Fractionation by molecular weight of organic substances in Georgia coastal water1 John R. Wheeler University of Georgia Marine Institute, Sapelo Island 31327 Abstract Ultrafiltration membranes were used to separate organic substances in Georgia coastal water into fractions to which approximate molecular weight (MW) ranges could be assigned. Salt marsh water contained high concentrations of dissolved organic carbon between 1,000 and 30,000 MW, as well as high concentrations of dissolved carbohydrate in all MW classes. Concentrations of dissolved carbohydrate in offshore waters were low, and in some MW fractions ( especially those >30,009 MW) carbohydrate was at times undetectable. Absorbances at 254 mn by MW fractions roughly paralleled concentrations of dissolved organic carbon. Most of the color of marsh, sound, nearshore, and offshore waters, once the particles were removed, was due to substances between 1,000 and 30,000 MW. All organic material in seawater that passes through a filter with a 0.8-p or 0.45-p pore size has commonly been considered dissolved. Recently attempts have been made to fractionate what should be termed filter-passing (rather than dissolved) organic substances. Ogura ( 1970) showed that some organic material in the East China Sea that passed a 0.45,~ poresize filter was retained by a 0.1-p pore-size filter, indicating the existence of particles <0.45 p in diameter. He also suggested that some organic substances that pass a 0.45-p pore-size filter may be colloidal rather than dissolved. Sharp (1973) used a 0.025-p pore-size filter and an ultrafiltration membrane with a cutoff of 50,000 molecular weight (a nominal pore size of about 0.003 ,x) to fractionate organic substances in the western North Atlantic Ocean; he found that the 0.025-0.8-p size fraction and the 0.003-0.025~ fraction each contained about 8% of the total organic material, and that about SO% of the total passed the ultrafiltration membrane. Ogura ( 1974) used ultrafiltration membranes to fractionate by molecular weight ( MW) dissolved organic carbon (DOC) at two stations in Tokyo Bay. There was a tendency for the [DOC] in the >lOO,OOO-MW 1 The research was supported in part by National Science Foundation grant DES72-01605 A02. Contribution 299 from the University of Georgia Marine Institute. LIMNOLOGY AND OCEANOGRAPHY fraction to be less than the [DOC] in the <500-MW fraction in surface water; also at least 40% of the DOC in all samples was between 500 and 100,000 MW. I have used ultrafiltration equipment similar to that of Ogura (1974) to fractionate by molecular weight organic substances in Georgia coastal water and have analyzed these fractions chemically and spectroscopically. The coast of Georgia is bounded by barrier islands. Between the mainland and the islands are salt marshes and estuaries, while on the seaward side of the islands is continental shelf water extending off shore about 100 km. Semidiurnal ebbing tides flush water out of the marsh-estuarine system through the sounds between the islands and onto the continental shelf. After some mixing on the inner shelf much of the water that flushed out of the marshes and estuaries returns on the incoming tide. In contrast to the high [DOC] and intense yellow color of the marsh water, the shelf water has a lower [DOC] and is a less intense color. Methods Samples were collected at three marsh stations, two sound stations, two nearshore stations, and three offshore stations (Table 1) . Some stations sampled in the fall were sampled again the following spring. All samples were collected at slack high tide. Marsh, sound, and 3-km-offshore stations 846 NOVEMBER 1976, V. 21( 6) Fractionation Table 1. Sample G stat ion No. 1A ik 5B 6~ ik 9F 10 11 12 13 14 15 I act Ott act May May Ott Ott 20 May 30 SCP 6 Ott 2.1 May 18 Fel, 18 Feb 20 May 21 day 3 15 16 5 7 G F H 1 I-I J and locations N Lat Date 2 20 Dates 74 74 74 75 75 74 74 75 74 74 75 75 75 75 75 jl'~3.8~ 31'23.7' 31"23.6' 31°23.8' 31°23.71 31'23.2' 31'23.1' 31'23.2' 31'20.8~ 31'21.4l 31'20.8~ 31'30.0' 31'23.0' 31'30.0' 31°15.01 of 847 sampling. Type OF station W Long 81'17.1' 81'17.2# 81°17.21 81°17.1' 81'17.2' 81'17.3' 81'17.3l 8lOl7.3' 81'14.5) 81'14.5' 81°14.51 80'57.0~ 81°06.01 80'57.0' 81°03.01 by MW 3 3 3 15 15 15 23 Marsh Marsh Marsh Marsh Marsh Sound Sound Sound km offshore km offshore km oFFshore km oFFshore Ikm offshore km offshore km oFFshore were sampled by immersing to a depth of 10 cm a precombusted (400°C) glass flask sealed with a silicone stopper. Polyethylene gloves that were prewashed with soap and water were worn. Care was taken to avoid contamination from the small boat or its engine by pointing the boat into the wind and collecting water from the bow. The stopper was released by hand, the bottle was allowed to fill, and the stopper was reinserted before the bottle was brought up through the surface to prevent collection of surface film substances. It is important to avoid collecting surface film in any sampling whero concentrations of subsurface particulate or dissolved organic substances are wanted because substances that accumulate at water-air interfaces tend to be different (frcqucntly higher molecular weight and may tend to easily coalescc into particulate material) from those that remain in solution (Wheeler 1972a,b, 1975). Samples at the- 15- and 23-km-offshore stations were collected in 1.5-liter PVC Niskin samplers at a depth of 1 m. Samples from marsh, sound, and 3-km-offshore stations wero stored at 0°C and taken immediately to the laboratory, where they were filtered through precombusted (450°C) Rcevc Angel 984-H glass fiber filters. The filtrates were then poured into the ultrafiltration cell, Samples from the 15- and 23-km-offshore stations were fil- Fig. 1. Retention of organic substances by ultrafiltration membranes UM-2 (A), PM-30 (B ), and XM-10OA ( C ) . “/o Retention = { 1 - ( [DOC] ultrafiltrate/[DOC]total )} X 100. tercd on board ship immediately after collection; the filtrates were kept at 0°C and ultrafiltercd in the laboratory on land within 4 h. An Amicon model 402 (400-ml capacity) ultrafiltration cell was used with 76-mmdiameter Amicon ultrafiltration membranes. The membranes used were (with their nominal MW cutoffs in parentheses) : WM-2 (1,000 MW), PM-30 (30,000 MW), and XM-1OOA (100,000 MW). Thcsc membranes are organic polymers (polyelectroon porous, lyte complexes ) supported spongy polyethylene bases. They were calibrated with the following organic compounds dissolved in doubly distilled water: urea (60)) dextrose ( ISO), sucrose ( 342)) raffinose ( 594)) inulin ( 5,000)) cytochromc c ( 12,384)) and bovine serum albumen (65,000) (Fig. 1). The mcmbranes were, operated at pressures of 4.0 atm ( UM-2)) 0.5 atm (PM-30), and 1.0 atm (XM-IOOA). These pressures yielded ultrafiltrates at the rates lof about 1, 2, and 2 ml min-‘. The ultrafiltration mcmbrancs are reusable but must bc handled carefully bccause their surfaces can be easily scratched. Before a membrane was used for the first time it was placed in distilled water in a 4-liter beaker and stirred with a magnetic stirring assembly for several hours, with frequent changes of water. Then it was Wheeler 848 Ultrafiltrate volume (ml) Fig. 2. Concentration of organic substances in the ultrafiltrate is in part a function of degree of concentration of sample above membrane as ultrafiltration proceeds. placed in the ultrafiltration cell, and 400 ml of distilled water was flushed through it. This procedure had the effect of reducing the leaching of carbon from the membrane into the ultrafiltrate to a level of <l mg C liter-l. The amount leached during each experiment was determined, and this amount was subtracted from the ultrafiltrate values of standards and samples. I found that the amount of organic material that passed a membrane was a function not only of molecular weight, but also of the degree of concentration of the sample above the membrane (Fig. 2). Sample 9 was passed through the PM-30 membrane, and the ultrafiltrate was collected in 50-ml increments and analyzed for DOC. After 350 ml had passed the membrane the final 50 ml was collected in two &ml fractions. The results illustrate the leakage of high MW substances through the membrane on concentration of substances above the membrane as ultrafiltration proceeded. Leakage was found with all three membranes (Ogura 1974), but not until the volume of water above the membrane was reduced by at least 50%. Therefore, each sample and standard had an initial volume in the ultrafiltration cell of 400 ml. The first 50 ml of ultrafiltrate was discarded, and the next 150 ml was collected for analysis. The remaining 200 ml above the membrane in the cell was discarded. The percentage of each standard retained by each membrane was calculated ( Fig. 1). After filtration with a glass fiber filter each sample was split into four aliquots : the first was ultrafiltered with the UM-2 membrane, the second with the PM30 membrane, and the third with the XM1OOA membrane. The fourth aliquot was not ultrafiltered, but was analyzed to provide total values for DOC, dissolved carand UV-visible absorption. bohydrate, Concentrations of substances were calculated as by Ogura ( 1974). I was concerned that any delay in ultrafiltration might result in detectable microbial utilization of organic substances, so I used the Reeve Angel 984-H filter, which is highly efficient for removing small particles from marine samples (Sheldon 1972). In a separate experiment, filtered water from samples 1 and 9 was allowed to stand at in situ temperatures for 8 h with no detectable decrease in concentrations of either total dissolved organic carbon or total dissolved carbohydrate, indicating that the 4-5 h required to ultrafilter the samples did not allow detectable microbial decomposition. This does not mean that the organic substances investigated are refractory, but only that over this short period the natural filter-passing heterotrophic population did not use a detectable amount of either total carbon or total carbohydrate. Dissolved carbohydrate was determined after Strickland and Parsons ( 1972)) dissolved organic carbon by the method of Menzel and Vaccaro ( 1964). Visible and ultraviolet absorption spectra were obtained by placing each ultrafiltrate in a IO-cm spectrophotometer cell and scanning from 700-240 nm against doubly distilled water as a reference with a Cary 14 recording spectrophotometer. Total absorption in the visible region of the spectrum (to which the Gelbstoff or yellow color of many coastal waters contributes) was integrated with a planimeter into units of square inches of spectrophotometer chart paper, and the percentage contribution of Fractionation 400 I 500 Wavelength (nm) I 300 I 600 700 Fig. 3. Fractionation of visible and ncar-ultraviolet absorption spectrum of sample 7. Initial glass fiber filtered water sample (a); XM-100A ultrafiltrate, which contains substances <lOO,OOO MW (b ) ; PM-30 ultrafiltrate, which contains substances <30,000 MW (c); and UM-2 ultrafiltrate, which contains substances <l,OOO MW (d). each MW fraction to the total visible absorption (380-700 nm) of each sample was calculated. Absorbance at 254 nm, a wavelength sometimes used to estimate concentrations of dissolved organic carbon in seawater (Mattson et al. 1974), was also noted. Resdts and discussion It can be seen in Fig. 1 that the membranes used do not retain 100% of substances with a MW above a precise cutoff and 0% of substances below it. Molecular shape is important in determining whether a molecule will pass a given membrane, a long narrow molecule having more tendency to pass than a spherical one of the same MW (Amicon Corp. 1972). I-Iowever, inulin (a linear chain of about 30 p1:2-linked fructofuranosc units with a MW of about 5,000) was retained 100% by the UM-2 membrane, indicating that the reported results serve as a guide to molecular weight despite retention differences due to shape. Figure 3 illustrates spectrophotometer tracings from sample 7. Absorption in the visible and near-UV regions illustrated is by MW 849 due to organic substances ( Kalle 1966)) and thus the spectra illustrate the fractionation by MW of these in the sample. The greatest absorption in both the visible and UV regions was due to substances in the l,OOO-30,000-MW fraction. This fraction was responsible for the greatest absorption in all samples except sample 9, where the >lOO,OOO-MW fraction predominated; sample 91also had a high [DOC] in the latter fraction. Figure 4 shows the percentage contribution of each fraction to the total visible absorption of the filtered water. Despite large variations in total DOC and in the intensity of color, the l,OOO-30,000-MW fraction was responsible for at least 50% of the total visible absorption in all except sample 9. The <l,OOO-MW fraction consistcntly contributed less to the absorption than the l,OOO-30,000-MW fraction. The abundance of yellow substances (Gelbstoff) in the >l,OOO-MW range was vividly illustrated during ultrafiltration with the UM-2 membrane of water from the marsh, sound, and 3-km-offshore stations: the ultrafiltrate was colorless, while the initially very pale yellow color of the water above the membrane became more intense as ultrafiltration proceeded and substances of higher molecular weight were concentratcd. Water above the UM-2 membrane from the stations 15 and 23-km offshore apparently remained colorless after concentration because color-forming substances were present initially in very small amounts. Figure 5 illustrates the absorbance at 254 nm for each MW fraction of each sample. These absorbances have a distribution similar to [DOC], but [DOC]/Abs254 is not constant between samples or within MW fractions of given samples, indicating differences in extinction at 254 nm for different organic substances. Results of the chemical analyses are illustrated in Figs. 6 and 7. As shown by the absorption data much of the organic material at the marsh and sound stations was in the l,OOO-30,000-MW fraction. High concentrations of carbohydrate greater than abiout 30,000 MW were found only in 850 Wheeler I MARSH n I5 and 23 km ’ OFFSHORE 1 SOUND 4 !i f- 60 50 J J&&...; I 3 5 7 9 II I3 I5 MOLECULAR WEIGHT Fig. 4. Percentage contributions of MW fractions to visible absorption due to filter-passing substances. This absorption plus scattering and absorption by particulates is responsible for color of the water. [Leftmost group should read <l,OOO; rightmost group should read >lOO,OOO.] c I * l.50- : l.25- 2 l.OO- E 075- 3 0.500.25 - l il n NO. I 3 5 7 9 ll 13 15 1000 l 3 5 7 9 tl IOOO- 30,000 MOLECULAR Fig. 5. Absorbance as in Fig. 4. [Leftmost 13 15 1 3 5 7 9’: 30,000-100,000 13 15 l 3 5 7 9 II 13 15 100,000 WEIGHT at 254 nm by MW fractions of organic substances, group should read <l,OOO; rightmost group should lO-cm cell. Samples read >lOO,OOO.] same Fractionation 851 by MW 5- =3Ez of - a O2- L I NO. < 1000 IOOO-30,000 30,000- Distribution by MW of dissolved organic the marsh. The marshes are dominated by Spartina alterniflora Loisel., and the high MW carbohydrates in the marsh may be breakdown products of its cellulose cell walls. Whatever their origin, high MW > 100,000 WEIGHT MOLECULAR Fig. 6. 100,000 carbon. Samples same as in Fig. 4. carbohydrates may be removed from the water (possibly through lysis into smaller molecules by extracellular enzymes) rather than transported out of the marsh; alternatively, much of the decrease in the con- 040 F 0.36 .c, T 0.32 E 0.28 7 5 r iif 0.24 - 4 020- W s E g O.l60.12- mo 5 0.08 u 8 88 --8 88 0.04- L_.NO. I 3 5 7 9 II I3 I5 < 1000 I 3 5 I 3 5 7 9 II I3 I5 IO00 - 30,000 MOLECULAR Fig. 7. Distribution by MW of dissolved carbohydrate 30,000-100,000 I 3 5 7 9 II I3 15 > 100,000 WEIGHT carbon. Samples same as in Fig. 4. 852 Wheeler centration of high MW carbohydrate between marsh and sound could be due to dilution of these substances in the larger volume of sound water. Seaward from the marsh stations there was a tendency for concentrations of DOC and dissolved carbohydrate carbon (DCC ) to decrease. Much of the decrease in DOC was in the 1,000-30,000-MW substances; high concentrations of these substances were peculiar to the marsh water. The offshore samples were variable with respect to which MW class contained the most DOC. The concentration of DCC decreased in all MW classes, and at times there was no detectable DCC in the >30,000-MW classes seaward from the sounds. This study illustrates the usefulness of molecular filtration in studying the organic chemistry of coastal water and in determining regional differences. Fractionation by MW can be a useful tool for studying such diverse phenomena as the chelation of metals, and the utilization of organic matter as food by suspended heterotrophs, as well as for concentrating high MW substances such as free enzymes from seawater. References AMICON CORP. 1972. ington, Mass. Applications manual. Lex- KALLE, K. 1966. The problem of the Gelbstoff in the sea. Oceanogr. Mar. Biol. Annu. Rev. 4: 91-104. MATTSON, J. S., C. A. SMITH, T. T. JONES, S. M. GERCHAKOV, AND B. D. EPSTEIN. 1974. Continuous monitoring of dissolved organic matter by UV-visible photometry. Limnol. Oceanogr. 19: 536-535. MENZEL, D. W., AND R. F. VACCARO. 1964. The measurement of dissolved organic and particulate carbon in seawater. Limnol. Oceanogr. 9 : 138-142. OGURA, N. 1970. On the presence of 0.1-0.5 F dissolved organic matter in seawater. Limnol. Oceanogr. 15: 476-479. -. 1974. Molecular weight fractionation of dissolved organic matter in coastal seawater by ultrafiltration. Mar. Biol. 24: 305312. SHARP, J. H. 1973. Size classes of organic carbon in seawater. Limnol. Oceanogr. 18: 441-447. SHELDON, R. W. 1972. Size separation of marine seston by membrane and glass-fiber filters. Limnol. Oceanogr. 17: 494-498. STRICKLAND, J. D. H., AND T. R. PARSONS. 1972. A practical handbook of seawater analysis, 2nd ed. Bull. Fish. Res. Bd. Can. 167. WHEELER, J. R. 1972a. Some effects of solar levels of ultraviolet radiation on the dissolved organic constituents of surface waters. Ph.D. thesis, Dalhousie Univ. 111 p. -. 1972b. Some effects of solar levels of ultraviolet radiation on lipids in artificial seawater. J. Geophys. Res. 77: 5302-5306. -. 1975. Formation and collapse of surface films. Limnol. Oceanogr. 20: 33S342. Submitted: Accepted: 8 #September 1975 27 April 1976