-
and monitoring programs to protect seagrasses. NOAA ton blooms in the sea. Elsevier Science Publishers, New
Tech Mem NMFS-SEFC-287, NOAA, p 106-113 York, p 693-698
Kenworthy WJ, Haunert DE (1991) The light requirements of Ward LG, Kemp WM, Boynton WR (1984) The influence of
seagrasses. proceedings of a workshop to examlne the waves and seagrass communities on suspended particu-
capability of water quality criteria, standards and monitor- lates in a n estuarine embayment. Mar Geol 59:85-103
This article was presented by G. W. Thayer (Senior Editorial Manuscript first received: A p r ~ 25, 1995
l
Advisor), Beaufort, North Carolina, USA Revised version accepted: January 24, 1996