Sustainable Mussel Farming

AIMAP 2009-M22

Table of contents

• A: Background & Project Intent
• CARRYING CAPACITY
• Carrying capacity of St. Ann’s Bay for mussel aquaculture in the existing lease configuration
• Introduction
• The Model
• Mussel biomass
• Chlorophyll in the absence of aquaculture
• Mussel biomass and growth
• Chlorophyll and mussel biomass
• Assessment
• Confidence in the results
• Remaining work
• References

A:  Background & Project Intent

St. Ann’s Harbour is the location of the single largest mussel farming activity in one single location in Nova Scotia. A total of 1399 acres is leased by two mussel farms, Bounty Bay Shellfish Inc., owned by the Dockendorff family of Morell, PEI; and 5M Aqua Farms Ltd. (5M), owned by local fishermen of St. Ann’s Harbour. The farms do not utilize all the area of the leases at present. This utilization varies with harvesting schedules and site development. Lease 1189 which has not yet been developed by 5M,  has 270 lines of the 870 lines or 33 % of the total lease. In addition Bounty Bay has not used four sections on Lease 1186 and 5M has not  used one  section. These unused sections represents five of the sixteen sections. This represents 106 acres on 1186 and 166 acres on 1189 or 20 % of the total leased area.

This study was funded by 5M and as a result the work was done primarily on 5M leases. The carrying capacity component involves all mussels grown in the harbour over the past eight years.

A number of hurdles face the company as it attempts to successfully grow mussels sustainably;  both from an ecosystem perspective and an economic perspective. This work attempts to provide some of the tools to enable 5M to succeed. Poor yields, inconsistent meats, slow growth, costly seed, and inventory management problems all contribute to higher costs of production and reduced marketable product. Poor marketing arrangements and costly shipping and controls to processing were major contributing factors. The company requires a significant change in approach to farm management using the tools learned here combined with a more independent approach to processing and marketing.

From November 1, 2009  until October 1, 2010, data was collected and trials conducted to attempt to demonstrate scientifically the objectives of each study. However, there is never such a thing as a normal year. Trends can only be determined over several years. The period of this study proved to be far from a normal situation. For the first time since 1951 there was no ice cover in the harbour to provide stratification of the water column. This enabled the farms to harvest all winter without having to contend with ice. This lack of ice cover appeared to result in higher levels of phytoplankton in all areas of the harbour as increased light transmission resulted in greater primary production. Higher precipitation combined with higher temperatures also acted as a fertilizer for primary production to occur. Colder conditions would have resulted in less organic input throughout the entire winter. Meat yields were reflective of these conditions as they soared to levels in the mussel farm never seen in ten years of production (>50 % North American).

Mussel harvests from January 1 until June 15 from the two farms were continuous every week with often 6-8 tractor trailer loads being removed from inventory each week. It is expected that total harvest from the two farms exceeded 3 million pounds in this period. This is far greater than the harvest in any comparable time frame since the farms began.

The weather anomaly that worked in favour of St. Ann’s Harbour did so at the expense of PEI producers that, because of mild climate, were unable to ice harvest as usual because the ice cover was poor for machinery to travel over the ice. As a result market opportunities presented themselves much more accommodating and less competitive in the winter of 2010.

Due to the size of the farms there was significant public opposition and as a result the environmental monitoring done both before the inception of the new leases and yearly thereafter by many government agencies and by the farms themselves under the mandatory Environmental Monitoring Program was more extensive than for any mussel farm in Canada as well as globally. Huge amounts of data were collected for both the water column and the benthic environment throughout the entire harbour. In retrospect 5M felt that this could make a strong case to achieve environmental certification if the work in this study expounded on the environmental eligibility criteria for the environmental certification agencies. The carrying capacity model, unlike most models, includes over eight years of actual farm performance making it far more than an academic exercise.

As the work is composed of different study components, each will be dealt with on an individual basis in this report. The objectives were attempted and in some cases were successful but in other cases the objectives were not achieved. This will be evident in the details of this report. A general summary of key points discovered from all components will follow the detailed work.

At the end of the study descriptions and results there will be a section which will make recommendations of change for 5M and or possible future research that may be required as follow up to this project.

CARRYING CAPACITY

Carrying capacity of St. Ann’s Bay for mussel aquaculture in the existing lease configuration

Introduction

In both agriculture and aquaculture, the production of animal protein is based on management of the nitrogen cycle. For example, livestock such as cows (herbivores) are grown by supplying nitrogen fertilizer to grazer foliage, providing their food supply. For farmed fish (predators), marine or vegetable protein is provided as pellets. Farmed bivalves rely on nitrogen contained in phytoplankton, a natural food source already on the water column. Nonetheless, the supply of phytoplankton may be limiting if the stocking of shellfish is too dense. In this case, the capacity of bivalves to filter phytoplankton exceeds the renewal of phytoplankton within the system. Renewal occurs in two ways: the import of cells from the coastal ocean via exchange and circulation, and growth of cells arising from primary production via light. In the latter process, nitrogen sources either from rivers or the coastal ocean provide nutrients required for photosynthesis. The relative significance of these processes including bivalve feeding may be enclosed in a simulation model used to determine carrying capacity.

The goal of this component of the project is to evaluate the carrying capacity of mussel culture in St. Ann’s Bay using an ecosystem simulation model. This is a widely accepted approach among the aquaculture research community. Both Grant and Filgueira are expert practitioners in this research area, and have published extensively in the scientific literature on this topic. Although the present report could be written purely as a technical document, this would reduce its accessibility to regulators and growers. In addition, the technical aspects, though not specifically pertaining to St. Ann’s are published in the literature. The technical specifics for St. Ann’s are being compiled into a separate document.

The term carrying capacity (CC) is popular in the context of aquaculture, but requires careful definition. For the purpose of St. Ann’s Bay, we define two aspects of carrying capacity, economic and ecological, each with different criteria.  They share the concept of food limitation, expressed as phytoplankton (measured as chlorophyll) depletion. Economic CC is the maximum bivalve stocking density that can be cultured without excessively compromising individual growth rate. Ecological CC is the highest stocking density that can be cultured without compromising other components of the ecosystem. The latter is estimated by maintaining chlorophyll depletion at a level within its natural range during the year, a criterion that is explored in detail below.

The Model

The CC simulation is a fully spatial model, meaning that there is a detailed grid underlying its structure. This approach allows mapping of results rather than lumping them into large regional boxes. Moreover, mussels are located specifically in the designated lease areas. The grid cells are inter-dependent, so that water and materials removed from one cell enter another, i.e., everything is accounted in the grid. Material and water are also exchanged with the coastal ocean, which is not itself modelled, but forms the model boundary. The model consists of the following components:

1. Physical circulation model written in Aquadyn software (www.synexusglobal.com)
2. Ecosystem model written in Simile software (www.simulistics.com) consisting of
3. Phytoplankton and primary production, vertical and horizontal distribution of chlorophyll (detrital food is present but far less important than phytoplankton)
4. Input of nutrients from the North River
5. Input of nutrients from the ocean
6. Mussel biomass, feeding and growth

The model is in fact much more complex than indicated by this list, but its details are not essential in interpretation of the following results. The coupling of the physical and ecosystem models consists of a customized scheme to do the budgeting between cells according to physical exchange as well as the biological transformations that occur within a cell such as mussel feeding or primary production. The finite element mesh generated to describe the bay consists of 652 nodes and 1063 connections. Simulations were run for 195 days covering the period from May 1 to December 31, which includes the ice-free period.

The units of the model (mostly carbon-based) are determined by the necessity of keeping the terms comparable. For example, a gram of phytoplankton and a gram of mussel tissue are not often compared as pure mass, since they may have different chemical composition. Similarly, rate processes such as primary production are expressed as carbon production. We do not work with wet weights for the same reason. Water content of living organisms is a variable quantity which does not specifically interact in chemical processes, i.e., we do not require water in the calculations. Finally, we do not model shell growth or length since they are not coupled to organic carbon cycling, but rather to the carbonate complex in seawater. As seen below, it is possible to return to mass units from organic carbon in order to make the results more interpretable by users.

Mussel biomass

Due to thorough records of the distribution of mussel biomass on the leases, we were able to place the leases and initialize these values in the model. It should be noted that, we focused on dry meat yield, converted back from organic carbon units in the model. The conversion of our biomass results in metric tonnes (MT) into traditional units (shell-on whole animal wet weight) is a factor of about 10. In all cases, we consider that there are equal numbers of juvenile (year 1) and adult mussels (year 2) stocked. Due to the size difference, year 2 mussels are 4x of the standing stock of year 1 mussels. At the end of the model year, there is a cohort entering their second year, and a two year old group ready for harvest. Mortality is considered too low to impact the results.

Chlorophyll in the absence of aquaculture

Because we are attempting to balance chlorophyll depletion through feeding against chlorophyll renewal through circulation (physical process) and photosynthesis (biological process), it is necessary to characterize circulation in the bay. One way of doing this is to introduce a model tracer continuously at the bay mouth and follow its course through time. Eventually the bay will achieve equilibrium with the outside, but prior to that time, the distribution of tracer is a measure of flushing. It can be seen that there is a strong gradient of exchange from mouth to inner harbor. The penetration of tracer is slightly asymmetrical, and the average conditions used in Simile indicate that there is reduced exchange on the southern side of the harbour compared to the northern side. These results mean that lease 1188 has the best exchange, followed by 1187, 1189, and 1186 respectively, although the top of 1186 juts into a higher exchange region.

Exchange rates do not tell the entire story. Phytoplankton grows via photosynthesis, but if exchange is too great their biomass cannot accumulate because it is flushed from the system. Chlorophyll in the outer bay is controlled by exchange with the coastal ocean. Closer to the North River there is increased chlorophyll due to the enrichment by river nutrients. Nutrients do not enter the model system via the North and South Gut, but there is reduced flushing and phytoplankton can accumulate via primary production. Supplements from the river affect phytoplankton locally, but do not increase the overall biomass of the bay, as indicated by extra simulations where the river is ‘turned off’.

Mussel biomass and growth

Mussel biomass 'forces' the model and any error in its representation will lead to incorrect interpretation of model results. Fortunately, there are accurate records of cultured biomass from which to conduct the modelling. Proceeding from 2003, there have been changes in stocked biomass, but since 2007 all of the leases have been stocked. Defining typical or average conditions has been challenging, but we start with 2009 during which cultured biomass (especially 1186) was far greater than in previous years. From here, we examined fractions of this biomass with most emphasis on 25, 50, and 75 % of this value, equal to 62, 125, and 188 MT respectively. It is critical to recognize that as mussels grow during the year, their biomass increases, such that there may be little food limitation at the time of seeding, yet severe food limitation at the time of harvest.

Consideration of a single annual value of biomass is thus not appropriate. This can be seen in model output from the point of view of individual growth where both year 1 and year 2 mussels show density-dependence, i.e., reduced size at the end of the year for higher stocking densities. For example, year 2 mussels have a growth penalty in individual weight of up to 30 % for increased stocking density. The detriment is even greater for year 1 mussels. This effect can be observed more clearly in the entire population which first demonstrates the huge change in cultured biomass that occurs during the year. For example, an initial stocking density of 30 tonnes produces about 100 tonnes (combined year 1 and 2 mussels) for a return of 3x. Doubling that stocking density produces a yield of only 150 tonnes, and as stocked biomass increases, the resultant return in terms of growth is ever decreasing. The ratio of final yield to initial stocking declines with higher stocking density. One might conclude that CC is a simple choice of acceptable meat yield, and that the growth penalty might be worth the higher stocking density. However, it is likely that considerations of the ecosystem come into play such that the decision is not purely economic. In fact, substantiation of sustainability by ecosystem-based criteria may be an important aspect of the business model given recent trends toward certification by groups like WWF.

Chlorophyll and mussel biomass

When mussels are added to the bay according to the recorded biomass distribution, the distribution of chlorophyll undergoes large changes, especially in the leases. We thus examine not only the effect of food on mussels, but also the effect of mussels on their food.  As mussels grow during the year, they become increasingly food-limited. Maps of chlorophyll are the most effective way to demonstrate this effect. For an initial stocking density of 62 MT, the average level of chlorophyll is reduced by ~30 % compared to the case without mussels. For an even higher stocking density (75 MT which grows to 188 MT), the resulting reduction in chlorophyll is severe; with levels around the leases depleted by 60 %, and in the center of the lease even further decreased. We show absolute levels of chlorophyll because the values are necessary to compare between scenarios where farms are present and absent. 

We thus revisit the previously mentioned idea of variation in chlorophyll as an assessment standard. Phytoplankton undergoes large changes in biomass over the year, varying by a factor of 4-5 fold. It is however the spatial distribution that must be considered. The question is posed: is the absolute level of chlorophyll resulting from mussel culture at various stocking densities too low? There are limited data available for determining spatial variance in St. Ann’s Harbour, but the offshore distribution of chlorophyll from satellite remote sensing provides a measure of variation at the 1 km scale. These data indicate that the coefficient of variation for chlorophyll within any month is 26 %. This value means that that chlorophyll reduction by suspension-feeders in this range (-26 %) is within the natural ‘noise’ of the system, 120-200 mg C m-3. In contrast, a stocked biomass of 75 MT eventually yields 188 MT, resulting in average chlorophyll values (60 mg C m-3)  much lower than those observed in the absence of mussels.

More insight into these results is obtained from maps of relative chlorophyll depletion. As indicated above, 62 MT grows to about 160 MT, and at the initial stocking level, percentage depletion is below 25 %. At double this stock (125 MT) depletion values are approaching 50 %. For a stocked biomass of 75 MT which grows to 188 MT, the mature biomass always causes greater than 50 % depletion, and up to 75 % in the leases.

Assessment

These results can be consolidated by looking at a curve of average depletion versus cultured biomass, i.e., the average depletion that occurs during the growth year. With increasing crop size, chlorophyll depletion reaches a maximum (asymptote). It should be noted that the asymptote is less than 80 % depletion since all of the water and its chlorophyll is not accessible to the farm sites. In terms of recommendations, first it would seem prudent to stock at a level below the asymptote, since it represents the most severe form of seston depletion that can occur. Secondly, a stocked biomass of 75 MT seems higher than is desirable since there is an average annual depletion of 55 % and depletion near the leases of 75 %. In this case, the entire bay has lower average chlorophyll than occurs naturally.

It is hard to typify stocked biomass in the bay, partially because 2009 included a large increase in biomass at 1186. Accounting for this increase, we estimate that initial density over the last few years (2 year classes) has been 60-70 MT dry tissue weight. The present degree of culture maintains phytoplankton biomass at a level close to the natural variation of the system; at initial stocking of 62 MT, average chlorophyll depletion is <30 %,and by the time of harvest <50 %. In this sense, present aquaculture in St. Ann’s Bay is sustainable, because according to an ecosystem view where natural levels of chlorophyll are maintained, the ecosystem has not been compromised. However, we would not recommend increasing overall biomass in the bay because food depletion would become excessive by the criterion employed here. 

In terms of crop yield, conditions of greater chlorophyll depletion (75 MT) have only a 17 % individual growth penalty compared to lower levels of culture (i.e., 15 MT). However, this may be significant in economic terms, an evaluation that is beyond the scope of our analysis.

One important message is that resulting size of the crop from any stocking plan must be considered in terms of yield, since although the initial stock is below carrying capacity, food limitation will take hold during the year.

Confidence in the results

Because this analysis is based on a simulation model, it is important to consider the potential sources of error in the prediction. A number of generalities apply. First, the model has been published in the peer-reviewed literature (see References), meaning that its aspects have been evaluated by international experts. In addition, it has been ‘groundtruthed’ with mussel growth in other locations. We do not have individual growth trajectories for this purpose in St. Ann’s Bay, but the model growth curves are similar to those observed elsewhere. The levels of mussel biomass considered to be sustainable are in fact those now present in culture. We would assume that expertise in husbandry would lead to stocked biomass that is close to optimal. We would have in fact been surprised if existing culture resulted in extremely low or high levels of chlorophyll depletion.

There are few approaches that can yield both production carrying capacity and an additional assessment that considers the ecosystem. Primary production is the mainstay of the ecosystem and is thus a suitable candidate for measures of sustainability. In addition, the model is fully spatial and capable of representing the leases and their zone of influence.

Remaining work

We include this perspective because the model is considered evolutionary. Based on feedback from regulators and growers, aspects of the model might be changed as well as other scenarios simulated. It would be useful to alter the biomass per lease from the present conditions to see if greater production might be attained not through reduction of overall stocking, but to more effective local planning. The model is ideal for this sort of use, requiring optimization calculations that are more involved than the present set. We intend to produce these extra simulations in the coming months.

The model is meant as a management tool, but this can only be accomplished in an iterative sense via interaction with users. For example, we would expect to receive questions regarding model structure or interpretation of results. After this report has been digested, we would be amenable to making a presentation of results, followed by discussion. Although we have produced the report for this project, it is a primary area of research which only achieves value if it is useful to regulators and growers. As previously mentioned, efforts like WWF standards attempt to use chlorophyll depletion in a crude way to assess sustainability. The present effort is orders of magnitude more effective, and could satisfy any standard in this context. Moreover, a predictive tool of this type has been demonstrated for other locations in Canada (although rarely), but never with so much attention to spatial planning and baywide application.

In addition to model use, its parameters will evolve. Specifically, additional groundtruthing of the primary production and nutrients submodel are possible, especially with continued refinement of remote sensing applied to the bay. We will continue to sample nutrients and chlorophyll in the bay, attempting to further strengthen model input. Extended mussel growth studies will produce a specific data set for comparing growth trajectories between model and field results.

References

Ferreira, J.G., J. Aguilar-Manjarrez, C. Bacher, K. Black, S.L. Dong, J. Grant, E. Hofmann, J. Kapetsky, P.S. Leung, R. Pastres, Ø. Strand, C.B. Zhu. 2010. Progressing aquaculture through virtual technology and decision-making tools for novel management. FAO Fisheries and Aquaculture Tech. Doc., in press.

Filgueira, R. and J. Grant. 2009. A box model for ecosystem-level management of mussel culture carrying capacity in a coastal bay. Ecosystems 12: 1222-1233.

Filgueira, R., J. Grant, Ø. Strand, L. Asplin and J. Aure. in press. A simulation model of carrying capacity for mussel culture in a Norwegian fjord: role of induced upwelling. Aquaculture.

Filgueira, R.,  J. Grant, C. Bacher, M. Carreau. submitted. A physical-biogeochemical coupling scheme for coastal ecosystem models. Env. Model. Software.

Grant, J., K. J. Curran, T.L. Guyondet, G. Tita, C. Bacher, and V. Koutitonsky. 2007. A box model of carrying capacity for suspended mussel aquaculture in Lagune de la Grande-Entrée, Iles-de-la-Madeleine, Québec. Ecol. Model. 200: 193-206.

Grant, J. and R. Filgueira. in press. The application of dynamic modelling to prediction of production carrying capacity in shellfish farming. In S.E. Shumway, ed. Shellfish Environment Interactions. Academic Press.

Grant, J., Bacher, C., Ferreira, J.G., Groom, S., Morales, J., Rodrigues-Benito, C., Saitoh, S.I., Sathyendranath, S., & Stuart, V., 2009. Remote sensing applications in marine aquaculture. In IOCCG (2009), Remote Sensing in Fisheries and Aquaculture.