Bottom-Up Scenario Planning Boosts Inclusive Marine Spatial Strategies

All sectors of society benefit from the various goods and services provided by the world’s oceans and coasts, for example, through sustenance, cultural, religious or commercial interests¹. Many coastal regions, however, are focusing on developing their ocean economies, which is placing increased pressure on marine ecosystems and the services they provide. These pressures include overexploitation of marine resources, pollution, ocean noise, alien introductions, disease and habitat destruction, as well as indirect pressures from climate change. The impacts of these pressures can change the state of marine ecosystems, risking the potential for long-term benefits^2,3,4,5. To support sustainable development, marine spatial planning (MSP) has emerged as an ecosystem-based approach to ocean governance and is rapidly being adopted by countries globally^6,7,8,9. MSP is intended to be a cooperative process to develop a sustainable blue economy (i.e., sustainable ocean development) by achieving ecological, economic and social goals, integrating sectors and government agencies, applying adaptive management and reducing conflicts through iterative input from all affected sectors and stakeholders¹⁰. Here, we acknowledge that the term ‘blue economy’ has been critiqued by authors such as Voyer et al.¹¹ and Germond-Duret et al.¹². These authors argue that the term is being used in competing and often conflicting ways, by including certain stakeholders and excluding others, and negatively impacting both environmental and social sustainability.

Globally, the importance of growing the blue economy using MSP has been recognized and is being reinforced by initiatives and commitments of the United Nations (UN) such as the 17 Agenda 2030 Sustainable Development Goals, the Decade of Ocean Science for Sustainable Development (2021-2030) and the Treaty for Marine Biodiversity of Areas Beyond National Jurisdiction (2023). Current MSP approaches have, however, been critiqued by many authors for not addressing power imbalances¹³, prioritizing economic objectives over environmental or social ones^14,15, having superficial stakeholder participation processes^14,16 and providing limited adaptive capacity¹⁷, for example. Therefore, developing a clear understanding of the interdependence among environmental, economic and socio-cultural dimensions becomes key when growing a sustainable and long-lasting blue economy¹⁸. To address these concerns, we aimed to develop a marine spatial plan for Algoa Bay that was inclusive, collaborative and followed an ecosystem-based approach. A spatial plan that is developed through a bottom-up process, encouraging participation from all affected stakeholders throughout the process, including non-commercial groups such as local and traditional affected parties, can benefit from various positive consequences. Inclusivity throughout the MSP process can lead to a more comprehensive understanding of diverse knowledge systems (e.g., refs. ^19,20,21,22), build trust among participants and enhance compliance and/or stewardship and thus sustainability of agreements (e.g., refs. ^23,24,25) as well as reduce current or avoid future conflict among ocean users (e.g., refs. ²⁶). However, the effects of inclusiveness are not uniformly positive; in some cases, it can also lead to delays, increased conflict, or superficial engagement if not managed effectively^27,28.

South Africa has a large exclusive economic zone (~1.5 million km²) and the national government has embarked on an economic development program to grow its blue economy through Operation Phakisa²⁹. Launched in 2014, Operation Phakisa aims to fast-track the implementation of actions that would unlock the economic potential of the country’s ocean and is supported as one of the solutions to resolve the ‘triple threat’ of high levels of unemployment, poverty and inequality in South Africa³⁰. However, some have criticized the ‘big results fast’ methodology of Operation Phakisa, noting that it is failing to incorporate values and views of local communities because public participation and knowledge sharing has been severely inadequate^31,32,33. This has already led to mistrust and a lack of buy-in into the process, given that local communities have experienced the process as inequitable and biased towards the commercial sectors such as fishing and mining^33,34.

In 2018, the MSP Act No. 16 of 2018 was gazetted³⁵ and came into force in April 2021. The Act established the regulatory framework for equitable and fair designation and use of space in the maritime domain for South Africa. As the lead authority of the MSP process in the country, the Department of Forestry, Fisheries and the Environment (DFFE) is responsible for facilitating collaboration with all authorities that have a spatial presence in the marine environment. As such, the DFFE chairs the Marine Spatial Planning National Working Group (MSP-WG, established in 2015). The MSP-WG is composed of members from departments responsible for defense, energy, environmental affairs, fisheries, mineral resources planning, monitoring and evaluation, public enterprises, science and technology, telecommunications, tourism, transport, rural development, land affairs and marine cultural heritage through the South African Heritage Resources Agency. No civil society representatives are included in this group, including local or Indigenous knowledge holders or academics with the data and expertise to contribute most to the MSP process (see Reed and Lombard³⁶ for suggestions of how civil society could be formally incorporated into this process).

In 2017, a community of practice (CoP) was established, the ‘Algoa Bay Project’, to pilot the MSP process in a high-use bay situated in the Southern Area (one of four defined marine areas)³⁷. The CoP aimed to develop appropriate locally relevant methods and products to inform the MSP process and was initially funded by the National Research Foundation (South African public funds) with additional funding raised for the second phase of the CoP (see funding statement). The Algoa Bay Project consists of researchers with transdisciplinary expertise gained from working in the Bay and fosters a diverse, equitable and inclusive approach to MSP²¹. Algoa Bay was chosen as a pilot site to explore the legislative, biophysical and socio-economic practicalities involved in developing a marine area plan because the site has the longest-standing biophysical monitoring along the country’s shoreline and has well-established port infrastructure with diverse recreational and sport activities that attract local and international tourists³⁷. Supplementary Note 1 provides a detailed description of marine activities and current area-based conservation measures within the bay.

Following the principles of the MSP process stipulated both internationally¹⁰ and nationally (MSP Act), to conduct a transparent cross-sectoral planning process, the initial phase of the project identified and collected spatial data from Algoa Bay’s marine stakeholders. This included approaching typical ocean users such as the tourism, fisheries, research and recreation sectors³⁸ as well as individuals from different cultural groups within the local and Indigenous communities in the bay to understand how the ocean is important to them^19,20. The Algoa Bay Project’s MSP process differs from the National process in that it is co-developed with stakeholders from the start, rather than including stakeholders merely to comment on government documents. National legislation does provide text indicating that stakeholders should be meaningfully consulted from the outset, but this has not been implemented in the unfolding national MSP process to date³⁴.

Using data collected from and curated by stakeholders, the Algoa Bay project paid specific attention to two of the six foundational principles of MSP highlighted by Reimer et al.³⁹: i.e., the MSP process should be ‘strategic’ and ‘participatory’. This implies that the process should be inclusive and should actively engage stakeholders during the development and implementation stages, while also being ‘evidence-based’ and reliant on the best available scientific information, including various forms of knowledge systems and priorities. This is a difficult task because the evidence-based approach is rooted in specialist knowledge gained through experience and/or tertiary education and is often tied to complex data analyses, which is beyond the knowledge of different stakeholders.

To integrate inclusive participation and the scientific method, decision-support tools have become indispensable to the MSP process by assisting stakeholders and decision makers to consider different options and the consequent trade-offs to sector-based interests^9,40,41. These tools aim to visualize the consequences of ocean use decisions over time, space or both, and to simulate a greater understanding among stakeholders. This leads to cohesive discussions and increased participation in decision-making, ultimately leading to shared ownership, stewardship and co-management options and subsequently implementation of MSP. To this end, the Algoa Bay Project developed the ‘Algoa Bay Marine Systems Analysis Tool’ or MSAT, a System Dynamics Model (SDM) used to explore trends over time and the underlying feedback effects between marine uses and the environment⁴². The structure of the MSAT model was compiled from expert opinion through stakeholder engagement processes, including representatives from academia, business, civil society and municipal and government officials. The tool is publicly available online (https://exchange.iseesystems.com/public/esteevermeulen/the-algoa-marine-systems-analysis-tool-algoamsat-user-interface) and allows any user to explore how different combinations of cross-sectoral management interventions can impact the socio-economic dimensions of the blue economy in the Bay and the sustainability of the marine environment over time. The MSAT was built with a bottom-up participatory process with stakeholders and therefore represents a participatory, trade-off and predictive modeling tool that can be used within the MSP process to investigate the outcomes of different scenarios⁴⁰.

MSP is a complex and complicated process that must consider the needs of numerous and diverse ocean users to create a management plan that is considered inclusive, and it is recommended to follow an ecosystem-based approach to promote sustainability^10,43. Producing scenarios of how management decisions will unfold is an excellent way to integrate and package vast amounts of data and knowledge from various stakeholders into interpretable outputs that can support decision-making and coherence among stakeholders. One of the main aims of the Algoa Bay Project was to complement the temporal MSAT decision-support tool with an equally rigorous spatial decision-support tool to integrate socio-cultural connections and sustainable development in a spatial zonation of ocean users. Here, we use a systematic-site prioritization tool, prioritizR⁴⁴, to develop four zoning scenarios for the Bay, in an effort to understand the spatial limitations of human uses. We attempt to provide insight into how ocean space can be zoned to foster ocean health that considers the interconnected social-ecological systems. We do this by incorporating spatial layers of commercially valuable activities (e.g., living and non-living resource extraction) with spatial layers that were collated during the initial stage of the Algoa Bay Project, which show the distribution of biodiversity, nature-friendly human activities³⁸ and areas important to local and traditional communities^19,20.

Systematic-site prioritization is closely synonymous with ‘systematic conservation planning’ (SCP), which aims to identify representative, well-connected networks of priority sites for biodiversity protection while minimizing a cost⁴⁵. Systematic-site prioritization tools, such as prioritizR⁴⁴ or Marxan with Zones (Marzone⁴⁶), identify representative spatially defined zones for multiple marine uses (e.g., refs. ^47,48,49). Various attributes within a site-prioritization algorithm can be altered to meet different objectives and thus produce scenario zonation maps that can be used during stakeholder engagement as well as provide evidence-based support to final zonation maps. South Africa has long used SCP to identify and support the designation of candidate sites for marine protected areas (MPAs)^38,50,51,52. SCP has subsequently been used to delineate Ecologically and Biologically Significant Areas⁵³ and priority areas for shark and ray conservation⁵⁴ to contribute to further expansion of South Africa’s MPA estate. However, the current MSP Framework in South Africa^55,56 does not suggest using the same or similar methodology for the zonation schemes within its marine area plans, even though the methodology has been used as a successful support tool for zoning plans elsewhere^57,58. This may be due to the complexity of analyses or that the methods and tools are viewed as a ‘black box’ (i.e., complex analyses, whose inner workings are hidden or not easy to understand). This study thus aims to increase transparency of the algorithm’s inner workings, key assumptions and parameters by providing a detailed step-by-step guide on how scenario zoning plans are created within prioritizR. The method specifically allows for an inclusive and participatory approach by developing different scenarios to explore the complex trade-offs that decision makers face.

Spatial zoning solutions for the four zoning scenarios we developed for Algoa Bay using the prioritizR tool⁴⁴ are shown in Fig. 1. By altering various attributes in prioritizR’s algorithm (see ‘Methods’), we explored scenarios based on different levels of the three axes of environmental health, sustainability of governance and the level of socio-economic prosperity. Two scenarios were biased toward allocating more space to economically beneficial activities such as tourism, fisheries and mining (i.e., the Business as Usual and Empty Bay scenarios), one scenario was biased towards allocating more space toward conservation and cultural/heritage needs (i.e., the Untouched Paradise scenario) and one scenario allocated space to all activities without bias (i.e., the Bay of Plenty scenario). To further investigate spatial allocation, zoning scenarios were run with and without the mandatory allocation of planning units (PUs) to a zone.

Scenario zoning schemes for Algoa Bay

By altering feature targets, feature layers, permitting or restricting mining and/or ship-to-ship bunkering activities, the four zoning scenarios showed strikingly different marine zonation in Algoa Bay (Fig. 1). For example, the area allocated to the conservation priority zone decreased as targets of extractive activities (such as mining and fishing) increased through the scenarios from Untouched Paradise, to Bay of Plenty and to Business as Usual (Fig. 2). In the Empty Bay scenario, where the current MPAs were removed from the solution, more space was allocated to the conservation priority zone compared with other scenarios, even though biodiversity features received lower targets compared with other feature layers (Fig. 2). However, the area allocated to the conservation priority zone in the Empty Bay scenario was less than the area of the current MPAs in the Bay. Through the scenarios, as more preference was given to human activities, the zones became more fragmented (indicated by a lower aggregation index; Fig. 3a, c) and less costly (as determined by the cost layer per zone, the cost of assigning a planning unit to a zone), except for the commercial fisheries zone (Fig. 3b, d).

Mandatory versus non-mandatory allocation of planning units to zones

In the two zoning scenarios, Untouched Paradise and Bay of Plenty, not all planning units had to be allocated to a zone for features to meet their targets, resulting in unallocated areas of 558 km²and 1432 km², respectively (Figs. 1 and 2). However, in the Business as Usual and Empty Bay scenarios, almost all planning units had to be allocated to zones, with only 24 km² and 2 km² area remaining unallocated, respectively (Figs. 1 and 2). The Bay of Plenty zoning scenario had the most unallocated area even though this scenario included equitable target setting for all features and included the current MPAs as well as current ship-to-ship bunkering and mariculture areas (Fig. 1). All other zoning scenarios were biased towards allocating more area to certain zones as higher targets were set for either biodiversity and cultural heritage, or towards commercially beneficial activities. For the Untouched Paradise and Empty Bay zoning scenarios, the algorithm had more area available to allocate to different zones to meet feature layer targets, as ship-to-ship bunkering and MPAs were not locked into their solutions, respectively. However, regardless of more space being available, the algorithm still required more area to meet these scenarios targets, as less area remained unallocated compared with that of the Bay of Plenty zoning scenario (Fig. 2). The Bay of Plenty scenario had the most unallocated area (Fig. 2) and high aggregation scores for all zones (Fig. 3). This indicates that when scenarios bias their zoning towards certain activities (by awarding higher targets toward either biodiversity and cultural heritage or commercially rewarding activities) there will be higher competition for space in the Bay compared with when an equitable and ecosystem-based approach is taken to create a zoning scheme.

Creating a zoning scheme in MSP is both complex and complicated. It requires decision makers to consider the spatial extent and distribution of numerous activities and integrate knowledge from a diverse array of stakeholders to spatially zone these activities. If national^55,56 and international¹⁰ guidelines for MSP are adhered to, this zonation must be considered inclusive across various stakeholder groups as well as promote sustainable development by incorporating an ecosystem-based approach. This study demonstrates how developing zoning scenarios using the spatial decision-support tool, prioritizR⁴⁴, can consider and integrate a large data set (142 spatial layers) and explore different stakeholder interests by producing single-layer outputs that are driven by different narratives. By complementing the temporal decision-support tool, ‘Algoa Bay Marine Systems Analysis Tool’, created by Vermeulen-Miltz et al.⁴², this study shows the potential of decision-support tools to enable inclusive stakeholder input and involvement during the decision-making process of MSP by increasing transparency and encouraging informed engagement. The zoning scenarios presented here provide the insight that the entire planning area does not need to be zoned to inclusive and ecosystem-based targets (i.e., the Bay of Plenty zoning scenario). Additionally, in scenarios that emphasize extractive use (i.e., the Business as Usual and Empty Bay scenarios), space is more contested and there is no room for development of new sectors (e.g., renewable energy) or growth of present sectors.

The Bay of Plenty zoning scenario, resulting from an attempt to ‘fairly’ divide the marine space among users, allows each ocean use to meet its spatial requirement. This provides evidence that, in the context of this study, if marine spatial plans deviate from the Business as Usual scenario, they can result in ‘win-win’ scenarios that are inclusive, environmentally sustainable and economically rewarding. To further explore this, we present two ‘extreme’ scenarios, Empty Bay and Untouched Paradise, representing potential zoning plans if exploiting the bay for economic gain is undertaken unsustainably or if conserving the bay is prioritized with no economic growth plan, respectively. These ‘extreme’ scenarios exhibit the potential losses in ocean use to stakeholders if the development of a marine spatial plan does not follow an inclusive and participatory approach. These scenarios illustrate how spatial decision-support tools, like prioritizR, can be utilized during the marine spatial planning process to integrate and distill large amounts of quantitative data with qualitative storylines into ‘easy-to-understand’ visualizations. These visualizations can facilitate discussion and cohesion among ocean users as well as encourage participation of more diverse stakeholders.

In all four zoning scenarios, targets specified for each feature layer were met, albeit with or without mandatory allocation of planning units to a zone. However, as marine activities increased through the scenarios, the complexity of zoning scenarios increased, leading to greater fragmentation and complexity of zones, as was found by Mazor et al.⁵⁹. A zoning scheme that consists of multiple fragmented zones, each permitting or prohibiting certain activities, will be difficult, if not impossible, to implement and regulate. Furthermore, the large amount of unallocated area in the Untouched Paradise and Bay of Plenty scenarios (that had no mandatory allocation of planning units) should caution planners to avoid trying to zone entire marine areas. Instead, this ‘unzoned area’ can be zoned as a ‘Future Development Zone’ that can be allocated to emerging marine activities (such as renewable energy). In practice, as outlined by South African^55,56 and international¹⁰ frameworks, MSP processes should not be static but should be subject to monitoring and evaluation to allow for adaptive management and improvements over time^60,61.

Both this study and the system dynamics model (i.e., a temporal decision-support tool) for Algoa Bay by Vermeulen-Miltz et al.⁴² modeled four future scenarios of the Bay following similar axes to develop the scenarios, i.e., socio-economic wealth/prosperity, environmental health, and governance for sustainability. Here, the most equal, inclusive or best ‘win-win’ scenario, the Bay of Plenty, attempted to ‘fairly’ divide the marine space between biodiversity features and human activities by setting area-based targets following Holness et al.³⁸. This ‘winning’ scenario followed the narrative of an ecosystem-based approach to grow the blue economy, within a governance regime that supports sustainable use of the environment and allocates space among all ocean users equally. The equivalent scenario in Vermuelen-Miltz et al.⁴² was the Sea of Plenty. Similarly, this scenario included ocean governance for sustainability and followed an ecosystem-based approach. It surfaced as the best (‘win-win’) scenario because it showed the highest socio-economic gains as well as the best marine health over the study’s 40-year projection, with many socio-economic gains being directly correlated with healthy ecosystems (e.g., increased mariculture yields, fish biomass and tourist numbers). Together, the temporal decision-support tools of Vermuelen-Miltz et al.⁴² and the spatial decision-support tool presented here, are powerful tools for understanding the consequences of decision-making and how governance can shape the future. The result of either product is useful as a standalone, where the zoning spatial decision-support tool lends itself to understanding spatial limitations, and the system dynamics model highlights future economic gains, a strong driving tool behind decision-making in MSP, especially in Global South countries⁶². By supplying easily understood economic information and projections, scenarios can inform and support decision-making allowing ecological and socio-economic goals to be met. Future work should build on existing examples of spatial-temporal coupling of models for decision-making (see refs. ^63,64,65,66). Coupled models can allow planners to consider the impacts of zonation schemes on environmental and socio-economic well-being, and the ecosystem services provided by healthy environments⁶⁷ (Orolowitz et al. (submitted)).

Spatial decision-support tools, such as Marzone or prioritizR are sensitive to input data, feature targets and cost layers^67,68,69,70. Although Algoa Bay is very well sampled with respect to environmental parameters, social, cultural and economic parameters were more difficult to source and presented a limitation to our study. In addition, we had to alter some of the data sets to conform to the spatial regulations put in place in 2019 by the declaration of the Addo MPA (for more detail see ‘Methods’ and Supplementary Note 1) and the no-take area for the small-pelagic fisheries put in place in 2023 to protect African penguins⁷¹. However, rather than discouraging planners from using spatial models when data are imperfect, we encourage planners to develop scenarios with available data that are transparently and explicitly incorporated into scenarios. No matter the data quality, scenarios are very effective at starting conversations and exploring options across diverse stakeholder groups⁷². By increasing the transparency of input data and parametrization of spatial decision-support tools, effectively removing the black box, diverse stakeholders are able to interpret the resulting maps (e.g., ref. ⁴⁹). Transparency also increases the legitimacy of the spatial decision-support tool among stakeholders and encourages them to provide data so they are not ‘forgotten’ within the site selection procedure⁶⁹. To demonstrate the value of contributing data and knowledge to stakeholders, planners can produce scenarios with and without certain data. For example, Kockel et al.⁶⁸ used Marzone⁷³ to investigate potential sites for new MPAs. They presented three zoning scenarios that incorporated various levels of data set completeness for small-scale fisheries. The data poor scenarios showed how small-scale fishers would incur inequitable costs in the MPA design compared with scenarios that incorporated high quality data collected through stakeholder participation workshops. Although spatial zoning tools are not expected to produce a final solution⁷⁴, by allowing for stakeholder participation in scenario development, greater understanding of zoning outcomes can be reached across border audiences⁵⁸, leading to large areas of agreement and more effective implementation of marine spatial plans⁷⁵.

Depending on the goals and objectives of different zoning scenarios, important information can be provided to planners to allow for the evaluation of trade-offs, synergies and conflicts between activities. Both prioritizR and Marzone decision-support tools have been successfully used to inform final zonation maps (Grenadine islands⁷⁶; St. Kitts and Nevis⁵⁷; Malaysia⁵⁸) and their potential to inform the MSP process is increasingly being shown in academic studies^{49,59,67,68,77,78,79,80}. Despite their usefulness in MSP processes, zoning scenarios require the analysis of large quantities of heterogeneous, multi-sourced and spatially explicit data. Tools such as prioritizR can be used to collate these large data sets and integrate them in a way that produces zoning maps that meet users objectives and are easier to digest and understand compared to being confronted with the raw data alone. Resulting scenario zoning maps are an ideal discussion tool to weigh trade-offs and help end users by providing a systematic and defendable result that can support decision-making^40,41,49,81.

A zoning scheme that is developed concurrently with a defensible, systematic approach and genuine and effective public and stakeholder engagement can facilitate an integrated MSP that can enable effective implementation. The next steps of the national MSP process in South Africa will be to incorporate all information in the draft marine sector plans to produce the first of four marine area plans in 2025. These marine area plans will consist of general development guidelines, sector development guidelines and a zoning scheme^55,56. In South Africa, the MSP process of collating spatial data, understanding various sectors objectives, future growth goals as well as synergies and conflicts between sectors is already far progressed. A National MSP Data and Information Report has been published (detailing how various sectors are using the South African marine space⁸²) and draft marine sector plans (detailing national development objectives, priorities and spatial interests of each marine sector⁸³) are currently under review following public comment. However, it is unclear how the Marine Spatial Planning National Working Group (MSP-WG) intends to collate all these data and objectives into a comprehensive zoning scheme. South Africa has a long history of using systematic conservation planning to identify priority sites for conservation, e.g., conservation plans/ MPA design^50,51,52,84, sharks and rays⁵⁴, and identifying Ecologically and Biologically Significant Areas⁵³. Here we show the value of using a similar tool to produce zoning scenarios for the MSP process and recommend that a similar approach is used to produce the marine area plans for South Africa.

An equitable and ecosystem-based MSP process requires integration of knowledge across disciplines (e.g., policy and science), sectors and power structures within the affected communities^{75,85,86,87,88}. Integration is facilitated through stakeholder engagement at various intervals throughout the MSP process. Using traditional and local knowledge from those who are frequently using the marine space, including commercial, subsistence and recreational fishers, tourist operators, as well as those using the space for religious, spiritual or recreational purposes can provide different perspectives and invaluable knowledge about the marine environment which can complement and should inform the knowledge and practice of coastal managing authorities and researchers⁷⁴. A bottom-up approach to MSP that is inclusive and ecosystem-based and seeks to include and collaborate with all stakeholders, from government to civil society and grassroots groups, is both possible and necessary if we are to support the UN Sustainable Development Goals in a time of rapid climate change, increased environmental destruction, human rights abuses and economic uncertainty.

In conclusion, we have provided evidence that using decision-support tools to explore scenarios of ocean use can integrate large amounts of spatial data and stakeholder input to produce relatively easily understood outputs (i.e., maps). Once inputs of a prioritizR algorithm have been collated, it is relatively easy and quick to adjust and produce scenarios that are guided by stakeholder input. These scenario outputs are a powerful tool, and even though they are not final marine spatial plans, they can guide the decision-making process and increase the inclusivity of stakeholders and transparency of the process.

We used the spatial decision-support tool, prioritizR⁴⁴, to explore four zoning scenarios for Algoa Bay. PrioritizR with multiple zones is a systematic-site prioritization tool that allocates a user-defined proportion of features or activities to several user-defined zones at the lowest possible cost. Here, the proportion of features (i.e., feature targets) allocated to zones was altered to create the four zoning scenarios: two scenarios were biased toward allocating more space to economically beneficial activities such as tourism, fisheries and mining (i.e., the Business as Usual and Empty Bay scenarios), one scenario was biased towards allocating more space toward conservation and cultural/heritage needs (i.e., the Untouched Paradise scenario) and one scenario allocated space to all activities without bias (i.e., the Bay of Plenty scenario). We explain the creation of the inputs needed to run a prioritizR algorithm (i.e., zones, cost layer and feature target development) and how these inputs were altered to create zoning scenarios.

All spatial manipulation and analyses were conducted in the R statistical environment (R version 4.3.3⁸⁹). Code and data are available in the GitHub repository: https://github.com/Tegan151/AlgoaBayZoningScenarios. Analyses described hereafter were performed with data projected in the Africa Equal-Area Conic projection.

Planning domain

Using Marxan software⁴⁶, Holness et al.³⁸ developed a systematic conservation plan for biodiversity conservation and nature-based activities in Algoa Bay. To facilitate the transfer of their spatial layers to our analysis, we used the same planning domain and planning units. The planning domain extended 150 km along the shore from Sardinia Bay to Cannon Rocks and was divided into 4108 x 1 km² planning units (Fig. 4). The landward extent of the planning domain follows the dune base⁹⁰ and extends to the outer edge of the territorial sea, 12 NM offshore⁹¹.

Collation of spatial data

A total of 142 spatially explicit data layers (referred to henceforth as ‘feature layers’) of marine biodiversity and human activities were collated (Table 1). Feature layers of commercial, subsistence and recreational fishing, active mariculture, transport and related activities, biodiversity features and non-extractive recreational activities were sourced from Holness et al.³⁸, who collated the data from various sources including published information, expert mapping by local and scientific stakeholders and the National Biodiversity Assessment 2018⁹². Supplementary Tables 1 and 2 provide details of the data sources and maps of data layers. In addition to data provided by Holness et al.³⁸, we sourced spatial layers of the mining license block, legal precincts for mariculture, dredge dumping areas, ballast water discharge areas, waste-water outfall points and a military practice area (Table 1). To assign these spatial layers to the 1 km² planning units, the spatial polygon was intersected with the planning units using the st_intersection function within the sf R package^93,94. If the areas of the resulting intersected planning units were >50% of the original planning units’ area, the activity was assigned as being present in that planning unit (i.e., a value of 100). Values of all feature layers (e.g., fishing effort) were scaled between 0 and 100.

To include socio-cultural and heritage information, the following feature layers were developed: (1) livelihood, work, subsistence; (2) spiritual and/or religious connections; (3) recreational uses, fun and family time; (4) learning and knowledge; and (5) cultural heritage and history. These features were mapped by Strand et al. (in review) using a participatory community mapping approach during stakeholder engagement workshops with local and Indigenous people residing in Algoa Bay. Strand et al. (in review) co-created the types of categories of socio-cultural and heritage features to map. They did this with local and Indigenous stakeholders by discussing the main themes emerging from transect walks, arts-based participatory research in the form of in-situ photovoice and digital storytelling, collaborative workshops and semi-structured interviews collated during their previous research^13,19. To account for accessibility to areas by different communities across the bay, these five feature layers were subdivided into four sub-feature layers determined by their geographic locations: (1) western extent of planning domain to the port of Port Elizabeth, (2) port of Port Elizabeth to the port of Ngqura, (3) port of Ngqura to Colchester, and (4) Colchester to eastern extent of planning domain (Supplementary Tables 1 and 2). Offshore areas within the livelihood, work, and subsistence feature layer were extracted and included as a separate feature layer and included as a separate ‘small-scale fishery’ layer. Small-scale fishers make a small profit from their catch and were thus separated from the livelihood, work and subsistence feature layer (Mia Strand, pers. comm.). Changes to spatial layers after the declaration of the Addo Elephant National Park Marine Protected Area in 2019.

Data collection for the analysis by Holness et al.³⁸ and the National Biodiversity Assessment 2018⁹² occurred before the establishment of the Addo MPA⁹⁵. The MPA is divided into eight marine (three restricted and five controlled zones, Fig. 1) and three estuary management areas (two restricted and one controlled) that restrict or allow certain activities⁹⁶. In general, no person may fish or attempt to fish in restricted zones, and no diving is allowed in the Sunday’s inshore and Bird Island offshore restricted zones (Fig. 4). Fishing with a permit is allowed in the Addo MPA controlled zones, with the exception of fishing from a vessel in the Sundays inshore and Cape Padrone inshore control zones (Fig. 4). It was not possible to source data for some human activities, such as commercial, small-scale and subsistence fisheries and diving, after the Addo MPA was declared, thus these activities were removed from within the boundaries of the Addo MPA zones which now restrict them (Supplementary Table 2).

The distribution of commercial fisheries intensities across the Bay was calculated using Vessel Monitoring System (VMS) and/or logbook data, assigning intensity of use per planning unit between 0 and 100⁹⁷, with the exception of the small-pelagic fishery feature layer, which is an interpolated spatial layer of catches scaled between 0 and 100⁹². To simulate displaced fishing intensity after the declaration of the Addo MPA, fishing intensity was extrapolated to planning units outside of the MPA’s restricted areas (Supplementary Table 2) by allocating all planning units of a feature layer within a prohibited a value of zero and then applying Eq. (1) to calculate the extrapolated value of each planning unit (PU_extrapolated).

Where: PU_revised is the value of a planning unit once all planning units that are no longer allowed in the Addo MPA have been allocated a value of zero, ΣPU_original is the sum of the original planning unit values before the allocation of zero to those values no longer allowed in the Addo MPA and ΣPU_MPA is the sum of the original planning unit values that were allocated a zero within the Addo MPA.

In addition, no-take zones for the commercial small-pelagic fishery were declared for 10 years around the two African penguin Spheniscus demersus colonies on St Croix and Bird islands in Algoa Bay to reduce potential predator-fisheries resource competition (Fig. 4 and Supplementary Note 1^71,98). To account for the no-take zones for the small-pelagic fishery, the commercial small-pelagic feature layer was further altered, using the formula above, to remove this fishery from the newly formed no-take areas.

Zone types

Seven types of zones were defined for this study, and each planning unit can be attributed to only one type of zone by the prioritization algorithm. Following the ‘National Framework for Marine Spatial Planning in South Africa’⁵⁶, zones were named after their primary use (i.e., the activity that is given priority in a particular zone) or their exclusive use. A zone may allow activities (consent-use activities) that do not impede the primary use, and it may also prohibit certain activities that conflict with the primary (prohibited use activities). Table 2 lists the zones defined for this study. Each feature layer was assigned to a zone type based on its primary use (Tables 1 and 2). Zones which may potentially allow additional consent uses have a ‘priority zone’ included within their naming to indicate that the zone may allow certain other consent uses. Consent and prohibited uses of each zone should be included in the management regulations of each zone and determined with appropriate stakeholders.

Scenarios are future projections used to explore the consequences of management decisions made today. Vermeulen-Miltz et al.⁴² developed a systems dynamics model (MSAT) to explore the consequences of decisions made for the management of Algoa Bay along three axes: one represents a continuum from good to bad ocean health, indicating whether ecosystems and biodiversity are thriving or dying; another represents good to bad ocean wealth, indicating the potential financial gain from the bay through industry and tourism; and the third represents good to bad governance, indicating whether sustainable governance practices such as an ecosystem-based approach are implemented or not. They describe four scenarios placed at different extremes of these axes; the ‘Sea of Plenty’ and ‘Dead man’s Chest’ scenarios result when management is at the ‘good’, or ‘bad’ end of all three axes, respectively. In-between scenarios result from management practices placed at different positions along the axes. The MSAT produces ‘behavior over time graphs’ that are useful for exploring the values, over time, of variables of interest to the system (e.g., income from mariculture), but the models are not spatially explicit. In order to investigate the consequences of different spatial zoning scenarios for the bay, we used the same three axes as Vermuelen-Miltz et al.⁴² and developed four similar zoning scenarios: the ‘Bay of Plenty’ and ‘Empty Bay’ scenarios are at the ‘good’ or ‘bad’ extremes of all three axes, respectively, and the ‘Business as Usual’ and ‘Untouched Paradise’ are two alternatives (Fig. 5 and Table 3).

Development of zoning scenarios

Human activities in Algoa Bay were zoned with prioritizR R package⁴⁴. PrioritizR is similar to the commonly used Marxan with Zones or Marzone⁷³ software, but differs in that it uses integer linear programming and exact solvers to produce a single optimal solution (rather than a set of near-optimal solutions as produced by Marxan). By using prioritizR’s minimum set objective function to solve prioritization problems, solutions reflect the minimum set of planning units that meet or exceed targets, while minimizing the costs associated with the planning units selected, where:

• Targets are the user-defined proportion of each feature layer to be represented within a zone in the final zonation solution. To reduce analytical complexity, targets were set to be met exclusively by one zone type. Features were allocated to zones based on the zone’s priority use (Fig. 6).

• A cost is a spatial layer representing the actual or surrogate cost of assigning a planning unit to a zone, for example, the potential loss of catches by fisheries when a no-take zone is introduced⁸⁰. PrioritizR uses one spatial layer of cost per zone.

  a Zones and b graphical description of the prioritization setup for four zoning scenarios for Algoa Bay, South Africa.

  Full size image

Zone development

Within prioritizR, inputs for zones one to five (Table 2) were relatively straightforward, that is each zone had targets for each feature layer defined and a cost layer assigned. However, for zones 6 and 7 (ship-to-ship bunkering and mariculture), the existing legally designated (anchorage areas) or proposed zones (proposed Aquaculture Development Zone (ADZ) precincts) for these activities were used (see Supplementary Note 1). Within prioritizR, to account for planning units a-priori as belonging to zones, the planning units can be ‘locked-out’ of all other zones thus making these planning units unavailable for these zones. However, prioritizR’s algorithm prefers the user to delete planning units that are not available for solving the zonation problem, but we did not wish to present maps with any planning units deleted. Rather, we chose to combine these activities in a sixth zone and ‘locked in’ the planning units to represent the presence of these activities in the zone’s solution. A cost layer was created for this zone by assigning a cost of one to the ‘locked in’ planning units, and all other planning units received a cost of 100).

The existing Sardinia and Addo MPAs are already zoned for the conservation and protection of biodiversity and were therefore locked into the conservation priority zone. They are represented in results as MPAs, and any additional planning units needed for the biodiversity targets to be met are presented as the conservation priority zone. However, the management regulations of the Sardinia and Addo MPAs allow many non-extractive activities and the controlled zones of the Addo MPA allow fishing activities with a permit. Therefore, many of the feature layers that overlapped with the MPAs are already present as ‘consent uses’ in these MPAs (Supplementary Table 2). To account for this, the proportion of a feature layer’s target that was met in the MPA was taken into account so that the feature was not overrepresented in the final scenario run of prioritizR.

Scenario and target development

Scenarios were implemented by changing which zones to include, altering targets of feature layers, increasing or decreasing the size of the ship-to-ship bunkering and mariculture zones or including a mining priority zone (Fig. 6).

For the two scenarios where environmental health is low because of bad governance (Empty Bay) or a focus on economic growth dominates (Business as Usual, Fig. 5), a mining priority zone was included in the prioritization, both anchorage areas were zoned for ship-to-ship bunkering and current as well as ADZ precincts for mariculture were locked into the solution (Figs. 4 and 6). Based on the Kunming-Montreal Global Biodiversity Framework 30 x 30 conservation target⁹⁹, which aims to have 30% of the ocean and land protected by area-based management tools such as MPAs by 2030, biodiversity feature layers had a 30% target. In line with this international agreement, we chose a minimum target of 30% for biodiversity feature layers. Similarly, cultural/heritage feature layers also had a target of 30% because we viewed them more similar to biodiversity layers than to commercial and ‘profitable’ feature layers. Commercial and small-scale fisheries, as well as both types of recreation (i.e., non-extractive and extractive), had a 60% target, double that of the biodiversity and cultural/heritage features, because of their role in direct income generation. However, for the Empty Bay scenario (which ignores current MPAs), a solution could not be reached with these targets so they were reduced to 25% and 50%, respectively. The Empty Bay scenario did not use the extrapolated feature layers to account for the restriction of activities in the Addo MPA (see section ‘Changes to spatial layers due to the declaration of the Addo Elephant National Park Marine Protected Area in 2019’).

In the scenario that focuses on environmental health (Untouched Paradise), a target of 60% was used for the biodiversity, non-extractive recreational and culture/heritage features, and a lower 30% target was used for extractive activities such as recreational, commercial and small-scale fishing (Fig. 6). This scenario also excluded ship-to-ship bunkering activities from the bay and allowed only the existing mariculture zone with no expansion into the ADZ precincts (Fig. 6).

The fourth scenario (Bay of Plenty) assumes good governance and aims to zone the bay for both environmental and socio-economic health. In an attempt to set targets fairly or equitably for all features, the area-based targets developed by Holness et al.³⁸ were used. Area-based targets are based on the extent of a feature, with lower targets assigned to features with larger area coverage and vice versa and consider the intensity of the activity per planning unit. We refer to this as a feature’s Area x Activity value (AA_v), and it is calculated with Eq. (2).

where PU_area is the area (in km²) of each planning unit in which the feature occurs, and PU_value is the intensity of the activity in the planning unit (e.g., fishing intensity or at-sea distribution of African penguins, etc.). For features that are just a distribution map of where a feature is present or absent (e.g., non-extractive recreational activities such as boating areas) the PU_value is 100 for each planning unit in which that feature occurs and AA_v for that feature will simply be equal to its total area. Targets were then calculated as follows, using an additive process: 60% of the first 1000 AA_v; 50% of the next 1000–5000 AA_v; 40% of the next 5000–10,000 AA_v; 30% of the next 10,000–50,000 AA_v and 20% of areas greater than 50,000 AA_v. With this process, feature layers with small distributions receive higher proportional targets (because they are likely at greater risk than widely distributed features), and for feature layers with intensities, targets can be met in areas where the activity is concentrated.

Cost layers

To avoid potential spatial conflicts among human activities, a synergy-conflict matrix was developed to define which activities were compatible, compatible but at a cost, or not compatible, with the objectives of each zone (Table 4). Scores of 0, 0.5 or 1 were allocated, respectively, guided by the compatibility of activities as indicated by the different marine sectors within the National Data and Information Report for Marine Spatial Planning produced for South Africa⁸². For each zone, each feature layer was weighted by this compatibility score. All layers were then summed and rescaled between 0 and 100 to create a zone-specific cost layer. To avoid assigning a zero cost to any planning unit, we added one (+1) to each planning unit’s cost value. Equation (3) was used to scale cost layers between 0 and 100.

Where: Cost_scaled is the new scaled value of a planning unit in a cost layer, Cost_PU is the original value of the planning unit in the cost layer, and Min and Max are the minimum or maximum value of all cost layers in a scenario, respectively.

Mandatory allocation of planning units

Whether it is mandatory to allocate a planning unit to a zone can be set within the prioritization algorithm of prioritizR⁴⁴. Therefore, all zoning scenarios were run with and without the mandatory allocation of planning units to investigate what influence this may have on the final zonation product.

Boundary penalty

As fragmented zones consisting of one or a few planning units are not conducive to the enforcement of regulations and management, a boundary penalty was used to avoid overly fragmented zones. The optimal boundary penalty of each scenario was sought based on iterations of prioritizations with a 10-fold increases in the boundary penalty between 0 and 0.1 and identified as the inflection point in the trade-off curve between the total cost and sum of perimeters of the solutions’ zones, following the methods described by Ardron et al.¹⁰⁰. However, these values were inadequate to identify each scenario’s optimal boundary length value and therefore additional iterations with values between 0.13 and 0.01 at 0.015 intervals were run. The inflection points in the total cost and sum of perimeter trade-off plots were identified using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method¹⁰¹ using R package topsis¹⁰². For trade-off curves and optimal boundary values, see Supplementary Fig. 2.

Comparison of scenarios

Zoning scenarios were compared with each other using three indices:

1. 1.

   The total cost of each zone within a zoning scenario, i.e., the sum of costs to assign a planning unit to a zone.

2. 2.

   The total area that each zone covers in a zoning scenario is in km².

3. 3.

   The aggregation index¹⁰³ of each zone per zoning scenario was calculated with the R package landscapemetrics¹⁰⁴. The aggregation index, ranging between 0 and 100, is calculated as the ratio of actual edge to the total amount of edge possible and indicates maximal disaggregation (0; planning units assigned to a zone share no edges) and maximal aggregation of a zone (100; all planning units assigned to a zone share an edge).