GIS-based identification of suitable sites for tilapia fish farming in Gauteng

July 14th, 2014, Published in Articles: PositionIT

 

In October 2010, the Agricultural Research Council – Institute for Soil, Climate and Water (ARC-ISCW) was contracted to research the use of a geographical information system (GIS) for the identification of suitable sites for warm water fish aquaculture (in ponds and cages) in the Gauteng Province, to this end, suitability maps for Mozambique tilapia were produced for cage and pond culture in the province.

Aquaculture, or fish farming, is the fastest growing segment of agriculture around the world. As natural fisheries become depleted, fish farming is taking on a greater role in helping to feed world populations [1]. Globally, nearly half the fish consumed by humans is produced by fish farms [2], and this worldwide trend towards increased aquaculture production is expected to continue [3].

Fig. 1: Schematic representation of the methodology.

Fig. 1: Schematic representation of the methodology.

In South Africa, aquaculture can play a significant role in food security, by providing a cheaper source of protein. However, there has been limited growth in the aquaculture sector in South Africa [7]. The fact that aquaculture is globally growing faster than any other food production sector, seen together with the pervasive poverty in rural areas, creates a powerful policy incentive to promote rural aquaculture where suitable conditions exist [8].

A realistic and practicable way of supplying more food protein is to increase fish production through the extension of aquaculture and inland fisheries. Since production sites for these activities need to satisfy fairly complex location criteria, it is important that suitable areas are identified and preferably designated in advance [9].

The warm water fish species considered for this study were Nile tilapia (Oreochromis niloticus), Mozambique tilapia (Oreochromis mossambicus), Sharptooth African catfish (Clarias gariepinus) and Common carp (Cyprinus carpio). Mozambique tilapia was chosen due to restrictions on the other two species in terms of legislation. Carp is currently blacklisted [12] and one cannot obtain a permit for the Nile tilapia which is generally regarded as the most suitable of all the tilapia species for culturing as food fish and is also the fastest grower [12]. Although it is present in the rivers of the Mpumalanga lowveld and Limpopo Province it is not yet legal for culturing in South Africa [13], which is why the slower growing Mozambique tilapia was chosen for this project. Consumer preferences presented a difficulty in marketing catfish [12] and it was thus also not considered for this study.

Fig. 2: Dams and perennial rivers in relation to the temperature condition grid.

Fig. 2: Dams and perennial rivers in relation to the temperature condition grid.

Earth ponds and cages were considered as a culture environment [13] because they offer a relatively low capital cost and operational simplicity.

Site selection is a key factor in any aquaculture operation, affecting both success and sustainability, as described by Hossain et al. [16] to identify suitable urban water bodies (UWB) for carp farming development.

The importance of this GIS desktop exercise for the Gauteng Department of Agriculture and Rural Development (GDARD) project to identify places and conditions of potential for aquaculture production in Gauteng, lies in its cost- and time-effectiveness rather than making use of traditional methods of site selection which are expensive and time consuming.

GIS is a tool that allows for the processing of spatial data into information tied explicitly to, and used to make decisions about, some portion of the earth. GIS combines multiple data layers to find suitable sites for trout and other forms of aquaculture [17]. The GIS methodology will eliminate visits to unsuitable sites. According to Steer [17], the use of GIS is costly in the beginning, but in the long run will save time and money.

Steer [17] also concludes that the GIS site selection method is an improvement (faster and cheaper) on the physical (physically visiting possible sites) site selection method. The GIS site selection study also generates information which can be used in negotiations for possible aquaculture sites [17].

Fig. 3: Sites in relation to fish processors.

Fig. 3: Sites in relation to fish processors.

GIS has the ability to combine the many diverse and complex factors which may need to be considered to reach development and administrative decisions. It provides an excellent modelling tool for environmental issues as well as a superb means of linking between biology, physiology, environment, systems and socio-economics.

This GIS research study was conducted using a desktop GIS exercise in ArcGIS 9. Variables and their respective parameters (i.e. thresholds) were identified as conditions for inland aquaculture suitability as described later on. For example, for tilapia, the tolerable water temperature is 15-42°C whilst the optimal temperature range is 28-30°C throughout the year [13].

In addition to the variables described later in this article, many other variables may also be considered. For instance, instead of just considering the slope of the terrain, the aspect of the slope could be considered. In the Southern Hemisphere, north-facing slopes are warmer than south-facing slopes [13] and this will have an impact on the water temperature.

Another additional factor that might have an effect on water temperature is the amount of wind that occurs on site. A windy site will tend to be cooler than a protected site, but a small amount of wind on a stagnant earth pond will aid in circulating the water [13].

Study area

Gauteng was chosen as the study area as (GDARD) jurisdiction resides in the Gauteng Province. This GDARD project is in line with the Gauteng Agricultural Development Strategy (GADS). Gauteng comprises 1,4% of the total area of South Africa and is located between 27 to 29 degrees east and 24 to 26 degrees south in the north-eastern part of South Africa. It covers an area of 16 548 km2 [26]. Gauteng is a summer rainfall area and has hot summers and cold winters with frost. Gauteng is generally regarded as too cold for warm water fish and too hot for cold water fish but actual GIS evaluation to verify these statements had not been done prior to this research project.

Datasets utilised

The site selection variables and their parameters (or thresholds) were identified as conditions for inland aquaculture suitability. Hossain et al. [16] suggested variables (or factors) to be considered when choosing a suitable site for inland aquaculture. These variables were traditionally grouped as water quality, soil quality, topography, infrastructure and socio-economic conditions and should be specific to the desired culture system and species selected.

The variables chosen for this project were identified on some of the production factors mentioned by Meaden and Kapetsky [9], Viljoen [20], Ragbirsingh and Da Souza [27], Aquaculture Innovations [13] and Hossain et al. [16] and on the availability of data.

Fig. 4: High and low risk dams.

Fig. 4: High and low risk dams.

The following datasets which served as geographical representations of reality were collected for the Gauteng Province.

Water source input data

  • Springs and boreholes vector file
  • Rivers vector file
  • Dams vector file

Restrictions

  • Water temperature rasters [28]
  • 90 m DEM raster
  • Protected areas vector file
  • Urban areas vector file
  • Wetlands vector file
  • Roads vector file
  • Distribution lines vector file
  • Transmission lines vector file
  • Substations vector file
  • Fish processors vector file

Economic factors

  • Soil clay percentage raster
  • Population density vector file
  • Unemployment vector file
  • Soil pH raster

Methodology

ArcGIS 9 (Esri) and MS Excel (Microsoft) software ware used to import, analyse and display spatial information. A summary of the methodology is displayed schematically in Fig. 1. The datasets were projected to Transverse Mercator (TM) with a central meridian of 28 degrees east and WGS 84 Datum so that buffering and areas could be calculated in metres.

Fig. 5: Mean pH values map of high and low risk spring and borehole sites.

Fig. 5: Mean pH values map of high and low risk spring and borehole sites.

Water source input data

All existing springs, boreholes and perennial rivers in Gauteng were considered as sources of water for fish ponds. Spring and borehole sites were obtained from the Department of Water Affairs (DWA) [19] in tabular format and converted to a vector file. This vector file was buffered by 1 km and then converted to raster format and this raster was subjected to the masks described below. One km is considered the furthest economically viable distance from a water source.

River sites were derived by buffering all perennial rivers in the river vector file by 1 km. This river file was converted to raster format which was then subjected to the sieving process.

All existing farm dams, irrigation dams and other registered dams in Gauteng were considered as sources of water for cage culture. Dams can be registered or not depending on their safety level. All dams of 50 000 m³ and above are registered with the Department of Water Affairs [19]. Vector data of all existing dams were collected on the assumption that the dams were available for cage fish production. The vector files of all existing dams were converted to a single raster file which was subjected to the restricted area masks as discussed below.

Restrictions

Restricted areas include protected areas, urban areas and wetlands and should be avoided. Water temperature is the most important limiting factor for tilapia, and the threshold was set at the tolerable water temperature of 15 – 42°C [13]. A restriction in terms of water temperature was considered by collecting water temperature rasters from the African Water Resources Database [28], which is a set of data and custom-designed tools, combined in a GIS analytical framework aimed at inland aquaculture planning for Africa. It was developed by FAO Aquaculture Service (FIRA) [29] and is aimed at facilitating responsible inland aquatic resource management with a specific focus on inland fisheries and aquaculture. It thus provides a valuable instrument to promote food security. The data could deepen the understanding of inland aquatic resource management and be of immediate value in addressing a wide variety of management and research questions [29]. These rasters were created using minimum and maximum air temperatures and mean annual wind velocity data [14]. The data have a spatial resolution of 5 km.

Fig. 6: Clay percentage map of high and low risk river sites.

Fig. 6: Clay percentage map of high and low risk river sites.

Due to the low resolution, the effect of microclimate might be omitted. Areas suitable for warm water fish might also exist in the parts of Gauteng deemed not suitable temperature-wise. Viljoen [20] recommended a physical examination and water temperature readings over a representative period.

The average monthly water temperature rasters from the AWRD [28] database were used to produce a conditional raster. ArcGIS Spatial Analyst was used to create the suitability rasters via condition functions from the temperature rasters. The condition for warm water fish in terms of average monthly temperature was that for preferably 12 months of the year, the water temperature should be in the range of 15 – 42°C. These suitability rasters were multiplied with each other to create a temperature suitability raster for warm water. The water temperature suitability raster for warm water fish was multiplied by the river raster and the spring and borehole raster as well as the dams raster.

An in-house 90 m Digital Elevation Model (DEM) raster was used to create a slope raster in ArcGIS 10 Spatial Analyst to be used for ponds. The ideal site should be flat or almost flat with a slope of 0,5 – 1% recommended, up to a maximum of 2,5%.

A suitability raster was created from the slope raster using ArcGIS 10 Spatial Analyst. For pond suitability the river raster and the spring and borehole raster located in suitable water temperature areas were multiplied by the slope suitability raster.

Fig. 7: High and low risk spring and borehole sites.

Fig. 7: High and low risk spring and borehole sites.

A protected areas vector file was converted to a protected area raster file called a protected area mask.

An urban area raster was created from the National Landcover 2000 raster [21] and a town area vector file converted to a town area raster. By combining these two files, an urban area mask raster was created.

A wetland mask was created by converting wetland data files to a mask raster. The National Landcover 2000 [21] raster of wetlands was used for the wetlands mask of cage culture, whereas the SANBI wetland map 3 [30] vector was converted to a raster and used for the pond culture mask.

First the raster of all existing dams was subjected to the protected areas mask. The resulting raster was subjected to the wetlands mask. After the subjection to the wetlands mask, the raster of dams was overlaid with the urban mask. The resulting dams were subjected to the substantial roads mask. The remaining dams are now not in restricted areas, but a further step in the sieving process is added whereby the suitable temperature condition mask is multiplied by the dams raster (Fig. 2).

A minimum distance from substantial roads should be less than or equal to 200 m. Steer [17] recommends a distance of 50 m or less from substantial roads, but the criteria was relaxed in this project in order to include more possible sites. It is important that 35 t trucks, which deliver feed and transport fish, be able to access the aquaculture sites. The creation of a substantial roads mask was achieved by buffering the vector file of all existing roads with 200 m and converting this roads vector file to a substantial roads mask raster. Dams, river, spring and borehole sites that were more than 200 m from a substantial road were eliminated via multiplication by this sieving process.

Transmission and distribution line vector files were obtained from Eskom and buffered with 75 m and converted to a raster to create a power line mask raster. This power line mask raster was used to eliminate sites which fell within 75 m of a power line.

Fig. 8: High and low risk river sites.

Fig. 8: High and low risk river sites.

Using Google Earth [11] and the in-house substation vector file, substations were measured and buffered with the measured distances and converted to a raster to create a mask of power lines and substations raster. River, spring and borehole sites located within this mask were eliminated by means of multiplication. However, as an economical factor, the closer the potential site is to medium volt power supply, the better. Therefore the Euclidian distances of potential
sites to medium volt distribution lines were mapped.

The distance from the potential sites to fish processors was taken into account as this will have cost implications. Steer [17] suggests a maximum distance of 100 km. Locality information of existing fish processors was collected and converted into a vector dataset. This fish processors vector file was buffered by 100 km. Euclidean distances between fish processors and possible sites in Gauteng were within 100 km (Fig. 3).

As most of the Gauteng water sources were within the 100 km buffer shape file, no mask was created. The small area in the north of Gauteng that was outside this 100 km buffer zone might well be within 100 km of fish processors in the adjacent Limpopo, Mpumalanga or North West provinces (Fig. 3).

Economic factors

Steer [17], Meaden and Kapetsky [9] and Aquaculture Innovations [13] discuss security risks involved with having cage and pond culture enterprises. Theft and vandalism are risk factors for site security. Steer [17] recommends that sites within a distance of 5 km from areas with a population density of 5000 people per square km or higher, and an unemployment rate of 25% and above, should be avoided as these pose a security risk. But the other side of the argument is that these aquaculture ventures might find marketing of their product and employment for their enterprises easier in these regions.

Therefore, sub-places with a population density of 5000 people per square km or more and an unemployment rate of 25% or higher were buffered by 5 km and converted to a raster file, called the risk mask. The parts of a river, spring and borehole grid located within this risk mask were named high risk areas, while river, spring and borehole sites located outside the risk mask were named low risk areas. Dams that were located within the risk mask were rated as “high risk” dams while dams outside of the risk area was named “low risk” dams (Fig. 4).

Fig. 9: Suitability map for Nile tilapia, African catfish and Common carp [14].

Fig. 9: Suitability map for Nile tilapia, African catfish and Common carp [14].

Because soil plays an important role in the economic viability of a pond aquaculture enterprise, soil factors such as soil pH (see Fig. 5) and clay percentage (see Fig. 7) were mapped. Dense, heavy clay soil is best for digging ponds and preventing seepage. Stony or sandy soil should be avoided.

Soil pH can affect water pH. A preferred pH range is from 6,8 to 8,5 [13]. However, the pH of the ponds can be controlled by liming before and after every cycle and the seepage rates of the ponds can be controlled by the use of plastic linings.

Two in-house soil clay percentage vector datasets were combined to form a soil clay percentage raster, and the values extracted for the existing river, spring and borehole raster that remained after the sieving process. The in-house soil pH raster of Gauteng was used to determine the soil pH of the river, spring and borehole sites that were left after the sieving process.

The sieving process resulted in high risk and low risk spring and borehole sites suitable for Mozambique tilapia (Fig. 9). The sieving process of the river raster file resulted in high risk and low risk river sites suitable for Mozambique tilapia (Fig. 10). The result of the sieving process for cage culture concerning Mozambique tilapia in Gauteng can be seen in Fig. 4.

Results and discussion

Places and conditions (variables) for potential tilapia aquaculture pond and cage sites in the Gauteng Province were mapped and identified by a limited budget- and time-constrained GIS desktop exercise.

For four out of 12 months the average monthly water temperature was below 15°C. Tilapia cultivation is still possible, but at low temperatures the fish will not grow. However, if the farmers have good water quality and feed the fish properly they can get them ready within 12 months.

Suitable sites making use of springs and boreholes as their water source are located in the central, northern, eastern and south-western parts of Gauteng. Dams that might be suitable for cage culture are located in the central, northern, eastern and south-western parts of the province. Fig. 2 which displays the position of the temperature condition grid for January to April and September to December explains most of these trends in the suitability maps as water temperature was a condition in the sieving process.

The resultant suitability map in a strategic reassessment of fish farming potential in Africa (Fig. 9), derived using MCE by Aguilar-Manjarrez and Nath [14], indicates that most of Gauteng is unsuitable and is at a much larger scale than the suitability maps produced in this research project.

Conclusions and recommendations

The central, northern, eastern and south-western parts of Gauteng are more suitable for tilapia aquaculture due to water temperature constraints that Gauteng poses to warm water fish.

This research project contributed to the knowledge of possible suitable sites for tilapia aquaculture enterprises in Gauteng on a smaller scale and incorporating possible water sources.

Field verification was not required for this project and, according to Marshall [32], there are virtually no commercial fresh-water fish farms in Gauteng, apart from a few small-scale catfish and tilapia units, all of which are experimental. There are also no hatcheries in Gauteng; the last tilapia and catfish hatchery was dismantled in 2010.

Limited knowledge exists of two feed suppliers in Gauteng [32], but it is recommended that for follow-up studies these feed suppliers as well as the vector file which contains fresh produce markets be used in an infrastructure sub-model to determine the suitability of sites and dams.

Potential sites should not be in flood zones. The acquisition of 20-, 50- and 100-year flood zone geographical information is recommended whenever a potential site is investigated to ascertain the proneness to flooding of the area.

Sources of pollution should be investigated. According to Viljoen [20], it is important to ensure that agricultural and mining chemicals and elements have not polluted the soil. An example is chlorinated hydrocarbon insecticides used in cotton cultivation. This type of pollution is detrimental to the fish and therefore human consumption of the fish. According to Silberbauer [33] acid mine drainage should also be kept in mind. An in-house vector file containing sources of soil pollution was utilised to display the proximity of pollution to existing high risk and low risk dams, high risk and low risk spring and borehole sites and river sites.

Acknowledgements

The author would like to acknowledge the Gauteng Department of Agriculture and Rural Development (GDARD) for funding of this project in conjunction with the Agricultural Research Council – Animal Production Institute (ARC-API).

References

[1] University of Wisconsin: Community, Natural Resource and Economic Development Impact Report: Growing Wisconsin’s aquaculture industry for the global marketplace, April 2009. www.uwex.edu/ces/cnred/documents/aquacultureprogram.pdf. Cited 11 July 2011.
[2] Stanford University: Stanford Report, September 8, 2009 http://news.stanford.edu/news/2009/september7/woods-fishfarm-study-090709.html.
[3] California Green Solutions: NOAA’s 10-year marine aquaculture plan, 2011. www.californiagreensolutions.com/cgi-bin/gt/tpl.h,content=1371. Cited 11 July 2011.
[4] FAO 2014: www.fao.org/focus/e/fisheries/aqua.htm. Cited 6 January 2014.
[5] FAO 2013: Low-Income Food-Deficit Countries (LIFDC) – List for 2013, www.fao.org/countryprofiles/lifdc/en/.
[6] Northern Aqua Farms: From: Snippets of information from FAO Reports and internet research, 2011. www.northernaquafarms.com/. Cited 11 July 2011.
[7] DAFF (Department of Agriculture, Forestry and Fisheries): Policy for the Development of Sustainable Inland Aquaculture Sector in South Africa, 2011. www.nda.agric.za/doaDev/fisheries/03_areasofwork/Aquaculture/AquaPolGuidLeg/AquaPolicyGuide/FreshwaterAquaculturePolicy.pdf. Cited 11 July 2011.
[8] QA Rouhani and PJ Britz: Contribution of aquaculture to rural livelihoods in South Africa: A baseline study. Report to the Water Research Commission.2004.
[9] GJ Meaden and JM Kapetsky: Geographical information systems and remote sensing in inland fisheries and aquaculture. FAO Fisheries Technical Paper No. 318. Rome, FAO. 262pp. 1991 www.fao.org/DOCREP/003/T0446E/T0446E00.HTM.
[10] FAO 2011: ftp://ftp.fao.org/es/ESA/policybriefs/pb_02.pdf. Cited 14 July 2011.
[11] MP Masser: Southern Regional Aquaculture Center (SRAC) Publication No. 163, April 1997. Revised.
[12] L Ter Morshuizen: Personal communication at Aquaculture in Southern Africa course attended in February 2011.
[13] Aquaculture Innovations: Aquaculture in South Africa, 5th Edition, ISBN 0-620-36488-2.2010.
[14] J Aguilar-Manjarrez and SS Nath: A strategic reassessment of fish farming potential in Africa, CIFA Technical Paper 32. Rome, FAO. 170pp. 1998.
[15] AE Nesamvuni and MC Dagada: Limpopo Department of Agriculture, Chapter 9: Aquaculture in the Limpopo province. 2011. www.lda.gov.za/downloads/downloads_reports/Aquaculture.pdf. Cited 23 February 2011.
[16] MS Hossain and SR Chowdburry, GD Dass: Integration of GIS and multicriteria decision analysis for urban aquaculture development in Bangladesh. Landscape and urban planning, Volume/issue 3-4 90, pp 119-133. 2008.
[17] LA Steer: Site selection for the small-scale aquaculture farming systems in the Western Cape: A GIS application. 2006.
[18] LG Ross: The use of geographical information systems in aquaculture: A review. 1998. www.aqua.stir.ac.uk/GISAP/gis-group/conf-papers. Cited 11 July 2011.
[19] DWA (Department of Water Affairs): Pretoria, Gauteng, South Africa. 2010.
[20] RH Viljoen: Report: Factors that influence the choice of a site for an aquaculture enterprise. 2001.
[21] NLC 2000: M Thomson. (2005) South African National Landcover 2000.
[22] RM Gurure: Influence of two organochloride pesticides, Thiodan and Lindane on survival of fingerlings of Oreochromis niloticus (Linnaeus) and Tilapia zillii (Gervais). 1987.
www.fao.org/docrep/field/003/ac162e/ac162e00.htm. Cited 14 January 2014.
[23] Michigan Department of Nature Resources: DDT and other Chlorinated Hydrocarbon Pesticides. 2014.
www.michigan.gov/dnr/0,4570,7-153-10370_12150_12220-26633–,00.html. Cited 14 January 2014.
[24] United States Environmental Protection Agency:Climate Impacts on Water Resources. 2013. www.epa.gov/climatechange/impacts-adaptation/water.html.
[25] Esri: Esri ArcReader10.1., 2014. www.esri.com/software/arcgis/arcreader
[26] South Africa.info: Geography and Climate, Gauteng. 2012. www.southafrica.info/about/geography/gauteng.htm. Cited 7 January 2014.
[27] Y Ragbirsingh and G De Souza: Site suitability for aquaculture development on the Caroni river basin, Trinidad West Indies using GIS. 2005. http://procs.gcfi.org/pdf/gcfi_56-61.pdf
[28] AWRD data archive of the FAO FIRA. www.fao.org/fishery/gisfish/servlet/CDSServlet?status=ND0yMzg5JjY9ZW4mMzM9KiYzNz1rb3M~ www.fao.org/docrep/010/a1170e/a1170e00.htm www.fao.org/fishery/collection/awrd/en. Cited November 2010.
[29] J Jenness, J Dooley, J Aguilar-Manjarrez and C Riva: African Water Resource Database. GIS-based tools for inland aquatic resource management. 1. Concepts and application case studies. CIFA Technical Paper. No. 33, Part 1. Rome, FAO, pp167. 2007.
[30] SANBI: Data from South African National Biodiversity Institute (SANBI). 2010.
[31] Google Earth 5.1 Retrieved from Earth.google.com. 2011.
[32] N Marshall: AASA (Aquaculture Association of South Africa). Personal communication.2010.
[33] M Silberbauer (RQS at DWA): silberbauerm@dwa.gov.za. Personal communication. 2010.
[34] MN De Mers: Fundamentals of Geographic Information Systems, 2nd Edition. New York, John Wiley & Sons, Inc., pp498. 2000.
[35] J Malherbe: Agroclimatologist, ARC-ISCW, Pretoria, Gauteng, South Africa. Personal communication. 2010.

Contact Celéste Dekker, ARC-ISCW, Tel 012 310- 2622, celeste@arc.agric.za