Determination of Plankton Biomass in Rongelap Atoll, Marshall Islands, by the Use of a Multi-Net Plankton Pump

Page 1

Determination of plankton biomass in Rongelap Atoll, Marshall Islands, by the use of a multi-net plankton pump’ With !3 figures md 2 tables in the text The Laboratory of Radiation Biology at the University of Washington has been engaged in recent years in a radioecological study of Rongelap Atoll, Marshall Islands, in the west equatorial Pacific Ocean. A part of this work involves estimation of the total standing crop of plankton inside the atoll, and this report describes the methods and results of such a determination. PAULIKkindly reviewed and commented upon the statistical analysis of this report. Description of the multi -net plankton pump A preliminary sampling of plankton in Rongelap Atoll in 1958 with a standard ;~ m net indicated that a more efficient gear \vas desired because the plankton density in the lagoon was low. Also, a single mesh-sized net may at times clog and it would not indicate any size composition of the plankton. To solve the above problems, a multi-net plankton pump sampler was developed.z AROX (1958, 1961) gives a complete bibliography of previous experience with plankton pumps. The plankton pump consisted of two main parts: a collector and a water pump fugal type, with a rated capacity of 1,924 l/rein., but was usually operated at a rate of 1,500 l/mirL Intake was through a 10.2 cm hose which could be lowered to a maximum depth of 22.9 m. Water volume entering the collector was determined by a “Sparling” water meter installed between the pump discharge and the The collector consisted of two stainless steel drums, one installed within the other. The outer drum was 57.2 cm in diameter and 90.2 cm deep with a 10.2 cm intake valve at one side near the bottom. The inner drum was 43,8 cm in diameter and 66 cm deep fastened to the bottom of the outer tank. A 15.2 cm discharge pipe led from the bottom of the tank through a 90° elbow to a 15.2 cm run-off 1 Report No. UWFL-81 from Division of Biology and Medicine, United States Atomic z The pump was designed by E. E, HELD and built by JOHNSTEVENS.

Page 2

Net depths without buckets # 6 mesh net 27.9 cm. deep fxpe Sparling flow meter Fig. 1. Diagram of the plankton pump Fig. 2. Photograph of plankton pump in operation, showing intake hose (left) and filtering cylinder with discharge valve (right).

Page 3

esh net 27,9 cm. deep Sporting flow meter pipe. ,i 6.3 mm pipe and valve was placed between the discharge valve and the , lbo\\’to permit the continuous removal of a sample of the discharge water. Inside the inner drum \vere installed three nylon nets with an inner diameter ()[ 4:3.s cm, of mesh sizes INO.6, No. 12, and No. 20, respectively. The nets with c:~[l\as facing \vere supported in the inner barrel by three stainless steel rings. The %:~ \vater was pumped into the outer drum, allo\ved to overffo\v the rim of [IIc’ inner drum find thence flOW by <ravity through the three nets to a discharge ]hmgelap Atoll (Fig. 3) has a square-shaped surface area with an extension in (he northeastern corner. Inside the surrounding reefs the depth increases rapidlv tt) :1 maximum of about 61 m. The bottom is relatively even except for numerous [oral hcxfs which extend almost to the water surface. xc\v water is brought into the lagoon from the east side during high tide via (JY<vfk)\vover the reefs. Outflow takes place mainly through the western and southern passes. VON ARX (l!%ll) has estimated the water renewal of the lagoon to take about one month. The surface current moves therefore in a westerly direction, but deeper layer \v:lters have a counter current. Our sampling procedures were designed to study L]](?variability in plankton abundance from east to west. from south to north, and fn)m surface to deeper l~vers. AISO considered \vere diurnal and seasonal variations. While the ship was cruising norrnallv at a speed of from 4 to 5 knots, samples (t)[d[l only be taken from surface and down to a depth of 3 m. .At anchor, pumping (ould l)e done to a maximum depth of 22.9 m. The basic sampling was done at the 61 cm depth with the ship cruising to-andcate hauls \vere also made on the same day. The second principal sampling was done ~vith the ship cruising in a north-south or south-north direction and with the pump intake It the 3 m depth, and ~vith the course-lines 5 minutes apart. The pu]np \vm operated normally for one hour corresponding to a filtered volume of :LboUt 9,5,000 1 of water before the nets were emptied and the collected plankton prcsemml. The pump was equipped with a double set of plankton nets so that C]liIIlIJf2of nets could be accomplished in less than one minute without stopping Finally, three fixed stations were established with continuous ,24-hour sampling of \vatcr had been strained. The plankton received in the three nets was filtered through a nylon filter of the same mesh as the net mounted in a millepore filter holder with the aid of an

Page 4

electric vacuum pump. One-fourth of the sample of each net from every station was preserved in formalin for species indenti8cation and species composition. The remaining 3/4of the filter was dried in an electric oven and, after all moisture had evaporated, the samples were sealed in small plastic bags and shipped to the Seattle Laboratory, where the dry weight on each filter was ascertained. The work began September 4 and was concluded on September 21, 1959.

Page 5

0, A. .Mathisen, Use of a plankton pump 739 For each sample the following information was entered on IBNI cards: station :Ind sample number, position at start and stop, date, time at start and stop, depth, \.(Jlume of ivater strained, dry weight of sample from each of the three nets, and e of each net from every station ion and species composition. The : oven and, after all moisture had plastic bags and shipped to the 1 filter was ascertained. The work The following definitions were introduced: Plankton was defined as all organisms retained by the three nets having mesh sizes No. 6, No. 12, and NTO.20. The random variable, y, is the combined dry weight of all plankton obtained in the three nets in one sample calculated per cubic meter of water. position of each s:unpfe means the mid-point between stinting and stopping position. Likewise, time is taken as the mid-point between start and stop. In order to study the effect of the position and depth of each sample, the SLII+KWof the Lrgoon was divided into 9 sections by latitudinal and longitudinal lines as indicated in Figure 3. For each of these factors there are three levels Likewise, for depth there are three levels designated as Ck (k = 1,2,3) where (Iw observations from the 61 cm and 3 m depths were considered to be representative of the water layers from the surface to a depth of 4.3 m, the observations from 6.1 rn were considered to be representative of the water stratum from the 4.3 to 12.2 rn depths, and finally, the observations from the 22.9 m depth were taken to be representative of the third and last depth stratum extending from 12.2 m to the bottom. The dividing points at the 4.3 and 12.2 m depths correspond approximately [o the geometric means between the 3 and 6.1, and 6.1 and 22.9 m depths. The random variable can thus be described as ~rjjji where r or replications within each cell ranged from 2 to 13 and where the three main factors, latitude, longitude, and depth, each have three levels. However, observations from the levels of c, and c:, exist only in cells al b, c,; observations. There thus exists a strongly unbalanced incomplete 3 X 3 X 3 factorial design with the three main factors considered fixed. The basic model used was one with fixed variables \vhcre a stands for latitudinal position effect, ~ for longitudinal position effect and Y for depth effect. The second order interactions were considered non-significant and included in the experimental error, Eijk. The somewhat complicated anrdysis of the given data cti speedily be done with a BIMD # 14 (1961) program which analyzes the statistical Significanceof independent variables that can be formulated in terms of general linear hypothesis model. Since the variables may be either classification or analysis of variance variables or regression variables of

Page 6

covariates, it is possible to introduce as covariates Xijk or time of day of a samp]e andzijk or day of sampling. The expectation within each cell would then be: with the following linear restraints: A log,,, y transformation of the random variable was introduced as recommended by WI~SOR and CLARKE (1940). The results of this analysis are given in the following analysis of variance table: d. f. 2 and 105 2 and 105 2 and 105 5 and 105 No significant difference could be established in regard to the main factors and the first order interaction between latitude or longitude and depth. HOWever, the interaction between latitude and longitude or position within the lagoon proved to be highly significant P <.01. This result is not surprising considering the current regimen inside the lagoon, In looking at the cell means, pronounced differences were evident with the lowest values of the cell means found near the west and south passes with outward current. Consequently, the best total estimate of the plankton biomass within Rongelap Atoll will be derived by stratifying the lagoon and by using total volume within each stratum, Ni as weighting factors, Estimate of total plankton biomass The term “plankton” is again defined as the organisms retained by the employed nets, and weight refers to dry weight. Repeated observations exist for each of the 9 cells in the water layer from surface to 4.3 m. For the two deeper layers extending from 4.3 to 12.2 m and from 12.2 m to the bottom, respectively, all three cells within each row or latitudinal direction were combined in both cases giving a total of 15 strata. The necessary data are listed in Table 1. The weighted over-all mean or population mean of dry weight of plankton per cubic meter in Rongelap Atoll is then (COCHRAN1953), meter and IQi = volume in m3 within each stratum. Since the sampled volume is extremely small in relation to the total volume of each stratum the finite popula-

Page 7

~tes xijk or time of day of a sample ithin each cell would then be: e was introduced as recommended of this analysis are given in the led in regard to the main factors le or longitude and depth. Howtude or position within the lagoon esuh is not surprising considering ng at the cell means, pronounced of the cell means found near the nsequently, the best total estimate will be derived by stratifying the stratum, Ni as weighting factors. the organisms retained by the t. Repeated observations existd

Page 8

N = estimated sample variance of the y values in the ilh stratum, and ni = number. of replication with each stratum. By substitution of the values given in Table 1 into the general formula for V (J,t) the following estimate of variance is obtained omitting the fpc correction: The best estimate of the total biomass, B, in dry weight inside RongeIap Atoll is then B = NJ,t = 423.5 metric tons, and with 50/0 confidence limits Optimum sample allocation The obtained estimates of the standard deviation of the random variable, weight of plankton per cubic meter, in each stratum can be used to estimate the allocation of samples given a fixed sample size, n, which will produce minimum variance of ~~t by random sampling in each stratum. By selecting a Lagrange multiplier ?. the problem is to minimize Again, by omitting the fpc correction this Ieads to the following expression for ni and the obtained values are given in Table 2. It is quite evident that the employed sample design failed to give proper attention to the deeper and larger strata, which would have reduced the confidence limits of the estimated biomass of plankton with the same sampling effort. Minimum sample size needed to meet a specified degree of accuracy of total biomass estimate Finally, it may be asked what is the minimum sample size needed to give the estimate of biomass with a prescribed degree of accuracy.

Page 9

in subsequent developments. The ariance of the y values in the ith each stratum. By substitution of ), in dry weight inside Rongelap and with 50/0 confidence limits 6 metric tons and eviation of the random variable, -atum can be used to estimate the oblem is to minimize to the following expression for ni 2. It is quite evident that the ~ttention to the deeper and larger ~e limits of the estimated biomass meet a specified degree Im ass estimate Lm sample size needed to give the O. A. Mathisen, Use of a plankton pump 743 The objective chosen was to obtain a coefficient of variation of 5°/0. Since B = 423.5 tons, the desired standard variation is (.05) (423,5) = 21.177 tons and omitting the very small fpc correction I)Y substitution of the values developed above. Finally, the optimum sample distribution within the strata can be calculated for the minimum sample size n~in by multiplication of the sample sizes in Table 2 Table 2. Distribution of sample size. Actual Optimum sample dist. by minimum 1. Construction of a multi-net plankton pump sampler has been described. The sampler was used to assess the biomass of plankton in Rongelap Atoll in 1959. Nets having mesh sizes No. 6, No. 12, and No. 20 were used. 2. The volume of water in the lagoon was calculated to be 51,361.108 m3. 3. The mean dry weight of plankton weighted by the volume of each stratum 4. The total biomass of plankton was estimated to be 423.5 tons with 50/0confidence limits of 472.6 tons — 374.4 metric tons.

Page 10

,.. 5. The variances of the mean weight in each stratum were utilized to cal- i culate an optimum sample allocation. 1 6, The minimum sample size needed to give a 50/0 coefficient of variation was calculated to be 70 observations, taken randomly but with optimum distribution within each stratum. ARON, W. 1958. The use of a large capacity portable pump for plankton sampling, with notes on plankton patchiness. — J. Mar. Res., 16, NW-173. — In press. Some aspects of sampIing the macroplankton. — Rapp. Cons. Expzor. Mer, Symposium on Zooplankton Production, Copenhague, Oct. 1961. BIMD Computer Programs Manual. 1961. Division of Biostatistics, Dept. of Preventive Medicine and Public Health, School of Medicine, Univ. of Calif., Los VON ARX, W. S. 1954. Circulation systems of Bikini and Rongelap Lagoons. — Prof. WINSOR, C. P., and CLARKE,G. L. 1940. A statistical study of variation in the catch of plankton nets. — 1. Mar. Res. 3, 1-34. J. CLARK: Wasphytoplankton included in biomass estimate? MATHISEN: Yes, but only those forms retained by the nets used in …