The Solomon Islands National Ocean Planning Spatial Information System: Spatial Decision Support Tools for Seaweed Aquaculture Planning in the Solomon Islands

SINOPSIS – The Solomon Islands National Ocean Planning Spatial Information System: Spatial decision support tools for seaweed aquaculture planning in the Solomon Islands

Dan Breen, Antony Vavia, Clayton Chan, Wesley Garofe, Mahuri Robinson, Sebastian Misiga, Sylvester Diake, James Teri2, Rosalie Masu, Ronnelle Panda, Ruben Sulu, Anne-Maree Schwartz, Richard Walter, Jillian Fenigolo, Armagan Sabetian and Lindsey White

About NZIPR

The New Zealand Institute for Pacific Research (NZIPR) was launched in March 2016. Its primary role is to promote and support excellence in Pacific research. The NZIPR incorporates a wide network of researchers, research institutions and other sources of expertise in the Pacific Islands.

Funding for this project was provided by the New Zealand Ministry of Foreign Affairs and Trade through the New Zealand Institute for Pacific Research.

Published by

SINOPSIS – The Solomon Islands National Ocean Planning Spatial Information System: Spatial decision support tools for seaweed aquaculture planning in the Solomon Islands

2019

Dan Breen1, Antony Vavia1, Clayton Chan1, Wesley Garofe2, Mahuri Robinson4, Sebastian Misiga2, Sylvester Diake2, James Teri2, Rosalie Masu2, Ronnelle Panda2, Ruben Sulu3, Anne-Maree Schwartz3, Richard Walter5, Jillian Fenigolo1, Armagan Sabetian1 and Lindsey White1.

ISBN: 978-0-473-47786-8

1 School of Science, Auckland University of Technology 2 Solomon Islands Ministry of Fisheries and Marine Resources 3 Mekem Strong Solomon Island Fisheries (MISSIF) 4 Triland Trading 5 University of Otago

Page | 1

Abstract

Seaweed cultivation can provide important Island at two times, and photos from income in remote areas and significant unmanned aerial vehicles (UAV drones) to exports for nations like the Solomon Islands. build georeferenced photo-mosaics at five It may also provide a sustainable alternative sites in the Western Province. to more extractive industries like forestry, fishing and terrestrial agriculture. This report Findings from the project include that: 1) outlines the spatial decision support tools potentially suitable environments for developed to help planners and policy makers seaweed farms can be identified from GIS in the Solomon Islands identify sites suitable habitat maps produced for the “Millennium for seaweed cultivation. In collaboration with Coral Reef Mapping Project” (Institute for the Solomon Island Ministry of Fisheries Marine Remote Sensing, University of South and Marine Resources we identified physical, Florida (IMaRS-USF), 2005); 2) suitability biological, economic and social factors influ- was also affected by depth, substrata, encing the suitability of locations for seaweed current, wind and wave action and herbivory farms and mapped 35 existing, previous or by fish, dugong and turtles; 3) the market potential locations for seaweed farms. price of seaweed, buying arrangements and competing opportunities for employment in We integrated data from physical and other industries were important economic biological surveys with population census and social constraints; 4) surprisingly, data and suitability scores rated by seaweed distances to the capital city, towns, villages farming experts familiar with the region. The and land appeared less important in an open source geographic information system archipelago linked by networks of coastal QGIS was used to map over 100 variables waterways; 5) while existing databases approximating suitability for seaweed illustrated broad patterns in factors affecting farming in terms of marine habitat, distance suitability, there were subtleties in human to land, rivers, transport, and protected areas judgement and interpretation that were not and possible social and economic constraints. easily captured using GIS data alone; and Data were georeferenced to point locations therefore 6) an interactive exploratory for known seaweed farming areas and to approach combining data and local expertise habitat polygons for reefs throughout the is likely to be more successful; 7) drone Solomon Islands. Multivariate patterns photography and online satellite imagery can among farm sites were analysed and mapped be used to map farming areas at different and a hierarchical multiple criteria model spatial and temporal resolutions; and 8) GIS was used to estimate site suitability according can spatially integrate diverse data sources to standardised site scores and weighted and interactive modelling tools within criteria. We used high-resolution satellite intuitive visual analyses. imagery to map seaweed farms near Wagina

Page | 3 Acknowledgements

We acknowledge the Government of the Solomon Islands, the Solomon Islands Ministry of Fisheries and Marine Resources, Mekem Strong Solomon Island Fisheries (MISSIF), Triland Trading and the Solomon Islands Government ICTSU (Information Communications and Technology Support Unit) for their support. We thank the many people of the Solomon Islands including the village of Vavanga for their hospitality and logistic support. Funding for this project was provided by the New Zealand Ministry of Foreign Affairs and Trade through the New Zealand Institute for Pacific Research to the School of Science, Auckland University of Technology.

Contents

Abstract 3

Acknowledgements 4

1. Introduction 9

2. Methods 14 2.1 Location 14 2.2 Expert Knowledge 14 2.3 Site Selection Criteria for Seaweed Farms 14 2.3.1 Physical Selection Criteria 15 2.3.2 Biological Selection Criteria 16 2.3.3 Economic Selection Criteria 17 2.3.4 Social and Cultural Selection Criteria 18 2.4 Geographic Information System and Spatial Data 19 2.5 Geospatial Analyses 20 2.6 Multivariate Analyses 20 2.7 Multiple Criteria Analyses 20 2.8 Satellite Image Interpretation, Drone Photogrammetry and Ground Surveys 21

3. Results 22 3.1 Geographic information system 22 3.2 Marine habitats 22 3.3 Expert Knowledge 23 3.4 Geospatial Analyses 35 3.5 Multiple Criteria Analysis 35 3.6 Interpretation of satellite, drone imagery and site inspections 46

4. Discussion 53

5. Conclusion 57

6. References 58

Appendix 1. Key institutions and individuals consulted for this project 64

Page | 5 List of Figures Figure 1 Location of the Solomon Islands and the Exclusive Economic Zone. 13 Figure 2 SINOPSIS 3.5 (Solomon Islands National Ocean Planning Spatial Information System). A Quantum GIS 3.2.3 project containing ~100 physical, biological, economic, social and analytical formatted data layers to inform planning for sustainable seaweed aquaculture. 24 Figure 3 Locations identified as previous, current or potential sites for seaweed farming. These include some sites with marginal conditions which in some way make them less suitable. 25 Figure 4 Distribution of “shallow terrace” reef habitats in the Solomon Islands. Data from the Millennium Coral Reef Mapping Project, Remote Sensing Centre, University of Florida. (IMaRS-USF, 2005). 26 Figure 5 Percentage area of the most common reef habitats at identified seaweed farm locations in the Solomon Islands. Data from the Millennium Coral Reef Mapping Project, Remote Sensing Centre, University of Florida. (IMaRS-USF, 2005). 27 Figure 6 Area of “shallow terrace” marine habitat in provinces of the Solomon Islands. Data from the Millennium Coral Reef Mapping Project, Remote Sensing Centre, University of Florida (IMaRS-USF, 2005). 28 Figure 7. Overall suitability of identified sites for seaweed farming rated by Manager 1. 29 Figure 8. Overall suitability of identified sites for seaweed farming rated by Manager 2. 30 Figure 9. Comparison of overall rating of site suitability for seaweed farming for Manager 1 and Manager 2 (r=0.8, P<0.001). 30 Figure 10. Principal components analysis (PCA) for ratings of seaweed farm suitability by Manager 1. Principal component axes 1 and 2 (29 % and 23 % of variance explained). 31 Figure 11. Principal components analysis (PCA) for ratings of seaweed farm suitability by Manager 2. Principal component axes 1 and 2 (32 % and 22 % of variance explained). 31 Figure 12. Cluster analysis of site suitability ratings for Manager 1. 32 Figure 13. Cluster analysis of site suitability ratings for Manager 2. 32 Figure 14. Cluster analysis of Euclidean distances between seaweed farm sites for Manager 1 33 Figure 15. Cluster analysis of Euclidean distances between seaweed farm sites for Manager 2 34 Figure 16. Example of geospatial proximity analysis – distance of sites (km) to Honiara. 36 Figure 17. Principal components analysis (PCA) for GIS point estimates of seaweed farm suitability. Principal component axes 1 and 2 (33 % and 16 % of variance explained). 36 Figure 18. Cluster analysis of seaweed farm-site similarity based on geospatial variables. 37

Figure 19. Combined weighted multiple criteria score of site suitability for criteria from Manager 1. 39 Figure 20. Weighted multiple criteria score of site suitability from Manager 2. 39 Figure 21. Section of a Criterion Decision Plus multiple criteria model for just Manager 1 ratings nested within biological, physical, economic and social sub-criteria. 40 Figure 22. Combined multiple criteria scores of the overall suitability of potential seaweed farm sites for just Manager 1 (other data sources excluded). All criteria are weighted by the level of importance assigned to each criterion by the manager. 40 Figure 23. Section of a Criterion Decision Plus multiple criteria model for just Manager 2 ratings nested within biological, physical, economic and social sub-criteria. 41 Figure 24 Combined multiple criteria scores of the overall suitability of potential seaweed farm sites for just Manager 2 (other data sources excluded). All criteria are weighted by the level of importance assigned to each criterion by the manager. 41 Figure 25 Comparison of overall rating of site suitability for seaweed farming for Manager 1 with weighted multiple criteria scores for Manager 1 (r=0.28, P>0.1). 42 Figure 26 Comparison of overall ratings of site suitability for seaweed farming from Manager 2 with their weighted multiple criteria scores (r=0.4, P=0.07). 42 Figure 27 Section of a Criterion Decision Plus multiple criteria model for GIS point distances and areas nested within biological, physical, economic and social sub-criteria. 43 Figure 28 Combined multiple criteria site scores of suitability for seaweed growing derived from GIS point distances and areas (other data sources excluded). A spatial representation of these scores can be viewed in the QGIS project application. 43 Figure 29 Section of a Criterion Decision Plus multiple criteria model for mean GIS polygon distances and areas nested within biological, physical, economic and social sub-criteria. This part of the model is not weighted but this can be done interactively within the software. 44 Figure 30 Combined multiple criteria scores of suitability for seaweed growing derived from mean GIS polygon distances and areas. 44 Figure 31 Comparison of the multiple criteria scores for Manager 1 with unweighted multiple criteria scores for the GIS point data (r=-0.2, P>0.1). 45 Figure 32. Comparison of the multiple criteria scores from Manager 2 with the unweighted multiple criteria scores for GIS point data 45 Figure 33. Boundaries of seaweed farms west of Wagina Island digitised from WorldView-2 imagery from November 2013 (green) and October 2015 (blue). 47 Figure 34. Boundaries of seaweed farms east of Wagina Island digitised from WorldView-2 imagery from October 2015 (blue). 47

Page | 7 Figure 35. Georeferenced photo-mosaic of reef flat (left) and shallow terrace (middle) on the edge of the Vona Vona Lagoon. There are two trial seaweed farms visible to the west and north of the lagoon. 48 Figure 36. Photo-mosaic from drone (UAV) photogrammetry overlain on Worldview-2 satellite imagery at Sageraghi showing location of GoPro video transect through a previous seaweed farm (Vavia 2018). 49 Figure 37. Aerial drone image (upper) contrasted with satellite image (lower) at Boeboe 2, site of a previous floating longline trial. 49 Figure 38. Photo-mosaic from UAV drone overlaid on WorldView-2 satellite image at a previous farm site near Babanga. 50 Figure 39. Section of a drone photo-mosaic of previous single floating line near Logha showing boulder corals and other outcrops on sand. 50 Figure 40. Seagrass and Halimeda algae on seabed at Sageraghi. 51 Figure 41. Algae with trapped gas bubbles on seabed at Sageraghi. 51 Figure 42. Posts from previous seaweed farm at Sageraghi. 52 Figure 43. Change detection of coastal habitats at Kolombangara Island from UAV drone photo-mosaics flown in 2017 and 2018 (Chan 2018). Other data show the adjacent distribution of reef habitats and flood plumes from the river crossing these areas. 52

1. Introduction

This project developed geographical prices for seaweed fell by 20 % due to the information system (GIS) decision support increase in fuel prices which also affected tools to help plan for sustainable national and international freight costs. In development of seaweed farming in the 2007, natural catastrophes resulted in the Solomon Islands. Seaweed farming across loss of seaweed farms in Rarumana, where the Pacific has grown considerably in the last they remain out of operation, although few decades (Zemke-White and Ohno, farming continues at Wagina and some 1999, Zemke-White and Wilson, 2015) but other smaller areas. there are significant bottlenecks to development including the ability to keep Seaweed farming is a practice that does not track of existing farms and to identify require extensive investment and technology, suitable locations for farms (MRAG Asia and it may not impact heavily on the environ- Pacific, 2016). ment. However, the industry in the Solomon Islands, is constrained by finding suitable Seaweed farming trials began in the sites and maintaining employment consistency, Solomon Islands (Figure 1) in 1988 with production rates and market accessibility. cooperation from the MFMR (Ministry of Fisheries and Marine Resources) in the One of the advantages in farming K. alvarezii Western Province at Vonavona Lagoon and is its vegetative reproduction. It is a clonal Rarumana Village. The trials imported the species which can be propagated through red alga Kappaphycus alvarezii from Fiji and fragmentation (Hurtado et al., 2014; demonstrated good production. Buschmann et al., 2017). The farming Unfortunately, herbivorous fish affected process is also considered relatively simple most of the trials. In 2001, the Aquaculture with low capital costs, and quick production Division of MFMR began trials in cycles, and it has become a favourable Rarumana. The following year, trials in practice among coastal communities Rarumana produced over 600 kg of seaweed (Valderrama et al., 2013; Kumar et al., 2015; (dry weight). The RFEP (Rural Fishing Mantri et al., 2016). Enterprise Project) held seaweed farming workshops in cooperation with the MFMR Countries in the Pacific usually practise the and SPC (Secretariat of the Pacific off-bottom, floating-raft and floating longline Community) and in 2003, RFEP provided methods (Pickering, 2006). In Asia, the materials for seaweed farming, including shallow off-bottom mono-line technique outboard motors, internet for communi- is one of the most popular methods used cation and a warehouse in Rarumana. in recent times (Hurtado et al., 2013). Trono (1992) and McHugh (2006) In 2005, there were around 130 seaweed provide descriptions of this and other farmers in Rarumana and the Shortland farming techniques. Islands, and 300 seaweed farmers in Wagina. Farming then expanded to Makira- The off-bottom longline method involves Ulawa and Malaita. Operations also began in straight rows of wooden stakes driven into eastern Marovo Lagoon. A year later, the the substratum and connected by 3 mm

Page | 9 propylene line suspended approximately the Solomon Islands as a result of MSSIF 30 cm above the seafloor (Msuya, 2006). funding and made six recommendations, Small cuttings of seaweed are attached at four of which are germane to this project. 20 cm intervals on the propylene lines using Firstly, continue aid funding, as it has been raffia tie-ties. This is the main form of shown to be effective and critical to the seaweed farming in the Solomon Islands development of the seaweed industry. currently. Floating long lines, where seaweeds are suspended just below the Second, conduct a census seaweed farms, as surface, have also been trialled in the the poor temporal resolution of farm Solomon Islands and are less dependent on statistics produces uncertainties in the value depth and substrata. of the industry to the economy and may also impact on effective disaster relief. This According to Glenn and Doty (1990), one of project therefore, provides training and the main hindrances to developing seaweed geospatial mapping tools that enable farming is finding productive sites. As part of Ministry of Fisheries and Marine Resources this project, Vavia (2018) used GIS software staff to assess farm areas more frequently. to integrate environmental, social and geographical data to identify potential sites Third, improve methods to identify new for seaweed farming in the Solomon Islands. growing areas. The project integrates GIS, This project builds on the above study with analytical tools and spatial data to enable data and software that allow Solomon Island government managers, scientists, industry Fisheries staff to view and analyse data for and communities to visualise physical, different environmental and socio-economic biological and socio-economic influences on criteria and their effect on site suitability. farm site suitability and compare suitability scores for different locations, scenarios and The Solomon Islands National Ocean priorities. The tools integrate GIS data from Planning Spatial Information System many sources with heuristic knowledge from (SINOPSIS) project had two primary expert opinion. objectives. The first was to create decision support tools to allow managers and other Finally, geospatial analyses can be used to stakeholders to plan for sustainable proactively plan for sustainable development development in seaweed farming. The and avoid issues resulting from expansion second was to build capacity and training in and competition. The GIS databases and the use of GIS-based decision support tools tools include information on ecologically and remote sensing. important habitats such as seagrass, mangrove and coral reef, on marine The starting point for the research was a protected areas and on the distribution of socio-economic review of a 3-year seaweed villages, population, transport and aquaculture project undertaken by the infrastructure. The distance from farm sites consultants MRAG Asia Pacific funded to such features can be weighted in models under the New Zealand Aid Programme’s to reflect negative or positive contributions (NZAP) Mekem Strong Solomon Islands to farm site suitability. The models allow Fisheries (MSSIF) programme. The review managers and stakeholders to visualise analysed the socio-economic outcomes for spatial relationships between farm sites and

surrounding habitats and human activities workshop with approximately 15 staff, and interactively explore potential consultants and students. management options and outcomes. Consultation with managers, industry and In planning this project, we consulted with scientists were used to refine project Solomon Island Government managers, objectives and identify criteria to select areas field staff, consultants and local people suitable for seaweed farming. These (Appendix 1). In particular, two staff objectives and criteria were used to identify members of the Solomon Island Ministry of and acquire data sets and tools best suited to Fisheries and Marine Resources (MFMR) site selection. Ideally, existing GIS data sets and consultants worked on the team were sourced but other remote-sensing data, providing expert local knowledge of seaweed non-spatial data, literature and anecdotal farming in the region and input into the information were collated and converted to design of the decision support tools. spatial formats where possible. Environmental and other geospatial Three young Pacific researchers were constraints were formally summarised within investigators on our project. Antony Vavia multiple criteria models providing a basis for (of Fijian and Cook Island descent) quantitative and qualitative comparisons completed his BSc (Hons) dissertation on among sites. Proximity analyses of distances, “Selecting Seaweed Farm Sites in the transport links and access between farms, Solomon Islands using Geospatial coastal populations and markets were used Information Systems.” Much of the research to evaluate logistic feasibility for local and presented in this report is drawn from regional infrastructure, development, Antony’s dissertation. Clayton Chan (from production and the transport of goods and Samoa) completed a research project on employees. “Using Remote Sensing Technology to Analyse Land Use Change in Vavanga, There is a wide range of geospatial analysis Solomon Islands.” Jillian Fenigolo was tools that can be used to help select or cultural liaison and research assistant on the prioritise areas. We aimed to develop tools ethnographic investigations led by Prof. that could be easily and cheaply used and Richard Walter from Otago University. applied generally to similar projects in other areas. These include spatially explicit, We initially engaged in workshops with staff graphic, interactive GIS displays particularly and students at the Solomon Islands suited to community participation in National University (SINU) and staff at meetings, promoting public awareness and incorporating local knowledge and opinion. MFMR, MSSIF and the Solomon Islands Where necessary, we used spatial modelling, Government Information, Communications multivariate statistics, and multiple criteria and Technology Support Unit (ICTSU) on analysis to help identify potential farming the use of remote-sensing technologies and locations. Multiple criteria analysis evaluates GIS-based decision support tools. the performance of a set of options in Additional training occurred on field trips meeting an overall goal in terms of sets of with Ministry of Fisheries and Marine weighted objectives and criteria. In this Resources staff and later at a 3-day

Page | 11 project, multiple criteria analysis was used to combine data for many parameters and sources in an interactive model that can be used to rate alternatives according to existing data and expert opinion. In line with review objectives, we trialled the use of satellite and drone imagery to map and census seaweed farms and ground-truthed some locations using underwater video transects. Under increasing aquaculture development, it will become increasingly important to monitor the growth of the industry, its impact on surrounding areas and the potential for conflict with competing use. Remote-sensing tools can assist in this work.

All formatted data, software, analyses and a dedicated laptop computer were provided to the Solomon Islands Ministry of Fisheries and Marine Resources with a DJI Phantom 4 drone and training to allow Fisheries staff to map farm locations.

The following sections describe the methods used to incorporate expert knowledge on seaweed farming in the Solomon Islands, define criteria for site suitability and display and analyse data in a Quantum GIS (QGIS) project application and in multivariate and multiple criteria analyses. A results section displays examples of maps and quantitative analyses of seaweed farm-site suitability according to expert ratings, environmental data and analyses. Finally, we discuss the implications of the project outcomes and make recommendations for future research and management.

Figure 1 Location of the Solomon Islands and the Exclusive Economic Zone.

Page | 13 2. Methods

2.1 Location

The project was based in the Solomon Consultation occurred primarily with Islands, which lie east of Papua New Guinea government and university staff in the capital (Figure 1). The Solomon Islands include city of Honiara on the island of Guadalcanal nine provinces: Choiseul, Western Province, but also around Gizo in the Western Isabel, Guadalcanal, Rennell/Bellona, Province where field surveys of existing, Central, Malaita, Makira-Ulawa and Temotu previous and potential seaweed farm sites and have an exclusive economic zone of were made. around 1,600,000km2.

2.2 Expert Knowledge

Criteria for seaweed farm sites were a scale from 1, least suitable, to 10, most identified from previous studies (Glenn and suitable) for 20 potentially influential Doty, 1992; Trono, 1990, 1992; Msuya, variables and for each site’s overall 2006; Sousa et al. 2012; Dean and Salim, suitability. The relative importance of each 2013; Kapetsky et al. 2013; Cabral et al. variable for seaweed farming was also 2016) and from discussions with fisheries estimated on a scale from 1 to 10. staff, consultants and community. A manager, GIS officer and a consultant for This information was mapped, included in the Ministry of Fisheries and Marine the decision support tool and analysed in Resources worked with Auckland University multivariate ordinations, cluster analyses and of Technology (AUT) staff and students to multiple criteria analysis as described below. map the approximate boundaries of existing, Correlations between information sources, previous and potential other sites for managers and multiple criteria scores were seaweed farms (Figure 2). They also rated examined graphically and by Pearson’s r the suitability of the 35 sites for farming (on statistic.

2.3 Site Selection Criteria for Seaweed Farms

Physical, biological, social and economic according to their importance in general, variables likely to affect the suitability of and for specific farm sites, approximated in locations for seaweed farming were spatial GIS analyses and incorporated into identified from literature and from GIS displays, multivariate analyses and discussions with the Solomon Islands multiple criteria analysis. Ministry of Fisheries and Marine Resources staff and consultants. These were refined Under different conditions, many ecological during discussions with managers, and social parameters may impact on site consultants and stakeholders, rated suitability and there may be no hard and fast

rules for predicting the success or impacts of and may encourage the growth of farming. The decision support tools competitive algae. Temperature varies developed here are designed to be flexible to throughout complex reef lagoon systems and allow a wide range of possibilities to be in response to currents, wind and weather considered by making information readily and is therefore difficult to map at scales available in an intuitive geographic format. relevant to the location of individual farms. In GIS proximity analyses, we estimated the 2.3.1 Physical Selection Criteria distance of habitats and farm sites to the reef crest and open ocean as an approximate Depth indicator of water movement and its effect For the off-bottom longline seaweed farming on temperature and other water properties. technique, water depth is critical. At the lowest tides, the seaweed should remain Freshwater covered with water, but it should be shallow Seaweeds (especially K. alvarezii) are enough for workers to wade among and adversely affected by low salinity from rivers access the lines easily. Broad-scale and streams, runoff from the coast itself and bathymetry data within reef areas were of even heavy rainfall during low tides. relatively low resolution. However, specific Turbidity associated with freshwater runoff habitat types (shallow terrace) classified by also affects the transparency of sea water and the Millennium Coral Reef Mapping Project this can impact on photosynthesis. We used (IMaRS-USF, 2005) from satellite data distance to land and distance to the nearest provided a useful representation for factors river mouth to indicate the potential such as depths, substrata and shelter from influence of freshwater on farm suitability. ocean swell in areas most likely to support off-bottom seaweed farming. This report Ocean swell focuses on site selection for the off-bottom Large waves, especially from oceanic swells, method but the data and tools could also be exclude seaweed farming in its current form used to explore the potential for other from locations near reef fronts and passes. farming methods. The sheltered shallow terrace habitats Tidal range mapped in the Millennium Coral Reef Mapping Project are generally protected Tides naturally vary within and between from most ocean swells. Impacts from rare months of the year but the highest spring but catastrophic events like cyclones and tides at most stations in the Solomon Islands tsunamis are less predictable. are only 1–2 m (Sulu et al., 2000; https://www. tide-forecast.com/locations/Wagina-Island- Current Solomon-Islands/tides/latest) which is small Strong currents may damage seaweed farm relative to other regions and to the depths structures and the plants themselves, but low best suited for the off-bottom technique. to moderate flows appear to benefit seaweed growth. Backwater areas of enclosed lagoons Temperature with minimal water exchange are not Sea temperature can influence the growth recommended for farming. Glenn and Doty rate of seaweeds, cause stress in the cultivar (1990) observed that downstream seaweeds

Page | 15 were unhealthier than those upstream, devastated the Rarumana seaweed farms. tended to have less colour and were more Other concerns are increasingly high sea fragmented because of the disease, “ice-ice.” temperatures, prolonged precipitation and Mtolera (2003) showed that the growth rate the effect of strong currents and ocean swells. of K. alvarezii can be heavily impacted by In near-shore areas, rivers and runoff from the presence (or absence) of minerals such land can cause seaweed to become stressed. as Cu, Fe, Mn and Zn (Glenn and Doty, Rain also interrupts the drying process and 1992). Similar observations were made by may cause seaweed to disintegrate (McHugh, Hurtado et al. (2014) on intensive farming 2002, 2003, 2006). While these issues conditions near dense structures and cannot be controlled, they may be bamboo rafts. Distance from reef crest and anticipated, and risk built into the the open ocean provided a very coarse assessment of areas where seaweed farms indicator of water movement but expert might be developed. judgement, field surveys, and aerial photos and habitat maps can also provide an 2.3.2 Biological Selection indication of current strength and direction. Criteria Wind While abiotic parameters including temperature, light, pH and salinity are Msuya (2006) stated that strong winds can important factors that influence the distribu- break and wash away seaweed resulting in tion and growth of seaweeds (Borlongan et large losses for seaweed farms. This also al., 2016), biotic factors such as algal disease, means that farmers must spend time and grazing fish and competitive algae (epibionts) resources fixing the farms and replanting the affect the development and productivity of seaweed. seaweed farms as well as the quality of the products (Buschmann et al., 2017). Substrata Sandy areas without boulders are preferred Competition with other algae for the stakes used in the off-bottom Epiphytes compete with host organisms for method. The shallow terrace habitat dissolved inorganic carbon and nutrients mapped by the Millennium Coral Reef (Borlongan et al., 2016). Infected K. Mapping Project (IMaRS-USF, 2005) alvarezii can have lower photosynthetic rates appears to include mostly sandy substrata compared to healthy seaweed, particularly but may also include rocks above and below when epiphytes cover a large surface area the surface. This habitat tends to be located (Hurtado et al., 2014). Macro-epiphytic inshore of the reef flat, on the edge of algae incur more labour for seaweed farmers lagoons where the deposition of sand is as they must remove epiphytes to maintain more likely. the health of the cultivars. Filamentous epiphytes have caused issues for farms in Natural disasters Rarumana and the Shortland Islands and A major risk for seaweed farming is natural resulted in severe losses. This issue can be catastrophes such as earthquakes, tsunamis managed by placing the seaweed farms in and cyclones. An example from the locations where there is better water flow Solomon Islands is the 2007 tsunami that (Kronen, 2013).

Disease 2.3.3 Economic Selection The seaweed industry has suffered from Criteria declines in production from a disease known Economic factors such as the costs of inter- as ice-ice (Loureiro et al., 2009, 2015) island shipping and prices received overseas caused by ecological stress (unfavourable for seaweed have a major impact on the water temperature, pH, salinity changes, commercial operation and growth of high irradiances or UV radiation) (Kronen, seaweed farming in the Solomon Islands. 2013; Borlongan et al., 2016). Four forms of Some external influences, such as seaweed disease include: international seaweed prices, are not particularly affected by site selection, but 1. Ice-ice, generally characterised by the they can influence the relative importance of white flakes on the thallus of K. alvarezii. domestic costs. We therefore made 2. Pitting, identified by the cavities forming approximate estimates of site characteristics on the branches. likely to influence the cost of production at the site level. 3. Tip darkening, where the end segments of the branches lose colour. Fuel and transport costs 4. Tip discolouration, where the tips become soft and pinkish and dissolve. Fuel for outboard and other motors is expensive in remote areas but this may be Diseased K. alvarezii branches demonstrate offset by the proximity of farms to the shore a decrease in photosynthetic performance, and the use of unpowered canoes. Potential which may reflect the decrease in pigment fuel costs were approximated by GIS concentrations (Ganzon-Fortes et al., 1993). estimates of the distance from farms to the Again, the incidence of this problem may be main islands (coast), to small islands, villages, reduced where there is better water flow. wharves, airports, the capital city and to significant settlements. After discussions with Herbivory by fish, turtles and managers we also estimated the distance to dugong the nearest shipping route as an appropriate indicator of site suitability. Herbivorous fish, such as rabbit fish (family Siganidae), graze heavily in some seaweed Equipment costs farming areas to the extent that is it hard to Equipment costs include costs for lines, justify starting a farm (Kronen, 2013). The drying facilities, vessels and motors. These growth of seaweed can be stunted simply by costs are relatively low when compared to having the plant tips eaten (Msuya, 2006). other forms of aquaculture but are Seasonal fish grazing was an issue during significant for initial farm establishment and trials in the Solomon Islands and could be recovery in low-income communities. avoided by moving farms to areas with less grazing (Kronen, 2013). Grazing from turtles Production size and dugongs also causes a loss to production in some areas of the Solomon Islands (pers. The potential for a site to reliably supply com. Wesley Garofe, 2017; McHugh, 2006). large quantities of seaweed has economies of

Page | 17 scale for transport, buyers and, potentially, sources of income that provide more money drying and even processing facilities. than farming seaweed, such as fishing. Other examples of competing livelihoods are Maintenance costs producing copra, harvesting sea cucumbers and logging. While logging and fishing can Maintenance of seaweed farms (McHugh, create environmental problems, 2006) requires removing any competitive environmental issues may also arise from algae, repairing any damage to the lines or aquaculture (Ottinger et al., 2016). Seaweed stakes and replacing lost seaweed due to farming and tourism industries may conflict grazers, diseases or water action (McHugh, as both activities use shallow coastal areas. 2006). Sites may vary in their suitability in Conflict between users may also be this regard, with respect to transport of increased by the visual impact of either workers, exposure to high wind and waves practice (Falconer et al., 2013). and susceptibility to disease and nuisance algae. Customary tenure 2.3.4 Social and Cultural The customary tenure system in the Selection Criteria Solomon Islands governs land ownership and also includes the adjoining sea. Each Namudu and Pickering (2006) found that landowning group has a committee that social criteria were often more important makes decisions around environmental use than physical factors for successful farming and there is a strong sense of attachment to of K. alvarezii in the Pacific Islands. Their traditional rights passed down through survey in Fiji analysed community generations of family leaders. Integrating employment opportunities and the conservation efforts with the values of perspectives of seaweed farmers and found indigenous communities has sometimes that distances between farms, working been a challenge because of customary rights populations and ports can have important (Walter and Hamilton, 2014). However, a economic and social consequences for study by Hviding (1998), suggests that the communities. customary tenure system is compatible with seaweed farming in the Solomon Islands, Distance to village although the rules of management, including The distance of farms from villages can the traditional leaders’ ability to govern affect the availability of workers, housing and resource use, have been found to vary other facilities such as shops and schools. In between regions (Aswani and Hamilton, some areas, however, workers have moved 2004). to temporary accommodation on small Permitting access to resources can be islands adjacent to seaweed farms. difficult due to overlapping and conflicting Competition with other management approaches that arise due to the nature of descent and entitlement rights. activities for labour and space Seaweed farming in the Solomon Islands is Declines in seaweed prices and delayed done by families and communities from payments lead farmers to lose interest nearby villages (or distant locations where (Pickering, 2006) and resort to alternative people reside near the farms temporarily)

and they may also hire additional workers if particularly if they are economically valuable necessary (McHugh, 2006). Neighbours may (Aswani and Hamilton, 2004). also deny one another access to resources

2.4 Geographic Information System and Spatial Data

The FAO (2014, 2016) has recommended complex analyses, simulations and decision that GIS based studies be used to facilitate support tools and can be used in interactive, aquaculture development and assist with participatory planning with stakeholders, decision making at a national level (Dean communities, managers and scientists. and Salim, 2013). An advantage of using a GIS-based approach is its positive impact on Spatially referenced data likely to be decision making by providing useful, and representative of these factors were collated, often freely available, quantitative and formatted and mapped in the ArcGIS qualitative information (Radiarta et al. 2008, (Environmental Systems Research Institute 2010). There are many regions with (ESRI)) and Quantum GIS (QGIS) relatively low farming intensities, such the geographic information systems within a Solomon Islands, that would benefit greatly common coordinate system (Universal from planning based on good information Transverse Mercator 57S). These include and consultation, while there is still the data on islands, marine habitats, shipping opportunity to do so (Kapetsky et al., 2013). routes, infrastructure, seaweed farms, villages, towns, schools, roads, marine According to Kapetsky et al. (2007, 2013), protected areas, marine species and marine spatial analysis applications to assist in vegetation. The open source QGIS software aquaculture planning need to incorporate is free for all users and will allow many users multiple models (environmental, social and to access and manipulate the spatial economic) and be applicable to broad spatial information. scales. It is important to define potential sites that may be appropriate for sustainable The GIS project interfaces included aquaculture and also have the ability to approximately 100 GIS layers with monitor their development and their effect formatted legends nested within broad topics on environments and communities (Ottinger for clarity and ease of access. An atlas of et al., 2016). predefined layouts at specific locations was built to facilitate data exploration and is GIS can benefit planning for aquaculture by included within the QGIS project. The integrating information for many digital maps compiled in this research are environmental, economic, social and cultural intended to assist scientists, stakeholders, variables from sources ranging from remote managers and communities to assess sensing, to fisheries surveys and anecdotal geographic influences on the suitability of interviews. GIS can display complex data in different areas for seaweed farms but will readily interpretable, highly visual maps. It also be of use in other environmental builds on and integrates knowledge, rather planning and assessment. than reinventing databases. It enables more

Page | 19 2.5 Geospatial Analyses

Several national-scale GIS data sets for coral in the data set. Boundaries of potential reef habitats, seagrass, mangrove, marine seaweed farm locations were intersected with protected areas, seaweed farming areas and the IMaRS-USF marine habitat classification provinces were combined using the ArcGIS map to determine which habitats union function into one combined layer of corresponded most frequently with sites approximately 30,000 polygons. The identified by managers as suitable for ArcGIS “Near” command was then used to seaweed farming. A simplified GIS point estimate the distance between each polygon data version of this dataset was generated for and the nearest river, island, mainland, road, existing, previous or potential seaweed farm wharf, shipping route, protected area, sites and attributed with distances to the mangrove, seagrass meadow, village, town above features, areas of habitats, multivariate and capital city and include this information statistics and multiple criteria scores.

2.6 Multivariate Analyses

Variables describing the suitability of each PCA scores and cluster groups were mapped site according to each manager and and the results included within the decision according to distances to geographic features support system. The free paleontological were analysed by principal components statistical software PAST version 3 analyses (PCA of correlation matrices) and (https://folk.uio.no/ohammer/past/) was cluster analyses (unweighted pair-group used for multivariate and correlation method) based on their Euclidean distances. analyses.

2.7 Multiple Criteria Analyses

Multiple criteria analysis has been widely The simplest is the Simple Multi Attribute applied in business management and Rating Technique (SMART) which evaluates environmental impact assessment (Edwards, alternatives according to a hierarchical tree 1977; www.infoharvest.com; of criteria nested within more general www.expertchoice.com), with applications objectives under a single, broad goal. also in fisheries (Mardle and Pascoe, 1999) Variables describing the suitability of each and selecting protected areas (Bakus, 1982; site according to each manager or through Fernandes, 1996; Rothley, 1997; Guikema geospatial analysis were included within a and Milke, 1999; Villa et al., 2002; Breen et single hierarchical SMART model using the al., 2002, 2004). The techniques incorporate software Criterion Decision Plus weighting of multiple criteria, calculation of (Infoharvest Inc). Data were first imported trade-offs, representation of uncertainty, into the exploratory “Brainstorm” window of sensitivity analyses of the relative influence the software and variables grouped by data of different criteria, and the ability to source and by theme (physical, biological, combine and assess alternative models, data economic, social) before generating the and sources of opinion. quantitative model. Variables rated by

seaweed industry managers were weighted in according to alternative priorities. Overall the model according to their perceived scores of site suitability were graphed and importance or influence. The models allow mapped for a few examples; other scenarios for criteria to be weighted interactively can be explored using the software itself.

2.8 Satellite Image Interpretation, Drone Photogrammetry and Ground Surveys

Remote sensing is the process of obtaining 2013 and 2015. A plug-in was installed in the information about the earth and its surface QGIS project interface to allow imagery and by scanning from aircraft, satellites, ships or other data to be sourced online. other remote platforms. Sensors mounted on the satellites or other craft measure A Phantom 4 DJI Professional drone with a electromagnetic energy reflected or emitted 21-megapixel camera was flown over six loca- by the earth’s surface in the form of waves tions at an altitude of 50 m capturing RGB (Dean and Populus, 2013; Dean and Salim, colour aerial photographs every 2 seconds 2013). The readings received from these along a gridded flight path. The flights were emissions can be classified or processed into pre-programmed using the flight planning different features due to the objects reflecting software application MapPilot (Drones Made different wavelengths back to the sensors. Easy) running on an Apple iPad Air2. This The benefits of this information greatly produced a series of high-resolution aerial expand the spatial and temporal scope of photographs with an image overlap of greater many oceanographic and coastal studies that than 80 %. The software Pix4D v3.1 was cover areas and time scales unattainable by used to georeference, align and convert the standard methods of ground-based sampling series of aerial images into a large high-resolu- (O’Regan, 1996). Observations include tion mosaic for each area by matching several oceanography data such as surface tempera- thousand joint features in each overlapping ture, ocean winds, waves, currents, turbidity photo. Ortho-mosaics were imported into and primary productivity levels (chlorophyll- ArcGIS 10.5.1 for analysis and comparison a concentration) and bathymetry (Dean and with WorldView2 satellite imagery provided Populus, 2013). With a long-time series of through ESRI base maps in ArcGIS. information and data, users can assess seasonal patterns and trends at different periods At each of the six sites, a 30-minute visual (daily, weekly, monthly and annually) (Dean survey by snorkel of the immediate area was and Salim, 2013; Kapetsky et al., 2013). conducted of the seabed to verify the types of substrata and benthos represented in GIS Satellite images from a range of sources were habitat maps and aerial photos. At three sites, examined to determine whether seaweed a waterproof GoPro Hero 3+ camera was farms could be readily viewed and mapped. used to record video transects of the seabed Georeferenced tif files captured from the and associated benthic organisms. Estimates WorldView-2 satellite sensor were imported of percentage cover were made by randomly into ArcGIS and the boundaries of farms pausing the video footage and recording the visible in the Wagina Island region were substrata and/or organisms under 10 points mapped to compare locations farmed in marked on the viewing monitor.

Page | 21 3. Results

3.1 Geographic Information System

Spatial data and analyses were compiled in a and geoprocessing capabilities in addition to central folder accessible through a the standard viewing functions. customised QGIS project interface. The A major part of the decision support QGIS project includes approximately 100 capability is just having a range of relevant, layers formatted with legends and nested in spatially referenced and formatted data folder categories including physical, readily accessible in one location. A major biological, social and economic vector data, advantage of GIS as a decision support tool analyses, and satellite, drone and other is that complex data for many locations can imagery (Figure 2). It includes data layers be simultaneously displayed in a highly mapping existing, previous and potential intuitive visual environment. It is a decision sites (Figure 3), expert ratings of site support, rather than decision-making tool, as suitability, the results of proximity, principal it allows users to freely explore data and components and multiple criteria analyses different planning scenarios in many ways. and many environmental, economic and At regional scales, complex environments social features. We present only a few like the Solomon Island archipelago can be examples as there are too many layers to be challenging to navigate. The QGIS interface displayed here. Figure 2 shows a snapshot of therefore includes “bookmarks” referenced the QGIS project interface with a few groups to specific locations of interest (such as poten- and layers open. The group folders for the tial seaweed farm locations) and preformatted data layers are designed to collapse and layouts for different locations to print maps expand, be switched on and off, panned, for whichever layers users wish to present. zoomed, queried and analysed using a The project interface includes a preformatted diverse suite of geospatial functions. The “atlas” function loaded with sequential print project and GIS data are held by the layouts for 35 potential farm sites. The basic Solomon Islands Ministry of Fisheries and data structures also provide a generic platform Marine Resources and the QGIS software is for many other analyses and software tools. a freely available GIS with many analytical

3.2 Marine Habitats

The IMaRS-USF (2005) Millennium Coral terrace” or in a few cases, “shallow terrace Reefs Mapping project uses a combination with constructions” (Figures 4 and 5). The of spectral analysis and visual interpretation other areas of reef front, reef flat and lagoon of satellite images to classify coral reefs and in Figure 5 are artefacts from drawing surrounding shallow waters into GIS approximate boundaries around complex polygons representing different habitats. The reef topography as the drying stony reef flat, main habitat coinciding with areas identified exposed reef front and deeper lagoon were as suitable for seaweed farming was “shallow considered unsuitable for the off-bottom

farming method. Deeper habitats may terrace habitats throughout the Solomon however, be available for other methods Islands and Figure 6 shows the total area in such as floating longlines. square kilometres of shallow terrace in each province. While some areas have more of When the reef habitat polygons are dis- this habitat than others, there are large areas played over satellite images, the shallow of shallow terrace around most islands terrace habitat coincides with the many throughout the archipelago. Most of the area shallow, sandy, internal outer margins of reef is, however, in the Western Province, with lagoons sheltered by the outer stony reef flats the least area in Guadalcanal where the and reef fronts. The habitat includes shallow capital Honiara is located. areas that are still deep enough to retain some coverage of water on the lowest spring tides. The distributions of shallow terrace and the It is intermediate in depth between the other IMaRS-USF marine habitats at shallow back reef flat and deeper lagoon. potential farm sites are mapped in the Appendix of Vavia (2018) and can be viewed Inspection of 35 potential farm sites overlaid at any scale in the SINOPSIS QGIS project. on the IMaRS-USF polygons and satellite The QGIS project includes other relevant images confirms the prevalence of shallow layers such as surveys of coral reefs (UNEP- terrace and, in a few cases, shallow terrace WCMC, 2010), seagrasses (McKenzie et al., with constructions as a substratum with 2016), mangroves (Green et al., 2006), a potential for seaweed farming. Other high-resolution marine benthic classification physical, biological, social and economic from satellite imagery of Roviana Lagoon criteria will also determine how viable (Roelfsema et al., 2013) and the locations of farming might be, but the area of shallow marine protected areas and other managed terrace provides a reasonable indicator of areas (UNEP-WCMC 2016). These and potential locations. other ecological data are important if the seaweed industry is to be not just successful, Figure 4 shows the distribution of shallow but also ecologically sustainable.

3.3 Expert Knowledge

The approximate outlines of areas identified correlation between the two managers’ by managers as existing, past or potential ratings of site suitability (Figure 9). new seaweed farm sites are shown in Figure 3. Figures 7 and 8 map ratings between 1 Figures 10 and 11 are multivariate principal and 10 (least to most suitable) by each components ordinations of each manager’s manager for overall suitability as a seaweed scores for seaweed farm suitability. The farm site. Although several sites, for example ordinations aim to graphically summarise around Wagina Island and northern Kia, differences among potential farm sites across stand out for their high scores, there are all expert environmental, social and many other locations also highly rated. Not economic ratings on just two axes. In these all sites were familiar to both managers, but examples, the first and second principal where a site could be rated by both component axes explain about 50 % of the managers, there was a strong positive variance in ratings. Sites that lie close together

Page | 23

Figure 2 SINOPSIS 3.5 (Solomon Islands National Ocean Planning Spatial Information System). A Quantum GIS 3.2.3 project containing ~100 physical, biological, economic, social and analytical formatted data layers to inform planning for sustainable seaweed aquaculture.

Figure 3 Locations identified as previous, current or potential sites for seaweed farming. These include some sites with marginal conditions which in some way make them less suitable.

Page | 25

Figure 4 Distribution of “shallow terrace” reef habitats in the Solomon Islands. Data from the Millennium Coral Reef Mapping Project, Remote Sensing Centre, University of Florida. (IMaRS- USF, 2005).

Figure 5 Percentage area of the most common reef habitats at identified seaweed farm locations in the Solomon Islands. Data from the Millennium Coral Reef Mapping Project, Remote Sensing Centre, University of Florida. (IMaRS-USF, 2005).

Page | 27

Figure 6 Area of “shallow terrace” marine habitat in provinces of the Solomon Islands. Data from the Millennium Coral Reef Mapping Project, Remote Sensing Centre, University of Florida (IMaRS-USF, 2005). on the ordination are more similar across Harbour, Shortland Islands and the Reef most ratings, while sites further apart are the Islands score successively higher scores for most different. having less competition in the labour market from competing industries like logging and The biplots also indicate which variables are fisheries but lower for catastrophic disturbance most responsible for the differences between and distance to Honiara. Babanga, Epaanga, sites according to the length and direction of Sageraghi, Futuna, Boeboe and Langalanga the labelled lines radiating from the centre of score highly on the catastrophic disturbance the graph. The length of the line indicates and distance to Honiara variables. Boeboe, the strength of the correlation with the Langalanga, Chubikopi, Marovo, Epaanga principal component axes and how much and Babanga also score highly for distance variation is explained by that variable. to village and cost of fuel. In Figure 10, in the top right-hand quadrant At the top of Figure 10, Kia, Marulaon and for Manager 1, sites at Wagina, Hatodea, the four Wagina sites score highest for Ferafaalu and Munda score highly for depth, tidal range, production and overall overall suitability, tidal range, effect of suitability. In the ordination for Manager 2 freshwater, current, wind and swell, while (Figure 11), variables rated for overall only a few sites in the lower left-hand suitability, substrata, depth and fuel costs quadrant, Mangakiki, Marau, Toa and Seghe score low on these variables. explained most differences between sites with Wagina, the Reef Islands, Kia, Hatodea, On the left, Telina, Sandfly, Supasizae, Star Marulaon and Sageraghi scoring highest for

these variables. Babanga, Chubikopi, Cluster 3 includes more remote small island Epaanga, Rarumana 1, Star Harbour and locations in Kia, the Shortland Islands, Futuna rated lowest on these variables. Northern Choiseul, Star Harbour, the Reef Islands and small islands closer to Honiara. Tauba, Star Harbour, Futuna, Marulaon, Cluster 4 is a trial floating line site at Boeboe. Sageraghi and Rarumana 2 scored highest on distance to village and distance to drying The cluster analysis of similarity among sites facilities while the Wagina sites scored for Manager 2 in Figures 13 and 15 shows lowest on these variables but highest for lack five clusters. Cluster 1 is around Wagina of competition from competing industries. Island. Cluster 2 includes Kia, Sandfly, Reef Note that the differences between managers Islands and Sageraghi. Cluster 3 includes in the absolute positions of variables and most of remaining locations except for sites in the ordinations are an artefact and it Rarumana 1 and Chubikopi in Cluster 4 and is the relative relationships among sites and Babanga in Cluster 5. associated variables that are relevant. With some exceptions, ordinations for both A cluster analysis of multivariate similarity managers showed similar groupings of sites among sites in Figure 12 and Figure 14 for distinguished by perceived depth, substrata, Manager 1 groups similar sites into four fuel costs and competition for workers. clusters. Cluster 1 includes sites at Wagina, Some grouping of these sites in geographic and the Western Province, Cluster 2 space was also evident (Figures 12 and 13). includes many sites closer to Honiara and

Figure 7. Overall suitability of identified sites for seaweed farming rated by Manager 1.

Page | 29

Figure 8. Overall suitability of identified sites for seaweed farming rated by Manager 2.

Figure 9. Comparison of overall rating of site suitability for seaweed farming for Manager 1 and Manager 2 (r=0.8, P<0.001).

Figure 10. Principal components analysis (PCA) for ratings of seaweed farm suitability by Manager 1. Principal component axes 1 and 2 (29 % and 23 % of variance explained).

Figure 11. Principal components analysis (PCA) for ratings of seaweed farm suitability by Manager 2. Principal component axes 1 and 2 (32 % and 22 % of variance explained).

Page | 31

Figure 12. Cluster analysis of site suitability ratings for Manager 1.

Figure 13. Cluster analysis of site suitability ratings for Manager 2.

Figure 14. Cluster analysis of Euclidean distances between seaweed farm sites for Manager 1

Page | 33

Figure 15. Cluster analysis of Euclidean distances between seaweed farm sites for Manager 2

3.4 Geospatial Analyses

Two approaches were used to estimate and in distances from the country’s capital, assign measures of proximity for different Honiara. locations to features such as the nearest mainland, islands, rivers, villages, towns, A PCA of potential farm sites described by capital, seagrass beds, mangroves and geographic proximity to 20 features (e.g. marine protected areas. In the first, distances mainland, island, river, road, wharf, shipping were assigned to each IMaRS-USF reef route, town, seagrass bed, protected area) in habitat polygon. This allows subsets of Figure 17 is dominated by the area of habitat polygons to be queried according to ‘shallow terrace with constructions’ at the combinations of decision rules as used in Reef Islands and the distances to features Vavia (2018). For example, select all the reef found only on larger more central islands. In habitat poly-gons over 5 ha in size that are the top right-hand quadrant, Kia, Wagina 4 shallow terrace, within 1 km of an island, and the Shortland Islands stand out for their greater than 2 km from rivers and within relatively large distances to villages, shipping 200 km of Honiara. routes, and wharves, but relatively large area of shallow terrace habitat. For display at a national scale and displaying measures for specific potential farm sites, the Figure 18 shows approximately three to four mean values of all polygons were assigned to clusters of similar sites based on the a central point for each farm and the proximity data with the Reef Islands (Nifioli) distances of the points themselves from standing out from all other sites; Wagina, environmental and socio-economic features Kia and most Western Province sites in a also calculated. In most cases, the latter large group; the Shortlands, Taro and distances were sufficient for most regional Marovo splitting off; and Marau, Sandfly, and national overviews, as shown in Figure Ngela, Langalanga, Hatodea and other sites 16 which shows a mostly longitudinal trend forming another large group.

3.5 Multiple Criteria Analysis

The Criterion Decision Plus software allows 19 to 30. Although displayed in separate users to interactively query data for variables figures, the four data source categories and alternatives (potential farm sites in this (Manager 1, Manager 2, GIS point data, GIS case), explore how different priorities and polygon data) are part of the same model. At weightings effect outcomes and use each level of the hierarchy, sub-criteria can sensitivity analysis and trade-offs to explore be weighted, examined separately or options. Each model holds information excluded from the analysis. from many different sources, formats and scales in a central standardised environment Weighted multiple criteria scores for expert that can be easily manipulated. ratings and criteria weights from Manager 1 and Manager 2 are mapped in Figures 19 to The input variables from the expert ratings 24. For Manager 1, there are similar and geospatial analysis and the resulting groupings of higher suitability sites at multiple criteria scores are shown in Figures Wagina, Kia and Western Province and

Page | 35

Figure 16. Example of geospatial proximity analysis – distance of sites (km) to Honiara.

Figure 17. Principal components analysis (PCA) for GIS point estimates of seaweed farm suitability. Principal component axes 1 and 2 (33 % and 16 % of variance explained).

Figure 18. Cluster analysis of seaweed farm-site similarity based on geospatial variables.

Page | 37 lower scores for the smaller, more central islands and outlying locations in the Shortland and Reef Islands (Figures 19 and 22). For Manager 2 (Figure 20 and 24) there are also medium to high scores for Wagina, Kia and the Western Province but higher scores for Marau, Maeloboa, Marulaon and the Reef Islands (Nifioli).

While there are some scatter and several influential outliers, there is a weak positive correlation (Figures 23 and 24) between the overall suitability ratings from Manager 1 (r=0.28, P>0.1) and Manager 2 (r=0.4, P=0.07) and their weighted multiple criteria scores. This is not surprising given that the data sources are similar, but it does indicate some consistency in subjective scoring and weighting.

The geospatial point (Figure 28) and polygon (Figure 30) unweighted scores show quite different overall site scores for suitability and do not correlate well with the managers’ ratings (Figures 31 and 32). However, these criteria have not been weighted according to importance of their effect on suitability. The available geospatial data also differed markedly to the expert judgement scores both in the kinds of variables addressed and in the types of suitability measures used to score sites.

Figure 19. Combined weighted multiple criteria score of site suitability for criteria from Manager 1.

Figure 20. Weighted multiple criteria score of site suitability from Manager 2.

Page | 39

Figure 21. Section of a Criterion Decision Plus multiple criteria model for just Manager 1 ratings nested within biological, physical, economic and social sub-criteria.

Figure 22. Combined multiple criteria scores of the overall suitability of potential seaweed farm sites for just Manager 1 (other data sources excluded). All criteria are weighted by the level of importance assigned to each criterion by the manager.

Figure 23. Section of a Criterion Decision Plus multiple criteria model for just Manager 2 ratings nested within biological, physical, economic and social sub-criteria.

Figure 24 Combined multiple criteria scores of the overall suitability of potential seaweed farm sites for just Manager 2 (other data sources excluded). All criteria are weighted by the level of importance assigned to each criterion by the manager.

Page | 41

Figure 25 Comparison of overall rating of site suitability for seaweed farming for Manager 1 with weighted multiple criteria scores for Manager 1 (r=0.28, P>0.1).

Figure 26 Comparison of overall ratings of site suitability for seaweed farming from Manager 2 with their weighted multiple criteria scores (r=0.4, P=0.07).

Figure 27 Section of a Criterion Decision Plus multiple criteria model for GIS point distances and areas nested within biological, physical, economic and social sub-criteria.

Figure 28 Combined multiple criteria site scores of suitability for seaweed growing derived from GIS point distances and areas (other data sources excluded). A spatial representation of these scores can be viewed in the QGIS project application.

Page | 43

Figure 29 Section of a Criterion Decision Plus multiple criteria model for mean GIS polygon distances and areas nested within biological, physical, economic and social sub-criteria. This part of the model is not weighted but this can be done interactively within the software.

Figure 30 Combined multiple criteria scores of suitability for seaweed growing derived from mean GIS polygon distances and areas.

Figure 31 Comparison of the multiple criteria scores for Manager 1 with unweighted multiple criteria scores for the GIS point data (r=-0.2, P>0.1).

Figure 32. Comparison of the multiple criteria scores from Manager 2 with the unweighted multiple criteria scores for GIS point data (r=0.1, P>0.1).

Page | 45 3.6 Interpretation of satellite, drone imagery and site inspections

Worldview-2 satellite imagery from the scale mapping and quantifying areas within ESRI base maps in ArcGIS and other web sites perhaps identified from coarser scale services provided variable but high resolu- data. The resolution makes it possible to tion images of seaweed farms. These were distinguish between slight but important sufficiently detailed to hand-digitise and geo- differences in depth and substrata such as reference boundaries of farms near Wagina small boulders and stones (Figure 35), Island on two dates (Figures 33 and 34). It seagrass and, in some areas, sediments was, however, difficult to determine the age oriented along dominant currents. Drone of the farms and which, for example, were photogrammetry has other applications currently farmed. Images were also often including 3D modelling, mapping of affected by cloud and their availability was important ecological areas like seagrass, restricted by time and by cost. However, coral, and mangroves, change detection of they could feasibly be used to map changes human development including seaweed in the extent of farms. farms. Training and equipment to conduct ongoing drone surveys was therefore Aerial photographs captured by drone provided to the Ministry of Fisheries and provided high resolution, georeferenced, Marine Resources staff. photo-mosaics at five locations in the Western Province (Figures 35 to 39). The The seabed at seven sites (Vona Vona, drone was able to cover a few hectares and Boeboe1, Boeboe2, Babanga, Epaanga, complete a detailed photo survey in a short Logha and Sageraghi) were surveyed on 15 minute flight and the scope of the survey snorkel to help interpret relationships could easily be increased with higher between satellite data, GIS habitat maps, altitudes and flying speeds. aerial photo interpretation and the benthic substrata and epibenthos. These sites were Substrata, corals, seagrass, algae, depth also surveyed by drone with the exception zones and existing and previous seaweed of Boeboe 1 and Epaanga. farms could be more easily interpreted from the drone images when compared to Babanga and Epaanga lie sheltered behind the satellite imagery, but for a smaller area. islands on the leeward side of reefs. As at Both sources of imagery were useful in Logha, there were scattered corals and interpreting the IMaRS-USF marine habitat stony outcrops on patches of sand at these classification and the boundaries of poten- sites (Figures 38 and 39). At Rarumana tial sites. The resolution of the mosaics is (Figure 35) and Sageraghi (Figure 36) there sufficiently high to detect relatively short- was less coral and more large sandy areas term changes in coastal habitat as evident in and at Sageraghi there were patches of the study by Chan (2018) on land use at a seagrass, Halimeda and filamentous algae coastal village (Figure 45). It was evident (Figures 40 and 41). that drone imagery could provide useful site information for farm development, planning, The snorkel survey was able to confirm the permitting and monitoring. location of the remains of the previous farm at Sageraghi, visible on the drone photo- The imagery is especially suitable for finer mosaic (Figure 42). The trial floating

longline farms at Boeboe (Figure 37) were have been affected by the slower rates of further inshore on the edge of narrow chan- water exchange. With some small excep- nels and passes in an almost estuarine inner tions, the abundance (<5-10 % cover) and network of small islands. The seabed was diversity of corals and other invertebrates at mostly fine sand with patches of algae and all sites surveyed was low but patches of seagrass. The trial seaweed farms there may seagrass and algae occurred in several areas.

Figure 33. Boundaries of seaweed farms west of Wagina Island digitised from WorldView-2 imagery from November 2013 (green) and October 2015 (blue).

Figure 34. Boundaries of seaweed farms east of Wagina Island digitised from WorldView-2 imagery from October 2015 (blue).

Page | 47

Figure 35. Georeferenced photo-mosaic of reef flat (left) and shallow terrace (middle) on the edge of the Vona Vona Lagoon. There are two trial seaweed farms visible to the west and north of the lagoon.

Source: Esri, DigitalGlobe, GeoEye, Earthstar Geographics, CNES/Airbus DS, USDA, USGS Figure 36. Photo-mosaic from drone (UAV) photogrammetry overlain on Worldview-2 satellite imagery at Sageraghi showing location of GoPro video transect through a previous seaweed farm (Vavia 2018).

Source: Esri, DigitalGlobe, GeoEye, Earthstar Geographics, CNES/Airbus DS, USDA, USGS Figure 37. Aerial drone image (upper) contrasted with satellite image (lower) at Boeboe 2, site of a previous floating longline trial.

Page | 49

Source: Esri, DigitalGlobe, GeoEye, Earthstar Geographics, CNES/Airbus DS, USDA, USGS Figure 38. Photo-mosaic from UAV drone overlaid on WorldView-2 satellite image at a previous farm site near Babanga.

Figure 39. Section of a drone photo-mosaic of previous single floating line near Logha showing boulder corals and other outcrops on sand.

Figure 40. Seagrass and Halimeda algae on seabed at Sageraghi.

Figure 41. Algae with trapped gas bubbles on seabed at Sageraghi.

Page | 51

Figure 42. Posts from previous seaweed farm at Sageraghi.

Figure 43. Change detection of coastal habitats at Kolombangara Island from UAV drone photo-mosaics flown in 2017 and 2018 (Chan 2018). Other data show the adjacent distribution of reef habitats and flood plumes from the river crossing these areas.

4. Discussion

Decisions in environmental management large areas of shallow terrace habitats and the use of natural resources are complex mapped in the Millennium Coral Reef and context dependent. Detailed spatial Mapping Project (IMaRS-USF, 2005). information is rarely collected specifically to “Shallow terrace” was the dominant habitat address these issues and best mapped within identified seaweed farm sites. approximations are often derived from data It is recognisable on satellite and drone intended for other purposes. What matters imagery as those shallow sandy areas on the though, is whether data are fit for purpose edge of sheltered lagoons. Though shallow, and this will vary with the situation and how these areas are likely to retain some depth of users evaluate the relevance and credibility water during low tides and may therefore be of the source. Data analyses alone are poor suitable for the off-bottom method of substitutes for the subtleties of human seaweed farming. There are hundreds of judgement. The information tools presented square kilometres of shallow terrace habitat here are intended as an aid to support not spread throughout all provinces of the replace human decision making. There is a Solomon Islands, with the greatest area in deliberate focus on data exploration rather the Western Province. than confirmatory analyses with a primary objective to help inform and engage All areas may not however, be suitable for stakeholders in participatory management. seaweed farms. Habitats mapped from We illustrate some preliminary analyses and satellite images, although of a relatively high make some broad observations but the resolution, may not always distinguish sand intention is for managers and stakeholders to from other substrata and habitats are often a use the tools and their own judgement as an complex mosaic of different substrata and integral part of the process. The intention is marine life. Other factors such as water also to provide a foundation to add new movement and temperature are also critical information from other GIS and non-spatial for seaweed growth and ultimately, data, remote sensing, literature, analysis and economic and social factors are probably the anecdotal knowledge from scientists, primary determinants of whether a farm managers, stakeholders and the community. succeeds.

The project was fortunate to have managers We made simple estimates of the relative and consultants with regional knowledge of importance of these variables and their seaweed farming in the Solomon Islands on values at over 30 potential or existing the research team. Some of this knowledge, seaweed farm sites using expert judgement in a rudimentary form, was successfully from managers and measures describing the included in databases. There was also some geographic relationships between farms and consensus on broad patterns in surrounding environments, infrastructure environmental, social and economic and people. Multivariate ordinations and influences on farm suitability, and many cluster analyses indicated that while physical potentially viable locations for seaweed factors like depth, substratum and water farming were identified. The extent of movement were critical, there were potential farm sites was also evident from the important differences between sites for

53 combinations of other factors like fuel, The multiple criteria analysis tool used in distance to village and alternative sources of this project integrated most of the data from employment. However, their influence on the GIS databases and expert scores within farm suitability is likely to vary markedly one hierarchical model. It allowed with location, time and other factors difficult alternatives (in this case farm sites) to be to control or predict, such as international quantitatively assessed against a range of seaweed prices. specific criteria according to user-defined weights. One advantage of the approach is Malczewski (2006, 2010) reviewed the that it standardises data to similar scales so increasing interest in the use of GIS and that data from different sources, formats and multiple criteria analysis from as early as measurement units can be directly 1990. Cabral et al. (2016) assessed the use of compared. The results presented here are a GIS integrated with a multiple criteria simplified examples, but the main strengths analysis to quantify compensation costs for of the tool are realised through an interactive seaweed farm installations in France. It approach with users able to incorporate their included socio-economic and biophysical personal knowledge, opinions and intuition. indicators in decision-making processes and The software also provides for other models considered the concerns of local to be built for other kinds of planning and communities with effects on other industries, evaluation wherever there are alternatives recreational use and marine ecosystems. and criteria involved. Sousa et al. (2012) identified sites for seaweed farming in Brazil from multiple The underlying databases also provide a criteria. The criteria used were mainly foundation for other analyses and decision distances to markets and other features, support tools. We provide a directory of unlike the study by Gimpel et al. (2015) additional free decision support tools where criteria were based on physical and including EMDS 6.0 (Environmental chemical measurements. Management Decision Support. https://emds.mountain-viewgroup.com/), Buitrago (2005) used a GIS to select sites for Marxan (Game et al. 2011) and Marzone oyster aquaculture in Venezuela. It (http://marxan.org/), Zonation employed 35 experts to evaluate criteria in a (https://www.helsinki.fi/en/researchgroups/di multiple criteria analysis. Variability among gital-geography-lab/software-developed-in- the opinions of the experts was surprisingly cbig), NatureServe Vista, large for factors such as bathymetry, use (http://www.natureserve.org /conservation- conflicts and tidal range. Nath et al. (2000) tools/natureserve-vista) and PAST statistical described subjectivity among experts when software (https://folk.uio.no addressing weightings and warned that /ohammer/past/). There is considerable decision support tools cannot alone be used potential to apply these kinds of decision to completely facilitate decisions, but should support to other applications in marine rather be used to stimulate discussion in resource management in the Solomons and conjunction with other analyses, maps, elsewhere (Breen et al. 2002, 2004; human intuition, ethics, knowledge, Fernandes et al 2005, 2009, 2010; Kerrigan consensus and experimentation. et al. 2010; Breen 2011; Game et al. 2011).

In addition to planning for the economic can be both expensive and time consuming. and social benefits of seaweed farming, there The data sets that are freely available may is a need to assess the potential have coarse resolutions but can, however, be environmental and social consequences of fit for purpose. Pasqualini et al. (2005) seaweed farms. For this reason, data on found their 10 m resolution map was 73 % seagrass, mangrove, marine protected areas accurate, and their fine-resolution 2.5 m and other managed areas were included in maps were 96 % accurate. The need for the QGIS project and in the proximity accuracy depends on the nature of the analyses. Additional information on research, and what ecological and social vulnerable ecosystems from other data sets, information is considered necessary. anecdotal knowledge and literature (e.g., Green et al., 2006; Lauera and Aswanib, More detailed aerial photography at specific 2008; Albert et al., 2010; Roelfsema et al., sites can be obtained from UAV drones and 2013; Roelfsema, Phinn, et al., 2013) could software that combines hundreds of also be mapped and incorporated in individual photos into georeferenced photo- decision support tools including those mosaics and topographic models. In designed specifically for conservation addition to being high resolution and low assessments, such as Marxan, Marzone cost, drone photography has the advantage (zoning for multiple use) and Zonation. of being able to strategically target specific times and locations. We used a small, off- Decision support tools can also incorporate the-shelf DJI Phantom4 quadcopter to information monitoring the growth and successfully survey five sites including a small administration of aquaculture activities. We trial seaweed farm and several previous farm digitised boundaries of seaweed farms sites. These photogrammetry tools have around Wagina Island in ArcGIS from many applications in the seaweed farming WorldView-2 satellite images. This industry and in other natural resource information contributes to a better management. understanding of site suitability and may be useful in the planning and administration of High spatial resolution imagery is useful but new farms and permits. Challenges here data for turbidity, sea surface temperature, include the cost of high-resolution satellite water quality and other variables are also imagery, atmospheric interference, and the desirable (Dean and Salim, 2013). However, periodic (but ongoing) availability of data. wave action, winds and currents mean that such measurements can be highly variable, One of the limitations addressed in a recent and although long term averages can be review by Ottinger et al. (2016) is the spatial used, the spatial scale is often coarse relative resolution of satellite sensors for to farm size. identification of single aquaculture sites, as the sites can be relatively small. Dean and Challenges that limit the growth and Salim (2013) noted that getting access to development of the seaweed industry in the such data means contacting suppliers of Solomon Islands stem from multiple satellite derived data and that one of the sources, many of which may not be readily issues with collecting and compiling data for described in a GIS. Some economic and spatial analyses is that although data may social criteria for site suitability were exist at national levels, the data processing assessed by expert judgement and very

55 approximate geospatial measures. However, post-harvest, having sand and plastic some of most important economic materials mixed with the seaweed, high determinants of seaweed farming success moisture content and low carrageenan yields operate at a national or international level. (Trono and Lluisma, 1992). The farm gate prices are determined by the buyers and One of the main issues for the Pacific dependent on the international markets Islands is the small contribution they make (Msuya, 2006). to global production, and the distances to markets and ports. Compared to Asian It is important for Pacific Island countries to seaweed cultivators, the Solomon Islands be producing seaweed of a higher quality and other Pacific producers struggle within and consistent output to stabilise their the carrageenan market. A shortage of position in the market, as the market for international buyers for Pacific seaweed has carrageenan should tend to expand if the limited the overseas market and growers and demand continues to grow (Cai et al., 2013). exporters are forced to accept lower price Pickering (2006, 2007) stated that seaweed negotiations (McHugh, 2006; Kronen, farming easily competes with copra, 2013). agriculture and fishing as a source of revenue, although not with revenues Valderrama et al. (2015) found that remote received from logging. Seaweed farms can places like the Pacific have difficulty because benefit from employing local communities of their lack of proximity to processing and families but fluctuations in the price of centres and their low market power due to seaweed can drive workers to more relatively low production and increasing fuel profitable industries. costs for shipping. McHugh (2006) noted the instability of farm gate prices that may While seaweed farming in the Solomon result from fewer farmers or seaweed Islands is less extensive than in Asia and shortages resulting from natural disasters. other regions, growth in the extent of farms has been strong (FAO, 2014) and there is International market prices are dependent much potential for farming in other parts of on the quantity and quality of product the Pacific. An ability to identify potential supplied, as well as the demand and farming areas and anticipate planning issues structure of the market. Producing low using tools such as those described here can quality seaweed results from poor handling help ensure that such growth is sustainable.

5. Conclusion

This study reveals substantial opportunities for the sustainable development of seaweed farming in the Solomon Islands but these need to be considered in conjunction with social and economic values and local knowledge and experience.

Discussions of farming techniques, the responses of seaweed cultivars to abiotic and biotic influences and economic influences on the industry (Trono, 1990; Trono, 1992; Hurtado et al., 2001; Pickering, 2006; Reis et al., 2017) indicate that there is potential to develop the seaweed industry in the Solomon Islands within its extensive shallow, sheltered, clean coral reef, island and lagoon systems. Many rural coastal communities scattered through the islands already rely on marine and other diminishing natural resources (Kronen, 2013) and there is an impetus to introduce opportunities for sustainable employment and income, especially when employment in other more extractive industries such as forestry, mining and fishing may be less sustainable.

GIS-based decision support tools can integrate data from many different sources, formats and fields of research and interactively display realistic representations of complex environments at many spatial scales in an intuitive visual format. They provide a useful focus for aquaculture policy and participatory planning, including workshops with managers, communities and other stakeholders. This project is another step in enabling a strategic and inclusive approach to planning for sustainable aquaculture in the Solomon Islands and the Pacific region.

57 6. References

Albert S., Grinham A., Bythell J., Olds A., Buschmann, A. H., Camus, C., Infante, J., Schwarz A., Abernethy K. and Gibbes B. 2010. Neori, A., Israel, Á., Hernández-González, Building social and ecological resilience to M.C., Critchley, A. T. (2017). Seaweed climate change in Roviana, Solomon Islands. production: Overview of the global state of Brisbane: The University of Queensland. exploitation, farming and emerging research activity. European Journal of Phycology, 52(4), Aswani, S. and Hamilton, R. J. 2004. Integrating 391–406. indigenous ecological knowledge and customary sea tenure with marine and social science for Buitrago, J., Rada, M., Hernández, H. and conservation of bumphead parrotfish Buitrago, E. (2005). A single-use site selection (Bolbometopon muricatum) in the Roviana technique, using GIS, for aquaculture planning: Lagoon, Solomon Islands. Environmental Choosing locations for mangrove oyster raft Conservation, 31(1), 69–83. culture in Margarita Island, Venezuela. Environmental Management, 35(5), 544–556. Bakus, G. J. 1982. The selection and management of coral reef preserves. Ocean Cabral, P., Levrel, H., Viard, F., Frangoudes, K., Management, 8, 305–316. Girard, S. and Scemama, P. 2016. Ecosystem services assessment and compensation costs for Borlongan, I. A., Luhan, M. R., Padilla, P. I, installing seaweed farms. Marine Policy, 71, 157– and Hurtado, A. Q. 2016. Photosynthetic 165. responses of ‘Neosiphonia sp. epiphyte-infected’ and healthy Kappaphycus alvarezii (Rhodophyta) Cai, J., Hishamunda, N. and Ridler, N. 2013. to irradiance, salinity and pH variations. Journal Social and economic dimensions of carrageenan of Applied Phycology, 28(5), 2891–2902. seaweed farming: a global synthesis. In D. Valderrama, J. Cai, N. Hishamunda and N. Breen, D. A. 2011. Coastal marine habitats and Ridler, eds. Social and economic dimensions of marine protected areas in the New Zealand carrageenan seaweed farming, pp. 5–59. Territorial Sea: a broad scale gap analysis. Fisheries and Aquaculture Technical Paper No. Wellington: Department of Conservation and 580. Rome: FAO. Ministry of Fisheries. Chan, C. 2018. Using remote sensing technology Breen, D. A., Avery, R. P. and Otway, N. M. to analyze land use change in Vavanga, Solomon 2002. Broad scale biodiversity assessments for Islands. SCI801 Specialist Readings. Auckland: marine protected areas in New South Wales. In School of Science, Auckland University of J. P. Beumer, A. Grant and D. C. Smith, eds. Technology. Proceedings of the World Congress on Aquatic Protected Areas. Cairns, Australia - August 2002. Dean, A. and Populus, J. 2013. Remote sensing Australia: ASFP. and GIS integration. In G. J. Meaden and J. Aguilar-Manjarrez eds. Advances in geographic Breen, D., Avery R. and Otway, N. 2004. information systems and remote sensing for Biodiversity assessment of the Manning Shelf fisheries and aquaculture - Summary version, pp. Bioregion. Report for the. NSW Marine Parks 35–41. Rome: FAO. Authority and Environment Australia.

Dean, A. and Salim, A. 2013. Remote sensing Ganzon-Fortes, E. T., Azanza-Corrales, R. and for the sustainable development of offshore Aliaza, T. 1993. Comparison of photosynthetic mariculture. In J. M. Kapetsky, J. Aguilar- responses of healthy and ‘diseased’ Manjarrez and J. Jenness, eds. A global assessment Kappaphycus alvarezii (Doty) using P vs I Curve. of offshore mariculture potential from a spatial Botanica Marina, 36(6), 503–506. perspective, pp. 123–181. Rome: FAO. Gimpel, A., Stelzenmüller, V., Grote, B., Buck, Falconer, L., Hunter, D., Telfer, T. C. and Ross, B. H., Floeter, J., Núñez-Riboni, I. and L. G. 2013. Visual, seascape and landscape Temming, A. 2015. A GIS modelling analysis to support coastal aquaculture site framework to evaluate marine spatial planning selection. Land Use Policy, 34, 1–10. scenarios: Co-location of offshore wind farms and aquaculture in the German EEZ. Marine FAO. 2014. The state of world fisheries and Policy, 55, 102–115. aquaculture 2014: Opportunities and challenges. Rome: Author. Glenn, E. P. and Doty, M. S. 1990. Growth of the seaweeds Kappaphycus alvarezii, K. striatum FAO. 2016. The state of world fisheries and and Eucheuma denticulatum as affected by aquaculture 2016: Contributing to food security environment in Hawaii. Aquaculture, 84(3–4), and nutrition for all. Rome: Author. 245–255.

Fernandes, L. 1996. Evaluating proposed uses of Glenn, E. P. and Doty, M. S. 1992. Water coral reefs: Comparison of economic and motion affects the growth rates of Kappaphycus multiple criteria approaches. PhD Thesis, alvarezii and related red seaweeds. Aquaculture, Department of Geography, University of Hawaii. 108(3–4), 233–246.

Fernandes, L., Day, J., Kerrigan, B., Breen, D., Green, A., Lokani, P., Atu, W., Ramohia, P., De'ath, G., Mapstone, B. and Kenchington, R. Thomas P. and J. Almany (eds.). 2006. Solomon 2009. A process to design a network of marine Islands marine assessment: Technical report of no-take areas: Lessons from the Great Barrier survey conducted May 13 to June 17, 2004. Reef. Ocean and Coastal Management, 52, 439–447. Guikema, S. and Milke, M. 1999. Quantitative decision tools for conservation programme Fernandes, L., Day, J., Lewis, A., Slegers, S., planning: Practice, theory and potential. Kerrigan, B., Breen, D., and Stapleton, K. 2005. Environmental Conservation, 26(3), 179–189. Establishing representative no-take areas in the Great Barrier Reef: Large-scale implementation Hurtado, A. Q. 2013. Social and economic of theory on marine protected areas. dimensions of carrageenan seaweed farming in Conservation Biology, 19(6), 1733–1744. the Philippines. In D. Valderrama, J. Cai, N. Hishamunda and N. Ridler, eds. Social and Game, E. T., Lipsett-Moore, G., Hamilton, R., economic dimensions of carrageenan seaweed Peterson, N., Kereseka, J., Atu, W., and farming, pp. 91–113. Fisheries and Aquaculture Possingham, H. 2011. Informed opportunism Technical Paper No. 580. Rome: FAO. for conservation planning in the Solomon Islands. Conservation Letters, 4(1), 38–46.

59 Hurtado, A. Q., Agbayani, R. F., Sanares, R. and Kerrigan, B., Breen, D., De’ath, G., Day, J., De Castro-Mallare, M. R. 2001. The seasonality Fernandes, L., Tobin, R. and Dobbs, K. 2010. and economic feasibility of cultivating Classifying the biodiversity of the Great Barrier Kappaphycus alvarezii in Panagatan Cays, Reef World Heritage Area. Research Caluya, Antique, Philippines. Aquaculture, publication No. 104. Townsville, Australia: 199(3–4), 295–310. Great Barrier Reef Marine Park Authority.

Hurtado, A. Q., Reis, R. P., Loureiro, R. R. and Kronen, M. 2013. Social and economic Critchley, A. T. 2014. Kappaphycus dimensions of carrageenan seaweed farming in (Rhodophyta) cultivation: problems and the the Solomon Islands. Social and economic impacts of Acadian marine plant extract powder. dimensions of carrageenan seaweed farming. In L. Pereira and J. M. Neto, eds. Marine algae: Fish Aqua Tech Paper, 580, 147–161. Biodiversity, taxonomy, environmental assessment, and biotechnology, pp. 251–291. Kumar, K. S., Ganesan, K., Subba Rao, P. V. Boca Raton, FL: CRC Press. and Thakur, M. C. (2015). Seasonal studies on field cultivation of Kappaphycus alvarezii (Doty) Hviding, E. 1998. Contextual flexibility: Present on the northwest coast of India. Journal of status and future of customary marine tenure in Applied Phycology, 28(2), 1193–1205. Solomon Islands. Ocean and Coastal Management, 40(2–3), 253–269. Lauera, M. and Aswanib, S. 2008. Integrating indigenous ecological knowledge and multi- Institute for Marine Remote Sensing-University spectral image classification for marine habitat of South Florida (IMaRS-USF). 2005. mapping in Oceania. Ocean and Coastal Millennium Coral Reef Mapping Project. Management, 51(6), 495–504. Validated maps. Cambridge, UK: UNEP World Conservation Monitoring Centre. Loureiro, R., Gachon, C. M. and Rebours, C. 2015. Seaweed cultivation: potential and Kapetsky, J. M. and Aguilar-Manjarrez, J. 2007. challenges of crop domestication at an Geographic information systems, remote sensing unprecedented pace. New Phytologist, 206(2), and mapping for the development and 489-492. management of marine aquaculture. Rome: FAO. Loureiro, R. R., Reis, R. P. and Critchley, A. T. Kapetsky, J. M., Aguilar-Manjarrez, J. and 2009. In vitro cultivation of three Kappaphycus Jenness, J. 2013. Global assessment of potential alvarezii (Rhodophyta, Areschougiaceae) variants for offshore mariculture development from a (green, red and brown) exposed to a commercial spatial perspective. Rome: FAO. extract of the brown alga Ascophyllum nodosum (Fucaceae, Ochrophyta). Journal of Applied Kasim, M. and Mustafa, A. (2017). Comparison Phycology, 22(1), 101–104. growth of Kappaphycus alvarezii (Rhodophyta, Solieriaceae) cultivation in floating cage and Malczewski, J. 2006. GIS-based multicriteria longline in Indonesia. Aquaculture Reports, decision analysis: a survey of the literature. 6, 49–55. International Journal of Geographical Information Science, 20(7), 703–726.

Malczewski, J. 2010. Multiple criteria decision Msuya, F. E. 2006. The impact of seaweed analysis and geographic information systems. farming on the social and economic structure of International Series in Operations Research and seaweed farming communities in Zanzibar, Management Science, 369–395. Tanzania. In A. T. Critchley, M. Ohno and D. B. Largo, eds. World seaweed resources: An Mantri, V. A., Eswaran, K., Shanmugam, M., authoritative reference system. Amsterdam: ETI Ganesan, M., Veeragurunathan, V., Thiruppathi, BioInformatics. S. and Seth, A. (2016). An appraisal on commercial farming of Kappaphycus alvarezii in Mtolera, M. S. 2003. Effect of seagrass cover India: Success in diversification of livelihood and and mineral content on Kappaphycus and prospects. Journal of Applied Phycology, 29(1), Eucheuma productivity in Zanzibar. Western 335–357. Indian Ocean Journal of Marine Science, 2(2), 163–170 Mardle, S. and Pascoe, S. 1999. A review of applications of multi-criteria decision-making Namudu, M. T. and Pickering, T. D. 2006. techniques to fisheries. Marine Resource Rapid survey technique using socio-economic Economics, 14, 41–63. indicators to assess the suitability of Pacific Island rural communities for Kappaphycus McHugh, D. J. 2002. Prospects for seaweed seaweed farming development. Eighteenth production in developing countries. Rome: International Seaweed Symposium, 15–23. FAO. Nath, S. S., Bolte, J. P., Ross, L. G. and Aguilar- McHugh, D. J. 2003. A guide to the seaweed Manjarrez, J. 2000. Applications of geographical industry. Fisheries Technical Paper 441. Rome: information systems (GIS) for spatial decision FAO. support in aquaculture. Aquacultural McHugh, D. J. 2006. The seaweed industry in Engineering, 23(1–3), 233–278. the Pacific Islands. Canberra: Australian Centre for International Agricultural Research. O'Regan, P. R. 1996. The use of contemporary information technologies for coastal research McKenzie, L. J., Campbell, S. J. and Lasi, F. and management: A review. Journal of Coastal 2016. Seagrass meadows of the Solomon Research, 12(1), 194–204. Islands, derived from field surveys conducted 13 May and 16 June 2004. Townsville: Centre for Ottinger, M., Clauss, K. and Kuenzer, C. 2016. Tropical Water and Aquatic Ecosystem Aquaculture: Relevance, distribution, impacts Research, James Cook University, Townsville, and spatial assessments – A review. Ocean and PANGAEA, Coastal Management, 119, 244–266. https://doi.org/10.1594/PANGAEA.868773 Pasqualini, V., Pergent-Martini, C., Pergent, G., MRAG Asia Pacific (fisheries and aquatic Agreil, M., Skoufas, G., Sourbes, L. and Tsirika, resource consultants). 2016. An economic and A. 2005. Use of SPOT 5 for mapping socio economic review of MSSIF’s funding for seagrasses: An application to Posidonia seaweed farming in the Solomon Islands. Report oceanica. Remote Sensing of Environment, for the New Zealand Aid Programme’s (NZAP) 94(1), 39–45. Mekem Strong Solomon Islands Fisheries (MSSIF) programme.

61 Pickering, T. 2006. Advances in seaweed Sousa, F. E., Moura, E. A. and Marinho- aquaculture among Pacific Island countries. Soriano, E. 2012. Use of geographic information Eighteenth International Seaweed Symposium, systems (GIS) to identify adequate sites for 18(3–5), 227–234. cultivation of the seaweed Gracilaria birdiae in Rio Grande do Norte, Northeastern Brazil. Pickering, T. D., Skelton, P. and Sulu, R. J. Revista Brasileira de Farmacognosia, 22(4), 868– 2007. Intentional introductions of commercially 873. harvested alien seaweeds. Botanica Marina, 50(5/6). Sulu, R., Hay, C., Ramohia, P. and Lam, M. 2000. Coral reefs in the Pacific. The status of Radiarta, I. N., Saitoh, S. and Miyazono, A. Solomon Islands coral reefs. In Status and 2008. GIS-based multi-criteria evaluation models monitoring, resources and management. for identifying suitable sites for Japanese scallop Proceedings from a regional symposium, (Mizuhopecten yessoensis) aquaculture in Funka International Coral Reef Initiative (lCRI), 22–24 Bay, southwestern Hokkaido, Japan. May 2000. Noumea IRD Centre, New Aquaculture, 284(1–4), 127–135. Caledonia.

Radiarta, I. N., Saitoh, S. and Yasui, H. 2010. Trono, G. C. Jr. 1990. Seaweed resources in the Aquaculture site selection for Japanese kelp developing countries of Asia: Production and (Laminaria japonica) in southern Hokkaido, socio-economic implications. In Culture and use Japan, using satellite remote sensing and GIS- of algae in Southeast Asia. Proceedings of the based models. ICES Journal of Marine Science, Symposium on Culture and Utilization of Algae 68(4), 773–780. in Southeast Asia, 8–11 December 1981, Tigbauan, Iloilo, Philippines (pp. 1-7). Roelfsema, C. M., Albert, S. and Phinn, S. R. Aquaculture Department, Southeast Asian 2013. Habitat map of Roviana Reef, Solomon Fisheries Development Center. Islands derived from a high-spatial-resolution multi-spectral satellite image using object-based Trono, G. C. Jr. 1992. Eucheuma and image analysis. PANGAEA. Kappaphycus: taxonomy and cultivation. https://doi.org/10.1594/PANGAEA.831039. University of the Philippines Research Digest.

Roelfsema, C., Phinn, S., Jupiter, S., Comley J. Trono, G. C. and Lluisma, A. O. 1992. and Albert S. 2013. Mapping coral reefs at reef Differences in biomass production and to reef-system scales, 10s–1000s km2, using carrageenan yields among four strains of farmed object-based image analysis. International carrageenophytes in Northern Bohol, Journal of Remote Sensing 34(18), 6367–6388. Philippines. The Ecology of Mangrove and Related Ecosystems, 223–227. Rothley, K. D. 1999. Designing reserve networks to satisfy multiple, conflicting demands. Ecological Applications 9, 741–750.

UNEP-WCMC, WorldFish Centre, WRI, Aquaculture Economics and Management, TNC. 2010. Global distribution of warm-water 19(2), 251–277. coral reefs, compiled from multiple sources (listed in “Coral_Source.mdb”), and including Vavia, A. 2018. Selecting seaweed farm sites in IMaRS-USF and IRD(2005), IMaRS-USF the Solomon Islands using geospatial (2005) and Spalding et al. (2001). Cambridge information systems. BSc Honours dissertation. (UK): UNEP World Conservation Monitoring School of Science. Auckland University of Centre. Technology.

UNEP-WCMC 2016. World database on Villa, F., Tunesi, L. and Agardy, T. 2002. protected areas: User manual 1.4. Cambridge, Zoning marine protected areas through spatial UK: UNEP-WCMC. Available at: multiple-criteria analysis: The case of the http://wcmc.io/WDPA_Manual Asinara Island National Marine Reserve of Italy. Conservation Biology 16(2) 515–526. Valderrama, D., Cai, J., Hishamunda, N. and Walter, R. K. and Hamilton, R. J. 2014. A Ridler, N. (eds.). 2013. Social and economic cultural landscape approach to community- dimensions of carrageenan seaweed farming. based conservation in Solomon Islands. Ecology Fisheries and Aquaculture Technical Paper No. and Society, 19(4). 580. Rome: FAO. Zemke-White, W. L. and Ohno, M. 1999. Valderrama, D., Cai, J., Hishamunda, N., World seaweed utilization: A summary. Journal Ridler, N., Neish, I. C., Hurtado, A. Q. and of Applied Phycology, 11(4), 369–376. Fraga, J. 2015. The economics of Kappaphycus seaweed cultivation in developing countries: A Zemke-White, W. L. and Wilson, P. 2015. comparative analysis of farming systems. World seaweed utilization. Seaweed Sustainability, 7–25.

63 Appendix 1. Key institutions and individuals consulted for this project

The team met with several management staff at the Ministry of Fisheries including the Deputy Directors, Ms. Rosalie Masu (Inshore Fisheries) and Ms. Ronnelle Panda (Policy, Planning and Project Management), the Chief Fisheries Officer, Mr. Sylvester Diake and research collaborators, Mr. Wesley Garofe (MFMR seaweed farming) and Mr. Sebastian Misiga (GIS analyst) accompanied us to locations in the Western Province where seaweed farms had previously been trialled and provided input on factors influencing the success of farming and existing sources of spatial data and geospatial software.

Discussions were also held with a consultant for the Ministry of Fisheries, Mr. Mahuri Robertson and Rarumana seaweed farming agent, Mr. Lamu Vara Livingstone. Discussions with Fisheries staff and consultants and existing GIS data were used to map and update existing and previous sites for seaweed farming and assess the suitability of these sites according to different physical, biological and social variables.

List of institutions consulted:

• Solomon Islands Ministry of Fisheries and Marine Resources (MFMR) • Gizo Fisheries Office (MFMR) • Mekem Strong Solomon Island Fisheries (MISSIF) • Solomon Islands Government ICT Support Unit • Rarumana Seaweed Farming • Solomon Islands National University (SINU) • University of the South Pacific (USP) • The Pacific Community (SPC) Sustainable Aquaculture Development project • New Zealand Ministry of Foreign Affairs and Trade (MFAT) • New Zealand High Commission • New Zealand Institute for Pacific Research (NZIPR) • Auckland University of Technology (AUT) • University of Otago.

List of individuals consulted:

• Solomon Islands Ministry for Fisheries and Marine Resources Head Office (MFMR)

> Ms. Rosalie Masu – Deputy Director, Inshore Fisheries > Ms. Ronnelle Panda –Deputy Director – Policy, Planning and Project Management > Mr. Sylvester Diake – Chief Fisheries Officer, Head of Aquaculture > Mekem Strong Solomon Island Fisheries (MSSIF)Dr. Anne-Maree Schwartz – Director > Dr. Ruben Sulu – Aquaculture Consultant • Solomon Island Government Information Communication Technology Support Unit (ICTSU)

> Mr. Smith Iniakwala – Director > Mr. Steve Erehiru – Deputy Director (Strategy and Planning) • New Zealand High Commission, Honiara, Solomon Islands

> Mr. Steve Hamilton, First Secretary (Development), NZ High Commission, Honiara. • Solomon Island National University (SINU)

> Prof. Prem P. Rai – Dean of School of Natural Resources and Applied Sciences (SNRAS) > Mr. Alex Makini – Head of Environmental Sciences, SNRAS • University of the South Pacific (USP)

> Dr. Patricia Rodie – Director Honiara Campus > Mr. Primo Ugulu –Geography Dept, Honiara • Gizo Fisheries (MFMR)

> Mr. Kolo Hivu – Fisheries Officer • Rarumana Village – seaweed farmers

> Mr Lamu Vara Livingston – Village spokesperson • Triland Trading

> Mr Mahuri Robertson

T: +64 9 923 9563 W: nzipr.ac.nz

ISBN: 978-0-473-47786-8