Spatial and temporal distribution of small pelagic fishes in the territorial waters of Tanzania.

Study Extent

Twenty one landing sites were selected from Frame survey to represent ring-net fishery along the coastal waters of Tanzania mainland and Zanzibar. In each landing site ring-net fishers were selected and trained. The training involved the use of GPS to mark the position where fishing event took place and fisheries data like time to start fishing, time finish fishing, weight and number of a catch, species, number of fishers in a boat, fishing ground name, and engine size (Hp) were recorded. Once the boat has arrived at the landing site fishers submit a data sheet to the enumerator who then enters that information into the electronic catch assessment (eCAS) system using a mobile phone. Once the data is in the database it is downloaded for further analysis. The eCAS database contains the fishery data that were fished within the territorial waters in 2020.

Sampling

For satellite data, daily and monthly SST and Chl-a composites data that cover the entire EEZ of Tanzanian from Plymouth Marine Laboratory (PML) archived at TAFIRI e-Station were downloaded. Daily data has a spatial resolution of one kilometer and monthly composites has four kilometers. Daily data intended for looking for association between catch rate and bio-physical variables and monthly composites for computing seasonal and spatial differences of SST and Chl-a in EEZ. To match fishery data, satellite data covering a period between 2011-2020 were used. For Potential fishing zone data, daily fronts derived from satellite sea surface temperature which serves as potential areas for fishing were obtained from Plymouth Marine Laboratory (PML) archived at TAFIRI e-Station were downloaded. Three days composite data of one kilometer resolution covering a period between 2015 to 2020 were used to derive SST fronts’ data

Method steps

1. In the present study, trend analysis has been done by using non-parametric Man- Kendall test. This is a statistical method which is being used for studying the spatial variation and temporal trends of SST and Chl-a. Man-Kendall test was chosen for monotonic seasonal and annual trends because it is a non-parametric test, capable of handling even data that are not normally distributed (Millard et al. 2020).Man-Kendall test is preferred because the study region is influenced by monsoon seasons of warm water and low Chl-a concentration (North East) and cooler water and high Chl-a concentrations (South East). Mann (1945) formulated a non-parametric test for trend detection and Kendall (1975) improved the technique by introducing an algorithm that correct for non-linear trend and turning point. We used Mann-Kendall test to detect if there are seasonal and monotonic trends of SST and Chl-a in the coastal and marine water of Tanzania.
2. We use Hovmoller diagrams to monitor SST and Chl-a concentration in coastal and marine waters between 2011 to 2019. We aligned month–year against variable of interest to plot 3-Dimension of SST and Chl-a. The ideal of using month–year cross section is able to reveal a tongue of relative higher or lower values of SST or Chl-a, which is invaluable for understanding changes in bio-physical variables. Satellite data was used to monitor spatial and seasonal variations of SST and Chl-a. the downloaded data contains acquisition date that was used to decompose into month and year. Computed month variables were then used to sort values of SST and Chl-a to their respective monsoon seasons—northeast (November to March), southeast (May to September).
3. To associate catch rates from ring-net, longline and purse seine with environmental variables, the location and time of fishing events were used to match the location and date of SST, Chl-a and primary productivity (PP). Once the time and location of fishing events matched then variable of satellite data the value of SST, Chl-a and PP were extracted. Extracted environmental variable were used to access the influence of SST, Chl-a and PP on the spatial and seasonal variation of catch rates. Catch rates served as depended (response) variable and SST, Chl-a, PP, longitude, and latitude as independent (explanatory) were fitted using Generalized Additive Model (GAM).
4. To understand a spatial variation in catches and compare them with environmental variable we standardized fisheries catch data to catch rates. Recorded catch and time spend for fishing was used to compute the catch per unit effort (CPUE). CPUE was obtained by dividing total weight of fish with time spent for each fishing event. The equation for computing CPUE is given below. CPUE= W(vd)/t(vd) Where W is total weight and t is an effective fishing time and v is a vessel on that particular day.
5. To compute the spatial distribution of fishing effort, the area of territorial waters was divided into 10km grids, and in each grid a total number of fishing events were counted. Grids without any fishing events were removed from dataset and the remaining were grouped into North East and South East monsoon seasons. The same technique was used to compute the spatial distribution of fishing effort in the EEZ. the area of EEZ was divided into 10km grids, and in each grid a total number of fishing events from dominant species of tuna (bigeye, yellowfin, and skipjack) and tuna-likes (swordfish) were counted. Any grids with empty fishing events were excluded from dataset and then grouped based on North East and South East monsoon seasons. Daily sea surface fronts were used to compute monthly potential fishing zones. The date of each front was used to derive a month that correspond to that date. Those months were used to aggregate a total number of potential fishing zone and a five years climatology of gridded PFZ was derived. Triangulate methods were used to analyze and summarize categorical data collected through questionnaires. Cross-tabulation was used to compute frequencies, percentage and marginal sums between pairs of discrete variables.
6. Data obtained were processed in R language (Team 2020). Because these data came from different sources, they were in different format. To use these data, we had to process and format them in a consistency structure for easy further analysis. The processed data were compiled and stored in comma separated delimited (.csv). Data were checked for outlier and we never checked for normality because the statistical test used are non-parametric—the test is independent of mean and standard deviation—measure of parametric. Seasonal and annual trends test proposed by Kendall (1975) was used to quantify the rate at which SST and Chl-a change over period of ten years. The influence of SST, Chlorophyll-a and Primary Production on catch rates from territorial waters and Exclusive Economic Zone (EEZ) was analyzed using Generalized Additive Model (GAM). GAM model was used because of it can handle data that are polynomial in nature (Wood 2015).Plotting, mapping and visualization of data was done using ggplot2 (Wickham 2016).

All fishes were identified to family or species.

1. Stolephorus commersonnii
   common name: Anchovy rank: species
2. Rastrelliger kanagurta
   common name: Mackerel rank: species
3. Auxis thazard
   common name: Mackerel rank: series
4. Caranx tille
   common name: Caranx rank: species

At the United Republic of Tanzania-Coastal marine waters, 800km stretch from the boarder of Kenya to the border of Mozambique between latitude 4.8°S to 10.4°S and longitude 38.3°E to 43.8°E.