Cost–benefit analysis of beach-cast harvest: Closing land-marine nutrient loops in the Baltic Sea region

Harvesting beach-cast can help mitigate marine eutrophication by closing land-marine nutrient loops and provide a blue biomass raw material for the bioeconomy. Cost–benefit analysis was applied to harvest activities during 2009–2018 on the island of Gotland in the Baltic Sea, highlighting benefits such as nutrient removal from the marine system and improved recreational opportunities as well as costs of using inputs necessary for harvest. The results indicate that the activities entailed a net gain to society, lending substance to continued funding for harvests on Gotland and assessments of upscaling of harvest activities to other areas in Sweden and elsewhere. The lessons learnt from the considerable harvest experience on Gotland should be utilized for developing concrete guidelines for carrying out sustainable harvest practice, paying due attention to local conditions but also to what can be generalized to a wider national and international context. Supplementary Information The online version contains supplementary material available at 10.1007/s13280-021-01641-8.


S1. Summary of earlier literature on CBA of beach-cast harvest
presented an overview of the costs associated with green tides in the Chinese Yellow Sea. In this setting, damage costs include damage on tourism, recreation and aquaculture, as well as management costs related to harvesting and the protection of aquaculture. They showed that preventing blooms by lowering nutrient load from local aquaculture is more profitable to society than mitigating the consequences of blooms by harvesting and protective measures. The prevention of eutrophication has been shown to be economically profitable also in the Baltic Sea (Gren et al. 1997;Hyytiäinen et al. 2015;Scharin et al. 2016). However, still remaining is the more niched question on whether the benefits of beach-cast harvest as a specific eutrophication mitigation measure is an economically profitable venture, taking into account the costs of harvesting and the benefits of, for example, nutrient uptake and improved recreational opportunities. Concerning this more specific issue, Gisselman (2014) applied CBA for assessing the use of macro algae and reed from the bay of Burgsviken in Southern Gotland for biogas production. The monetized benefits and costs gave a social loss, but all benefits were not monetized and it was judged that a more complete monetization could result in a social gain. In contrast to the present study, the macro algae collection concerned free-floating macro algae, and no data on nutrient contents of collected algae were used; Gisselman (2014) cautiously regarded his study as a pre-study. Risén et al. (2017) applied the contingent valuation method for estimating nonmarket values of harvesting beach-cast for the purpose of removing nutrients from the Baltic Sea and producing biogas and biofertilizer in the case of Trelleborg Municipality in Southern Sweden, but did not analyze costs of such harvest and production. An overview by Blidberg et al. (2012) included such costs, based on the case of Trelleborg, and the benefits of nutrient uptake and indicated a net gain for society. However, this overview was not based on a strict CBA framework, which implies risks of overlapping benefit and cost items. Ofori and Rouleau (2020) conducted a willingness to pay study in Elmina, Ghana, for periodically removing Sargassum seaweeds from beaches. Mossbauer et al. (2012) gave examples of costs of beach-cast management for the case of the German Baltic Sea coast and also the potential commercial value of beach-cast, but did not include a full CBA. Charlier et al. (2007) presented cost estimates of beach-cast harvesting in Bretagne, France. Mossone et al. (2019) studied preferences of beachgoers for beaches with or without beach-cast at a number of Mediterranean seaside resorts, but not in a eutrophication context or with including costs of removing beach-cast. The benefits and costs of beach cleaning have been studied (e.g., Zielinski et al. 2019;Cruz et al. 2020;Mouat et al. 2010), but such cleaning is often about collecting litter and not related to eutrophication mitigation.

S2. Information from interview study with stakeholders
S2a (referred to in Section 3). Complementary data for the cost-benefit analysis was obtained from another research project conducting a stakeholder analysis of the beach-cast system on Gotland and applying the same system boundaries and scope. In this project, semi-structured interviews were carried out in December 2019−May 2020 with 18 stakeholders of Gotlandbased beach-cast harvesting activities, representing six non-governmental organizations (NGOs), six actors in the public sector and six actors in the private sector. The interviews provided stakeholder insights to the practical aspects and functioning of the beach-cast system, which served as a knowledge base that could be used for shedding light on consequences studied in the CBA. The interviewees were selected through snowball sampling to represent all scales of operation within the beach-cast system and grouped accordingly into NGOs, the public sector and the private sector. The interview results were scanned for statements relating to 'ecosystem service provision' and these findings are referred to in this study, relating the statements to respective stakeholder groups.
S2b (referred to in Section 4.2). Stakeholders from all three groups, NGOs, public and private sector, emphasized the positive effects on recreational opportunities from beach-cast removal, both on land and at sea. They perceived the collection of beach-cast in piles as making the water more accessible, as well as being a way of concentrating the inevitable malodour to certain spots, a conduct that they viewed as increasing people's pleasure of visiting the beaches.
S2c (referred to in Sections 4.2 and 4.3). All stakeholder groups, NGOs, public and private sector, stated that beach-cast harvesting is primarily limited to certain beaches and periods of the year, foremost fall, winter or spring. The public sector claimed that the reason for this is the uncertainties regarding the habitat effects, wherefore a precautionary stance is applied.
NGOs and the private sector, however, believed or speculated that the reason for this procedure is administrative restrictions without tying them to concerns for habitats.
S2d (referred to in Section 4.8). Stakeholders perceived local networking and collaborations between and among actors as a positive consequence of the LOVA system. All stakeholder groups expressed the emergence of this social cohesion, which was a reoccurring topic throughout the interviews, explicitly stated or embedded in the responses.

S3. Nutrient removal from the marine system
The nutrient removal reported by the LOVA projects are based on mandatory chemical analyses of beach-bast carried out by professional laboratories. The reported removal can still be expected to be uncertain, because there is no detailed information available about how beach-cast samples were taken and a procedural variability can be expected among the projects. We therefore also make use of scientifically published results on N and P content in beach-cast from 15 sites situated in the southernmost part of Gotland (Franzén et al. 2019).
While 15 sites are a small sample, a consistent protocol for sampling and chemical analysis were followed and indicated a mean concentration of 5.42 kg N t -1 FW and 0.55 kg P t -1 FW (i.e., 3.96 kg PO4-eq t -1 FW) (Franzén et al. 2019, Table 1). Applying these estimates on the total harvest of the LOVA projects (about 90 000 t FW) results in a somewhat higher nutrient removal than that reported by the projects: 487 800 kg N and 49 500 kg P. This indicates that the LOVA projects have not overestimated the nutrient removal.

S4. Benefits of nutrient removal from the marine system
The benefits of reduced eutrophication effects in the Baltic Sea are relatively well-studied thanks to a number of valuation studies from early 1990's onwards. We have chosen to focus on four valuation studies published in peer-reviewed scientific journals during the last ten years (after 2010) in order to have as recent and scientifically reliable results as possible on people's willingness to pay: questions about willingness to pay (WTP). The total WTP for all nine countries for reaching the eutrophication objectives of the Baltic Sea Action Plan (BSAP) was estimated to €2011 3603 million yr -1 (ibid., Table 7, p. 21). The corresponding total WTP among Swedes was found to be €2011 572.7 million yr -1 (ibid., Table 7, p. 21).
• Czajkowski et al. (2015) studied recreational values associated with the Baltic Sea by using the travel cost method. Data were obtained from a questionnaire to households in the nine littoral countries of the Baltic Sea. The annual benefits of recreation enjoyed at an improved environmental status of the sea was compared to those enjoyed at current status. The change in total consumer surplus for all nine countries because of a reduced eutrophication was estimated to €2011 1969 million yr -1 (ibid.,  Kinell et al. (2012, pp. 16-17) estimated the local population being sampled for the survey to 52 308 households, which implies a total WTP amounting to €2009 6.4 million yr -1 (10.2 × 52 308 × 12).
Linking the results of the valuation studies to amounts of nutrient removal is a challenge because the studies do not specify what nutrient reductions are necessary to accomplish the reduced eutrophication effects subject to valuation. However, the situation is relatively straightforward for Ahtiainen et al. (2014) and Czajkowski et al. (2015), because these studies valued reduced eutrophication effects for the whole Baltic Sea and reduction objectives have been defined in BSAP for the sea as a whole, see more below. While there is some uncertainty concerning whether these objectives are sufficient for achieving the improved environmental status that BSAP is aiming at, the complexity increases in the case of the valuation studies focusing on particular parts of the Baltic Sea: Nieminen et al. (2019) and Östberg et al. (2012). This is because eutrophication effects in these sea areas might be dependent not only on the local nutrient load but also on nutrient loads from other parts of the sea.
The Helsinki Commission (HELCOM 2015, p. 98) estimated that the annual load of nitrogen and phosphorus to the Baltic Sea must be reduced by 70 988 t N and 12 132 t P in comparison to a baseline period of 2008−2010 for achieving BSAP objectives by 2021, i.e., an annual load reduction of 67 060 t PO4-eq (70 988 × 0.42 + 12 132 × 3.07). We therefore interpret these reduction figures as the necessary annual reduction required for achieving the improvement that was valued in Ahtiainen et al. (2014) and Czajkowski et al. (2015) and for which estimates of annual total benefits are thus available. The results of Ahtiainen et al.
(2014) thus suggest a benefit of €2011 54 kg -1 reduced PO4-eq (€2011 3 603 000 000 yr -1 divided by 67 060 000 kg PO4-eq yr -1 ), whereas the corresponding benefit based on Czajkowski et al. (2015) is €2011 29 kg -1 reduced PO4-eq (€2011 1 969 000 000 yr -1 divided by 67 060 000 kg PO4-eq yr -1 ). The latter estimate thus constitutes 54 % of the former one, and one reason for this is likely to be methodological: Czajkowski et al. (2015) was a travel cost study able at estimating recreational benefits only, whereas the contingent valuation method used by Ahtiainen et al. (2014) can estimate total benefits in the sense that the method can capture both use and non-use values.
We now proceed by comparing the results above to those of Nieminen et al. (2019) and Östberg et al. (2012). As mentioned above, these two studies were about particular parts of the Baltic Sea, which means that the task of relating their valuation results to nutrient reductions becomes more challenging. We approach this task in the following way.
For Nieminen et al. (2019), there is a need to relate to nutrient reductions necessary for Finnish marine waters to reach GES. This should not be confused with nutrient reduction targets for Finland, because nutrients reach Finnish waters also from other parts of the Baltic Sea. We therefore take the nutrient reduction targets for the Baltic Sea mentioned above as a point of departure and assume that reasonable targets for Finnish waters is proportional to the area of Finnish exclusive economic zone (81 553 km 2 , Marineregions.org 2020a) in relation to the area of the whole Baltic Sea (415 266 km 2 , HELCOM 2006, p. 6), i.e. 19.6 %. This implies an annual reduction for Finnish marine waters amounting to 0.196 × 67 060 = 13 144 t PO4-eq. Relating the average estimate of total benefits in Nieminen et al. (2019) to this target results in €2017 36 kg -1 reduced PO4-eq (€2017 470 500 000 yr -1 divided by 13 144 000 kg PO4-eq yr -1 ). It should be noted that relating the estimated benefits in Nieminen et al. (2019) to reduced eutrophication only implies a risk for overestimating the benefits of reduced eutrophication, because achieving GES is also about improving other environmental aspects.
On the other hand, Nieminen et al. (2019) reported that reduced eutrophication effects were prioritized by survey respondents together with reducing toxic substances.
Table S1 summarizes all results with respect to benefits per reduced kg PO4-eq, including conversion to €2018 and USD2018. The relatively wide range of USD2018 17−73 kg -1 reduced PO4-eq (mean: 38) is not surprising, given the variation in valuation approaches, geographical scope and the challenge of associating benefits with necessary nutrient reductions. We also assume that average benefit estimates are applicable to beach-cast harvest, which suggest a situation where beach-cast harvest is considered as integrated into an assembly of many different measures for combatting eutrophication in the Baltic Sea. A considerably more refined analysis, such as applying meta-analysis function benefit transfer (e.g., see Johnston et al. 2021), is left here as a suggestion for future research. Finally, it should be observed that the estimates of Hasselström et al. (2020) are partly based on older valuation studies than the other ones in Table S1 and might therefore be less relevant. However, excluding those estimates does not affect the range of USD2018 17-73 kg -1 reduced PO4-eq, and the influence on the mean value is limited to a 10 per cent increase from USD2018 38 to 42.

S5. Emission estimates for the harvesting of beach-cast on Gotland
This section details the process of estimating emissions (combustion emissions of CO2, NOx, NH3, plus aggregated climate impacts last taking a life cycle perspective) resulting from typical beach-cast harvesting activities on Gotland. The first part of the section presents the case that forms the basis of the estimations. Thereafter, emissions data extracted from the Ecoinvent database (v.3.3) and USLCI database are presented and used to estimate average emissions per day. Ecoinvent and USLCI are two of the most established databases that provide documented process data and associated emission/impact factors used in environmental assessments such as life cycle assessment (LCA). Each database is compiled using different methods, data sources, and geographical contexts, and as a result, emission/impact factors for processes often vary to some degree. They are used side-by-side in the present calculations for reference, because they both contained suitable processes for diesel combustion.
The amount and types of emissions resulting from the combustion of diesel depends to a large extent on the type of combustion engine being used (i.e., which are subject to varying degrees of efficiency, filtration, etc.) and on how one delimits the system boundaries (e.g., end-of-pipe emissions from the motor's exhaust vs. full life cycle perspective) and functional unit of the study (e.g., emissions resulting from 1 km of transport in city vs. 1 km of transport on motorway). The emissions calculations hereafter presented should not be referred to as accurate modelled predictions nor as measured emissions data, but rather, these estimates should be considered as indicative of potential emissions from beach-cast harvesting activities.
Case summary. The case information (i.e., harvesting process description, fuel consumption, transport estimates, biomass yield estimates, etc.) is based on data received by Linus Hasselström and Jean-Baptiste Thomas during an interview carried out with Ulf Smedberg of Smedbergsgård AB, in December 2018. Ulf Smedberg is a major actor on Gotland with respect to carrying out beach-cast harvesting and his services were employed in many of the LOVA projects. Key information necessary for the calculations includes: (i) Approximately 30 km of beaches cleaned per year by Ulf Smedberg, perhaps totaling between 10k tons and 12k tons fresh weight of harvested biomass yr -1 . (ii) Total travel is usually around 150−200 km day -1 consuming around 20 L diesel per hour, travelling at 20−65 km hr -1 (circa 50 km hr -1 on average) (iii) 10-15 hrs of harvest time per day, diesel consumption around 8−10 L (9 L on average) hr -1 when at work (during harvest), harvest around 300−400 tons on a good day, usually between 100−300 tons, conservative estimate would be circa 150 tons day -1 .
From this information one can estimate average diesel consumption: ≈70 L for transport to and from the harvest location ≈113 L for harvesting Total ≈ 183 L day -1 (i.e., for around 150 tons of harvested beach-cast) Emissions data for combusted diesel. The emissions from two alternative diesel combustion processes were identified using SimaPro (V8.5). SimaPro is one of the commonly utilised LCA modelling tools. For each of these processes, emissions of CO2, NOX and NH3 resulting only from the combustion of diesel were exported. In addition, the climate impacts (GWP100, expressed in kgs of CO2 equivalents) for the full life cycle of each of these processes were calculated using the CML 2 Baseline 2000 (V2.05) method, for reference.
The first process, "Diesel, burned in agricultural machinery {GLO}| diesel, burned in agricultural machinery | Alloc Def, U", was selected from the Ecoinvent database (V3.3). In addition to the diesel's production, distribution, storage and combustion, the dataset includes contributions for heavy road transport with tractor, an agricultural trailer, tyres, agricultural shed for the vehicles when not in use. The unit for this process is 1 MJ.
The second process, "Diesel, combusted in industrial equipment/US", was selected from the USLCI database. This dataset was selected to provide additional perspective specifically on diesel combustion, and is described as representing diesel combustion in industrial applications, such as mobile refrigerator units, generators, pumps and similar equipment. The unit for this process is 1 L of diesel.
Given the different units used for each of these processes, the first was also converted* to 1 L by multiplying by (1/0.026). and per ton of harvested beach-cast (i.e., total divided by 150), following practices undertaken at Smedbergsgård AB. These are calculated using the emissions extracted from both the Ecoinvent process and the USLCI process, separately, for reference. The results from each are similar, with the only major difference being in the GWP100 (full life cycle) emissions estimates, owing to the fact that the Ecoinvent process includes a range of additional contributing sub-processes (e.g., trailer, shed, tyres, etc.).

*Conversion factors from MJ to kg to L for common diesel oil
Assuming use of Ecoinvent process "Diesel, burned in agricultural machinery {GLO}| diesel, burned in agricultural machinery | Alloc Def, U": 1456 kg CO2 eq (GWP100) 9.7 kg CO2 eq (GWP100) Assuming use of USLCI process "Diesel, combusted in industrial equipment/US":

S6. Harvest costs
Harvest cost figures for the LOVA projects are given in USD2018 in Table 3 in the article.  Table 1 in the article. f Based on a total nutrient removal of 301 949 kg PO4-eq, see Table 1 in the article.
An alternative way of estimating average harvest costs per kg removed PO4-eq is to use the nutrient concentration results in Franzén et al. (2019), see also Appendix S3. Dividing the cost t -1 FW in Table S2 by the mean nutrient concentration in Franzén et al. (2019), i.e., 3.96 kg PO4-eq t -1 FW, results in SEK2018 15 kg -1 removed PO4-eq excl. volunteering (60 divided by 3.96) and SEK2018 28 kg -1 removed PO4-eq incl. volunteering (111 divided by 3.96), which is only a few SEK lower than the corresponding cost estimates in the last row of Table S2.
Substantial variability in harvest costs per ton harvest can be expected across beaches because of different harvest conditions due to the degree of accessibility for tractors, personnel skill and experience, depending on volunteering, etc. (Mossbauer et al. 2012), and there are even more potential sources of variability for costs expressed per kg removed PO4-eq, because the projects' beach-cast sampling procedures also come into the picture as well as the natural variability in nutrient concentrations and the increased risk for the LOVA projects to make computational errors in the reporting.
For illustrating this, Table S3 reports the results of an analysis of costs per ton harvest and per kg removed PO4-eq based on such costs for individual LOVA projects, where each project is equally weighted (=1). It is evident from the coefficient of variation that there is a substantial variability, in particular for the costs per kg removed PO4-eq. This is to some extent explained by one project that reported very low nutrient concentrations: 0.52 kg N t -1 FW and 0.035 kg P t -1 FW, which is about one tenth of the mean concentrations found by Franzén et al. (2019).
The fact that this project also had relatively high costs per ton harvest implies that the project experienced extremely high costs per kg removed PO4-eq (i.e., the maximum values reported in Table S3). If this outlier is excluded from the analysis, the mean cost excluding volunteering is reduced from SEK2018 88 to 58 kg -1 removed PO4-eq, and the mean cost including volunteering is reduced from SEK2018 225 to 150 kg -1 removed PO4-eq. While the average situation for the LOVA projects indicates a cost including volunteering below SEK2018 100 kg -1 removed PO4-eq (USD2018 12), it is important that this cost can be considerably higher for individual projects. September 2019) and for Trelleborg, Blidberg et al. (2012) reported € 140 t -1 DW as a beachcast removal cost, i.e., about € 30 t -1 FW if assuming the mean dry matter content of 24 % found by Franzén et al. (2019). While those cost estimates are found within the min-max interval of Table S3, they still illustrate the substantial variation. Gisselman (2014) used a macro algae harvest cost of SEK 1260 t -1 FW for the case of the bay of Burgsviken in Southern Gotland. This cost is higher than the max value in Table S3, but was for collecting free-floating macro algae, which can generally be expected to be more expensive than beachcast harvest.

S7. Transaction costs
The costs for administrating the LOVA project system are identified as transaction costs. We delimit those costs to the administrative costs at the County Administrative Board of Gotland, whose officials are handling LOVA applications and reporting, and they also take part in informing project applicants and others about the LOVA system. LOVA project owners and other project participants can also be expected to incur administrative costs, but it is assumed that they are included in the volunteering reported in harvest costs (see Section 4.6 in the article). Salary costs related to administration of LOVA projects at the County Administrative