Development of a Technique for Taking Inventory of Stocks and a Software Tool for Managing Mussel Farming Operations and Equipment

Table of contents

• INTRODUCTION
• Glossary
• METHODOLOGY
• General methodology used
• Magdalen Islands sampling campaign
• Carleton sampling campaign
• PROPOSED INVENTORY METHODOLOGY
• Description of the collection methodology on a longline
• Description of the collection methodology for the entire culture
• Minimum sampling
• Improved sampling
• Description of the weight measurement methodology
• COMMENT
• POSSIBLE EXTRA OPTIONS
• Study of the precisions to plan in future samplings
• Spring 2010 sampling campaign
• PLANNING THE CONTINUATION OF THE PROJECT: FIELD WORK AND ANALYSES

INTRODUCTION

As part of the project to develop a standard methodology for the assessment of mussel stocks, this report presents the results of the work carried out during the sampling campaign that took place in September and October 2009 and presents a proposal for the project's final sampling plan.

The 2009 fall sampling campaign had two main objectives:

• To continue studying different levels of variability when taking samples in mussel breeding, and
• To continue to evaluate various problems relating to sampling in mussel farms.

This report presents the various results from the statistical analyses conducted and the final methodology proposed.

Glossary

In this report, specific definitions for some terms used are as follows:

Cohort: In this report, a cohort corresponds to a group of longlines that belong to the same producer, that were socked at the same time and that have the same spat size (for continuous breeding) and that are expected to be harvested in the same year.

Loop: The term "loop" is used to indicate a sock dropper loop in the case of continuous breeding and to define a collector in self-managed breeding.

Section: The term "section" is used to determine a portion of a sock dropper loop. Three sections are considered: top, middle and bottom.

METHODOLOGY

General methodology used

The general methodology used for the 2009 fall sampling campaign differs from the methodology proposed following the spring 2009 pre-sampling campaign. Following the pre-sampling campaign, the planned methodology consisted in taking one sample of mussels in each loop sampled, but the weight of the mussels had to be high enough to enable use of declumpers. In this case, the work on the vessel was more important than taking small samples declumped manually, but the total mussel handling time was much lower. However, this methodology encountered a lot of resistance from marine farmers, mainly because:

• of the excessive cost associated with loss of mussels. Assuming a minimum of 150 kg of mussels per loop to make the declumper work and a minimum of 20 to 30 loops per inventory, it was estimated that a minimum of 3 000 to 4 500 kg of mussels would be lost through such an inventory methodology.
• Using the declumper to separate the different mussel samples would be difficult
  because this equipment must work continuously (the samples must be inserted one after the other to enable proper use of the equipment).

Faced with these major concerns and issues, we developed a methodology that is much closer to that used during the pre-sampling campaign. The changes were made to enable an assessment of the methodology in order to  reduce handling by marine farmers as much as possible.

The methodology used for the fall 2009 sampling campaign therefore proposes taking three one-foot samples of mussels from each loop sampled. With this methodology, marine farmers must manually declump and clean the sampled mussels. It was therefore necessary during the fall 2009 sampling campaign to complete the variability information to reduce the number of samples to be collected and mussel handling (cleaning, fall-off, weighing).

Magdalen Islands sampling campaign

Sample taking for the sampling campaign on the Magdalen Islands was planned for September 23 to 25 with two marine farmers (one collecting offshore and one collecting in a lagoon). Given the cold temperature and the difficulties exiting offshore, only cultures in the lagoon were sampled on September 24. All weight and size measurements were taken in the MAPAQ laboratory on September 25.

A total of 36 samples (4 longlines x 3 loops x 3 sections) were planned for the collection in the lagoon. However, a mechanical problem with the vessel delayed the start and 27 samples (3 longlines x 3 loops x 3 sections) were collected in the end. All 27 samples were collected in a span of about four hours (including travel time).

Despite thevessel's mechanical problems, collecting the 27 samples allowed us to study the variabilities in this type of culture because the samples were collected in different cohorts, three loops per longline and three sections per loop in each (with the exception of a missing piece of information at the top of a collector). The smaller number of samples compared to the number expected will, however, decrease the precision of the variance estimators obtained.

The 27 samples collected each correspond to a length of one foot on the section of the loop. Sampling sites on the loops were selected randomly before the longlines were hauled.

Carleton sampling campaign

The sampling campaign in Carleton took place from October 26 to 28 with two marine farmers with continuous cultures. At the outset, it was deemed preferable to sample a continuous and a self-regulated culture, but the closure of the self-regulated sites did not allow this type of culture to be sampled. However, note that with one of the marine farmers, measurements were taken of the collectors (data therefore comparable to a self-regulated culture). The cultures were sampled on September 27. All weight and size measurements were taken in Carleton on September 28.

A total of 24 samples (4 longlines x 2 loops x 3 sections) were planned for collection by each marine farmer. All 24 samples were collected on the first marine farmer's site, whereas 21 of the 24 samples were collected on the other marine farmer's site. One loop could not be sampled (tangled longline). All 45 samples were collected in a span of one day (with a change of vessel offshore near the harvests).

The 45 samples collected each correspond to a length of one foot on the section of the loop. Sampling sites on the loops were selected randomly before the longlines were hauled.

In addition to the data related to the cultures, information was collected about the sample collection time in order to optimize the methodology for the work required of the marine farmers.

PROPOSED INVENTORY METHODOLOGY

With regard to the work done during the pre-sampling and fall sampling campaigns, it seems evident that the final sampling method must reduce the work required of the producers as much as possible. The time required for the following tasks (done in the laboratory during the pre-sampling and fall sampling campaigns) must be decreased in the final inventory method.

• Manual fall-off of the mussels
• Manual cleaning of the mussels
• Mussel sample weight measurements
• Individual mussel size measurement

The methodology proposed in the pre-sampling report (using the declumper) raised many concerns from marine farmers because of the considerable loss of mussels that it entails and the difficulty using this equipment for distinct measurements on a set of samples. Faced with this choice, we therefore have to consider a manual method, but we therefore must reduce the required handling as much as possible.

Description of the collection methodology on a longline

The methodology discussed in this section targets the inventory on a longline. The inventory of an entire culture will be discussed in the next section.

Steps proposed for each longline selected:

1. Random selection, on this longline, of five loops or collectors
2. Taking a 30-cm sample from each loop to a random predetermined site. If the predetermined site corresponds to a bare patch along the mussel line, take another 30-cm sample at another site (select random replacement sites in advance). The sample is placed in a bag or container marked with the longline and the loop.
3. If the total length of the longline or the number of loops on the longline is not known, take the length of line represented by the loop (take the length top to bottom and multiply by 2).
4. Take the length of line on the loop that does not contain any mussels (except for the 30-cm section that was used to take the sample).

Description of the collection methodology for the entire culture

The variability components study and the visual observations taken in the field show that it is important to take measurements on all sections of the loops in a longline and to take longlines in the highest possible number of cohorts in a culture. Sampling several longlines in one cohort must be considered an option that gives a plus-value regarding specification of the inventory and must be considered only when all cohorts have been sampled for at least one longline.

Although the number of cohorts differs from one culture to another, the information provided by harvesters during the 2009 sampling campaigns led us to believe that there are about 6 to 12 cohorts for one culture.

Minimum sampling

The minimum acceptable sampling to be representative of an entire culture must include samples from one longline in each cohort in a culture. Therefore, the sampling plan involves:

• One longline selected randomly for each cohort in a culture
• Taking a 30-cm sample on five different loops in the longline
• The five 30-cm samples are taken in different sections of the loops (top, bottom and middle) defined randomly, but only one sample is taken per loop out of five loops.
• Of the five loops sampled, length measurements of those shorn of mussels
• The loops selected must be the least adjacent possible; the best representativity method would be to take a loop in each 1/5 of the longline; for example, for a longline of 100 loops, one loop in the first 20, one loop in the next 20, etc. This proposal, although optimal for representativity, can be very difficult to apply in the field (especially for collectors of self-regulated cultures); it will therefore be important to consider this issue in the final report after the final sampling campaign and after discussions with the various stakeholders. Lastly, it is important to indicate that samples could be taken in conjunction with other operations associated with production (for example, floatability adjustment), which requires longlines to be hauled.
• The five 30-cm samples are weighed separately to get a gross weight before they are
  grouped into one composite sample from which a sub-sample weighing about one kilogram is taken. This sample will be cleaned and stripped to get an estimate of the net weight of the composite sample and the number of mussels in that sample.

Improved sampling

Improved sampling, which will provide a more accurate estimate of a culture, involves collecting samples on two longlines in each cohort in a culture. Therefore, the sampling plan involves:

• Two longlines selected randomly for each cohort in a culture
• For the rest of the sampling, taking samples on a longline is the same procedure as the process described in minimum sampling.

The study of the precision based on the number of longlines will be provided in the final report; it will then be possible to determine whether an improved sample is required to satisfy the various stakeholders. However, it must be noted that a work day should result in an inventory of 8 to 10 longlines, which generally corresponds to minimum sampling for the entire culture. As improved sampling will require twice as much time, can marine farmers spend two full days on stock inventory?

Description of the weight measurement methodology   

The proposed steps for measuring samples are as follows:

1. Weigh (gross weight) the five samples from one loop individually.
2. Group the five samples into one composite sample.
3. Take a sub-sample weighing about 1 kg that is the best possible representation of the composite sample; in other words, take mussels that are still stuck together and noticeably individual mussels in the proportions of the composite sample. Taking only individual mussels when the composite sample contains several clusters of mussels must therefore be avoided.
4. Weigh (gross weight) the sub-sample of about 1 kg.
5. Clean the sub-sample and strip the mussels in it.
6. Weigh (net weight) the sub-sample once it is cleaned and stripped.
7. Count the mussels in the sub-sample. Do not include mussels from a second set.
8. Enter the quantitative data from the inventory into the Excel file to be provided to marine farmers. This file will calculate the yield estimates (kg of mussels per foot), the density (number of mussels per foot), the weight and the size (from the weight-size regression).

Note: All calculations after information was gathered will be done in the Excel file to be provided to marine farmers. The only quantitative information to be requested from the marine farmers will be the gross weight of all samples, the gross and net weight of all sub-samples and the number of mussels in the sub-sample. The marine farmers will, of course, have to identify the longline associated with each weight measured if one estimate per longline is desired.

COMMENT

The methodology presented is the same as that targeted for the inventories following the project. The May 2010 sampling campaign will use improved sampling (two longlines per cohort) in at least some cohorts in order to complete the sample variability study and to provide a quantitative assessment of the advantage of using two longlines per cohort instead of one.

To reduce the inventory work required of the marine farmers as much as possible, use of minimum sampling (one loop per longline) seems the most interesting. The next sampling campaign will determine whether minimum sampling yields acceptable precision and representativity for stakeholders in the field.

The main issue with using only one loop per longline comes from the possible non-representativity, for the longline, of the sample in this loop. This issue is important if an estimate is desired for the longline, but is less important for an estimate for an entire culture because the effect of a loop underestimating the yield, density or weight of a longline can be cancelled out by the effect of a loop overestimating these same variables. In addition, the non-representativity in the loop seems much less of an issue than non-representativity on the longlines.

POSSIBLE EXTRA OPTIONS

It may be possible to suggest an extra option to the marine farmers: the process of conducting an inventory with the help of divers. This option is only complementary and its use would be left at the full discretion of the marine farmers to improve the quality of the inventory or reduce the costs to marine farmers and divers.

Study of the precisions to plan in future samplings

Given that the methodology proposed uses a new measurement method and, for now, considers only one measurement of count per longline, it will be necessary in the May 2010 sampling campaign to consider the effect:              

• of measuring only the gross weight of all five samples;
• of counting only one sub-sample per longline (sub-sample taken from a composite sample containing the five samples);
• of the measurement of the proportion of line shorn of mussels;

Also, the use of the final estimators obtained by multiplying different intermediate estimators (for example, the number of mussels in a longline will be obtained by multiplying the number of mussels per foot by the proportion of line containing mussels, all multiplied by the length of the longline), makes it difficult at this stage of the study to determine a precision for these final estimators.

Lastly, it is difficult to determine the gain in precision by using a sub-sample from five samples grouped into one vs. one measurement in one 30-cm sample.    

The May 2010 sampling campaign, however, will aim to assess this precision from measurements to be taken using the new methodology. This will therefore be used to assess:

• whether taking five samples per loop is enough or whether three or four will be sufficient.
• whether taking five samples must necessarily be done in five different loops or whether the number of loops can be reduced.
• whether taking one sub-sample weighing 1 kg is enough or whether one or two extra samples would be necessary to get an adequate precision.
• the number of loops to assess with regard to the proportion of lines shorn of mussels.

Once all of this extra information is gathered and analyzed, it will be possible to determine the best sampling strategy to optimize the precision of the inventories. However, it must be noted that the adjustments should be minor and that, as a result, the inventories to be done in May will have precisions that are close enough to the inventories that will be done thereafter using the final inventory method. Therefore, it can be assumed that the inventories to be done in May can be considered clearly equivalent in quality (representativity and precision) to the inventories produced using the final method.

Spring 2010 sampling campaign

Following analysis of the results from the 2009 sampling campaigns and the final inventory method proposal, the May 2010 sampling campaign should focus on conducting inventories with various marine farmers using the proposed methodology. These inventories will aim to validate the practical interest of this methodology by providing participating marine farmers with a full inventory of their cultures. While these inventories are being conducted, it will also be possible to gather the latest information required for the final minor adjustments to the inventory methodology.

Of the last adjustments to be considered, the next sampling period will target:

• a final verification of the existing relationship between the weight and size of the mussels.
• an improvement of knowledge regarding the variability of the samples and the resulting precision.   
• the initial studies of the measurement of the length of the stripped loop.
• the study into the feasibility of taking 15- to 20-cm samples instead of 30-cm samples.
• the study of the significance of the advantages of taking net weight measurements on only the composite samples (decreases handling for measurements by 80%).   
• determining the need to take more samples from the collectors because of the greatest measurement variability (number of mussels per foot and yield).

PLANNING THE CONTINUATION OF THE PROJECT: FIELD WORK AND ANALYSES

February and March:   

• Discussions with stakeholders (SODIM, MAPAQ) to, if necessary, make the final adjustments to the inventory method proposed in this report.
• Contacts with the marine farmers in order to plan the May collection period.
• Development of the Excel file to be given to the marine farmers for calculating their future inventories.

May:   

• Sampling campaign in Carleton and the Magdalen Islands: The purpose is to test the final inventory method in the field by conducting an inventory of all cultures visited.  
• Project planning anticipated the visit of five marine farmers: two socked harvests in
  Carleton, one socked culture in the Magdalen Islands, one self-regulated culture in Carleton; i.e. one socked culture in theMagdalen Islands or one self-regulated culture in Carleton. Regarding the last choice, samples from a second marine farmer in the Magdalen Islands are more beneficial at this stage of the study than those from a second marine farmer that are self-regulated.
• The number and location of marine farmers can be reconsidered depending on the environment's needs and the availability of these farmers. Put another way, the number of inventories can be increased if the environment (SODIM, MAPAQ, etc.) requires it and can be decreased if the number of marine farmers available is limited in May.