Laolao Bay Road and Coastal Management Improvement

Project: Ecological and Water Quality Assessment

Phase II Report: Integration of water quality and ecological data to characterize

coral-reefs and the drivers of change since 1991

A Report Prepared by the Pacific Marine Resources Institute for the CNMI Division of

Environmental Quality

December 2012

2

Authors:

1Dr. Peter Houk,

2Dr. Ryan Okano,

2John Iguel,

3Dave Benavente, and

2Steven Johnson

1Pacific Marine Resources Institute, PO Box 10003, PMB #1156, Saipan, MP. 96950

(www.pacmares.com); 2CNMI Division of Environmental Quality, PO Box 501304, Saipan, MP.

96950 (www.deq.gov.mp); 3CNMI Coastal Resources Management Office, PO Box 10007,

Saipan, MP. 96950 (www.crm.gov.mp)

Project affiliations:

This assessment was funded by an American Recovery and Reinvestment Act (ARRA) grant,

administered by the National Oceanic and Atmospheric Administration (NOAA) and the CNMI

Division of Environmental Quality. This assessment is a component of the Laolao Bay Road &

Coastal Management Improvement Plan, and is associated with road and drainage improvements

and upland re-vegetation components of the project

3

Table of Contents

Executive Summary: ................................................................................................ 4

Introduction: ............................................................................................................. 7

Environmental regimes ............................................................................................................. 8

Study focus ................................................................................................................................. 8

Study Design and Data Collection Methods: ......................................................... 9

Ecological data collection ......................................................................................................... 9

Water quality data collection ................................................................................................. 11

Ecological data analyses ......................................................................................................... 11

Water quality data analyses ................................................................................................... 12

Examining the contribution of localized stressors ............................................................... 13

Results and Discussion: .........................................................................................13

Water Quality .......................................................................................................................... 13

Reef flat assemblages .............................................................................................................. 14

Causes of change on the reef flat............................................................................................ 15

Reef slope assemblages............................................................................................................ 15

Causes of change on the reef slope ......................................................................................... 17

Summary and conclusions: ...................................................................................17

References: ..............................................................................................................19

Figures:....................................................................................................................23

Tables ......................................................................................................................41

4

Executive Summary:

Due to a favorable natural setting, Laolao Bay is one of Saipan’s most desirable coastal areas,

with some coral reefs valued at over 10 million dollars per square kilometer based upon a recent

economic study. However, in recent times Laolao’s environmental integrity has been declining,

leading CNMI’s natural resource management agencies to prioritize revitalization and

rehabilitation activities. The present study was conducted as part of the American Recovery and

Reinvestment Act (ARRA) grant awarded to the Division of Environmental Quality (DEQ) to

conduct the Laolao Bay “Road and Coastal Management Improvement” project. Briefly, the

project included road construction and drainage improvements in the western portion of the bay

that have decreased land-based sediments being washed into the bay during storm events. The

project also included multi-year, re-vegetation efforts that replanted barren savannah uplands

with native trees and shrubs. In association with these activities, ecological and water quality

monitoring were conducted by the Pacific Marine Resources Institute, in collaboration with DEQ

and CNMI’s marine monitoring program, to evaluate the current ecological and environmental

conditions, and their expected improvement due to project activities.

The present study was designed from historical surveys conducted in 1991 that provided

extensive baseline data to place modern ecological assemblages into context. Designs consisted

of re-establishing six monitoring stations across Laolao Bay, where both ecological and water

quality data were collected using standardized protocols described within the report.

Water quality. Water quality profiles conducted in the shallow waters outside the reef margin

showed two predominant patterns in freshwater discharge across Laolao Bay. During full and

new moon periods when negative tides became evident, the easternmost bay had significantly

lower salinity levels that gradually increased moving westward. It appears this natural gradient

in salinity is an artifact of greater connectivity with the island aquifer, as groundwater seepage

through the reef matrix in eastern Laolao was the proposed mechanisms of delivery during

negative tide events. In contrast, the opposite trends were evident during periods of steady

rainfall, as surface discharge was highest in the western portion of the bay, where the largest

volcanic watersheds exist, and salinity increased moving eastward.

Monthly sampling of reef flat waters at six stations revealed that ~60% of the variance in nutrient

and turbidity data could be explained by a single principle components analysis axis, thus

providing a single variable that accurately indicated ‘pollution loading’ across Laolao. Among

the six sampling stations, pollution loading was significantly predicted by minimum daily tide

height for the easternmost, sheltered reef flat waters. These results agree with the water quality

profiles, and confirm the isolated contribution of nutrient-rich groundwaters in easternmost

Laolao Bay. Elsewhere, significant ties between rainfall and pollution were strongest,

suggesting that non-point source pollution from surface runoff was a major driver of water

quality patterns.

Temporal changes in water quality were also evident. PCA analysis of historical water quality

data revealed similar overall spatial trends, but differences in the relationships between sampling

stations. Historical data also highlighted the greatest pollution loading in the easternmost Laolao

Bay, where groundwater discharge was most prolific, followed by the sampling stations

5

associated with the two largest watersheds. However, multivariate analyses revealed that spatial

patterns were not statistically differentiated in 1991, while based upon the present data, trends

have now become statistically differentiated. The enhanced distinctions between monitoring

sites suggests that pollution loading has grown in a spatially inconsistent manner over the past 20

years, described in greater detail within.

Reef flat assemblages. Reef flat benthic substrates were dominated by the brown macroalgae

Sargassum polycystum, along with several persistent red algae, including Laurencia and

Palisada. While macroalgae were ubiquitous throughout Laolao, the species abundances

patterns differed between sites, and even among transects within a site. Comparative analyses

between 1991 and present indicated significant increases across numerous taxa, most notably

persistent red macroalgae and the seasonal brown alga, Sargassum. Through time, the three

sampling stations associated with the highest ‘pollution loading’ (described above) have become

most distinct, suggesting negative changes were greatest for these sites. Overall, reef flat

macroalgal assemblages were well predicted by water quality patterns. Multivariate correlation

examination revealed that the similarity matrices associated with the algal assemblages and water

quality were similarly structured, supporting that nutrients and turbidity were predictors of algal

growth, diversity, and relative abundance on the reef flats.

Reef slope assemblages. Benthic surveys reported lower coral coverage in the eastern portion

of Laolao, where groundwater influences were noted. Beyond coverage, the composition of

coral assemblages was also significantly different. Where groundwater influence was noted,

coral assemblages were comprised of smaller Porites, Leptoria, Favia, and Galaxea colonies,

genera that provide for reduced structure compared with other sites where Montipora, Porites,

Acropora, Pocillopora, and other branching corals were more common.

Perhaps the strongest indication that negative change had occurred through time comes from

examining coral population structures. We report a universal increase in coral population density

across the entire bay, as many small corals continue to replace fewer larger ones. The nature of

these comparisons suggested high partial mortality of large colonies over the years, and a failure

of new corals to grow to adult sizes. Both are consistent with studies reporting increased

localized stressors (i.e., water pollution and/or reduced grazing) as potential causes. Notably,

two stations had coral assemblages contrary to the main findings, where colony sizes and

assemblage evenness remained more similar through time.

Reef slope fish assemblages were dominated by small-bodied acanthurids (i.e., herbivorous

surgeonfishes) with a near absence of piscivore and benthivore consumers observed. Mean food-

fish biomass observed during each 3-minute, 5 m radius, stationary point count was less than 2

kg. One site showed contrasting findings, similarly noted above as where coral and benthic

assemblages had the least amount of negative change through time. Overall, the low abundance

and diversity of food-fish assemblages were indicators of reduced ecological function (i.e.,

grazing and nutrient processing). In support, comparisons of fish assemblages between 1991 and

the present found significant declines in abundances and diversity since 1991, whereby

functionally diverse assemblages have been replaced mainly by small acanthurids that grow and

reach reproductive maturity across much shorter time periods. We caution that while robust,

temporal comparisons represent two snapshots in time that provided one indication of

6

compromised fish assemblages. Further evidence comes from examining spatial gradients across

Laolao Bay.

In contrast to the reef flat benthic assemblage that were spatially predicted by water quality

trends, reef slopes coral and benthic assemblages were influenced by a suite of localized

stressors and natural environmental regimes. Hierarchically, the negative change and current

status of reef slopes were predicted by herbivore/detritivore fish assemblages, water quality, and

wave energy. Regression modeling indicated that a consistent suite of independent predictor

variables explained 57% to 83% of the variance in coral and benthic assemblages on the reef

slopes. The strongest individual contributor to all predictive models was mean

herbivore/detritivore fish size, however, best-fit models were interactive in most instances. It

follows that water quality appears to be an influential, but secondary stressor for reef slope

assemblages, synergistically driving negative change alongside herbivory. Wave exposure

represented a tertiary predictor of favorable benthic and coral assemblages. Exposure is a

continuous, natural environmental driving regime that has long been known to facilitate the

flushing of nearshore waters, as well as nutrient transfer and metabolic exchange of shallow reef

assemblages. Thus, moderate exposure regimes enhance the growth of calcifying organisms

such as corals on the reef.

Conclusions. This study provided quantitative, measurable data that described the current

condition of Laolao bay’s coral reef ecosystem and how it has change over time. The main

purpose of these data are to provide a baseline upon which (positive) change can be detected due

to ARRA-project activities that addressed water quality concerns. The present evidence suggests

that reef flat algal assemblages are expected to show the greatest improvement in integrity due to

project-related road drainage improvements, however, the ecological response may require

several years before becoming evident, as the ecological assemblages we see today are formed

through the time-integrated responses of individual species to driving environmental regimes and

localized stressors. Beyond the present project, these data also provide guidance for

conservation planning activities pertaining to the marine habitats found within Laolao Bay.

7

Introduction:

Throughout the past decade, the Commonwealth of the Northern Mariana Islands (CNMI) long-

term marine monitoring program has been documenting change over time on 30 of CNMI’s

nearshore coral reefs. The most notable trends over this time period included a series of

disturbance years (2004 to 2006) when high abundances of Crown-of-Thorns starfish

(Acanthaster planci) caused extensive coral mortality across CNMI’s reefs. Interestingly, coral

reef recovery following predator starfish activity has not been uniform. Particularly relevant, the

nearshore coral reefs in Laolao Bay have experienced limited to no recovery over the past 5 to 6

years despite successful coral recovery elsewhere (http://www.cnmicoralreef.net/monitoring.htm,

Houk and van Woesik 2010). These trends corroborate public opinion and insight from CNMI’s

resource management agencies suggesting that water quality and coral/fish assemblages continue

to decline in Laolao Bay (CNMI Coral Reef Initiative 2009). In support, compromised coral reef

assemblages in Laolao have previously been attributed to watershed development (Houk and van

Woesik 2010), and corroborated with decreased herbivory as a result of unsustainable fishing

practices that have been described for the CNMI (Houk et al. 2012). It is well known that

together, poor water quality and reduced herbivory provide an advantage for macroalgae to

outcompete corals for space on the reefs, and through time, result in less architecture to support

desirable, biologically diverse assemblages (Smith et al. 2001; Lapointe et al. 2004; Burkepile &

Hay 2006; Mork et al. 2009). However, the relative influence of water quality versus herbivory

in driving negative change, and reducing recovery capacity following disturbances, remains

unclear (Houk et al. 2010b). This lack of clarity is problematic because management strategies

and evaluations of management success hinge upon understanding the nature and magnitude of

localized stressors on coral reefs, especially in terms of their prioritization of activities given

limited resources.

The negative trends that have emerged from Laolao bay over the past 10 years provided a

foundation for existing management plans and strategies aimed at mitigating further decline

(CNMI Coral Reef Initiative 2009). Using the developed management plans, the CNMI Division

of Environmental Quality (DEQ) applied for and received funding from the American Recovery

and Reinvestment Act (ARRA) to conduct the Laolao Bay “Road and Coastal Management

Improvement” project. Other documents describe the project in greater detail, but briefly, the

project included several rehabilitation activities within the watershed: 1) road construction along

the steep entrance road in the western portion of the bay, 2) the installation of stream crossings

for six major intermittent streams, and 3) revegetation of native plants in a denuded western

portion of the bay. In perpetuity, the project goals are to decrease the sediment load being

washed into the bay during storm events.

In association with the ARRA project, extensive ecological and water quality monitoring were

conducted by the Pacific Marine Resources Institute (PMRI), Division of Environmental Quality

(DEQ), and Coastal Resources Management Office (CRM) to evaluate the current status of

marine resources and their expected improvement due to project activities. Here, we summarize

integrated findings from both ecological surveys and water quality monitoring activities, and use

comparative datasets developed in 1991 as part of the Laolao Bay golf course construction

(Cheenis Pacific Company 1992), to assist in evaluating the magnitude and cause of negative

change in Laolao Bay over time. Historical data provided a robust snapshot of the coral-reef

8

ecosystem in Laolao in 1991. Comparisons with modern fish, coral, macroinvertebrate, and

algal assemblages, as well as water quality, provide a rarely available context to indicate changes

that have likely occurred over longer time periods.

Environmental regimes

Modern reef assemblages in Laolao Bay are influenced by natural environmental regimes that

deserve attention prior to assessing ecological condition or describing ecological dynamics.

Most influentially, watershed geology differs greatly across the bay, as karst limestone bedrock

dominates the watersheds in the easternmost portion of the bay, while moving westward,

volcanic soils and bedrock become more pronounced (Cloud 1959). As a consequence, there is

significantly greater connectivity with the freshwater aquifer in the eastern side of the bay, while

the thick soils and volcanic basement facilitate the transfer of surface water discharge in the

western side of Laolao. Differences in the nature of freshwater delivery to the nearshore marine

environment have consequences for coral reef assemblages, which shift predictably with the

amount and frequency of freshwater influence (Houk & Starmer 2010). Thus, we expect a

greater influence of watershed pollution where continuous groundwater discharge occurs in

comparison with intermittent flows due to storm events.

The degree of wave exposure also changes throughout the bay. Reefs in the easternmost part of

the bay are sheltered from prevailing northeast, tradewind-generated waves, while moving

westward, wave exposure gradually increases. Similar to freshwater discharge, wave energy also

structures coral reef assemblages (Houk & Starmer 2010). Generally, moderate wave exposure

(i.e., energy) and shallower depths are associated with more vigorous coral growth due to

increased metabolic rates (Birkeland 1997; Yentsch et al. 2002) and flushing of suspended

particulate matter that reduce the impacts of localized pollution discharge (Fabricius 2005;

De'ath & Fabricius 2010). However, coral growth can be inhibited by the destructive force of

continuous, high wave exposure (Grigg & Maragos 1974; Storlazzi et al. 2005), while extremely

low exposure facilitates the retention of waters being discharged from land (Wolanski et al.

2003). With respect to the CNMI, wave exposure in eastern Laolao bay can be characterized as

‘low’, while moving westward, ‘moderate’ wave exposure becomes evident.

Study focus

The present study describes the current condition of Laolao Bay’s coral reef ecosystem and

provides a foundation for anticipating the benefits of the Laolao ARRA project that was

completed in July 2012. Using an extensive snapshot of coral-reef assemblages from 1991

described above, we first quantify major shifts that likely occurred over the past two decades.

The temporal comparisons provided an indicator of change, and also where negative change was

most pronounced within Laolao. In order to study the causes of negative change, we next

perform spatial investigations to quantify how localized stressors (i.e., water pollution and/or

grazing potential by fish and urchins) and environmental regimes dictate the modern coral-reef

assemblages in Laolao. Spatial investigations provided a foundation for comprehending the non-

uniform, negative changes reported by the temporal comparisons, as well as the modern

gradients in coral and algal assemblages across Laolao Bay. Within, we isolate upon three likely

drivers of negative change, and note that their influence differs by habitat (i.e., reef flat and reef

9

slope). We last summarize our findings to provide a framework for assessing the effectiveness

of the recently completed ARRA project into the future, and better focus overall management

strategies for Laolao Bay.

Study Design and Data Collection Methods:

Laolao Bay represents a unique geological feature on the east coast of Saipan, CNMI, where

well-developed coral reef assemblages exist due to their protection from prevailing northeast

tradewind-generated swells and the presence of large volcanic watersheds that have created

gentle bathymetric slopes conducive for reef development. The present study examined coral

reef assemblages and water quality within six sub-drainages across Laolao Bay, following

sampling designs that were established in 1991 as part of an extensive ecological assessment

conducted for the Laolao Bay golf course (Figure 1). Ecological data were collected from two

key habitats: 1) the outer reef flat, and 2) the reef slope at 3–5 m depth. In addition, one new

station was established further south of existing sites in order to evaluate the watershed drainage

associated with the San Vicente access road into the future, which was a major component of

construction activities (Station D1, Figure 1). Only reef slope assemblages were examined from

station D1 as no reef flat habitat exists. Finally, in both instances benthic and coral surveys were

conducted during October, in the midst of the rainy season.

Data from historical fish surveys were only available from a spring survey event in 1991, while

the present data were collected in the fall of 2011. However, this study is focused upon temporal

comparisons for food-fish only, and thus smaller non-food fish, known to have a larger seasonal

variation in comparison (Galzin 1987; Thompson and Mapstone 2002) were excluded from

analyses. Thompson and Mapstone (2002) found significant diurnal variation for some food-fish

families on the Great Barrier Reef, Australia, most notably the surgeonfishes and parrotfishes.

However, when examining variation across days and months, disproportional influences of daily

versus monthly variation were noted for most food-fish families. In sum, variation in fish

assemblages stemmed mainly from short term sampling related errors or behavioral phenomena.

This represents a random component of variation included in statistical models, prior to assess

significance. For the present study, we appreciate the presence of short term variation, however,

we focus upon a comparison across 20 years, whereby the magnitude of change was much larger

in comparison. Temporal comparisons are used as one indicator of changing fish assemblages.

Temporal comparisons are augmented with spatial comparisons described below.

Ecological data collection

In 1991 benthic substrate abundances were estimated along each transect by haphazardly tossing

five, 1 x 1 m quadrats at ten-meter intervals. Benthic substrates were recorded under each of 16

intersecting cross-lines within the quadrat, using the highest taxonomic resolution possible,

typically at the genus or species level. Turf and crustose coralline algae were separated, while

rubble and sand were combined into a single category. These methods yielded a total of 80 data

points per transect, and three transect to generate statistical estimates at the station level. Current

effort used the same number of transects, and transect placement, but increased the level of

replication to improve statistical confidence. Fifty, 0.5 x 0.5 m quadrats were placed at 1 m

intervals. On the shallow reef flats, substrate was recorded under each of 6 intersecting cross-

10

lines within the quadrat using the highest taxonomic resolution possible, typically at the generic

level. On the reef slope, a digital photograph was taken at each 1m interval, and the benthos

under each of five random points were assigned a pre-defined category using the freely available

computer software, Coral Point Count (Kohler & Gill 2006). These methods yielded a total of

250 (reef slope) or 300 (reef flat) data points per transect, and three transect to generate sound

statistical estimates at the station level (Houk & Van Woesik 2006). The benthic categories

chosen for analysis were corals (to genus level), turf algae (less than 2 cm), macroalgae (greater

than 2 cm, to genus level if abundant), fleshy coralline algae known to overgrow coral

(Peyssonnelia, Pneophyllum) (Keats et al. 1997; Antonius 1999, 2001) crustose coralline algae,

sand, and other invertebrates (genus level if abundant). Means, standard deviations, and standard

errors were calculated based on the three 50 m replicates, with n = 300 individual points per

transect, n = 900 data points per site for the reef flat, and n = 250 points per transect, n= 750 data

points per site for the reef slope.

For both survey periods coral population data were collected in a similar manner. In each

instance, replicate 1 x 1 m quadrats were tossed at equal intervals along the three, 50 m transect

lines. In 1991, fifteen replicate quadrats were sampled while the present surveys used ten

following an initial inspection of the data showing sufficient statistical power to meet survey

demands, as well as species saturation considerations. Within each quadrat, all corals were

identified to the species level, the maximum diameter and the diameter perpendicular to the

maximum were recorded. Surface area was calculated from these measurements considering

colonies were circular in nature.

Algal occurrence and richness data were collected by haphazardly tossing 18, 1 x 1 m quadrats

along the three 50 m transect lines at ~8 m intervals. Within each quadrat presence/absence data

were collected for all algal species that were visually distinguishable. In the event that species

names were uncertain, collections were made and/or unique field names were recorded. Beyond

replicated, quadrat-based presence/absence data, these methods also provided for frequencies of

occurrence by pooling data from the individual quadrats to the site level.

For both survey periods macroinvertebrate abundances were collected using identical methods.

All conspicuous macroinvertebrates that were observed along 50 x 2 m belt transects were

identified to the species level and recorded. Statistical estimates were calculated based upon the

three replicate transects surveyed for each station.

For both survey periods fish population data were collected in a similar manner. Twelve

stationary-point-counts (SPC’s) were conducted at equal intervals along the transect lines at each

station (Bohnsack & Bannerot 1986). During each SPC, the observer recorded the name and size

of all food-fish within a 5 m circular radius at varying taxonomic resolution based upon

functional similarity (i.e., large-bodied acanthurids, large-bodied scarids, Naso unicornis, N.

lituratus, small-bodied acanthurids, etc.). Large-bodied fish were defined as species that have a

mean reproductive size > 25 cm, while small-bodied were defined below 25 cm. Food fish were

defined by acanthurids, scarids, serranids, carangids, labrids, lutjanids, balistids, kyphosids,

mullids, and holocentrids that are a known to be harvested. Statistical estimates were calculated

based upon twelve replicate SPC’s at each station. In 1991 no size estimates were recorded, so

historical comparisons are based upon population densities, while the contemporary baseline

11

includes biomass estimates as well. Size at maturity estimates and length-to-biomass

coefficients were collected from fishbase (www.fishbase.org).

Water quality data collection

Spatial profiles were conducted on a quarterly basis to characterize discharge patterns for the

nearshore reef slopes adjacent to each study region (1-3 m depth, outer reef) using a YSI 6600

EDS that was attached to a slow moving boat (n = 6 total events). In each instance, an auxiliary

bilge pump system was used to generate a continuous flow of nearshore marine waters (~0.5 m

in depth) through the YSI sensor casing system, and the boat was driven along the reef line

following a pre-defined track that was confirmed with GPS locational data collected at 10 second

intervals. High resolution data (±0.01 in all cases) were collected for salinity (ppt), conductivity

(S/m), turbidity (NTU), chlorophyll-a (mg/L), pH, and temperature (degree C). Based upon

preliminary insight from our phase I report, two types of freshwater discharge events were

focused upon: 1) groundwater discharge during the lunar maximum and minimum when tidal

exchanges were largest, and 2) surface water discharge following major storm events.

In addition to conducting profiles on the outer reef slopes, monthly sampling was conducted at

six, permanent reef flat monitoring stations that were established during the 1991 study (Figure

1). Reef flat sampling was conducted over the course of a year for all stations, during mid-to-

high tides in the morning hours (9 to 11 AM). During each event, sampling consisted of taking

YSI measurements (same parameters noted above) as well as grab samples from the top 0.5 m of

water that were later processed for total suspended sediments (CNMI Division of Environmental

Quality Laboratory) and nutrients (University of Guam Water and Energy Research Institute

Laboratory). All samples for nutrient analyses (nitrate, nitrite, total nitrogen) were filtered

immediately, stored on ice or frozen, and processed within 48 hours.

Ecological data analyses

Historical data from the 1991 ecological assessment of Laolao bay were only available in printed

format, as appendices to the submitted report. Combined efforts from the authors transcribed

these data for processing and analysis. However, in some instances historical data consisted of

summaries in lieu of raw data themselves. This artifact introduced some limitations for our

ability to conduct statistical examinations of change over time.

Raw data pertaining to historical coral colony size distributions were only available in

summarized formats. Species abundance and colony-size summaries for each site were used to

generate randomized, truncated, log-normal distributions to estimate size-frequency distributions

(Preston 1980). For each site the known means, standard deviations, and number of colonies

observed served as inputs to generate size distribution estimates using the freely available

software R (R Development Core Team 2008). We performed 1000 iterations of this process

and a representative of the most frequent distribution (i.e., the convergent distribution) was

selected for use. In order to test the accuracy of this assumption, the same process was

conducted for the 2010 colony size datasets, and pairwise Kolmogorov-Smirnov (KS) tests

revealed no significant differences between the generated distributions and the raw

12

measurements. To compare coral colony size distributions between 1991 and 2010 similar KS

tests were conducted between colony-size distributions at each site between years.

Coral-colony data were also assessed in multivariate space, using site-level summary statistics.

In general, multivariate statistics described herein are well summarized by Anderson et al.

(2008). Bray-Curtis similarity distances were calculated between each pair of sites, for both

timeframes, and resultant relationships were visualized in non-metric, multi-dimensional scaling

plots (n-MDS) (Anderson et al. 2008). Tests of significance were conducted to examine the

change in coral assemblages through time, and across Laolao bay, due to gradients in water

quality. If data had similar multivariate dispersion among the a priori groups being investigated

(i.e., time frame and nature of freshwater discharge, measured by a PERMDISP test (Anderson et

al. 2008), parametric multivariate testing was conducted (PERMANOVA). If dispersion was

significantly different, non-parametric tests were used to examine separation (ANOSIM).

Historical fish population density data were available in raw format and provided an opportunity

to draw comparisons through time based upon numeric density. Prior to analyses, fish data were

grouped by genus, and trophic function within relevant genera, to provide a more informative

interpretation of characteristic difference between the two time frames. Multivariate analyses

were used to highlight the extent and nature of temporal comparisons. Bray-Curtis similarity

distances were calculated between each pair of SPC’s, for both timeframes, and resultant

relationships were visualized using n-MDS plots. Tests of significance were conducted to

examine the change in fish assemblages through time, and across Laolao bay due to gradients in

water quality, as described above.

Algal diversity data were assessed in multivariate space, using site-level summaries from quadrat

surveys. No historical data were available that provided adequate estimates of species richness

or frequency of occurrence, thus, detailed insight regarding algal assemblages are based upon the

present findings. Initially, presence/absence simple matching similarities were calculated

between each pair of sites, and these measures of multivariate differences were visualized in

MDS plots (Anderson et al. 2008). Subsequently, tests of significance were conducted to

determine if algal assemblages differed across the bay. Described above, statistical testing

included initial tests to determine multivariate normality (PERMDISP). If the within-site

variance significantly differed, data were transformed into rank-based numbers, and ANOSIM

tests were conducted for pairwise examination. Else, PERMANOVA pairwise testing was

conducted.

Standardized pairwise tests were utilized to examine benthic and algal assemblage change

between timeframes at each monitoring station. In all instances, initial homogeneity of variance

tests were conducted to ensure that dependent variables met assumptions of normality, and data

transformations were conducted as necessary to validate assumptions. If data were not normal,

non-parametric testing was conducted to examine for differences.

Water quality data analyses

Nutrient and turbidity data from monthly reef flat water sampling were available from both the

historical and present data collection effort. Because of shifting analytical techniques used to

13

process the samples and technological improvements to nutrient sampling in general, no direct

temporal comparisons were appropriate. In lieu, relative comparisons across the sampling

stations were made for each time period. Data were compared using ANOSIM testing on rank-

transformed data, and visualized using principle components ordinations.

The 2011/2012 nutrient constituent data were also analyzed for their relationships with

groundwater and surface discharge events. Regressions were conducted between the ranked

nutrient data and: 1) maximum low tide during the date of sampling, and 2) 48-hour rainfall

values collected from the Saipan airport rain gauge.

Examining the contribution of localized stressors

Multiple regressions were conducted to examine the relative influence of natural and human

influences in driving spatial trends in coral and benthic substrate reported. Independent

predictors included: 1) water quality, measured by the score of the first axis of a principal

components analysis that explained ~60% of the variance, 2) mean herbivore abundance, body

size, and multivariate assemblage distinction (i.e., distance between sampling centroids, which is

an indicator of how different any particular fish assemblage was compared to all others), and 3)

wave exposure, calculated by the angle of exposure and mean wave height from each quadrant

(Houk & van Woesik 2010). Dependent variables examined included a suite of ecological

metrics that were available from the collected datasets: 1) coral cover, population density,

skewness/kurtosis, and multivariate distinction (i.e., centroid differences), 2) benthic substrate

ratio’s, 3) algal species richness and multivariate assemblage distinctions. Our goal in

examining a suite of dependent variables was to determine if consistent drivers of ecological

change existed, and if so, what generalizations they might infer. Only single variable regression

models were considered given the small number of sampling stations (n=6), and goodness-of-fit

was evaluated using the Akaike Information Criterion.

Results and Discussion:

Water Quality

Water quality profiles defined inherent differences in the nature of watershed discharge into

Laolao bay. Profiling activities conducted during full and new moon periods with no rainfall

found comparatively lower salinity levels in the eastern portion of the bay where karst bedrock

exists in the watershed (Figures 2 and 3). Thus, it appears the natural gradient in salinity is an

artifact of greater connectivity with the island aquifer, as freshwater seepage through the reef

matrix is the proposed mechanisms of delivery during large, negative tide events associated with

full and new moon periods (Corbett et al. 1999; Umezawa et al. 2002; Houk et al. 2005; Houk &

Starmer 2010). In contrast, the opposite trends were seen during periods of consistent rainfall

and limited tidal exchange to promote connectivity with the aquifer (Figures 2 and 3). Surface

discharge was highest in the western portion of the bay, where the largest volcanic watersheds

exist.

Principle components analysis (PCA) of the reef flat water quality data revealed that ~60% of the

variance in nutrient and turbidity data was explained by the first PCA axis (Figure 4 and 5).

14

These results confirmed that the first PCA axis provided a useful, univariate indicator of

pollution loading that could be used for subsequent regression examinations with rainfall and

groundwater discharge. Among the six sampling stations, pollution loading was significantly

predicted by groundwater discharge for the easternmost, sheltered reef flat waters (station B2,

Figure 1), supporting the findings from water quality profiles, and confirming the isolated

contribution of nutrient-rich groundwaters in eastern Laolao Bay (Table 1). Elsewhere,

significant ties between rainfall and pollution were found where major hydrodynamic channels

were absent, and reef flat waters had longer retention times (i.e., all stations except for B5 and

C5). While channels may appear to facilitate the dispersal of reef flat waters, both B5 and C5

consistently had the highest nutrient and turbidity levels, except for where groundwater

influences were noted (B2). Thus, the lack of any relationship with rainfall suggests that these

waters may be continuously enriched, a notion which is supported through high algal biomass

and diversity (Figure 4 and 5), and/or augmented by nutrient addition through enhance biological

decomposition and cycling of macroalgae. In support, ammonia was consistently highest for

these two stations as compared with others, indicating rapid biological cycling of available

nitrogen. In sum, the cumulative evidence suggests that non-point source pollution is related to

rainfall intensity across the majority of Laolao Bay, and notably in the vicinity of recently

completed ARRA activities. However, non-point source pollution was not uniform across all

monitoring stations due to differing algal growth and the nature of freshwater discharge.

In addition to spatial differences in water quality across Laolao Bay, temporal changes were also

evident. PCA analysis of historical water quality data revealed similar overall spatial trends, but

differences in the magnitude of inter-site differences (Figure 5). Historical data similarly

highlighted the greatest pollution loading in the easternmost Laolao Bay, where groundwater

discharge was most prolific, followed by the sampling stations associated with the two largest

watersheds (B5 and C5) where hydrodynamic channels exist. However, multivariate analyses

revealed that while patterns existed, they were not statistically differentiated in 1991 (ANOSIM

R-statistic < 0.3). In contrast, the water quality datasets collected during the present study

revealed that the same spatial trends have now become statistically differentiated (Table 2). The

enhanced distinctions between monitoring sites suggests that pollution loading has grown in

magnitude over the past 20 years, with the most notable negative changes at stations B2, B5, and

C5, respectively (Table 2).

Reef flat assemblages

Benthic data collected from the reef flat habitat showed three dominant categories covered ~90%

of the substrate; brown, red, and turf algae (Figure 6). Brown algal growth was comprised

mainly of Sargassum polycystum, a seasonal algae common to reef flats of the Marianas (Tsuda

2003). Other notable contributions of Padina tenuis were also evident in a spatially inconsistent

manner (Figure 7). Red algal growth was comprised mainly of Laurencia spp. and Palisada sp.,

more persistently growing algae, known to proliferate where poor water quality and reduced

herbivory exist (Burkepile & Hay 2009; Houk & Camacho 2010). While red algae were prolific

throughout Laolao, the relative abundance of individual species varied between sites, and even

between transects within a site.

15

Comparatively, the algal assemblage at site B1 was most distinct from all others (R-Statistic >

0.45 for all but one pairwise comparisons, P<0.05, ANOSIM) based upon high abundances of

small gelids (i.e., red turf algae), Boodlea (i.e., siphonous green algae), as well as low

abundances of Laurencia and Palisada (SIMPER, genus that accounted for the top 30% of

differences). Examinations of intra-site heterogeneity (i.e., diversity of the algal assemblages

within each monitoring station) support this trend. Lowest heterogeneity was found at site B1

(P<0.05, PERMDISP), suggesting that strong, continuous environmental regimes existed at site

B1. The continuous addition of poor water quality appears to be most advantageous to a limited

number of algal species, resulting in low heterogeneity. Elsewhere across the reef flats,

coverage, species richness, and heterogeneity were higher, with maximal values being reported

for site C4, where watershed size is largest.

Comparative analyses between 1991 and present indicated significant increases in macroalgae on

Laolao reef flats (P<0.05, pairwise t-test for all sites grouped). Largest increases were found for

persistent red algae, described above, however increases in seasonal Sargassum growth were also

evident (Figure 8). The largest increases in macroalgae were found at sites adjacent to road

improvement activities (sites C1 – C9).

Causes of change on the reef flat

Modern reef flat assemblages were predicted by water quality patterns. Multivariate correlation

examination revealed that the similarity matrices associated with the algal assemblages and water

quality were similarly structured (RELATE, Rho = 0.64, P<0.05; Figure 9), supporting that

nutrients and turbidity were good predictors of algal growth, diversity, and relative abundance.

In support, numerous studies have found increased algae proliferation in association with

enhanced groundwater influence and/or nutrient delivery from runoff waters (Umezawa et al.

2002; Lapointe et al. 2004; Houk et al. 2010a).

Reef slope assemblages

Benthic surveys reported lower coral coverage in the groundwater influenced, eastern portion of

Laolao, where turf algae dominated the substrate (pairwise t-test for coral and turf algae

abundances at site B19-24 compared with others, P<0.05 in both instances, Figure 10 and 11).

The coral assemblage composition was also significantly different between the two regions, but

also through time as well (PERMANOVA, F-Statistic >3.5, P<0.05, for both comparisons,

Figure 12). Significant interactive effects indicated that the strongest changes through time

occurred where groundwater influences were high (F-Statistic = 4.29, P<0.01, Figure 12). Here,

coral assemblages were mainly comprised of smaller Porites, Leptoria, Favia, and Galaxea

colonies. These encrusting and massive growth forms provided reduced coral-reef architecture,

or three-dimensionality, compared with other survey stations. Elsewhere, larger colonies of

encrusting Montipora, massive Porites, corymbose Acropora, Pocillopora, and Stylophora, and

an increased richness of a variety of corals were found in greater abundance.

Comparisons with historical benthic substrate data found non-significant changes in coral

coverage when grouping all stations together, a potential artifact of noted Crown-of-Thorn

starfish activity just prior to the 1991 surveys (Cheenis Pacific Company 1992), and a lack of

16

any major disturbance over the five years prior to the present surveys (CNMI Marine Monitoring

Team long-term datasets). While the frequency of disturbance and recovery cycles since 1991 is

not well understood, deeper examinations of the coral assemblages beyond coverage provides

insight into their changing integrity. We report a universal increase in population density at all

stations, especially in the western portion of the bay (P<0.05, grouped and pairwise t-tests,

Figure 13), as encrusting Montipora, massive Porites, Leptastrea, and Favia corals have all

shifted from fewer large colonies to higher densities of juveniles. These significant trends infer

partial mortality of larger coral colonies, as well as selection against coral colony growth and

survival. Both are consistent with published reports describing ecological states under the

influence of high sediment loads, enhanced nutrients, and decreased herbivory (Rogers 1990;

Fabricius 2005; Dikou & van Woesik 2006; Houk et al. 2010a). Notably, there were two

monitoring stations where contrasting findings emerged, and coral assemblages remained

statistically similar in terms of composition and population density through time (B25-27 and

C25-27, Figure 13). These stations represent localized areas of refuge where coral population

structure showed minimal change. However, when considering the cumulative changes in coral

assemblages across Laolao Bay, a significant reduction in integrity has become evident since

1991.

Reef slope fish assemblages were dominated by small-bodied acanthurids (i.e., herbivorous

surgeonfishes) with a near absence of piscivore and benthivore consumers observed (Figure 14).

Mean food-fish biomass per 5 m, 3-minute SPC was less than 2 kg. One site showed differences

from these generalized findings, B25-27, noted above as where coral and benthic assemblages

have had non-significant change through time. Elsewhere, the low abundance and diversity of

food-fish assemblages are characteristic of high fishing pressure and suggest that a compromised

grazing function may exist (Jennings et al. 1995; Bellwood et al. 2003; Elmqvist et al. 2003). In

support, modern food fish abundances and diversity were significantly lower than they were in

1991 (Figures 15 and 16). The temporal comparison highlights that more diverse assemblages of

large-bodied fish (i.e., species with reproductive sizes greater than 25 cm) have been

disproportionally replaced by small acanthurids (i.e., faster growing species with reduced grazing

potential). The extent, nature, and consequences of compromised fish stocks are furthered

below, where they are investigated alongside water quality trends and wave exposure regimes to

provide a more holistic understanding of the coral ecosystem condition across Laolao Bay.

Dominant macroalgae that were found in more than half of all survey quadrats on the reef slopes

included Amansia glomerata, Amphiroa fragilissima, and Dictyota friabilis. Pairwise

examinations of site-level differences across the bay were mainly non-significant, especially in

comparison to the reef flat algal assemblages that showed the opposite trends. The only

exception was a significant pairwise difference between the algal assemblages at sites B19 and

D1 (R-Statistic >0.45, P<0.05, ANOSIM, Figure 17), where extreme difference in wave

exposure existed. However, heterogeneity within each site differed (i.e., intra-site diversity).

Sites B19, C19, and C22 had significantly lower diversity among replicate quadrats compared

with other sites (P>0.05 for pairwise comparisons between these three sites, while P<0.05 for

comparisons with others, PERMDISP). In contrast, macroalgal assemblages were most evenly

distributed at sites C25 and B22, where large watersheds drain onto reef flats that are well

connected by hydrodynamic channels in the reef crest.

17

Macroinvertebrate assemblages on the nearshore reef slopes were expectedly dominated by

grazing Echinometra and Echinothrix sea urchins. Echinometra urchins were found in high

densities (~40 per 100m2) at all sites except C19-21, where relatively strong, negative changes in

reef status were noted. In support, historical data show a major reduction in Echinometra at

C19-21, from over 100 individuals per 100 m2 to approximately 10. The larger, black-spined

Echinothrix urchins have increased through time (mean of 0.8 compared to 3.2 per 100 m2,

Mann-Whitney U-test, P<0.05), however current densities are less than 10 individuals per 100

m2, well below Echinometra abundances. Urchins obviously have key functional roles on

Laolao’s reefs, documented as keystone species elsewhere (Edmunds & Carpenter 2001; Mork et

al. 2009), and with the exception of the localized decline at C19-21, this study does not suggest

compromised urchin densities exist for 5 m reef slope assemblages. Other trends in

macroinvertebrate abundances include an increase in sea cucumber densities in the calmer,

eastern side of Laolao, where northeast tradewind generated waves have less influence, and

sandy bottoms are more prevalent. Here, the increase in sea cucumber density has been

accompanied by a shift in dominance from Holothuria edulis to Stichopus chloronotus. No

contemporary scientific evidence exists to corroborate if compositional shifts in sea cucumbers

reflect changing reef integrity, however the increase in abundance of sea cucumbers (grouped)

suggests more detritus material now exists compared to 1991, supporting enhanced nutrient and

sediment delivery.

Causes of change on the reef slope

Regression modeling indicated that a consistent suite of independent predictor variables

explained 57% to 83% of the variance in coral and benthic assemblages on the reef slopes

(Figure 18). The strongest individual contributor to all predictive models was mean

herbivore/detritivore fish size, however, best-fit models were interactive in most instances,

inclusive of fish size, wave exposure, and water quality. These findings support that mean

herbivore size, rather than biomass, may be a good indicator of grazing potential, and the ability

to regulate undesirable benthic substrates. In support, ecological function is known to increase

exponentially with fish size (Lokrantz et al. 2008). It follows that water quality appears to be an

influential, but secondary driver of benthic and coral assemblages across Laolao Bay reef slopes.

Similar findings have been reported elsewhere for reef slopes with varying watershed discharge

patterns and fish abundances (Houk et al. 2010b), whereby herbivory was the strongest driver of

recovery patterns in American Samoa, and water quality represented a synergistic, but secondary

driver. Last, wave exposure represented a tertiary predictor of favorable benthic and coral

assemblages in Laolao. Exposure is a continuous, natural environmental driving regime that has

long been known to facilitate flushing and nutrient transfer of shallow water reef slope

assemblage (Yentsch et al. 2002). Under moderate exposure, the growth of calcifying corals is

expected, (Grigg & Maragos 1974), however, when extremely high exposure exists, calcifying

crustose coralline algae becomes more prolific (Yamano et al. 2003).

Summary and conclusions:

Since the inception of CNMI’s coral reef initiative program in 2000, numerous local scientists

and resource managers have anecdotally suggested that Laolao’s coral reef environment shows

negative signs of human influence. However, this study is the first to provide a quantitative

18

assessment of the coral-reef assemblages and the causes of their status and change through time.

We summarize that both water quality and compromised fish assemblages were strong drivers of

negative change that has occurred over the past two decades, yet their influence differed by

habitat. Coral reef assemblages on the reef slopes were best predicted by herbivore/detritivore

fish size, while on the reef flat, water quality was most influential to the distribution and

abundance of macroalgae. Coupled with published studies elsewhere (Schaffelke et al. 2005;

Houk & Camacho 2010), the present evidence suggests that macroalgal abundances and diversity

on the reef flat are likely to be the best ecological indicators of successful construction and

revegetation activities associated with this project. However, the ecological response may

require 2 to 3 years before becoming evident, as modern assemblages are formed through the

time-integrated responses of individual species to driving environmental regimes over time

(Connell et al. 1997; Nyström et al. 2000; Brown et al. 2002; Mumby et al. 2005).

Within each habitat (i.e., reef flat and reef slope), the influence of groundwater discharge in

eastern Laolao bay was clearly a stronger driver of modern coral and algal growth compared

with surface runoff in the western bay. Groundwater discharge is known to influence coral-reef

assemblages due to its frequency and magnitude of discharge (Umezawa et al. 2002; Costa Jr et

al. 2008). Elsewhere across Laolao, watersheds that contributed most to pollution runoff were

associated with stations B5 and C5. In both instances, large watersheds drained into reef flat

waters associated with hydrodynamic channels, and diverse and abundant macroalgal growth

was present on the reef flats. ARRA construction and rehabilitation activities are expected to

improve upon this situation. Specifically, construction activities associated with site C5 included

paving and improving the drainage of LaoLao Bay Drive, the road entering Laolao from San

Vicente. Additionally, numerous impermeable stream crossings were installed to reduce the

transport of sediments to the ocean from key locations. Revegetation activities associated with

site B5 have successfully rehabilitated denuded areas in eastern LaoLao watersheds. Current

monitoring of the revegetated areas by DEQ suggests that a high survival rate exists thus far.

Through time then, the adjacent reef flats are ideal locations for future assessments of project

effectiveness. Both ecological monitoring and continued nutrient sampling are recommended

given their coupling.

In contrast with the reef flat habitat, a more complicated situation exists for the reef slopes.

Here, herbivory was the largest driver of negative change. Further studies to describe the

characteristics of the fish assemblages across Laolao Bay are desirable to understand the

magnitude and spatial nature of fisheries issues. Based upon the current evidence, the reef slopes

are expected to show diminished improvement as a result of the ARRA project in comparison to

the reef flats. Holistically, the findings offered here provide a framework for further study of the

nearshore fisheries in Laolao Bay, and offer that overall management strategies for Laolao Bay

should include addressing the suite of localized stressors.

19

References:

Anderson M., Gorley R. & Clarke K. (2008). PERMANOVA+ for PRIMER: Guide to Software

and Statistical Methods. Plymouth, UK, 214pp. PRIMER-E, Plymouth, UK, 214pp.

Antonius A. (1999). Metapeyssonnelia corallepida, a new coral-killing red alga on Caribbean

Reefs. Coral Reefs, 18, 301-301.

Antonius A. (2001). Pneophyllum conicum, a coralline red alga causing coral reef-death in

Mauritius. Coral Reefs, 19, 418-418.

Bellwood D.R., Hoey A.S. & Choat J.H. (2003). Limited functional redundancy in high diversity

systems: resilience and ecosystem function on coral reefs. Ecology Letters, 6, 281-285.

Birkeland C. (1997). Life and Death of Coral Reefs. In. Chapman and Hall, New York,(1997).

Bohnsack J.A. & Bannerot S.P. (1986). A stationary visual census technique for quantatively

assessing community structure of coral reef fishes. In. National Oceanic and Atmospheric

Administration Technical Report.

Brown B.E., Clarke K.R. & Warwick R.M. (2002). Serial patterns of biodiversity change in

corals across shallow reef flats in Ko Phuket, Thailand, due to the effects of local

(sedimentation) and regional (climatic) perturbations. Marine Biology, 141, 21-29.

Burkepile D.E. & Hay M.E. (2006). Herbivore vs. nutrient control of marine primary producers:

Context-dependent effects. Ecology, 87, 3128-3139.

Burkepile D.E. & Hay M.E. (2009). Nutrient versus herbivore control of macroalgal community

development and coral growth on a Caribbean reef. Marine Ecology Progress Series,

389, 71-84.

Cheenis Pacific Company C.P.C. (1992). Quantitative marine baseline survey. Biological

resources and water quality. Bahia Laulau and Unai Laulau Kattan areas, Saipan,

Mariana Islands. In: (ed. Office CCRM) Saipan, MP.

Cloud P.E. (1959). Geology of Saipan, Mariana Islands; Submarine topography and shoalwater

ecology. Geological Survey Professional Paper, 361-445.

CNMI Coral Reef Initiative C.R.I. (2009). Laolao Bay Conservation Action Plan. In: (ed.

Agencies CNRM) Saipan, MP.

Connell J.H., Hughes T.P. & Wallace C.C. (1997). A 30-year study of coral abundance,

recruitment, and disturbance at several scales in space and time. Ecological Monographs,

67, 461-488.

20

Corbett D.R., Chanton J., Burnett W., Dillon K., Rutkowski C. & Fourqurean J.W. (1999).

Patterns of groundwater discharge into Florida Bay. Limnology And Oceanography, 44,

1045-1055.

Costa Jr O.S., Nimmo M. & Attrill M.J. (2008). Coastal nutrification in Brazil: A review of the

role of nutrient excess on coral reef demise. Journal of South American Earth Sciences,

25, 257-270.

De'ath G. & Fabricius K. (2010). Water quality as a regional driver of coral biodiversity and

macroalgae on the Great Barrier Reef. Ecological Applications, 20, 840-850.

Dikou A. & van Woesik R. (2006). Survival under chronic stress from sediment load: Spatial

patterns of hard coral communities in the southern islands of Singapore. Marine Pollution

Bulletin, 52, 7-21.

Edmunds P.J. & Carpenter R.C. (2001). Recovery of Diadema antillarum reduces macroalgal

cover and increases abundance of juvenile corals on a Caribbean reef. Proceedings of the

National Academy of Sciences of the United States of America, 98, 5067-5071.

Elmqvist T., Folke C., Nyström M., Peterson G., Bengtsson J., Walker B. & Norberg J. (2003).

Response diversity, ecosystem change, and resilience. Frontiers in Ecology and the

Environment, 1, 488-494.

Fabricius K.E. (2005). Effects of terrestrial runoff on the ecology of corals and coral reefs:

review and synthesis. Marine Pollution Bulletin, 50, 125-146.

Galzin R. (1987). Structure of fish communities of French Polynesian coral reefs. II. Temporal

scales. Marine Ecology Progress Series, 41: 137-145.

Grigg R.W. & Maragos J.E. (1974). Recolonization of hermatypic corals on submerged lava

flows in Hawaii. Ecology, 387-395.

Houk P. & Camacho R. (2010). Dynamics of seagrass and macroalgal assemblages in Saipan

Lagoon, Western Pacific Ocean: disturbances, pollution, and seasonal cycles. Botanica

Marina, 53, 205-212.

Houk P., Didonato G., Iguel J. & Van Woesik R. (2005). Assessing the effects of non-point

source pollution on American Samoa's coral reef communities. Environ. Monit. Assess.,

107, 11-27.

Houk P., Musburger C. & Wiles P. (2010a). Water quality and herbivory interactively drive

coral-reef recovery patterns in American Samoa. PLoSONE, 5, e13913.

Houk P., Musburger C. & Wiles P. (2010b). Water quality and herbivory interactively drive

coral-reef recovery patterns in American Samoa. PLoS One, 5, e13913.

21

Houk P., Rhodes K., Cuetos-Bueno J., Lindfield S., Fread V. & McIlwain J. (2012). Commercial

coral-reef fisheries across Micronesia: A need for improving management. Coral Reefs,

1-14.

Houk P. & Starmer J. (2010). Constraints on the diversity and distribution of coral-reef

assemblages in the volcanic Northern Mariana Islands. Coral Reefs, 29, 59-70.

Houk P. & Van Woesik R. (2006). Coral reef benthic video surveys facilitate long-term

monitoring in the commonwealth of the Northern Mariana Islands: Toward an optimal

sampling strategy. Pac. Sci., 60, 177-189.

Houk P. & van Woesik R. (2010). Coral assemblages and reef growth in the Commonwealth of

the Northern Mariana Islands (Western Pacific Ocean). Marine Ecology-an Evolutionary

Perspective, 31, 318-329.

Jennings S., Grandcourt E.M. & Polunin N. (1995). The effects of fishing on the diversity,

biomass and trophic structure of Seychelles' reef fish communities. Coral Reefs, 14, 225-

235.

Keats D.W., Chamberlain Y.M. & Baba M. (1997). Pneophyllum conicum (Dawson) comb nov

(Rhodophyta, Corallinaceae), a widespread Indo-Pacific non-geniculate coralline alga

that overgrows and kills live coral. Botanica Marina, 40, 263-279.

Kohler K.E. & Gill S.M. (2006). Coral Point Count with Excel extensions (CPCe): A Visual

Basic program for the determination of coral and substrate coverage using random point

count methodology. Computers & Geosciences, 32, 1259-1269.

Lapointe B.E., Barile P.J., Yentsch C.S., Littler M.M., Littler D.S. & Kakuk B. (2004). The

relative importance of nutrient enrichment and herbivory on macroalgal communities

near Norman's Pond Cay, Exumas Cays, Bahamas: a "natural" enrichment experiment.

Journal Of Experimental Marine Biology And Ecology, 298, 275-301.

Lokrantz J., Nyström M., Thyresson M. & Johansson C. (2008). The non-linear relationship

between body size and function in parrotfishes. Coral Reefs, 27, 967-974.

Mork E., Sjoo G.L., Kautsky N. & McClanahan T.R. (2009). Top-down and bottom-up

regulation of macroalgal community structure on a Kenyan reef. Estuarine Coastal and

Shelf Science, 84, 331-336.

Mumby P.J., Foster N.L. & Fahy E.A.G. (2005). Patch dynamics of coral reef macroalgae under

chronic and acute disturbance. Coral Reefs, 24, 681-692.

Nyström M., Folke C. & Moberg F. (2000). Coral reef disturbance and resilience in a human-

dominated environment. Trends in Ecology & Evolution, 15, 413-417.

Preston F.W. (1980). Noncanonical distributions of commonness and rarity. Ecology, 61, 88-97.

22

R Development Core Team R.D.C. (2008). R: A language and environment for statistical

computing. R foundation for statistical computing, Vienna, Austria. http://www.R-

project.org.

Rogers C.S. (1990). Responses of coral reefs and reef organisms to sedimentation. Marine

ecology progress series. Oldendorf, 62, 185-202.

Schaffelke B., Mellors J. & Duke N.C. (2005). Water quality in the Great Barrier Reef region:

responses of mangrove, seagrass and macroalgal communities. Marine Pollution Bulletin,

51, 279-296.

Smith J.E., Smith C.M. & Hunter C.L. (2001). An experimental analysis of the effects of

herbivory and nutrient enrichment on benthic community dynamics on a Hawaiian reef.

Coral Reefs, 19, 332-342.

Storlazzi C., Brown E., Field M., Rodgers K. & Jokiel P. (2005). A model for wave control on

coral breakage and species distribution in the Hawaiian Islands. Coral Reefs, 24, 43-55.

Thompson A.A., & Mapstone B.D. (2002). Intra- versus inter-annual variation in counts of reef

fishes and interpretations of long-term monitoring studies. Marine Ecology Progress

Series, 232: 247-257.

Tsuda R.T. (2003). Checklist and bibliography of the marine benthic algae from the Mariana

Islands (Guam and CNMI). University of Guam Marine Laboratory.

Umezawa Y., Miyajima T., Kayanne H. & Koike I. (2002). Significance of groundwater nitrogen

discharge into coral reefs at Ishigaki Island, southwest of Japan. Coral Reefs, 21, 346-

356.

Wolanski E., Richmond R.H., Davis G. & Bonito V. (2003). Water and fine sediment dynamics

in transient river plumes in a small, reef-fringed bay, Guam. Estuarine, Coastal and Shelf

Science, 56, 1029-1040.

Yamano H., Abe O., Matsumoto E., Kayanne H., Yonekura N. & Blanchon P. (2003). Influence

of wave energy on Holocene coral reef development: an example from Ishigaki Island,

Ryukyu Islands, Japan. Sedimentary Geology, 159, 27-41.

Yentsch C.S., Yentsch C.M., Cullen J.J., Lapointe B., Phinney D.A. & Yentsch S.W. (2002).

Sunlight and water transparency: cornerstones in coral research. Journal Of Experimental

Marine Biology And Ecology, 268, 171-183.

23

Figures:

Figure 1. A satellite image of Laolao bay showing major watershed delineations, intermittent

streams, water quality sampling locations, and ecological transects surveyed.

24

a)

b)

Figure 2a-b. Satellite image of Laolao bay with water quality and ecological monitoring stations

shown, as well as the results from two characteristic salinity profiling events. Profiles show a

greater influence of freshwater (i.e., lower salinity, lighter colors, see Fig. 3) in the eastern

portion of the bay (a), a probable consequence of limestone bedrock in the watershed promoting

enhanced connectivity with the aquifer during full moon, maximal tidal exchange periods.

During minimal tidal exchange periods and high rainfall, the opposite pattern was found (b), as

freshwater delivery was more proportional to watershed size where volcanic bedrock and soils

exist.

25

Figure 3a-b. Boxplots summarizing freshwater discharge patterns associated with two

characteristics freshwater discharge regimes that existed in Laolao Bay (black lines represents

the median, the boxes indicates the 25th (lower) and 75th (upper) percentiles, and the error bars

indicate the 5th (lower) and 95th (upper) percentiles). Groundwater discharge associated with

lunar extremes (full and new moons and the associated negative low tides) created an east-to-

west gradient (a) of freshwater delivery to Laolao Bay. In contrast, significant rain events

created a west-to-east gradient (b) of freshwater delivery.

26

Figure 4. Map of Laolao Bay showing landuse patterns associated with each sub-watershed

(dashed lines). Intermittent streams that carry runoff water into the nearshore environment are

shown as blue lines. Circles represent reef flat water quality monitoring stations used both in

1991 and the present study.

27

a)

b)

Figure 5a-b. Principle components analysis and ordination of water quality (nutrient and

turbidity) data from monthly samples at six monitoring locations across Laolao Bay (see Figure

1). Constituents used for historical water quality data analysis (a) included NO3-N,PO4-P, and

turbidity. Constituents used for the present dataset (b) were NO3-N, NH4, total N, and turbidity.

28

C7-9 C4-6 C1-3 B7-9 B4-6 B1-3

Pe

rcent

cover

(SE

bars

)

0

20

40

60

80

Brown Algae

Red Algae

Green Algae

Turf

CCA

Figure 6. Benthic substrate abundances on Laolao reef flats. Algae are grouped by phylum

and/or functionality.

29

C7-9 C4-6 C1-3 B7-9 B4-6 B1-3

Pe

rcent

cover

(SE

bars

)

0

10

20

30

40

50

60

70

Acanthophora

Laurencia/Palisada

Padina

Sargassum

Tolypiocladia

Figure 7. Macroalgal composition on Laolao reef flats. Algae are grouped by genus, with

dominant taxa shown.

30

Station

C7-9 C4-6 C1-3 B7-9 B4-6 B1-3change in m

acro

alg

ae p

erc

ent

covera

ge (

2010 -

1991)

-10

0

10

20

30

40

50

Brown algae

Red algae

Figure 8. Change in two dominant algal phyla since 1991. Significant increases in red algae, as

well as all macroalgae grouped together, were found across Laolao (P<0.05, pairwise t-test

between years for all transects across Laolao combined).

31

Figure 9a-b. Ordination plots of water quality (a) and reef flat algal assemblage (b) data with

vectors showing the water quality constituents (a), or algal genus (b), that were most responsible

for the plot structure. Site B2 had the greatest concentrations of nutrients and highest turbidity,

with a steady decrease moving from left to right in the plot (a). Algal growth was highest at site

B2, and decreased moving from right to left, while diversity was highest at site C5, and

decreased moving from top to bottom. The two plots were significantly correlated based upon

inter-site differences (RELATE, Rho = 0.64, P<0.05), depicting that water quality was a strong

driver of reef flat algal assemblages.

32

D1-3 C25-27 C22-24 C19-21 B25-27 B22-24 B19-21

Perc

en

t cover

(SE

bars

)

0

10

20

30

40

50

60

Coral

Crustose coralline

Fleshy coralline

Turf algae

Macroalgae

Figure 10. Percent cover of dominate benthic substrates on Laolao reef slopes. Clear difference

can be seen between survey stations associated with enhanced groundwater connectivity (B19-21

and B22-24) compare with others. These conditions are less favorable for coral growth and more

favorable for turf algae persistence.

33

Station

C25-27 C22-24 C19-21 B25-27 B22-24 B19-21

change in c

ora

l cover

and p

opula

tion d

ensity (

2010 -

1991)

-10

0

10

20

30

% Coral cover

Population density

Figure 11. Change in percent coral cover and colony density per m2 since 1991. Significant

increases in population density are noted throughout Laolao (P<0.05, pairwise t-test), however

changes in coral coverage were spatially inconsistent. Similar to other ecological data collected

on the reef slope, inherent difference can be seen between sites associated with enhanced

groundwater delivery (B19-21 and B22-24) and others.

34

Figure 12. Multi-dimensional scaling plot highlighting spatial and temporal trends in Laolao

Bay coral assemblages. Significant differences were attributed to both groundwater influence

(low vs. high) and change through time (2010 vs. 1991) (PERMANOVA, F-Statistic >3.5,

P<0.05, for both comparisons). Additionally, there was a significant interactive effect (F-

Statistic = 4.29, P<0.05) highlighting that greater changes occurred through time where high

groundwater influence was evident. Vectors display corals that were significant drivers of these

trends (spearman correlation coefficients >0.5).

35

B19-21

Pe

rce

nt

of

tota

l co

lon

ies m

ea

su

res

0

25

50

0

25

50

1991

2010

0

25

50

C19-21

0

25

50

C22-24

0

25

50

C25-27

Colony diameter (cm)

0-2 2-4 4-6 6-8 8-10 10-15 15-20 20-30 30-50 >50

0

25

50

B22-24

B25-27

*

*

*

*

Figure 13. Trends in coral population demography since 1991 (see methods). Each graph shows

comparative colony-size distributions. At four survey stations denoted with an (*) pairwise

Kolmogorov-Smirnov tests showed a significant reduction in mean colony size, and a larger

proportion of juvenile corals, through time.

36

D1-3 C25-27 C22-24 C19-21 B25-27 B22-24 B19-21

Fis

h b

iom

ass p

er

SP

C (

kg

) (S

E b

ars

)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Large-bodied parrotfish

Naso lituratus

Naso unicornis

Small-bodied surgeonfish

Small-bodied parrotfish

Figure 14. Mean fish biomass per stationary point count survey (SPC) on Laolao reef slopes.

Dominant fish, or functionally-similar fish groupings, are shown.

37

Station

C25-27 C22-24 C19-21 B25-27 B22-24 B19-21ch

an

ge

in

me

an

fis

h a

bu

nd

an

ce

pe

r S

PC

(2

01

0 -

19

91

)

-15

-10

-5

0

5

10

Large Acanthurids

Large Emperor/Snapper

Large parrotfish

Unicornfish

Rabbitfish

Small Acanthurids

Small parrotfish

Small emperors/snappers

Figure 15. Change in the population density of dominant fish in Laolao since 1991. A

significant reduction in fish density, evenness, and functional group abundances was evident, as

2010 fish assemblages have been reduced to high densities of small acanthurids.

38

Figure 16. Multi-dimensional scaling plot summarizing findings for Laolao fish assemblages.

More diverse and abundant fish assemblages in 1991 have been replaced by high densities of

small acanthurids. Significant declines in fish density, evenness, and functional groups were

evident since 1991 (ANOSIM R-Statistic = 0.62, P<0.01). Vectors display fish groups that were

significant drivers of these trends (spearman correlation coefficients > 0.5).

39

Figure 17. Multi-dimensional scaling plot highlighting relationships between Laolao reef slope

algal assemblages. Each data symbol represents a replicate presence/absence quadrat (n=16 at

each site, see methods). Comparatively, the algal assemblages at sites B19 and D1 were

statistically distinguished (R-Statistic > 0.45, P<0.05, ANOSIM). However, no other pairwise

differences were noted, as overall algal occurrences were similar throughout most of the bay on

the reef slopes.

40

Figure 18. Graphical summaries of best-fit multiple regression models (see Table 3 for a full

description). Fish size is the mean size for herbivores/detritivores, fish assemblage distinctness

refers to the mean distance between multivariate sampling centroids (i.e., a measure of how

different fish assemblages at any particular site were in comparison to all others, see methods),

multivariate distance between a site-specific fish assemblage and the overall mean, water quality

is the principle component axis-1 score from 2011-12 data (Figure 15), wave is wave exposure,

and remaining variables are self-explanatory.

41

Tables

Table 1. Summary of best-fit regression models describing the relationships between 2011-12

water quality trends, measured by the first principal component axis which explained 56% of the

variance in 2011 reef-flat water quality variation, and two environmental predictors, daily low

tide and rainfall.

Site Equation Slope (SE) Intercept (SE) R2 P-Value

B2 Y = -1.8(low_tide) – 2.7 ±0.7 ±0.3 0.42 0.04

B5 none significant -- -- -- --

B8 Y = -1.7(rainfall) + 1.1 ±0.4 ±0.2 0.63 0.006

C2 Y = -0.6(rainfall) + 1.2 ±0.3 ±0.1 0.35 0.05

C5 none significant -- -- -- --

C8 Y = -2.1(rainfall) + 1.1 ±0.6 ±0.2 0.59 0.008

42

Table 2. ANOSIM results highlighting significant differences in the 2011-12 water quality data

across Laolao Bay reef-flat monitoring stations (ANOSIM Global R-Statistic = 0.45, P<0.001).

Significant groups are defined by pairwise R-Statistics > 0.45, P<0.05. Vectors indicate

constituents that were the strongest drivers of the trends (see methods).

Site Rank PCA axis-1 Significant groups

B2 6 -3.9 a

B5 5 -0.5 b

B8 1 1.9 c, d

C2 2 2.0 b, c, d

C5 4 0.1 b

C8 3 0.2 b, c, d

43

Table 3. Summary of regression models describing relationships between coral reef metrics and

independent variables pertaining to localized stressors and environmental gradients. Dependent

variables: 1) coral cover change since 1991, 2) coral population density change since 1991, 3)

coral population kurtosis change since 1991, 4) BSR is the ratio of coral and crustose coralline

algae to turf, macro, and fleshy corralling algae from 2011-12 benthic data (see methods), 5)

multivariate algal assemblage differences from 2011-12 quadrat data (see methods). Independent

variables were mean herbivore/detritivore size (herb_size) and multivariate differences

(fish_PCO), water quality principle component axis-1 score from 2011-12 data whereby higher

scores indicate better water quality (wq), and wave exposure (wave_exp). AIC refers to Akaike

Information Criterion, an overall goodness-of-fit measure for regression models where a lower

score indicates a better fit.

Dependent Equation Slope (SE) Intercept (SE) R2 P-Value AIC

1) Coral cover change Y = 2.9(herb_size x wq) -

7.7 ±0.6 ±3.1 0.83 0.007 38.3

1) Coral cover change Y = 9.4(herb_size) - 13.4 ±2.5 ±5.6 0.72 0.02 41.4

2) Coral population

density change Y = 5.1(fish_PCO) - 2.0 ±1.7 ±3.7 0.62 0.04 36.7

3) Coral kurtosis change Y = -3.1(wave_exp x wq x

herb_size) - 43.2 ±1.1 ±11.6 0.57 0.05 49.5

4) BSR (2010) Y = 0.1(wave_exp x wq x

herb_size) + 0.7 ±0.04 ±0.5 0.58 0.05 14.4

5) Algal assemblage

centroids (2010)

Y = -0.3(wave_exp x wq x

herb_size) + 11.6 ±0.08 ±0.9 0.60 0.04 37.8