Journal of Experimental Marine Biology and Ecology 448 (2013) 128–135

Contents lists available at SciVerse ScienceDirect

Journal of Experimental Marine Biology and Ecology

j ourna l homepage: www.e lsev ie r .com/ locate / jembe

Invasive mangrove removal and recovery: Food web effects acrossa chronosequence

Margaret C. Siple ⁎, Megan J. DonahueHawai'i Institute of Marine Biology, University of Hawai'i at Mānoa, Kāne'ohe, HI 96744, USA

⁎ Corresponding author at: School of Aquatic and FWashington, P.O. Box 355020, Seattle, WA 98195. Te

E-mail addresses: mcsiple@hawaii.edu, siplem@uw.e

0022-0981/$ – see front matter © 2013 Elsevier B.V. Alhttp://dx.doi.org/10.1016/j.jembe.2013.06.008

a b s t r a c t

a r t i c l e i n f o

Article history:Received 14 December 2012Received in revised form 7 June 2013Accepted 9 June 2013Available online xxxx

Keywords:ChronosequenceEcosystem engineersInfaunaInvasive speciesRhizophora mangle

Red mangrove (Rhizophora mangle) was introduced to Hawai'i in 1902 and has since overgrown many coastalareas in Hawai'i, transforming nearshore sandy habitat into heavily vegetated areas with low water velocity,high sedimentation rates, and anoxic sediments. Introduced mangrove forests provide habitat for exotic spe-cies, including burrowing predators, which can exert top-down effects on benthic communities. Removal ofmangrove overstory is a popular management technique, and, here, we study community change over achronosequence of mangrove removals from 2007 to 2011, investigating infaunal community structure,crab abundance, and response to predator exclusion in the presence of mangrove overstory and along thechronosequence of removals. Overstory removal results in gradual changes in community composition con-current with slow decomposition of sedimentary mangrove biomass (k = 0.36 ± 0.06 × 10−3 d−1).Changes over time after removal include an increase in total infaunal abundance, a decrease in sub-surfacedeposit feeders, and an increase in suspension-feeding worms. Burrowing crab densities are uniform acrossmangrove and removal sites, and, unlike in native mangroves, their effects on infaunal communities are sim-ilarly negligible in both mangrove and removal areas. These results show that recovery from invasion and re-moval occurs gradually and is not governed by top-down effects.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Red mangrove (Rhizophora mangle L.) was introduced to Hawai'iin 1902 and to He'eia Marsh on the island of O'ahu in 1922 to controlrunoff from upstream agriculture (McCaughey, 1917). While otherspecies of mangrove have been introduced to Hawai'i, R. mangle isthe most successful, occupying coastal habitats throughout the mainHawaiian islands, including estuarine fishpond sites developed foraquaculture by native Hawaiians as early as 1000 C.E. (Allen, 1992,1998; Kirch, 2007). In their native range, mangroves are ecosystemengineers, strongly modifying their environment and providing im-portant ecosystem services, including shoreline protection, entrap-ment of heavy metals (Clark et al., 1998; Harbison, 1986), sedimentstabilization (Posey, 1987), litterfall subsidy (Twilley et al., 1986),and nursery grounds (Mumby et al., 2004; Robertson and Duke,1987). In their introduced range, these potential ecosystem servicesmust be weighed against impacts on native ecosystems: In Hawai'i,mangroves create habitats dramatically distinct from the sandflatsinhabited by the few native coastal macrophytes (Allen, 1998).

Hawaiian mangrove sediments host a higher diversity and abun-dance of infauna than adjacent sandflats, but this higher diversity

ishery Science, University ofl.: +1 206 661 8403.du (M.C. Siple).

l rights reserved.

includes many alien species (Demopoulos and Smith, 2010). This dif-ference betweenmangrove sediments and adjacent sandflatsmay resultfrom enhanced productivity due to litterfall subsidy or increased micro-habitat heterogeneity (bottom-up effects) or from trophic effects, suchas changes in habitat use by mobile predators (top-down effects). De-spite R. mangle's substantial litterfall subsidy to coastal habitats inHawai'i (Cox and Allen, 1999), stable isotope analysis indicates that thisproductivity is not consumed by benthic invertebrates but, instead, subsi-dizes bacterial foodwebs (Demopoulos et al., 2007). The high detrital out-put and undigestedmangrove leaf detritus increase sediment anoxia andnegatively impact native infaunal assemblages (Demopoulos, 2004).

Removal has been a popular tool for the control of alien pest speciesin Hawai'i (e.g., Scowcroft and Conrad, 1992; Stone et al., 1992), includ-ing R. mangle (Chimner et al., 2006; Rauzon and Drigot, 2002). Themostcommon method of mangrove removal is to cut the prop roots belowthe high tide mark, flooding the roots with saltwater, and haul awaythe overstory, leaving the flooded roots to decompose. While thismethod is favored in Hawai'i, the time course of recovery following re-moval is poorly resolved. The only study examining recovery dynamicsafter removal compared sites in two different estuaries: a two-year-oldremoval in Pearl Harbor and a six-year-old removal in Kāne'ohe Bay(Sweetman et al., 2010). Sweetman et al. (2010) found that removal in-creased total macrofaunal abundance and shifted the infaunal commu-nity from sub-surface deposit feeders (mostly tubificid oligochaetes) tosuspension feeding spionid polychaetes. Even six years after removal,flooded mangrove roots remained above the sediment surface along

129M.C. Siple, M.J. Donahue / Journal of Experimental Marine Biology and Ecology 448 (2013) 128–135

with persistent changes in bacterial and macrofaunal carbon consump-tion (Sweetman et al., 2010).

Post-removal changes in infaunal community structure may be dueto changes in sediment chemistry, changes in the lability of organicmatter, or top-down effects of a changing predator community. Inmangroves, the physicochemical and biological processes that controlinfaunal abundance and community composition are interdependent:Burrows can enhance bacterial activity and algal productivity in man-grove sediments, altering nutrient availability (Mchenga et al., 2007),and burrowers structure infaunal communities through predation(Alongi, 1989). Because they can influence physicochemical processesin mangrove sediments in addition to preying on infauna, burrowingpredators constitute an important link between top-down and bottom-up forcing mechanisms for mangrove infauna.

Mangrove-associated predators may exert trophic effects on in-faunal assemblages. While mangroves have been characterized asrefugia from predators (Macia et al., 2003) and nursery habitat formany species (Laegdsgaard and Johnson, 2001), vegetation may alsoattract predators by providing habitat for benthic invertebrates(Nagelkerken and van der Velde, 2004). Therefore, despite the popu-lar notion that mangroves primarily function as refugia from largerpredators (e.g., Primavera, 1997), infauna in mangrove habitats mayexperience enhanced predation. Experimental exclusion of predatorycrabs in native mangrove forests in Thailand led to higher species di-versity, richness and biomass in the benthic infaunal community(Kon et al., 2009). In Hawai'i, R. mangle provides habitat for intro-duced crabs (Nakahara, 2007) that forage on infaunal communities(Cannicci et al., 1996; Hill, 1979) and may influence benthic commu-nity structure through predation and burrowing.

Taking advantage of ongoingmangrove removal in a native Hawaiianfishpond, this paper investigates the time course of shifts in communitystructure following removal and the effects of burrowing predators inmangrove and mangrove removal areas. We address three questions:1) How doesmangrove removal affect infaunal and epifaunal communi-ties? 2) What is the rate of mangrove decomposition and related com-munity shifts? 3) Do burrowing predators modify infaunal communitycomposition in introduced or removed mangrove areas?

2. Methods

2.1. Study site

This study tookplace at Loko I'a o He'eia, an 88-acrefishpond locatedin Kāne'ohe Bay, O'ahu (21°26′10.74″ N, 157°48′28.05″W). He'eia fish-pond is a shallow reef flat surrounded by a permeable rock wall built600–800 years ago. Mangrove was introduced to the area in 1922,and the wall was completely covered by mangrove as recently as2005, forming a large stand that is contiguous with surrounding forest(Fig. 1).

Mangrove was introduced to He'eia in 1922 and spread to form acontinuous stand around the mouth of He'eia stream. In the 1960s, R.mangle expanded past the stream to grow along the fishpond wall(Chimner et al., 2006). Mangrove completely covered the fishpondwall by the time managers began removing 50–75 m sections of man-grove in 2006 (Fig. 1). Likemost othermangrove areas in Hawai'i, He'eiafishpond is a modified coastline with mangrove stands less than100 years old. This fishpond is one of several in the main HawaiianIslands invaded by R. mangle, and the same management solution(above-ground removal) has been proposed for many of them. For thisstudy, sampling sites were chosen at the center of each removal section,for a total of eight sites along this well-documented chronosequence:two sites with fresh mangrove and six sites with mangrove removedbetween 2007 and summer 2011.

The fishpond is tidally dominated, exchanging seawater through foursluice gates (mākāhā) and the permeable rockwall; it also receives fresh-water input from He'eia Stream containing sediment and land-derived

nutrients (Young, 2011). Flow rates through mākāhā into the pondvary between 0.1 and 4.9 m3/s on a flood tide and flow rates out vary be-tween 0.1 and 4.6 m3/s, with most of the exchange occurring throughthe two east-facing mākāhā (Fig. 1; Young, 2011). The pond is shallow(average depth 0.413 m; 4 mmaximumdepth)with a soft bottomdom-inated by muddy sediments near stream inputs and sand/coral rubblenear the seawall (Vasconcellos, 2007).

2.2. Physical environment

To test for environmental differences across sites, we used datafrom an ongoing monitoring project on the physical oceanographyof the fishpond. Salinity, pH, dissolved oxygen (DO), chlorophyll aconcentration, and temperature were measured monthly at or nearour benthic sampling sites (McCoy, 2011). These parameters weretaken from April to August of 2011 at midtide using a pre-calibratedYSI® 6660 and averaged over the bottom 25 cm of the water column.Data were collected monthly. Physical variables were compared be-tween sites using Kruskal–Wallis tests.

2.3. Grain size

Cores for grain size were collected at each site in 2012, sectioned at5 cm, and frozen until analysis. In the lab, core subsamples were sievedthrough 2-mm, 500-μm, and 63-μm meshes. Sediment fractionsretained on each sieve were dried at 60 °C for 2–5 days, weighed, andpercent rubble (N2 mm), sand (63 μm b x b 2 mm) and silt/clay(b 63 μm) calculated. We used a log-linear model to test for indepen-dence between grain size distribution and site.

2.4. Decomposition rate

Sedimentary mangrove biomass (SMB) was measured from the top5 cm section of 6.5-cm (inner diameter) cores collected in December2012. Roots, bark and leaf material were rinsed with fresh water ona 250 μmsieve, picked out under a dissectingmicroscope at 60×magni-fication, and dried at 60 °C to constant weight (1–3 days) beforeweighing. Three nestedmodelswere compared formangrove decompo-sition: a single exponential model, SMB = ae− bt, where t is days sinceremoval (Poret et al., 2007; Twilley et al., 1997), a double exponentialmodel,SMB ¼ a ce−b1t þ 1−cð Þe−b2t

� �which assumes that a proportion,

(1 − c), of thematerial is less labile and decomposes at rate b2while themore labile material decomposes at rate b1 (Wieder and Lang, 1982),and an asymptotic model SMB = a(ce− bt + (1 − c)), which assumesthat a proportion, (1 − c), of the material will never decompose(Wieder and Lang, 1982).Modelswerefit using theminpack.lmpackagein R (Elzhov et al., 2013) for nonlinear regression and compared usinglikelihood ratio tests.

2.5. Chronosequence

Infaunal cores (6.5 cm i.d.) were collected May 10th–16th, 2011.This core diameter is standard for sampling small infaunal inverte-brates. These cores also provided the initial time point for the exclu-sion experiment described below. Twelve cores were collected atsites M1, R11, R10, and R09, and 9 cores were collected at sites M2,R08_Jul and R08_Feb and R07. Cores were sectioned at 5 cm, sievedthrough a 500 μm mesh, and preserved in 10% formalin with RoseBengal dye (0.05 g/L). In the lab, samples were rinsed gently withflowing tap water, then sorted under 60× magnification and storedin ethanol.

Polychaetes were sorted to family; other organisms were sortedinto the following taxonomic groups: oligochaetes, amphipods, ostra-cods, tannaeids, decapods (mostly small shrimp), and nematodes.Taxon richness (S), normalized taxon richness (d = (S − 1) / ln N),Shannon–Weiner diversity, and Shannon evenness were determined

Fig. 1. Map of study sites. The dotted line indicates where mangrove was at the beginning of the study. Rhizophora mangle used to grow around the entire pond. The dotted lineindicates the portion of the wall that was vegetated at the beginning of the study. M2 and M1 remained intact throughout the study. R11 had mangrove removed in July of2011. R10, R09, and R07 were removed in 2010, 2009, and 2007 respectively. R08_Feb and R08_Jul were removed in February and July of 2008. Pie charts show grain size distri-bution at each site. Mākāhā are indicated with arrows, with arrow thickness indicating the relative magnitude of tidal exchange through that mākāhā at spring tide according toYoung (2011).

130 M.C. Siple, M.J. Donahue / Journal of Experimental Marine Biology and Ecology 448 (2013) 128–135

for all cores and compared along the chronosequence using general linearmodels. In addition, taxa were assigned to trophic, domicile, and feedingguilds according to Barnes (1980), Fauchald and Jumars (1979), andSheridan (1997) (Supplementary Table 3). Since macrofauna were notidentified to the species level in this study, taxa were assigned to aguild only if all the species in that taxon were classified as belongingto that guild in the literature and in a previous study in Hawaiian man-groves performed near the study site (Demopoulos, 2004; see Supple-mentary Table 3 for guild assignments and Supplementary Table 4 fora list of species found around O'ahu). We tested for changes in thetotal abundance and abundance of taxa and trophic guilds over thechronosequence using linear regression on log-transformed counts tomeet the homogeneity of variance assumption. Finally, we used aSIMPER analysis to compare the overall similarity of community com-position between mangrove and removal sites.

We used canonical correspondence analysis (CCA) (ter Braak,1986) with the vegan package in R (Oksanen et al., 2012) to assessthe relationship between infaunal community composition and envi-ronmental variables (temperature, salinity, pH, dissolved oxygen,chlorophyll a, turbidity, and grain size) across the study site. Adetrended correspondence analysis indicated that taxa (gradientlevel = 3.0) showed unimodal responses across environmental gra-dients (Gauch, 1982). Contributions of environmental variableswere tested for significance using a Monte Carlo test with 3999permutations.

2.6. Predator exclusion experiment

We tested the effect of burrowing predators on the infaunal com-munity using a replicated exclosure experiment across the fivenorthern sites of the chronosequence (M2, M1, R11, R10, and R09;Fig. 1). Burrowing predators were excluded using wire mesh cages(36 × 36 × 30 cm, 1.27 cm mesh) with 10 cm of aluminum flashingbelow the sediment surface. Each set of exclosures included a preda-tor exclusion cage, a cage control with 26 × 26 cm openings on eachside, and an open control with no cage. Three (M2 and R09) or four(M1, R11, R10) replicates of each exclosure typewere randomly locatedwithin eachof thefive sites. An infaunal core (6 cm inner diameter)wastaken from each experimental unit in May (May 10–16, 2011) and inSeptember (Sept 24–29, 2011), corresponding to 1 month before and3 months after mangrove removal at site R11. Infaunal cores wereprocessed as described in the Chronosequence section.

To test for changes in infaunal community composition due totreatments and over the course of the experiment, we used permuta-tional analysis of variance (PERMANOVA; Anderson, 2001). Within-site replication allowed us to test for an interaction between Treat-ment and Site, i.e., whether predator effects differed between sites.In our study system, community richness is relatively low and abun-dance varies substantially between sampling sites; therefore, we usedthe modified Gower similarity index with log base 2 so that a singlespecies change in community richness was equivalent to a doubling

Fig. 2. Sedimentary mangrove biomass measured in 2012 vs. days since removal. Bio-mass measured in previous studies is marked with filled triangles. The line representsmodel predictions for the exponential decay model.

131M.C. Siple, M.J. Donahue / Journal of Experimental Marine Biology and Ecology 448 (2013) 128–135

of abundance (Anderson et al., 2006). Prior to multivariate analy-sis, taxa containing fewer than 15 total individuals (summed acrossall cores) were removed from the dataset (excluded taxa are indi-cated in Supplementary Table 2). We used the PERMANOVAmodel: Infauna ~ Site + CageTrt + Month + Site × CageTrt +Site × Month + CageTrt × Month + Block(Site × CageTrt) + Site× CageTrt × Month. Since the experimental design involved repeatedmeasures (cores taken in each cage at each time point), cages were treat-ed as blocks. The Block(Site × CageTrt) × Month term is excluded be-cause there is only a single sample per cage at each time point. Theresidual error term, therefore, represents the variation within a cagetreatment at each site. Significant multivariate patterns were examinedfurther with univariate PERMANOVA tests. Statistical analyses wereperformed using the vegan package in R (R Team, 2012; Oksanen et al.,2012) and Primer-E® software (Clarke, 1993).

2.7. Predator community

To compare the density of burrowing predators across sites in Mayand in September, we set large traps overnight (modified commercialcrab traps) for 3–4 nights and set small crab nets during the day for3–5 days. All traps were baited with skin and bones from two milkfish(Chanos chanos) per trap (~1 kg total). Crabs caught in all traps wereidentified, sexed, andmeasured for carapace length andwidth. Biomasswas calculated from carapace width using available species- andsex-specific parameters (Mohapatra et al., 2010; Roongrati and Iaitim,1994; Songrak et al., 2010; Sukumaran and Neelakantan, 1997). Scyllaserrata was removed from statistical comparisons because of itslow catch rate (seven individuals captured across the course of theexperiment) and large size (15–25 cm c.w.). We tested whethercommunity composition was independent of site and month using alog-linear analysis. Crab size (individual biomass) and crab density(number of individuals caught per unit effort) and crab biomass-density (biomass of catch per unit effort) were compared across sitesin each month using Kruskal–Wallis tests.

3. Results

3.1. Physical environment and grain size

The pond contains a natural gradient from fresh, turbid, productivewater on the west side near the existing mangroves to more saline,less turbid waters on the east side near the sea wall. All of the studysites were along the east side and experienced similar monthly meantemperature, salinity, DO, pH, turbidity and chlorophyll a concentrationover the course of the study (Kruskal–Wallis tests; df = 6, p = 0.981,0.217, 0.979, 0.966, 0.217, and 0.597 respectively). Grain size distribu-tion varied by site with less rubble and more silt/clay in the southernsites (Fig. 1).

3.2. Mangrove decomposition rate

The single exponential model of mangrove root decomposition wasmost parsimonious (R2 = 0.37; p N 0.17 for all likelihood ratio tests).Sedimentary mangrove biomass decreased over the chronosequencewith decay constant k = 0.36 ± 0.06 × 10−3 d−1 and an intercept of1285 ± 77 g dw m−2, equivalent to a loss of 12.6% (CI95% = (8.4,16.7)) per year (Fig. 2).

3.3. Community patterns across a chronosequence of mangrove removal

Total infaunal abundance increased with time since removal witha doubling time every 2.6 years (R2

adj = 0.159, F = 15.9, p b 0.001;Fig. 3). Taxon richness, normalized taxon richness, Shannon diversity,and Shannon evenness did not change over the chronosequence.

The abundance of suspension feeders increased with time since re-moval with a doubling time of 1.7 years (R2adj = 0.257, F1,78 = 28.3,p b 0.001; Fig. 4), and the abundance of omnivores doubled in2.3 years (R2adj = 0.109, F1,78 = 10.7, p = 0.001; Fig. 4). Therewere no clear patterns in domicile or mobility groupings over thechronosequence.

Overall, mangrove communities are 61.28% dissimilar from removalsites (Fig. 5), with higher abundances of oligochaetes (contributing22.19% of the dissimilarity), amphipods (19.33%), and sabellids (10.4%)in the removal sites contributing most to the dissimilarity.

Differences in environmental variables across sites explained 25%of the variance in community composition, with the first three canon-ical axes of the CCA explaining 20% of the variation (permutation test:p = 0.005; Fig. 5). Sites M1 and M2 were most clearly distinct fromthe other sites. M1 was distinguished by the presence of sternaspidpolychaetes and was positively associated with salinity and negative-ly associated with turbidity. M2 was distinguished by the presence ofcossurids and was positively associated with temperature and pH.Ostracods distinguished two cores at R07.

3.4. Predator exclusion experiment

The infaunal community differed between sites and over time butnot between treatments (Supplementary Fig. 1, PERMANOVA Time:F1,39 = 3.45, p = 0.004; Site: F4,39 = 6.17, p b 0.001; Treatment:F2,39 = 1.06, p = 0.379). Sites differed in the abundance of sabellidpolychaetes (Site: F4,39 = 3.49, p = 0.012), opheliid polychaetes(F4,39 = 7.30, p b 0.001), amphipods (F4,39 = 3.53, p = 0.013), and ol-igochaetes (F4,39 = 9.89, p b 0.001). Sites also differed in the pattern ofcommunity change over time (Site × Time: F4,39 = 3.60, p b 0.001). Ol-igochaetes changed in abundance differently across sites (Site × Time:F4,39 = 2.76, p = 0.043): decreasing at one mangrove site (M2; t =5.94, p = 0.015) and increasing at the 2011 removal site (R11; t =2.43, p = 0.037). Opheliids changed differently across sites, as well(Site × Time interaction; F4,39 = 4.37, p = 0.005): decreasing at onemangrove site (M2; t = 2.78, p = 0.036) and increasing in theone-year-old removal site (R10; t = 3.15, p = 0.014). Shifts in amphi-pod abundance also differed between sites (Site × Time: F4,39 = 5.16,p b 0.001), decreasing at both mangrove sites (M1, W = 112, p =0.02; M2, W = 65.5, p = 0.02).

Fig. 3. Total infaunal abundance vs. days since removal, with logarithmic regressionline. The model for this regression is log(A + 1) = 7.32 × 10−4(DSR) + 3.491,where A is the abundance and DSR is the days since removal.

Fig. 4. Suspension feeders (a) and omnivores (b) increase with days since removal. Noother trophic, domicile, or mobility groupings showed significant trends with dayssince removal.

132 M.C. Siple, M.J. Donahue / Journal of Experimental Marine Biology and Ecology 448 (2013) 128–135

3.5. Predator community

The crab community was similar in density, size, and communitycomposition across sites. There was no difference in crab density(individuals caught trap−1 day−1) or crab biomass-density (biomasscaught trap−1 day−1) between sites inMay (density:χ2

df = 42 = 2.92,p = 0.712; biomass-density: χ2

df = 42 = 5.16, p = 0.270) or inSeptember (density: χ2

df = 42 = 2.39, p = 0.792; biomass-density:χ2

df = 42 = 9.48, p = 0.05). Similarly, there was no difference in crabsize between sites inMay (χ2

df = 42 = 0.250, p = 0.992) or in Septem-ber (χ2

df = 42 = 0.670, p = 0.954). Although the crab communitychanged between months (χ2

df = 3 = 32.7, p = 0.001, N = 151), thecrab community was independent of site in both May (χ2

df = 12 =6.21, p = 0.90, N = 68) and September (χ2

df = 12 = 14.33, p = 0.27,N = 83). The community consisted of Samoan crab (S. serrata,nMay = 1, nSept = 4), the crenate swimming crab (Thalamita crenata,nMay = 66, nSept = 66), and two endemic Hawaiian species of swim-ming crab: the long-eyed swimming crab (Podophthalmus vigil,nMay = 0, nSept = 9) and the blood-spotted swimming crab (Portunussanguinolentus, nMay = 1, nSept = 4).

4. Discussion

4.1. Mangrove decomposition is slow in Hawaiian mangroves

Decomposition rate estimated from this chronosequence was 12.6%(CI95% = (8.4, 16.7)) per year after mangrove removal (Fig. 2, k =0.36 ± 0.06 × 10−3 d−1), which overlaps with previously reportedestimates for Oahu (12–36% per year, Sweetman et al., 2010). Sincedecomposition data is largely unavailable for the invaded range ofR. mangle in Hawai'i, we use nativemangroves as a point of comparison,and present the following results as a baseline with which to comparedecomposition in other invasive stands. In native mangrove forests,root decomposition was much higher at 52% per year in Belize (k =1.9 ± 0.38 × 10−3 d−1) (Middleton and McKee, 2001; k estimatedfrom their Supplementary Fig. 1 using an exponential model) and 38%and 65% at two sites in the Everglades (k = 1.3 ± 0.48 × 10−3 d−1

to 2.9 ± 0.15 × 10−3 d−1, Poret et al., 2007). This study suggests thatmangrove decomposition in its introduced range in O'ahu is muchslower than in its native range and that habitat recovery following

mangrove removal may take decades, as also noted by Sweetman etal. (2010).

The difference in mangrove decomposition rate between O'ahuand the native range may be due to differences in flow conditions,methodology, effects of mangrove removal, and/or macrofaunal di-versity. First, flow conditions affect decomposition rates and may bemodified by the fishpond walls. However, flow rates measured withinthe mangrove removal area (mean current magnitude 0.06 m/s over4 days; Ruttenberg et al. 2013) are comparable to bulk flow ratesthrough native mangrove forests (b0.2 m/s; Furukawa et al., 1997;neither Middleton and McKee (2001) nor Poret et al. (2007) reportflow rates for comparison). Second, Middleton and McKee (2001)and Poret et al. (2007) both used litter bags to experimentally mea-sure decay rates, while this study and Sweetman et al. (2010) esti-mated decomposition from in situ measurements of root biomass.Litter bag experiments are likely to overestimate decomposition ratesbecause root material is dried before burial, which causes tannins andphenols to leach out more rapidly than they would in situ, hasteningmicrobial colonization and decomposition (Bärlocher, 1997). Third,we measured decomposition rate at mangrove removal sites, where

Fig. 5. Ordination plots based on a taxonomic canonical correspondence analysis: (a) eigenvector plot of environmental characteristics from stepwise selection of variables in thetaxonomic process and (b) infaunal taxon biplot where each taxon is represented in the same ordination, both with all cores as points.

133M.C. Siple, M.J. Donahue / Journal of Experimental Marine Biology and Ecology 448 (2013) 128–135

mangrove roots are no longer oxygenating the sediments. Sedimentchemistry at the one-year removal (R10) had shallower O2 penetrationdepth and higher sulfide concentrations compared to the intact man-grove site (M1) (Ruttenberg et al. 2013). Finally, loss of mangrove bio-mass may be slow because its decomposition is primarily bacterial:Hawaiian mangroves lack coevolved macrofauna, which are responsiblefor much of the leaf and root decomposition in native mangroves (e.g.,Nordhaus and Wolff, 2007) and can oxygenate the sediments throughbioturbation.

Decomposition rate is highly dependent on the presence ofmangrove-consumingmacrofauna, and rates aremuch lower where bac-teria are solely responsible for organic carbon consumption (Kristensen etal., 2008; Middleton and McKee, 2001). In addition, larger detritivoresprefer conditioned (decomposed or excreted by crabs)mangrovemate-rial to fresh mangrove detritus (Giddins et al., 1986; Lee, 1989;Torres-Pratts and Schizas, 2007). However, unlike in its native range,macrofauna in Hawai'i do not consume mangrove-derived carbon(Demopoulos et al., 2007), and bacteria dominate short-term (2d) Cprocessing in six-year removals (Sweetman et al., 2010). In addition,Hawai'i has fewer burrowers and bioturbators than native mangrovesand their absence may reduce mixing and increase sediment anoxia,slowing decomposition.

4.2. Mangrove and removal areas host distinct infaunal communities

Differences in total infaunal abundance between mangrove and re-moval sites (Fig. 3), are similar in direction but differ in magnitudethan prior studies on O'ahu. Total abundances found at the northern-most mangrove site in this study (M2: 21,885 ± 2682 ind m−2) aresimilar to those found by Sweetman et al. (2010) in Pearl Harbor(21,597 ± 12,731 ind m−2), but other mangrove sites in this study(M1 and R11 inMay) had substantially lower total abundances. In addi-tion, we found a 1.5-fold increase in the first two years following re-moval (Fig. 3), while Sweetman et al. (2010) found an 8-fold increaseat their two year removal site in Pearl Harbor. Overall, total abundancesin this study were more similar to a six-year removal in Kāne'ohe Baythan a two-year removal in Pearl Harbor, suggesting that total abun-dance depends on location as well as recovery time.

While oligochaetes and amphipods were numerically dominant atall sites, there were shifts in less dominant groups betweenmangroveand removal sites (Supplementary Table 1) similar to prior studies.

Sites with intact mangrove were dominated by amphipods (39% ofabundance), oligochaetes (29%), nematodes (6%), and sub-surface de-posit feeding cossurid polychaetes (6%) while removal sites containedoligochaetes (46%), amphipods (26%), sabellid polychaetes (9%), andostracods (6%). Sweetman et al. (2010) also found sub-surface depositfeeders to be numerically dominant at intact mangrove sites, while re-moval sites and pre-invasion areas in Kāne'ohe Bay had a higher abun-dance of suspension and surface-feeders (e.g., corophiids, sabellids, andspionids). This pattern has been attributed to decreased particle size,high sedimentation, and high levels of organic enrichment in man-groves as opposed to sand or mud flats (Demopoulos and Smith,2010; Ellis et al., 2004) and is consistent with previous studies of nativeand invasive mangrove forests (Demopoulos and Smith, 2010). Afteroligochaetes and amphipods, suspension feeding sabellid polychaetescontributed the most to dissimilarity between mangrove and removalsites (Supplementary Table 1). Shifts in suspension feeder abundancehave been attributed to changes in water velocity (LaBarbera, 1984),which is much lower in mangrove habitats because of prop-root struc-ture, and the associated increase in sedimentation.

4.3. Community recovery after mangrove removal

Mangrove decomposition state is affected by the macrofaunal andbacterial communities present, but the decomposition state of the man-grove in turn affects faunal densities: previous work indicates thatmeiofaunal densities are higher on more decomposed R. mangle leaves(Torres-Pratts and Schizas, 2007), and laboratory experiments haveshown that larger detritivores prefer conditioned (decomposed or excret-ed by crabs) mangrove material to fresh mangrove detritus (Giddins etal., 1986; Lee, 1989).Whenmangroves decompose, total amounts of tan-nins and phenolic compounds inside the tissues decrease, potentially ren-dering mangrove roots and leaves more bioavailable (Alongi, 2009). Anincrease in microbial production may explain the gradual increase in in-faunal density after removal, with older sites containing higher abun-dances of infauna than more recent removals (Fig. 3).

The increase in total abundance (Fig. 3) was accompanied by a de-crease in sub-surface deposit feeders, which persisted over time. Arapid shift to “post-removal” conditions is also apparent from theMDS plot of all sites in September (Supplementary Fig. 1): The removalsite (R11), which was removed just three months before the collectionof September cores, already appears similar to the 2–4 year removal

134 M.C. Siple, M.J. Donahue / Journal of Experimental Marine Biology and Ecology 448 (2013) 128–135

sites. Rapid post-removal shifts in community structure and total infau-nal abundance coincide with rapid changes in short-term C-processingfound at 2-year removal sites by Sweetman et al. (2010).

Macrofaunal community structure of benthic environments hasbeen used in the past to characterize ecosystem health (Brown et al.,2000; Kremen, 1992). If suspension feeder abundance is an indicatorof lower sedimentation rates and a return to high-flow, unvegetatedhabitat, the long-term increase in suspension feeders following removalmay be evidence of recovery in the benthic community. At the four-yearremoval site in this study, sabellids were still at about half the abun-dance of sandflat controls in Kāne'ohe Bay surveyed by Sweetman etal. (2010), suggesting that they have not yet reached pre-invasiondensity.

4.4. Top-down processes do not regulate infaunal communities in Hawaiianmangrove or mangrove removal habitats

Changes in predation pressure or bioturbation rate are unlikely to ac-count for the increase in infaunal abundance along the chronosequence:crab density, biomass, and community structure were similar acrosssites. The fishpond hosts a diversity of introduced predators (Nakahara,2007), including the portunids T. crenata and S. serrata, as well as twopatchily distributed native portunid crabs, P. vigil and P. sanguinolentus.The range of the most abundant species, T. crenata, was smaller thanthe length of each study site, so differences in abundance across siteswere not due to differences in short-term aggregation. The taxa respon-sible for extensive bioturbation in nativemangroves (Uca spp. andmem-bers of the family Sesarmidae) are absent from Hawaiian mangroves(Demopoulos, 2004), while T. crenata and S. serrata, the most abundantand largest crabs at our sites respectively, inhabit shallow burrows(Williams, 1994). Further, total burrow density is much lower in man-groves on O'ahu than in native mangroves (Demopoulos, 2004;Kristensen, 2008) so burrowing is unlikely to have the dramatic impactson sediment chemistry seen in native mangrove habitats.

Burrowing predators did not exert top-down control (SupplementaryFig. 1): therewasno effect of predator exclusionon infaunal communities.Thiswas a surprise, because similar experiments in nativemangrove havedemonstrated negative effects of epifauna (sesarmid, grapsid, ocypodidand hermit crabs, as well as snails) on meiofaunal (Schrijvers et al.,1995) and macrofaunal abundance (Schrijvers et al., 1998) with similarlevels of replication (n = 2, Schrijvers et al., 1995, 1998; n = 4, Kon etal., 2009). Additionally, a power analysis of log-transformed total abun-dances from this study shows that a doubling of sample size would benecessary to detect a difference (α = 0.05) between cage treatments inmangrove sites. Additional cage replicates at the removal sites wouldnot substantially increase the power.

The low power calculated for this study indicates that predatoreffects, if present, were overwhelmed by spatial variation. Instead ofstrong top-down control, we found communities that differed betweensites and community shifts from May to September that depended onsite. In addition, 58% of the total community variation was ascribed totreatment replicates within each site, indicating high microhabitat het-erogeneity throughout the pond. Contrary to studies in native man-groves, this within- and between-site spatial heterogeneity was farmore important than predation or burrowing in structuring infaunalcommunities. This study highlights the importance of small-scale het-erogeneity in structuring infaunal communities, indicating the predom-inance of bottom-up processes. Possible top-down effects, if present,would need extensive replication in order to detect in this system.

5. Conclusions

Mangrove foodwebs in Hawai'i function differently than man-groves in their native range: Demopoulos et al. (2007) demonstratedthat mangrove-derived carbon is not taken up by macrofauna inHawai'i, and, in this study, we see no evidence of top-down control

on infaunal communities as is seen in native mangrove (Kon et al.,2009; Schrijvers et al., 1998). Instead, we see strong evidence of hab-itat heterogeneity, even at the sub-meter scale.

Invasive species eradication can have dramatic effects on invadedecosystems, and post-removal assessments have been recommendedas a way to monitor the effect of eradication. This study shows thatchanges in community composition occur gradually after an initialrapid transformation from living to decomposing mangrove. However,the slow rate of mangrove root decomposition and the lack of deepburrowing predators in Hawai'i suggest that a return to pre-invasionconditions may take decades without additional intervention.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jembe.2013.06.008.

Acknowledgments

MCS was supported by an NSF Graduate Research Fellowship.Many thanks to Lisa Hinano Rey (NSF/DBI-URM 0829272, PI M.Hadfield), Kaleonani Hurley (Kupu/Hawai'i Youth ConservationCorp), and three teams of Laulima A 'Ike Pono interns (NSF/GEO-OEDG #0914516, PI J. Lemus) for help collecting and processingsamples. Thanks to F. I. Thomas, C. R. Smith, K. Ruttenberg, R. Briggs,and S. Leon Soon for helpful discussions, and to F. I. Thomas and C.R. Smith for insightful comments on an earlier version of this manu-script. Mahalo to Hi'ilei Kawelo, Keli'i Kotubetey, and all our commu-nity partners at Paepae O He'eia. [ST]

References

Allen, J., 1992. Farming in Hawai'i from colonization to contact. N. Z. J. Archaeol. 14,45–66.

Allen, J.A., 1998. Mangroves as alien species: the case of Hawai'i. Global Ecol. Biogeogr.Lett. 7, 61–71.

Alongi, D.M., 1989. Ecology of tropical soft-bottom benthos - a review with emphasison emerging concepts. Rev. Biol. Trop. 37 (1), 85–100.

Alongi, D.M., 2009. The Energetics of Mangrove Forests. Springer, Brienigsville, PA (216pp.).

Anderson, M.J., 2001. A newmethod for non-parametric multivariate analysis of variance.Austral Ecol. 26 (1), 32–46.

Anderson, M.J., Ellingsen, K.E., McArdle, B.H., 2006. Multivariate dispersion as a mea-sure of beta diversity. Ecol. Lett. 9 (6), 683–693.

Bärlocher, E., 1997. Pitfalls of traditional techniques when studying decomposition ofvascular plant remains in aquatic habitats. Limnetica 13 (2), 1–11.

Barnes, R.D., 1980. Invertebrate Zoology, Saunders College, Philadelphia, PA. 4th ed.(1089 pp.).

Brown, S.S., Gaston, G.R., Rakocinski, C.F., Heard, R.W., 2000. Effects of sediment con-taminants and environmental gradients on macrobenthic community trophicstructure in Gulf of Mexico estuaries. Estuaries 23 (3), 411–424.

Cannicci, S., Dahdouh-Guebas, F., Anyona, D., Vannini, M., 1996. Natural diet and feedinghabits of Thalamita crenata (Decapoda: Portunidae). J. Crustac. Biol. 16 (4), 678–683.

Chimner, R.A., Fry, B., Kaneshiro, M.Y., Cormier, N., 2006. Current extent and historicalexpansion of introduced mangroves on O'ahu, Hawai'i. Pac. Sci. 60 (3), 377–383.

Clark, M.W., McConchie, D., Lewis, D.W., Saenger, P., 1998. Redox stratification andheavy metal partitioning in Avicennia-dominated mangrove sediments: a geo-chemical model. Chem. Geol. 149 (3–4), 147–171.

Clarke, K.R., 1993. Nonparametric multivariate analyses of changes in communitystructure. Aust. J. Ecol. 18 (1), 117–143.

Cox, E.F., Allen, J.A., 1999. Stand structure and productivity of the introduced Rhizophoramangle in Hawai'i. Estuaries 22 (2A), 276–284.

Demopoulos, A., 2004. Aliens in Paradise: A Comparative Assessment of Introduced andNative Mangrove Benthic Community Composition, Food-web Structure, andLitter-fall Production. University of Hawai'i at Mānoa.

Demopoulos, A.W.J., Smith, C.R., 2010. Invasive mangroves alter macrofaunal commu-nity structure and facilitate opportunistic exotics. Mar. Ecol. Prog. Ser. 404, 51–67.

Demopoulos, A.W.J., Fry, B., Smith, C.R., 2007. Food web structure in exotic and nativemangroves: a Hawaii–Puerto Rico comparison. Oecologia 153 (3), 675–686.

Ellis, J., Nicholls, P., Craggs, R., Hofstra, D., Hewitt, J., 2004. Effects of terrigenous sedi-mentation on mangrove physiology and associated macrobenthic communities.Mar. Ecol. Prog. Ser. 270, 71–82.

Elzhov, T.V., Mullen, K.M., Spiess, A.N., Bolker, B., 2013. R Interface to the Levenberg–Marquardt Nonlinear Least-squares Algorithm Found in MINPACK, Plus Supportfor Bounds.

Fauchald, K., Jumars, P.A., 1979. The diet of worms: a study of polychaete feedingguilds. Oceanogr. Mar. Biol. Annu. Rev. (SAUS) 17, 193–284.

Furukawa, K., Wolanski, E., Mueller, H., 1997. Currents and sediment transport in man-grove forests. Estuar. Coast. Shelf Sci. 44 (3), 301–310.

135M.C. Siple, M.J. Donahue / Journal of Experimental Marine Biology and Ecology 448 (2013) 128–135

Gauch, H.G., 1982. Noise reduction by eigenvector ordinations. Ecology 63 (6),1643–1649.

Giddins, R.L., Lucas, J.S., Neilson, M.J., Richards, G.N., 1986. Feeding ecology of the man-grove crab Neosarmatium smithi (Crustacea, Decapoda, Sesarmidae). Mar. Ecol.Prog. Ser. 33 (2), 147–155.

Harbison, P., 1986. Mangrove muds — a sink and a source for trace-metals. Mar. Pollut.Bull. 17 (6), 246–250.

Hill, B.J., 1979. Aspects of feeding strategy of the predatory crab Scylla serrata. Mar. Biol.55, 209–214.

Kirch, P.V., 2007. Hawai'i as a model system for human ecodynamics. Am. Anthropol.109 (1), 8–26.

Kon, K., Kurokura, H., Tongnunui, P., 2009. Do mangrove root structures function toshelter benthic macrofauna from predators? J. Exp. Mar. Biol. Ecol. 370 (1–2), 1–8.

Kremen, C., 1992. Assessing the indicator properties of species assemblages for naturalareas monitoring. Ecol. Appl. 2 (2), 203–217.

Kristensen, E., 2008. Mangrove crabs as ecosystem engineers; with emphasis on sedi-ment processes. J. Sea Res. 59 (1–2), 30–43.

Kristensen, E., Bouillon, S., Dittmar, T., Marchand, C., 2008. Organic carbon dynamics inmangrove ecosystems: a review. Aquat. Bot. 89 (2), 201–219.

LaBarbera, M., 1984. Feeding currents and particle capture mechanisms in suspensionfeeding animals. Am. Zool. 24 (1), 71–84.

Laegdsgaard, P., Johnson, C., 2001. Why do juvenile fish utilise mangrove habitats?J. Exp. Mar. Biol. Ecol. 257 (2), 229–253.

Lee, S.Y., 1989. The importance of Sesarminae crabs, Chiromanthes spp. and inundationfrequency on mangrove (Kandelia candel (L) Druce) leaf litter turnover in a HongKong tidal shrimp pond. J. Exp. Mar. Biol. Ecol. 131 (1), 23–43.

Macia, A., Abrantes, K.G.S., Paula, J., 2003. Thorn fish Terapon jarbua (Forskal) predationon juvenile white shrimp Penaeus indicus H. Milne Edwards and brown shrimpMetapenaeus monoceros (Fabricius): the effect of turbidity, prey density, substratetype and pneumatophore density. J. Exp. Mar. Biol. Ecol. 291 (1), 29–56.

McCaughey, V., 1917. Themangrove in theHawaiian Islands. Hawaii. For. Agric. 14, 361–366.McCoy, D., 2011. Improving Sample Collection and Time Series Analysis of Environ-

mental Data in He'eia Fishpond, O'ahu, Hawai'i, from 2007 to 2011, Global Environ-mental Science. University of Hawai'i at Mānoa.

Mchenga, I.S.S., Mfilinge, P.L., Tsuchiya, M., 2007. Bioturbation activity by the grapsidcrab Helice formosensis and its effects on mangrove sedimentary organic matter.Estuarine Coastal Shelf Sci. 73 (1–2), 316–324.

Middleton, B.A., McKee, K.L., 2001. Degradation of mangrove tissues and implicationsfor peat formation in Belizean island forests. J. Ecol. 89 (5), 818–828.

Mohapatra, A., Mohanty, R.K., Mohanty, S.K., Dey, S.K., 2010. Carapace width andweight relationships, condition factor, relative condition factor and gonado-somatic index (GSI) of mud crabs (Scylla spp.) from Chilika Lagoon, India. IndianJ. Mar. Sci. 39 (1), 120–127.

Mumby, P.J., Edwards, A.J., Arias-Gonzalez, J.E., Lindeman, K.C., Blackwell, P.G., Gall, A.,Gorczynska, M.I., Harborne, A.R., Pescod, C.L., Renken, H., Wabnitz, C.C.C.,Llewellyn, G., 2004. Mangroves enhance the biomass of coral reef fish communitiesin the Caribbean. Nature 427 (6974), 533–536.

Nagelkerken, I., van der Velde, G., 2004. Relative importance of interlinked mangrovesand seagrass beds as feeding habitats for juvenile reef fish on a Caribbean island.Mar. Ecol. Prog. Ser. 274, 153–159.

Nakahara, B., 2007. Utilization of Mangrove Habitat by Megafauna Along the SouthernCoast of Moloka'i, Hawai'i, Oceanography. University of Hawai'i at Mānoa, Honolulu.

Nordhaus, I., Wolff, M., 2007. Feeding ecology of the mangrove crab Ucides cordatus(Ocypodidae): food choice, food quality and assimilation efficiency. Mar. Biol.151 (5), 1665–1681.

Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O'Hara, R.B., Simpson,G.L., Solymos, P.M., Stevens, H.H., Wagner, H., 2012. Vegan: Community EcologyPackage. R package version 2.0-7.

Poret, N., Twilley, R.R., Rivera-Monroy, V.H., Coronado-Molina, C., 2007. Belowgrounddecom-position ofmangrove roots in Florida Coastal Everglades. Estuar. Coasts 30 (3), 491–496.

Posey, M.H., 1987. Influence of relative mobilities on the composition of benthic com-munities. Mar. Ecol. Prog. Ser. 39 (1), 99–104.

R Development Core Team, 2012. R: A Language and Environment for Statistical Com-puting. R Foundation for Statistical Computing, Vienna, Austria 3-900051-07-0(URL http://www.R-project.org/).

Rauzon, M.J., Drigot, C., 2002. Red mangrove eradication and pickleweed control in aHawaiian wetland, waterbird responses, and lessons learned. In: Veitch, C.R.,Clout, M.N. (Eds.), IUCN Species Survival Commission No. 27. IUCN — World Con-servation Union, Gland, Switzerland.

Robertson, A.I., Duke, N.C., 1987. Mangroves as nursery sites — comparisons of theabundance and species composition of fish and crustaceans in mangroves andother nearshore habitats in tropical Australia. Mar. Biol. 96 (2), 193–205.

Roongrati, M., Iaitim, W., 1994. Some Biological Aspects of Long-eyed Swimming Crab(Podophthalmus vigil) Along the Eastern Coast of the Gulf of Thailand. EasternMarine Fisheries Development Center, Rayong.

Ruttenberg, K.C., McManus, M.A., Glazer, B.T., Falinski, K., Mills, H., 2013. A PreliminaryStudy of the Impacts of Mangrove Removal on Flow, Turbidity, and Redox State ofthe Benthos in He'eia Fishpond, O'ahu. Hawai'i. SOEST Report # 14–01.

Schrijvers, J., Okondo, J., Steyaert, M., Vincx, M., 1995. Influence of epibenthos onmeiobenthos of the Ceriops tagal mangrove sediment at Gazi Bay, Kenya. Mar.Ecol. Prog. Ser. 128 (1–3), 247–259.

Schrijvers, J., Camargo, M.G., Pratiwi, R., Vincx, M., 1998. The infaunal macrobenthosunder east African Ceriops tagal mangroves impacted by epibenthos. J. Exp. Mar.Biol. Ecol. 222 (1–2), 175–193.

Scowcroft, P.G., Conrad, C.E., 1992. Alien and native plant response to release from feralsheep browsing on Mauna Kea. In: Stone, C.P., Smith, C.W., Tunnison, J.T. (Eds.),Alien Plant Invasions in Native Ecosystems of Hawai'i: Management and Research.University of Hawai'i Press, Honolulu, HI, pp. 625–665.

Sheridan, P., 1997. Benthos of adjacent mangrove, seagrass and non-vegetated habitatsin Rookery Bay, Florida. USA. Estuar. Coast. Shelf Sci. 44 (4), 455–469.

Songrak, A., Koedprang, W., Wangpittaya, A., 2010. Fishery Biology of Spiny Rock Crab(Thalamita crenata Latreille, 1829) in Sikao Bay, Trang Province, Thailand.

Stone, C.P., Cuddihy, L.W., Tunison, J.T., 1992. Responses of Hawaiian ecosystems to re-moval of feral pigs and goats. In: Stone, C.P., Smith, C.W., Tunison, J.T. (Eds.), AlienPlant Invasions in Native Ecosystems of Hawai'i. Management and Research Uni-versity of Hawai'i Press, Honolulu, HI, pp. 666–704.

Sukumaran, K.K., Neelakantan, B., 1997. Length–weight relationship in two marineportunid crabs, Portunus (Portunus) sanguinolentus (Herbst) and Portunus(Portunus) pelagicus (Linnaeus) from the Karnataka coast. Indian J. Mar. Sci. 26(1), 39–42.

Sweetman, A.K., Middelburg, J.J., Berle, A.M., Bernardino, A.F., Schander, C.,Demopoulos, A.W.J., Smith, C.R., 2010. Impacts of exotic mangrove forests andmangrove deforestation on carbon remineralization and ecosystem functioningin marine sediments. Biogeosciences 7 (7), 2129–2145.

ter Braak, C.J.F., 1986. Canonical correspondence analysis: a new eigenvector techniquefor multivariate direct gradient analysis. Ecology 67 (5), 1167–1179.

Torres-Pratts, H., Schizas, N.V., 2007. Meiofaunal colonization of decaying leaves of thered mangrove Rhizophora mangle, in southwestern Puerto Rico. Caribb. J. Sci. 43(1), 127–137.

Twilley, R.R., Lugo, A.E., Pattersonzucca, C., 1986. Litter production and turnover inbasin mangrove forests in southwest Florida. Ecology 67 (3), 670–683.

Twilley, R.R., Pozo, M., Garcia, V.H., RiveraMonroy, V.H., Bodero, R.Z.A., 1997. Litterdynamics in riverine mangrove forests in the Guayas River estuary, Ecuador.Oecologia 111 (1), 109–122.

Vasconcellos, S.M.K., 2007. Distribution and Characteristics of a Photosynthetic BenthicMicrobial Community in a Marine Coastal Pond. University of Hawai'i at Mānoa,Honolulu.

Wieder, R.K., Lang, G.E., 1982. A critique of the analytical methods used in examiningdecomposition data obtained from litter bags. Ecology 63 (6), 1636–1642.

Williams, D.G., 1994. Marine habitats of the Cocos (Keeling) Islands. Atoll Res. Bull.(406), 1–10.

Young, C.W., 2011. Perturbation of Nutrient Inventories and Phytoplankton Communi-ty Composition During Storm Events in a Tropical Coastal System: He'eia Fishpond,O'ahu, Hawaii, Oceanography. University of Hawai'i at Mānoa.