Multi-objective zoning for aquaculture and 6 biodiversity

Aquaculture is the fastest growing food production industry in the world yet research and guidance demonstrating strategic multi-objective zoning for sector expansion is scarce. Quantifying and mitigating conflicts and impact on sensitive coastal environments through jointly-optimized objectives for aquaculture and biodiversity simultaneously has not been tested yet. We here develop and evaluate six alternative planning scenarios for one of the European Union’s highest 32 priority mussel aquaculture areas, the Emilia-Romagna Region in Italy. We i) develop an aquaculture 33 profitability surface as a function of the distance from main ports, and in parallel build a fine-scale 34 aquaculture suitability distribution surface for important commercial species using multi-criteria analysis; 35 ii) prioritize protected areas for biodiversity while testing how different considerations of human impacts 36 influence priorities; iii) simultaneously plan for aquaculture and biodiversity while minimizing impacts on 37 other maritime activities. We compare results from different scenarios according to how well they capture 38 suitable aquaculture habitats and minimize impacts. We introduce a new evaluation method for scenario 39 comparison in spatial optimization using a nearest-neighbour analysis for spatial pattern similarities. Lastly, we test the “value of information” provided by our investment in developing the fine-scale 41 suitability surface to improve efficiencies. 42 We find that an integrated multi-objective zoning approach, which simultaneously optimizes for 43 biodiversity and aquaculture, supports more efficient planning than traditional sector specific growth 44 strategies. We also discovered that the fine-scale suitability model delivered an 8% more efficient solution 45 than the simple distance function, highlighting the role of proxy surfaces and diminished returns from 46 investing in comprehensive habitat suitability analysis in regions without much variation in key 47 parameters. 48 We offer evidence of improved efficiency and practical guidance for integrated planning in Blue Growth 49 agendas. Our analysis can be applied in any context where multiple objectives occur for aquaculture sector 50 growth and biodiversity conservation. 51 52 53

We included 33 conservation features (Appendix S.2) based on regional spatial data provided in the 142 Tools4MSP Geoplatform -the primary platform developed to support marine spatial planning in the 9 Adriatic Sea (data.tools4msp.eu) (Menegon et al. 2018

162
The suitability (S) for each planning unit (i) was calculated according to the following equation (Appendix

173
The primary socio-economic factor considered was the distance of a planning unit to any port, as we 174 assume aquaculture zones placed further away from shore will decrease net profits due to the operational 175 costs of travel (Mazor et al. 2014 (Table 2). To make our comparisons as objective as possible, we did not apply the 208 boundary length modifier in our analyses to preference compactness.

247
To compare scenarios, we were interested in understanding how the different cost surfaces (e.g. area,

255
Examining how the best solution or selection frequency changes from scenario to scenario is a common 256 way to evaluate differences across scenarios. This is often done graphically by identifying where planning 257 unit selection diverges or remains the same between two scenarios. Another method is to use statistical 258 analyses such as calculating the Jaccard statistic (Real and Vargas 1996)

266
We adapted the Nearest Neighbor Analysis approach (Ebdon, 1991), normally used for single scenario 267 analysis, to investigate the similarity in the spatial configuration across scenarios.

268
The best solution outputs from each scenario have been used to compute the Nearest Neighbour ratio

291
If the index is less than 1, the two scenarios exhibit more similarity in their patterns than random. If the 292 index is greater than 1, the patterns are more dissimilar than random.

294
To evaluate this new Nearest Neighbor method, we also calculated the Jaccard distance (0 (full similarity)

295
to 1 (no similarities)) for each pairwise scenario, defined as the ratio of the size of the symmetric difference 296 between scenarios to their union (Real and Vargas 1996). We compute (see Eq.4) how many times the

340
Best solutions from the multi-objective scenarios (Figure 3; S5-S6) show the nearshore coastline as the 341 most preferred placement for aquaculture, with biodiversity objectives needing to be met throughout.
382 Figure 5 shows that NN distances and Jaccard distances are not perfectly correlated so they reflect real 383 differences in similarities. The J ratio is always less than 1 which means all the plans are more similar than 384 random. Jaccard alone is not sufficient to study the comparison among scenarios, NN alone provides a 385 better understanding, the combination of both gives a more solid analysis.

429
We did not quantify the potential negative impacts from bivalve farms on the surrounding biodiversity as 430 part of this study. However, these impacts can be accommodated in the multi-objective zoning approach 431 using Marxan with Zones. This will be particularly important for integrated zoning for aquaculture species 432 that are known for causing more substantial environmental impacts (e.g. carnivorous crustaceans and 433 marine fish (Primavera 2006)).

435
An important part of our analysis was developing a fine-scale suitability surface for the target aquaculture

470
According to our statistical analysis (histograms and ratios), the S5 and S6 spatial plans are very similar. In

29
Plans S2 and S4 were expected to be similar because the aquaculture suitability has been built with strong 475 emphasis on socio-economics. However statistical analysis reveals that plan S4 is preferable because of 476 the more detailed socio-economic modelling.

478
However, the difference in using profitability and suitability is less marked in S5 and S6 in line with 479 crossplot output and also consistent with the distance between S5 and S6, which is lower than that one 480 between S2 and S4.

483
We provide practical guidance on how to conduct and evaluate multi-objective spatial prioritization for 484 aquaculture and biodiversity using a systematic planning approach. We believe our overall approach can 485 be adopted to any study where society needs to make trade-offs between biodiversity, aquaculture and 486 other industries. This work is a first step in helping bring aquaculture zoning into the planning dialogue 487 that is taking shape around the world but has yet to see the development of long-term growth strategies