Results Directory UMM :Data Elmu:jurnal:J-a:Journal of Experimental Marine Biology and Ecology:Vol245.Issue2.Mar2000:

J . Chang J. Exp. Mar. Biol. Ecol. 245 2000 215 –224 219 Table 2 Trichodesmium abundance estimated with bottle- and net-collected samples at various locations in the southern East China Sea 21 Cruise Station Sampling Trichomes l Volume filtered depth m by net l Bottle Net March 1994 25823.09N, 122815.99E 10 3 44.3 11 681 25838.59N, 123807.79E 10 0.9 39 628 25817.89N, 122830.19E 10 3.5 18 280 25814.09N, 122839.99E 10 11.9 12 646 25811.19N, 122850.79E 10 3 2.8 23 172 25804.59N, 122800.09E 10 3 9.5 15 052 25826.39N, 121834.69E 10 5 0.1 13 918 25852.19N, 121820.09E 1 9 0.1 48 431 May 1994 25855.99N, 121819.99E 10 99 142.3 9370 25840.29N, 121849.99E 10 48 5.3 28 572 25830.09N, 122810.29E 10 25 57.5 8602 25825.09N, 122820.09E 10 1 11.3 6911 25819.89N, 122830.29E 10 7 7.9 5793 April 1995 26800.69N, 121810.39E 15 2 14.6 7413 25810.19N, 122850.09E 2–5 138 41.6 11 726

3. Results

Most of Trichodesmium existed as individual trichomes in the samples. Even in net-collected samples, colonial forms were rarely seen. The observed colonies were usually small and contained 10 to 50 trichomes. Judging by the low abundance of colonial forms, variation caused by colony-formation was ignored in data analysis. In samples collected by Go-Flo bottles in July 1995 and concentrated by settling, 21 trichome concentrations ranged from 27 to 109 trichomes l with a mean of 58.6 21 trichomes l Table 3. The variance due to natural distribution in the vicinity of the 2 22 Go-Flo bottles was estimated to be 136.8 trichomes l , and the variance due to 2 22 subsampling from a Go-Flo bottle was 1177.3 trichomes l Table 4. If trichome Table 3 21 a Trichodesmium abundance trichomes l in individual subsamples Date Subsample Go-Flo number Mean ¯ method x 1 2 3 4 5 July 1995 1 54 27 36 42 56 settling 2 64 40 77 109 81 58.6 July 1995 1 49 56 99 55 90 filtration 2 41 49 22 54 87 60.2 August 1994 1 24 11 16 20 5 settling 2 4 4 59 14.4 a Original water samples were collected with 2.5-l Go-Flo bottles. 220 J . Chang J. Exp. Mar. Biol. Ecol. 245 2000 215 –224 Table 4 Trichodesmium mean squares and variance components at each subsampling level in various sampling a programs 2 2 2 Method MS MS MS s s s C.V. 1 2 1 2 2 July 1995 Settling – 509.1 706.4 – 136.8 1177.3 62 Filtration – 561.9 605.2 – 180.1 1008.7 57 Net tow 57.4 0.3 5.8 9.5 22.8 5.8 40 31 August 1994 Settling – 217.9 394.9 – 43.1 658.2 184 Net tow 156.0 50.8 5.1 17.5 22.9 5.1 34 a The calculated C.V. for a single-subsample estimate is also listed. Settling and filtration are two methods used to concentrate trichomes in bottle-collected samples. concentration at this station has to be estimated with a single subsample from a single Go-Flo bottle, as commonly practiced in oceanographic investigations, the calculated C.V. was 62. The subsamples initially retrieved from a Go-Flo bottle always had lower trichome concentrations Table 3. A paired t-test indicated that the initially retrieved subsamples contained significantly fewer trichomes at the P ,0.05 level. Additional evidence came from samples collected in August 1994. Vastly different trichome concentrations were found between the two subsamples from the same Go-Flo bottle Table 3. During this cruise, the sequence of subsamples was not recorded. Most likely, the subsamples with fewer trichomes were initially retrieved. The extremely uneven distribution greatly 2 increased s , and if a single subsample was used to estimate Trichodesmium abundance, 2 the C.V. would be 184 Table 4. To evaluate if trichomes failed to sink when the subsamples were settled, filtration was used as an alternative method to concentrate Trichodesmium. The mean trichome 21 concentration found on the Nuclepore filters, 60.2 trichomes l , was very close to the mean from the settling method Table 3. The magnitude of variance at each level was also similar, and a C.V. of 57 was calculated for a single subsample estimate Table 4. For net-collected samples, the variance caused by taking a 1-ml subsample from a scintillation vial was small Tables 4 and 5. Much of the total variance was contributed 2 by distinctive net tows s and by distributing the water sample in a receiving bottle 2 into scintillation vials s . If Trichodesmium abundance was estimated from a 1-ml 1 subsample obtained from a single net tow, the C.V. ranged from 34 to 40. C.V.s of net-collected samples did not vary much between the two cruises, and had lower values than did any of the bottle-collected samples Table 4. Trichodesmium abundance estimated with bottle-collected samples often failed to match that estimated with net-collected samples. For example, during the July 1995 21 cruise, the bottle cast method estimated a concentration of 58.6 trichomes l , but the 21 net tow method estimated only 8.8 trichomes l Tables 3 and 5. When data collected from 17 stations were compared Table 2, the correlation coefficient was 0.461, not significantly different from 0 using the a 50.05 criterion Fig. 1. However, along a short transect in the southern East China Sea, these two methods generated approximate- J . Chang J. Exp. Mar. Biol. Ecol. 245 2000 215 –224 221 Table 5 21 a Trichodesmium abundance trichomes l in individual subsamples Vial number Net tow number Mean ¯ x 1 2 3 2 July 1995 1 8.6 7.8 3.9 15.4 9.0 6.8 2 9.0 10.3 5.6 14.7 8.9 5.9 3 10.1 9.5 6.6 13.3 8.1 5.2 8.8 31 August 1994 1 24.7 25.3 19.1 27.4 21.2 16.2 2 11.0 24.9 17.3 11.2 26.4 16.0 3 11.5 26.1 20.6 12.4 29.7 14.2 19.7 a Original water samples were collected with a plankton net. Fig. 1. Trichodesmium. Correlation between trichome concentrations estimated with bottle- and net-collected samples. Solid circles d are single-subsample estimates, and open circles s are mean abundance of multiple samples collected in August 1994 and July 1995. The diagonal line marks the position that the same abundance is estimated by the two methods. 222 J . Chang J. Exp. Mar. Biol. Ecol. 245 2000 215 –224 Fig. 2. Trichodesmium. Abundance at 10-m depth estimated with bottle-collected d and net-collected samples s along a transect starting at 25855.99N, 121819.99E and ending at 25819.89N, 122830.29E. ly the same distribution pattern for Trichodesmium with high concentrations at the western end, decreasing toward the east Fig. 2.

4. Discussion