Study area description
Field sampling was conducted in Chwaka Bay and non-estuarine nearshore areas of Fumba on Zanzibar Island (Unguja), Tanzania, from January to December 2018 (Fig. 1). Chwaka Bay is a semi-enclosed embayment with a maximum average depth of 3.2 m at spring high tide and a total area of about 50 km2 at high water76. Mangroves are fringing the bay in the south with several creeks, while dense continued or disconnected seagrass meadows (commonly mixed but sometimes monospecific) of different complexity characterise large areas of the bay, which is bordered at the entrance by patch reefs77,78. Chwaka Bay is considered nursery ground for various fish species of economic and ecological importance79,80. In contrast to Chwaka Bay, the sampling area of Fumba is an open, non-estuarine environment in the coastal area of the Menai Bay Conservation Area (MBCA), where the main livelihood activities that surround the MBCA are fishing and agriculture81. The study area of Fumba is extensively covered by seagrass meadows, macroalgal belts, mangroves with small creeks and coral reefs with an average water depth of 10 m at high tide82. Fishing activities in both Chwaka Bay and Fumba are highly concentrated in nearshore areas with a subsequent pressure on the associated fish stocks81.
The southeast monsoon (SEM), lasting from April to October, drives the climatic conditions and is characterised by lower air temperatures, strong winds and rough sea, while the northeast monsoon (NEM) lasts from November to March and is characterised by higher air temperatures, lower wind speed and calm sea. There are two rainy seasons, including the long rain season from March to May and irregular short rains from September to November49. Mangroves in Chwaka Bay and Fumba are dominated by a muddy bottom substratum and turbid waters that fluctuate depending on runoff during different seasons, with average macroalgae coverage ranging from 3 to 29% (Table S3). Thalassia hemprichii dominated both inshore and nearshore seagrass meadows in the two study sites (Chwaka Bay and Fumba) (Table S3). All seagrass meadows, except inshore seagrass meadows in Fumba, were to some degree mixed with different seagrass species (i.e. T. hemprichii, Enhalus acoroides, Cymodocea rotundata and/or Syringodium isoetifolium) (Table S3). Calcareous algae (Halimeda spp.) as well as other macroalgae generally comprise a large part of seagrass meadows in Chwaka Bay77, while seagrass meadows of Fumba comprised macroalgae, such as Gracilaria spp. and Chaetomorpha spp.
Habitat characterization, sampling of fish larvae and environmental parameters
In each seascape area (Chwaka Bay and Fumba), sampling sites were established (0.5–5 km apart in each habitat) in mangrove creeks (Mang), inshore seagrass meadows (inSeag) (located adjacent to mangroves) and nearshore seagrass meadows (nearSeag) (located in-between mangroves and coral reefs). Habitat characterizations in terms of habitat cover (%), seagrass canopy height (cm) and seagrass shoot density (number of shoots per m−2) were conducted along transects (100 m in length) at one occasion per month in January, March, July and September 2018 to capture potential differences across seasons. In each seagrass habitat, two transects were laid parallel to the shoreline, while in mangrove creeks, two transects were laid from upstream towards the mouth of the creek. Approximately 10 m apart, a quadrat of 0.5 m2 was thrown randomly five times in each transect. In each 0.5 m2 quadrat (n = 5), the percentage cover of seagrass, macroalgae and unvegetated area was quantified and seagrass species composition determined. A quadrat of 0.0625 m2, placed inside the 0.5 m2 quadrat frame, was used to assess seagrass canopy height and shoot density (n = 5). In addition, substrate bottom type was also determined as either rocky, muddy or sandy.
Sampling of fish larvae was performed in the three habitats (mangroves and the two seagrass habitats) during daytime (between 6:30 and 11:00 h) at high tide on a monthly basis from January through December 2018. The sampling was carried out using an ichthyoplankton net (500-μm in mesh size and a cod end of the same mesh size) with a mouth diameter of 0.5 m and a length of 2.5 m, fixed with a flowmeter in the mouth frame to determine the filtered volume of water. The plankton net was towed horizontally (at an average depth of 1 m) behind a small boat for 15 min (with a very low speed of approximately 1–1.5 knots, which is equivalent to 1.9–2.8 km per hr) and replicated two times in each habitat. After each tow, the fish larvae specimens were placed in sample bottles, quickly fixed with 75% ethanol solution and transported to the laboratory for further analysis. GPS coordinates were taken at each sampling site to be able to follow up the same locations throughout all sampling occasions. In situ water environmental parameters were measured (in triplicate) at the water surface in each habitat of the two sites (Chwaka Bay and Fumba) during every sampling occasion and included pH, water temperature, dissolved oxygen (DO) and salinity. Water temperature and pH were recorded in the field using a multiprobe pH meter with a temperature sensor (Model STX-3). A portable refractometer (HHTEC 4-i-1) was used to measure salinity, and a DO meter was used to measure dissolved oxygen (Extech 407510). Triplicate samples of 1.5 L of water for chlorophyll-a analysis was obtained using a water sampler at a depth of one meter, placed in a cold box and transported to the laboratory for chlorophyll-a (phytoplankton biomass) determination.
In the laboratory, fish larvae were separated from other zooplankton and debris using a stereomicroscope (Zeiss Stemi 508). Using the specialized identification guides by Jeyaseelan83, Mwaluma et al.84 and Leis and Carson-Ewart85, each fish specimen was taxonomically identified to family level and measured for total length (mm). The growth stage of each specimen was determined as either preflexion, flexion, postflexion or juvenile, and in the case of syngnathiforms (seahorses and pipefishes), they were determined as either larvae, juvenile or adult because they do not have differentiated growth stages as larvae. Distorted fish larvae or very small larvae at the egg yolk stage, which were difficult to identify, were grouped as unidentified. At the end of the sampling, some few late juveniles and adults (about 2% of the catch) were sampled occasionally in the plankton net, particularly in seagrass habitats, and mostly from the family Syngnathidae (pipefishes and seahorses), which are slow swimmers, and a few individuals from the families Serranidae, Scaridae and Apogonidae. Chlorophyll-a concentrations were measured spectrophotometrically in the laboratory using a Shimadzu UV–visible spectrophotometer, following protocols by Strickland and Parsons86.
Before estimating the abundance of fish larvae (per 100 m3), all late juveniles (with all features of adult fish) and adult fishes were recorded and excluded from the catch. Habitat characteristics (i.e. habitat cover, seagrass canopy height and seagrass shoot density; Table S3) were compared among habitats, i.e. Mang, inSeag, and nearSeag, using one-way ANOVAs in SPSS version 20. Differences in fish larvae abundance and family richness were analysed using three-way model ANOVAs with Season (2 levels, fixed), Site (2 levels, fixed) and Habitat (3 levels, fixed) as explanatory factors. Prior to the ANOVA analyses, the assumption of homogeneity of variances was checked to discover if the data were normally distributed, and when it was heteroscedastic, the data were transformed using either log10 (for abundance data) or square root (for richness data) transformations. A posteriori multiple comparison tests were carried out on data from the significant interactions using the Holm-Sidak method. All analyses that concerned the three-way ANOVAs were performed in SigmaPlot version 14.0. Analysis of similarities (ANOSIM) was used to test for differences in assemblage structure across months, between monsoon seasons, among habitats, between sites and based on combinations of these factors. Since the postflexion growth stage comprised a large proportion of the catch, we did also separate multivariate analyses to test for differences in assemblage structure of fish larvae at this development stage. Patterns of similarities from one of the ANOSIM analyses were visualized using non-parametric multidimensional scaling (nMDS) based on Bray–curtis similarity index measures and calculated based on abundance data after square-root transformation. The multivariate analyses were performed using the PRIMER Software package87.
The study protocol was approved on the 23rd of January 2017 by the Department of Ecology Environment and Plant sciences (DEEP), Stockholm University, in collaboration with the Institute Postgraduate Studies Committee of the University of Dar es Salaam (UDSM) in compliance with the Tanzania Fisheries Act (2003) and the Wildlife Conservation Act (1974). We confirm that the study was undertaken with all procedures that minimize the pain and suffering, and improve animal welfare. The permit to sample and transport the larval fishes from the field to the laboratory was issued by the Ministry of Livestock and Fisheries and other local authorities for complying with the requirement of Fisheries Regulations (G.N. No. 308 of 2009).
Consent to participate
The authors declare their participation in the study.