Chesapeake Science Vol. 15. No. 4. p. 185-204 December, 1974 Growth and Dissipation of Phytoplankton in Chesapeake Bay. II. A Statistical Analysis of Phytoplankton Standing Crops in the Rhode and West Rivers and an Adjacent Section of the Chesapeake Bay1 H. H. SELIGER and M. E. LOFTUS McCoNum-Pratt Institute and Department of Biology The Johns Hopkins University Baltimore, Maryland 21218 ABSTRACT: It is possible to make statistically significant comparative measurements of similar sec- tions of subestuaries under conditions where the large natural variations would mask all but drastic changes in the systems if they were studied individually. The comparative study is proposed as a moditica- tion to the baseline study of a single system for the assessment of the effects of man?s activities in an estu- ary. We have made temporally coincident measurements of phytoplankton production, standing crops and a range of physical and chemical parameters in comparable sections of the Rhode and West rivers and in an adjacent section of the Chesapeake Bay for the 3-year period 197fL1972. We analyzed the data for standing crops and demonstrated that at least at the trophic level of phytoplankton, the judicious application of a paired comparative sampling protocol to the Rhode and West rivers is superior to a study of either system alone. We calculate that the paired comparison sampling protocol requires ap- proximately one tenth the sample size of the single-system sampling technique to achieve the same sig- nificance level. Introduction The ultimate goal of many estuarine studies is to identify relationships among the factors affecting the viability of the aquatic biota, particularly those species of eco- nomic, recreational or aesthetic importance, in order to furnish a basis for most efficient utilization of the system. Suggestions can be made relative to the management of a ?Contribution No. 775 of the McCollum-Pratt In- stitute, The Johns Hopkins University. Research sup- ported by U. S. Atomic Energy Commission Contract AT(I l-l)3278 and National Science Foundation Grant GI-321 IO. The authors would like to thank Mrs. Cath- erine Eisner who has been the mainstay of our sam- pling program for her devoted efforts in helping us to collect these data. Thanks are due to Dr. R. Ballentine for suggestions and comments on the manuscript. portion of an ecosystem, i.e., whether or not to direct a chlorinated waste water dis- charge directly into a spawning area, or to place a cooling water intake in a nursery area. Very often this general knowledge of life-cycle relationships and psysiology can provide the proper advice and so avoid catastrophic consequences. However, where the cause and effect relationship is not so evident or when there is a set of complex trophic level interactions, a predictive model does not yet exist. This is due in part to the complexity of the life cycles of the preda- tors of major importance (shellfish, finfish, crabs), in part to the complexity of the trophic interactions among all of the pelagic and benthic species, and in part to the large experimental variances of the natural sys- tems, daily, seasonal and annual. 185 188 H. H. Seliger and M. E. Loftus A major concern of our research program has been to study the natural phytoplankton community in a subestuary. We assume that the quantitative relationships among nutrients and nutrient turnover, salinity, temperature, turbidity, species selection and succession, predation and exchange with the bay can be determined. From these quanti- tative relationships it follows that specific parameters will emerge which can serve as diagnostic indicators of the physiological state and of the previous history and permit the prognosis of the stability of the phyto- plankton community. These relationships should permit the prediction of the direc- tion of changes in the community in re- sponse to proposed nutrient, sediment or heat loading. In the study of any natural system the experimenter may remove samples for study in the laboratory under controlled condi- tions. However, the natural system, with diverse community population is, at any time, the integral of all of the aperiodic climatic, biotic and chemical interactions that have occurred. Thus, experimental re- producibility in the natural system is very difficult to achieve. We are immediately faced with the problem of how to make statistically significant measurements in this variable system. We have applied the following line of reasoning: Consider any given natural sys- tem on which measurements are to be made. The total measured variance will be com- posed of the variance associated with the ?treatment? or the man-introduced stress whose effect it is desired to assess and the large natural variation of the system due to daily, seasonal and annual fluctuations in wind, tide, sunlight, rainfall, etc. In prin- ciple therefore a ?before? and ?after? base- line study of a single system will be subject to both of these sources of uncertainty and only ?treatments? which produce suffi- ciently large mean differences (before minus after) can be assessed with any degree of statistical significance. A further complica- tion exists because statistical parameters such as S. D., tests such as Chi Square, Student?s t, F variance ratio, Chauvenet?s Criterion and levels of significance have implicit in them the assumption that the data are normally distributed about their mean value. How then are we to assign levels of significance to differences in time averages of these quantities from one sea- son to another or from one year to the next? This latter assignment is at the heart of the baseline study. It should be pos- sible to choose a second system which is comparable (similar) to the first in its re- sponse to the natural fluctuations and dif- fers from the first in the absence of the particular ?treatment?. Under these condi- tions it should be possible to analyze differ- ences between the two systems and to re- move the large natural variations from the statistical analysis. By virtue of the com- parability of the systems, the expected value of the mean of the differences, properly normalized, should be zero. Non-homo- geneities within the individual systems and their varying responses to localized mete- orological changes in addition to measure- ment error will give rise to a normally dis- tributed spread of difference values which is amenable to statistical analysis. The trick is to work with comparable systems. This is what we have done in the present paper. Our sampling protocol has included tem- porally coincident (l-2 hours) measure- ments of comparable sections of the Rhode and West rivers and of an adjacent sec- tion of the Chesapeake Bay on approxi- mately a weekly basis for a three year period, 1970-1972. We have asked the fol- lowing questions: a) Are the mean differ- ences of phytoplankton parameters mea- sured in the Rhode and West rivers statis- tically significant (95% level of significance, Student?s t)? b) How might it be possible to relate such statistically significant differ- ences to man?s activities in both systems? c) Might the technique be applied to follow more precisely an annual or seasonal trend in the change of one system with respect to another? d) By how much must the param- eters in either of these subestuaries change, presumably as the result of a hypothetical perturbation to one system, in order that the mean differences may be considered statistically significant (95% level of sig- nificance, Student?s t)? We have analyzed the data both by the comparative technique and by the baseline technique. Growth and Dissipation of Phytoplankton in Chesapeake Bay 187 Description of the Area The Rhode and West rivers are small tributary estuaries with a common mouth, which enter the western Chesapeake Bay approximately 5 miles south of Annapolis and the Severn River. The Rhode and West river transects (1 and 2) and the adjacent bay transect (3) are shown in Fig. 1 and have been referred to previously (Loftus et al. 1972; Seliger 1972). Table 1 shows the volume and surface area data at mean low water for the Rhode and West rivers, ab- stracted from Pritchard and Han (1972). The sections RRl, 2, 3, WRl, 2, 3 re- ferred to in Table 1 and shown in Fig. 1 correspond to those used by Pritchard and Han (pers. commun.) in their preliminary model for exchange rate constants for waters in these subestuaries. From inspection, and on the basis of sa- linity transects and phytoplankton sampled in the Rhode and West rivers and in the Chesapeake Bay, we decided that transects 1 and 2 represented approximately equiva- lent sections of each river, and that transect 3 was representative of the main portion of the Chesapeake Bay adjacent to these rivers. There were several decided advantages accruing to us by virtue of our location at the Rhode River. 1) Approximately 2500 acres of the watershed surrounding Muddy Creek, the main tributary creek of Rhode River, are conserved by the Smithsonian Institution as the Chesapeake Bay Center for Environmental Studies (CBCES). This consideration together with a relatively low population density in the Rhode River watershed made the Rhode River the least disturbed subestuary on the western shore of the Chesapeake Bay. 2) The phyto- plankton study reported here is a part of an interdisciplinary research program on the entire Rhode River watershed-estuarine sys- tem (Anon. 1973:Vol. IV). 3) The Rhode and West rivers form a common mouth emptying into the bay. They have similar TABLE 1. Volumes and surface areas of Rhode and West rivers at mean low water. Segment Pritchard and Surface Area Volume Mean Depth Han Section (106m2) ( 106m3) (m) Cadle Creek Bear Neck Creek B Whitemarsh Creek W Sellman Creek Muddy Creek Total Rhode River Cheston Creek Scaffold Creek Popham Creek Cox Creek Tarthouse Creek Learch Creek Smith Creek Johns Creek South Creek Total West River 9.14 13.08 1.34 Rhode River C 0.20 RR1 1.65 S 0.25 RR2 2.38 RR3 0.81 5.9 West River PSC 1.06 WRl 2.90 CT 0.92 WR2 2.10 WR3 2.16 3.31 1.22 0.29 1.5 3.78 2.29 0.93 0.41 5.09 0.97 11.47 1.06 1.0 4.25 1.47 1.12 1.22 3.28 1.56 1.52 1.65 2.13 1.20 1.94 188 Ii. H. Seliger and M. E. Loftus Fig. 1. Chart of Rhode and West rivers showing transects 1, 2, 3 and 4 as thick dashed lines. The sec- tions designated by Pritchard and Han (1972) in their model for exchange rate constants are delineated by the dotted lines. volumes and essentially form a common system to which nutrients and plankton are delivered by the adjacent bay as the result of exchange due to tidal and density flows. The bay waters entering these rivers are modified as a result of the physical charac- teristics of these shallow basins, the nature of the bottom sediments, the delivery of nutrients in fresh water runoff from the uplands, the tidal flushing of the marshes, and the additional effluents due to the hu- man population of the watershed and shore- lines. The Rhode and West rivers together with their watersheds are comparable sys- tems, dominated by the exchange with bay water. We have not yet progressed to the stage of introducing our own experimental ?treatments?. We therefore limit the scope of the present paper to asking whether any significant differences among the transects might be explained by any of the factors listed above. The Rhode River is a special case of an estuary with two-layer flow; strong vertical mixing (Bowden 1967) due to tidal currents and wind gives rise to vertical and lateral homogeneity. The vertical salinity profiles show monotonic increases from top to close- to-bottom of cu. 0.1 to 0.3ofoo with in- creases in the bottom 0.5-l m of 0.5-l o/00. There is a very small gradient of surface water salinity between the mouth of the river and the mouth of Muddy Creek except following a period of heavy rainfall in the watershed. The upper limit of the Rhode River subestuary, where the salinity ap- proaches O.lo/oo and the chlorinity:total dis- solved solids approaches 1:lO to 1:20 (Pritchard 1967a, b), extends a significant distance up into Muddy Creek, the major tributary creek. This is the case for all of the tributary creeks of the Rhode and West rivers. Because of the small volume of the tidal section of Rhode River, rainfall produces relatively large excursions in the upper limit of the estuary. The tidal section of the subestuary encompasses essentially the re- mainder of Muddy Creek. There is a negli- gible ?river section? associated with the Rhode River. The land runoff into Rhode River consists of drainage directly into the tidal section of the subestuary. Winds play an important part in maintaining the well- mixed essentially isohaline character of this shallow subestuary. Under proper condi- tions, a strong northwest wind will rapidly exchange the estuary section of the Rhode River and its plankton populations with the bay. The delivery of nutrients to the estuary sections of the Rhode and West rivers is the result of tidal action (flushing of the marshes, remixing of soluble nutrients from interstitial water, and resuspension of in- terstitial sediments) and exchange with the Chesapeake Bay across the mouths of the rivers. However, subsequent to heavy rains the transition zone is subject to major changes in phytoplankton relative species compositions, coinciding with large increases in standing crops of chlorophyll a. Methodology SAMPLING PROGRAM AND PARAMETERS The parameters measured in our sam- pling program are listed in Table 2 together with the techniques used for each measure- ment, the literature references and the co- efficients of variation (C. V.) or the experi- mental standard deviations (S. D.). Subse- quent to July, 1972, when it was desired to exclude small dinoflagellates such as Proro- centrum minimum and Exuviella sp. from the ?nannoplankton? filtrate, 10 micron net- ting was used for nannoplankton filtration. We have found that 20 micron netting is suf- ficient to remove all predators including tin- tinnids, rotifers, copepod nauplii and veliger larvae when it is desired to examine short- term effects of additions of specific nutrients on rates of primary production. Just prior to Hurricane Agnes (June 21, 1972) a fourth transect was established at the mouth of Muddy Creek in Rhode River (Fig. 1). tainty of 4%. It is always the combined ex- perimental S. D. or C. V. which is reported. AVERAGE OF INTEGRATED TRANSECTS Water samples were delivered to 5-gallon translucent polyethylene carboys on deck by a peristaltic pump. As the boat proceeded (approx. 2 knots) at constant speed along a transect a vertical tube was raised and low- ered at a uniform rate between the surface and the depth of disappearance of a Secchi disc. The water sample collected was desig- nated as the whole integrated or ?A? sample. dinoflagellate species found in this area of the Chesapeake Bay. However since the major phytoplankton biomass, chlorophyll pigments and primary production in Chesa- peake Bay are due to phytoplankton which can pass through a 20 micron net (Seliger 1972; Loftus et al. 1972; McCarthy et al. 1974) and whose classification has not been determined, we have used this crude size filtration and extractable chlorophyll pig- ments to delineate the nannoplankton. The technique used and its limitations are described in Loftus et al. (1972). Early in 1972, continuous in viva chloro- phyll fluorescence records were made at 0.5 m and at 1.5 m depths along the regular transects. We were able to demonstrate that the upward and downward motion of the boat with the hose input fixed relative to the boat was equivalent to the manual raising and lowering of the sampling tube. Samples were therefore collected at a fixed ?still- water? depth of 0.5 m. The addition of the in vivo chlorophyll fluorescence continuous chart records to the sampling protocol pro- vided the further advantage of following the occasional patchiness of the surface waters caused by the differential phototactic migra- tion of the larger dinoflagellate species. The storage, retrieval, treatment and plot- ting of data were made compatible with the Hewlett Packard 9800 Series Programmable Calculator. The yearly data were stored on magnetic cards in pairs consisting of the day of the year and the measured or calculated value of the parameter. Programs were written to operate on as well as to print out the data or to plot the re- sults on the 9800 Series Plotter, either for in- dividual years or for a multiple year inter- val. The abscissa is the day of the year. EXPERIMENTAL STANDARD DEVIATIONS There are irreducible variations between our so-called comparable sections of the rivers due to short-term differential effects of local climate. One would therefore expect to observe a high frequency (t = days) jitter superimposed on the seasonal variations of parameters. We integrated this high fre- quency component by means of a smoothing program for drawing lines in the plotted data. A smoothed curve between data ordi- nate i, corresponding for example to day 120 and data ordinate i + 1, corresponding to day 127 is actually a line drawn between the ordinate. The coefficients of variation and the standard deviations shown in Table 2 are the combined uncertainties due to instrumental variance and the variance of repetitive field sampling. For example, in the case of the determination of dissolved inorganic carbon concentrations the C. V. of repetitive anal- yses of a single sample was 4%. However the C. V. using field samples was found to be 670, implying an additional sampling uncer- l/4 [(i - 1) + 2i + (i + l)] for day 120 and the ordinate Growth and Dissipation of Phytoplankton in Chesapeake Bay 189 PHYTOPLANKTON SPECIES IDENTIFICATION We have been able to identify the larger In Viva FLUORESCENCE DATA HANDLING l/4 [i + 2(i + 1) + (i + 2)] TABLE 2. Parameters in sampling program. c. v. Parameter Measurement Method of Measurement Units Reference so;. 00 Turbidity Secchi disc disappearance m-l 01 Salinity Beckman induction salinometer o/o0 02 Temperature Thermistor probe of salinometer ?C 03 Dissolved oxygen Yellow Springs Model 5 1 O2 probe mg mm3 04 Inorganic carbon Beckman carbon analyzer mgmm3 04 Inorganic carbon Alkalinity titration mg me3 05 Dissolved NO,- Optical density (543 nm) fig atom liter-? 06 Dissolved NO,- Cd reduction, optical density (543 nm) fig atom liter-? 07 Dissolved NH,+ Optical density (640 nm) fig atom liter-? 08 Dissolved PO,- - Optical density (883 nm) pg atom liter-? 09 Total dissolved N Ultraviolet oxidation to NO, Gg atom liter- I 10 Total dissolved P Ultraviolet oxidation to PO1 pg atom liter-? 11 Extractable chlorophyll a 90% Acetone extraction; fluorometry fig liter- ? 12 Extractable chlorophyll b 90% Acetone extraction; fluorometry fig liter- 1 13 Extractable chlorophyll c 90% Acetone extraction; fluorometry fig liter- ? 14 Extractable pheophytin a 90% Acetone extraction; fluorometry Gg liter- i 15 Rate of carbon uptake 14C bicarbonate tracer technique KgC literr?hr-? 16 Gross 0, evolution Winkler technique (modified) WgC liter-? hr-? 17 Rate of respiration Winkler technique (modified) PgC liter-? hr-? 18 Stimulable bioluminescence Mechanical stirrer [lo* photons] 19 20 21 22 23 Rainfall at Rhode River Hour of day at high water slack Surface sunlight intensity during incubation PH Rain gauge L: over 5 day intervals mm per 5 days Chlorophyll a from in viva fluorescence Derived parameters Assimilation rates Ratio Chl c/Chl a Ratio Pheo a/Chl a Dissolved inorganic N Dissolved organic N Dissolved oreanic P Ratio Chl a ; 20 p Chl a whole sample Eppley pyroheliometer (integral/hr) foot candle meter pH probe Continuous flow fluorometer Ly hrr? fig liter 1 Loftus et al. 1972; Lorenzen 1966 20% 30-39 30 31 32 33 34 35 (15)/(11) (13)/(11) (14)/(11) 2 (05) (06), (07) (09) - (33) (lo) - (08) -1 mgC mg Chl a-? hr- pg atom liter-? pg atom liter-? pg atom liter-? 36 Benshneider and Robinson 1952 Morris and Riley 1963 Solorzano 1969 Murphy and Riley 1962 Strickland and Parsons 1968 Strickland and Parsons 1968 Loftus and Carpenter 1971 Loftus and Carpenter 197 1 Loftus and Carpenter 197 1 Loftus and Carpenter 197 I Steemann Nielsen 1952 Carpenter 1965 Carpenter 1965 Seliger and McElroy 1968; Biggley et al. 1969 Yentsch and Lee 1966 Humphrey 1963 Beers and Stewart 1969 10% 0.2 of 00 0.2OC 5% 6% 6% 5% 7% 10% 4% 8% 7% - 10% - 6% 20% -12% Growth and Dissipation of Phytoplankton in Chesapeake Bay 191 for day 127, where (i - 1) represents the parameter value for the week prior to day 120 and i + 2 represents the value for the week subsequent to day 127. For the end points (1, . . . n) the smooth ordinates are % [2 datum 1 + datum 21 and % [datum (n - 1) + 2 datum n] respectively. Provision in the plotting pro- gram was made so that solid lines were not drawn between parameter values separated by more than two weeks. We drew dashed lines in these cases to extrapolate the trend and to indicate the absence of specific data. A smoothed plot consists of the actual data points through which the smoothed lines were drawn and is so described in the cap- tion. In some cases, the lines were drawn through the data points directly. Results and Discussion PHYSICAL CHARACTERISTICS The salinities and temperatures of surface waters in the Rhode River (transect 1) over the period 1969 through 1972 are shown in Figs. 2 and 3, respectively. Similar data were obtained for both West River (transect 2) and the Chesapeake Bay (transect 3) over the period 1970 through 1972. The data of Fig. 2 indicate the aperiodic component of the sea- sonal and annual salinity patterns in this area during the past four years. However, general features can be observed. There is a sharp spring drop, a more gradual rise beginning in May and reaching a maximum (in 1969 as high as 16?/,,) around October. The curves of Fig. 2 indicate the increasingly larger spring flows of the Susquehanna in 1970, 1971 and 1972, resulting in successively deeper troughs. The effect of Hurricane Agnes, subsequent to June 21, 1972 is seen as a drop from approxi- mately 7?/,, to 2?/,,, followed by a recovery during July and August. A more intensive time sequence of the salinity just prior to Agnes and extending to November, 1972 is shown in Fig. 4 (A. Place, pers. commun.), for the mouth of Muddy Creek (transect 4, see Fig. 1). The major troughs in Fig. 2 are the result of the delivery of spring runoff by the Sus- 192 H. H. Seliger and M. E. Loftus 2o.m T I RtlODE RIVER I _ :.:.::I ::::::: ,963. I 1970. . . . . . . . . ..I..:::___ ,971 I ,972. Fig. 2. Surface water salinities (O.lm) at the begin- ning of transect 1, designated by 1 1, in Rhode River for the period 1969 through 1972. The solid dots are true data points. The solid lines are smoothed curves as described in the text. The dashed lines are extrap- olated and are shown between data points separated in time by more than 2 weeks Fig. 3. Surface water temperatures (O.lm) at the beginning of transect 1, designated by 1 1, in Rhode River for the period 1969 through 1972. The solid dots are true data points. The small vertical bars represent the estimated standard error. The solid lines are smoothed curves as described in the text. The dashed lines are extrapolated and are drawn between data points separated in time by more than 2 weeks. quehanna River. The smaller oscillations are the result of a combination of the above with the effects of local winds and rainfalls. We have referred to the effect that a strong northwest wind can have on the exchange of Rhode River water. As can be seen from Fig. 1, a northwest wind will not affect the West River in the same way. It is possible therefore that under certain local wind conditions, even though phytoplankton growth conditions in both rivers were optimal and would lead to high standing crops, one or the other would exchange its waters more rapidly with the bay and not reflect this increase. These local and temporary differences in exchange rates are part of the irreducible experimental variance when standing crops are measured by the paired comparison technique. A second source of variance occurs because of the finite time required for the system to approach a new steady state as the result of a change in any part. Fig. 5 shows the rainfall in units of cm per 5 day interval at the CBCES. In this section and the following we shall use the abbreviations RR, WR and CB for Rhode River, West River and Chesapeake Bay, respectively. In February, 1972 there were heavy local rains in this area resulting in a marked lowering of the RR and WR surface water salinities in March. This can be seen by comparing Fig. 6, in which RR J *- g _ 3 _ $4- 0 I I I I I I JUNE JULY AUGUST SEPT OCT NO? 1972 Fig. 4. Surface water salinities (O.lm) for transect 4 (mouth of Muddy Creek) prior, during and subsequent to Hurricane Agnes. 300 T RHODE RIVER 2Li.m RAINFALL Fig. 5. Rainfall in Rhode River area over the time period 196991972 shown as sum over 5 day intervals (cm/5 days). Growth and Dissipation of Phytoplankton in Chesapeake Bay 193 20 r 16 3 16 i WEST RIVER ( - ) 14 RHODE RIVER ( - j :: 2 12 i ;: IO J FMAMJJASONDJ MONTH 1972 Fig. 6. Smoothed curves of surface water salinities of transect 1 in Rhode River (thin line) compared with those of transect 2 in West River (thick line), showing very close agreement. surface water salinities for 1972 are superim- posed upon the same parameter for WR with Fig. 7 where RR is compared with CB. In Fig. 6, the similarity between RR and WR salinities is quite evident, while in Fig. 7 differences between RR and CB occurred during March-April. The means of t&e paired salinity difference measurements, AS(RR-WR), for 1970, 1971 and 1972 were 0.2?/,, (*; n = 20), O.l?/,, (NS; n = 29) and -0.2O/,, (*; n = 26), respectively, where the asterisk indicates a P 5 0.05 level of significance, based on the Student?s t test of means to be described in the section on standing crops of chlorophyll. The symbol NS indicates ?not significant?, a P > 0.05 level of significance. We ordinarily take P < 0.05 as the cutoff for assigning statistical significance. F variance ratio tests between the 3-year combined data and any single year were not significant. For the complete set of data, therefore, s(RR- WR)3 Years = O.O2O/,,,, (NS; n = 75) with a standard deviation S. D. = 0.5O/,, and a standard error of the mean of 0.06?/,,. The large natural variations in surface water salinities in the Rhode and West rivers are temporally coincident in transects 1 and 2. Therefore these transects represent compa- rable salinity sections. The temperatures in the subestuaries are more periodic than the salinities. From Fig. 3, except for the slight fluctuations, the surface water temperatures at any day x in the Rhode River can be fitted approximately by the relation 7r(x - 30) T = 28 sin2__ 365 (1) The means of the paired temperature dif- ference measurements AT (RR-WR) for 1970, 1971 and 1972 were 0.5 C (NS; n = 20), -0.1 C (NS; n = 30) and 0.1 C (NS; n = 25), respectively. The pooled 3-year data gave fi (RR-WR)? year8 = 0.1 C (NS; n = 75) with a standard deviation, S. D. = 0.8 C and a standard error of the mean, 0.09 C. The data indicate that temporally coinci- dent measurements of salinity and tempera- ture in comparable sections of the two rivers show no differences or trends in salinity or temperature between the two rivers. The rivers are approximately the same depth and the exchange of .both mesohaline sections with the bay is approximately the same. This has been confirmed recently by Pritchard and Han (pers. commun.) on the basis of a preliminary model of the sections shown in Fig. 1. It would therefore be expected that the similar phytoplankton communities in the Rhode and West rivers would respond simi- larly to temperature and salinity changes. Since the bay exerts the major influence on the rivers, salinity differences between the 20 16 RHODE RIVER (--_) CHESAPEAKE BAY (- 1 J F M A M J JASONOJ MONTH 1972 Fig. 7. Smoothed curves of surface water salinities of transect 1 in Rhode River (thin line) compared with those of transect 3 in Chesapeake Bay (thick line) showing the large spring dilution of the river water as the result of local rainfall. 194 H. H. Seliger and M. E. Loftus surface waters of the rivers and the bay will largely reflect phase differences due to finite exchange times and only occasionally will be due to heavy local rainfalls. However the shallow and more protected rivers will heat up faster than the bay. As shown in Fig. 8, the river is significantly warmer than the bay for a major portion of the spring and summer. The temperature difference is relatively greater in the early spring. One might expect, therefore, that the differential effects of inso- lation would be manifested as an early spring increase in production rate in the rivers, leading to an earlier increase in the standing crop of chlorophyll relative to the bay. Fig. 9 shows smoothed curves for extractable chlo- rophyll a for both RR and CB for 1971. In 1971 we did not observe the ?usual? spring pulse of production and standing crops in the bay which occurred in 1969, 1970, 1972 and very markedly in 1973. However, what we shall call the spring insolation increase in RR was very marked in February and March as shown in the figure. The seasonal flow patterns of surface wa- ters in the two rivers can be inferred from the summary data on the differences in surface water densities, aT, between RR and CB and between WR and CB, in Fig. 10. These plots of Aa, over 3 survey years show that from March to December the river surface waters tend to overlay the adjacent bay surface water on most survey dates. rile! 32.m ! RHODE RIVER (-1 CHESAPEAKE BAY (-1 Fig. 8. Surface water temperatures at the begin- ning of transect 1 in Rhode River (thin line) com- pared with temperatures at the end of transect 3 in Chesapeake Bay (thick line) for the time period 1969- 1972. q!s. RHODE RIVER ( -_) -0 5 35. CHESAPEAKE BAY (---_) c- < 0 30 Fig. 9. Extractable chlorophyll a in transect 1 of Rhode River (thin line) as compared with the transect 3 of Chesapeake Bay (thick line) during 1971, showing a spring insolation increase in the river during February and March, 1971. These surface density gradients imply a two- layer flow also evident in vertical profiles of temperature and salinity. Since similar pat- terns in Au, existed between the bay surface waters and surface waters of both rivers, both would receive phytoplankton inocula from the same common source, via transport of near-surface bay water to deeper layers in the rivers. During sunlight hours these surface waters may contain relatively higher concen- trations of vertically-migrating, positively- phototactic dinoflagellate species such as Gymnodinium nelsoni and Prorocentrum minimum. In these cases, there will be an increased delivery of these dinoflagellates to the rivers relative to the smaller nannoplank- ton which do not migrate appreciably and are therefore more uniformly distributed in the water column. The density discontinuity in the rivers occurs near 2 m depth at below 1% surface light. The .nannoplankton delivered to the river at this depth must depend upon vertical mixing to reach optimum light levels for photosynthesis. Again the positively photo- tactic dinoflagellates would appear to have an advantage over the nannoplankton in being able to migrate to surface waters. Therefore density driven exchange between the bay and the rivers can promote the delivery and retention and even occasional dominance of the larger dinoflagellates in the rivers. The larger dinoflagellates are absent in the winter Growth and Dissipation of Phytoplankton in Chesapeake Bay 195 3.0 1.5 4.5 c 4 : 3.0 E ( wn-CE) I 1.5 0 - 1.5 . li_AL . . . 1 9 . 0% . . . . F I ?0 .-.L;I; . . .* f l -3.0 t ? I I97Lzi ,971. Fig. 10. Differences in surface water densities be- tween Rhode River and Chesapeake Bay (top) and West River and Chesapeake Bay (bottom), suggesting a two-layer flow. months when little or no density gradient exists. LIGHT INTENSITIES The photic zone in the subestuaries is quite limited in depth. The depth of visual disap- pearance of a Secchi disc, corresponding to two tenths2 of the surface sunlight intensity, varies during the year between 0.5 m and 1.5 m. There is a general pattern of increases in 2 The assumption is made that the 1% light in- tensity level is equal to three times the Secchi disc depth. absorption during spring and fall. Turbidities in Rhode River and West River showed a strong overlap and were consistently higher than in the bay. This may be the result of runoff or marsh flushing in which case the compositions of sediment and detrital suspen- sions could conceivably differ from those in the bay. Alternatively, the increased river turbidities could be due to the higher steady state resuspension of sediments by tidal ac- tion in the shallower river bottoms. This estimation of absorption by Secchi disc is at best a crude one. Since the method is a visual one, it measures the relative absorp- tion of a range of wavelengths for human photopic vision. The peak sensitivity for photopic vision, 555 nm, lies between the absorption peaks for chlorophyll and most of the blue-absorbing accessory pigments. One must therefore make the tenuous assumption that absorption coefficients do not change with depth or with sediment load. We have just begun a study of the spectral distributions of underwater sunlight in estuarine waters using a 4 T diffuser-detector in combination with a pressurized underwater spectrometer designed by W. G. Fastie (see Seliger and Fastie 1968). In Fig. 11 we show the relative spectral distribution of sunlight at the water surface and at a depth of 0.8 m corresponding on that day to the depth of disappearance of a Secchi disc. The surface sunlight intensity was 0.55 Langleys per minute. These studies will be presented separately. The major significance of Fig. 11 is to demonstrate the marked contraction of the photic zone in the subestuary as compared with coastal and oceanic waters. There is also a significant distortion of the original sunlight spectrum. The mean path length for absorp- tion (true absorption and scattering) of blue light changes from 33 m in the clearest ocean waters (Jerlov 1968) to 0.5 m in the Rhode River. During the months of April through Octo- ber light appears to be a limiting factor for phytoplankton growth in both rivers. Phyto- plankton mixed in the water column spend a significant fraction of their daylight hours at light intensity below their photosynthesis sat- uration values. A decrease in sediment load should in principle result in an increase in 196 H. H. Seliger and M. E. Loftus I I I I I I II 400 500 600 700 WAVELENGTH I nm) Fig. 1 I. Relative spectral photon intensities of sun- light incident on the surface (0.55 Langleys per min- ute) of the water and at a depth of 0.8 m in the Rhode River on 27 July 1972. The intensity data are nor- malized at 710 nm. production in the water column. However, the relationships between nutrient delivery, which accompanies sediment delivery, and the direct and indirect effects of sediments (including detritus) in providing nutrients to the phytoplankton and zooplankton are not completely understood. UNIFORMITY OF TRANSECTS In general in all three transects, the surface waters which comprised the photic zone were reasonably well mixed, as evidenced by the uniformity of depth profiles of salinity. The nannoplankton were uniformly distributed throughout the photic zone and any minor taxis or flotation related to light or nutrients was insufficient to produce significant differ- ential vertical distributions. Under these con- ditions grid sampling comparisons by extrac- tion of chlorophyll pigments of surface water samples (0.5 m) and as functions of depth showed the same shallow horizontal negative gradient of phytoplankton concentrations along transect 1 and in fact all the way out to the end of transect 3 (see Fig. 1). This was verified in much more detail by continuous in vivo chlorophyll fluorescence sampling along these same transects. At intervals throughout the year inocula of the larger dinoflagellates originating in the bay grow up in significant concentrations in the saline portions of the tributary creeks and drift as patches into the main river sections. We shall discuss in a subsequent paper the summer blooms of Prorocentrum minimum which occur in this area of the bay. This phenomenon occurred subsequent to Agnes during July and August, 1972. A subsequent blooming and dissipation of the dinoflagellate Gymnodinium nelsoni in Rhode River is shown in the in vivo chlorophyll fluorescence transect records of Fig. 12. The measure- ments were made at 0.5 m depth along the axis of the river, from the beginning of transect 4 straight through to the end of transect 1, a distance of 4.5 km. The axial stations on the abscissa refer to specific buoys or landmarks. Thus transect 4 com- prises axial positions from 0 to 4; transect 1 comprises axial positions 6 through 10. It is important for continuous in vivo fluorescence measurements to parallel any plankton sam- pling program, in order to avert the large statistical variations possible in grab samples due to patchiness. The continuous, or inte- gral sampling technique acts to average out the patchiness in the area. STANDING CROPS OF PHYTO- PLANKTON The measured standing crops of total chlo- rophyll a in the Rhode River, West River, and Chesapeake Bay for the period 1969 through 1972 are shown by the solid lines in Fig. 13. In these graphs the solid lines have been drawn through the actual weekly data points. Each point represents an average value of total chlorophyll a liter-? for a complete transect. Several points can be made: 1) Except for a trend toward high standing crops in summer and low standing crops in winter there existed no apparent reproducibil- ity of standing crops of phytoplankton from one year to the next. 2) Despite the averaging involved in the individual sampling transects the weekly measurements of standing crops showed large oscillations. Variations in average phyto- plankton concentrations can be introduced as the result of sampling at different phases of the tide, since there do exist gradients of phytoplankton concentrations along the tran- sects. These variations are averaged out somewhat by virtue of the fact that the linear Growth and Dissipation of Phytoplankton in Chesapeake Bay Fig. 12. In vivo fluorescence in units of extractable chlorophyll a (j.rg liter-?) preceding (top), during (middle), and following (bottom) a bloom of Gymno- dinium nelsoni in Rhode River. Recordings were made from samples drawn through fluorometer from OSm depth by means of a peristaltic pump. The axial dis- tance from the origin to position 10 was 4.5 km. extent of each of the integrated transects in the rivers was more than twice the tidal excursion. In the determination of a hydro- graphic model of conservative quantities such WEST RIVER Fig. 13. Standing crops of total chlorophyll a (un- shaded) and the chlorophyll a contained in a size fraction greater than 20 Jo (darkened areas) for the time period 196991972 for Rhode River (top), West River (middle), and Chesapeake Bay (bottom). The blank area enclosed between the solid lines and the darkened areas represents the nannoplankton. as salinity, it is reasonable and necessary to make measurements at the same arbitrary phase of the tide. However the phases of sunlight intensities, heating and turbulent mixing due to insolation and wind mixing are solar phases. In phytoplankton sampling one must solar choose between sampling at the same time, and sampling at the same lunar 198 H. H. Seliger and M. E. Loftus time. We chose the former since sunlight is man?s activities in both river sections were the ! the primary source for free energy and regula- same. tion in phytoplankton communities. The reversal of the trend of RR - CB > 0 3) The solid black portions in the figures occurs during May-June when Prorocentrum show the contributions to the total chloro- blooms originate in the bay north of Rhode ! phyll a pigments in the surface waters, of River (Seliger 1972). In our statistical analy- dinoflagellates greater than 20 p in cross sis we have divided the time periods of the sectional linear dimension. As can be seen, RR - WR comparison into winter (day l- the major contribution to chlorophyll pig- 90) spring-summer (day 91-240) and fall ments is due to phytoplankton which pass (day 241-365). The division corresponds to through a 20 p mesh net (Seliger 1972; Loftus the general trend in Fig. 13 of winter low, et al. 1972; McCarthy et al. 1974). In July spring-summer rise and peak and fall de- and August, 1972, significant fractions of cline. In the analysis of the RR - CB data rapidly growing G. nelsoni were able to pass we have separated the times of the Proro- through a 20 p net and in these cases a 10 h centrum blooms (day 151-200) from the rest mesh size was used for filtration. The differ- of the data and therefore have winter (day ence in timing between the appearance of l-90) spring-summer minus the Bloom Pe- large dinoflagellate standing crops in RR and riod (day 91-150; 201-240) Bloom (day WR, and the later appearance in CB, can be 15 l-200), and fall (day 24 l-365). seen by examining the solid black areas of We asked whether, during any of the time Fig. 15 for July through September, 1972. In periods, the differences between RR and WR the rivers the dinoflagellate blooms began to and between RR and CB were significantly grow immediately following Agnes while in different from zero. When there are two sets the bay the blooming did not occur until of measurements xi (Rhode River) and yi August, when the river standing crops were (West River) we may ask whether x is past their peak. significantly different from y. In this case the A pattern, if any, which might be developed Student?s t is defined as from these blacked areas is a minor appear- 5-y ante of larger dinoflagellates in summer and a significant growth in the fall. The relation- ship of these dinoflagellate successions and ?=&$- (2) nutrients will be discussed in a later paper. Despite the rather chaotic appearance of where S, and S, are the respective standard the 3-year chlorophyll a standing crop data it d eviations of the measurements (x,, xZ . . . x,) has been possible to obtain some valid statis- and (yl, yZ . . . y,>. As Can be seen from Fig. tical analyses by virtue of the temporal Coin- 13 the large variations in xiand y,thrOughOut cidence of the measurements. Owing to the th e year as well as from one week to the next residence times of the phytoplankton in the will give rise to large values of S, and S,. This rivers and the increased nutrient availability in turn will require the numerator of equation we might expect that even in the presence of (2) to be large in order to be statistically exchange with the bay, the steady state stand- significant. The random method of analysis ing crops of phytoplankton, and consequent of variance therefore tends to mask true chlorophyll a concentrations, would be. higher differences which may exist between systems in the rivers than in the bay (RR ~ CB > 0). The same line of reasoning applied to the x and y (see Sokal and Rohlf, 1969: 328). By the expedient of making essentially si- RR - WR differences would predict a zero multaneous measurements of xi and yi we can average difference provided that a) the ex- ask whether change rates relative to total volume in each river section were the same, b) local effects z = ; 2 (x, - yz) (3) such as wind differences, creek contributions, 1 depth differences, etc. tended to average out, is significantly different from zero. In this and c) the chemical inputs and nutrient case the Student?s t is defined (Sokal and delivery from the marshes including all of Rohlf, 1969) as Growth and Dissipation of Phytoplankton in Chesapeake Bay 199 z-y t=-.__.._ SF (4) where CL is the parametric mean of the population with which we are concerned and S, is the standard deviation of the mean of the difference measurements. The null hypothesis asks whether the data zi belong to the popula- tion (H,:p = 0). There are two sources of variance which contribute to the total variance Si, s; = -!._- [x ZT - f (Cs,.] n-1 t5) The first can be defined as the result of a large number of repetitive measurements made on x and y separately. The apriori variance of zi, assuming only measurement error, is s2, = a,? + u; = 2 a,2 (6) where the g?s are used to indicate the stan- dard deviations of repetitive measurements as we have discussed above. The second experi- mental variance is due to the patchiness of the systems, caused by local variations in mixing and in exchange rates caused by the wind, blooms of organisms developing in some creeks and not in others, local rainfalls, etc. The data on standing crops of chlorophyll a in RR, WR and CB were analyzed in the following ways: To conform with equation (3) we let zi = RR, - WRi in one analysis and zj = RR, - CB, in the second. In order to give equal relative weight to periods of the year when standing crops were low we also analyzed the relative differences zi-(RR1 - WRi)/(.RRi) and zj = (RR, - CBj)/(RRj). The summary of the paired variate statisti- cal analyses of the chlorophyll a standing crop data for 1970, 1971 and 1972 is given in Table 3 for both the relative differences (RR - WR)/RR, (RR - CB)/(RR) and the absolute differences RR ~ WR, RR - CB. The years have been subdivided into winter, spring-summer and fall periods in an attempt to detect seasonal patterns. There are several results from Table 3 which can be summarized. a) The standing crops of chlorophyll a in West River appear to be consistently higher than those in Rhode River. However only in spring-summer, 1972 did the mean differ- ences and the mean relative differences reach the 5% significance level. If we assume a normal distribution of error for the paired comparisons (RR, - WRi), we obtain, for 1972: z=-- i6 [$ (RR, - WRIT = -3.0 S, = 9.69 s, = 1.90 tz5 = - 1.57?? The mean difference was not statistically significant. A further advantage of the paired comparison sampling protocol comes from the ability to treat individual data. A fre- quency table (Sokol and Rohlf, 1969:549) of the twenty-six original zi values for 1972 is shown in Table 4. A Chi Square test of these data indicates a non-normal distribution. However upon inspection the datum in the +3.5 S, class appears to be the major source of the high Chi Square value. The justifica- tions for excluding this paired difference value within the +3.5 S, class were twofold: Chauvinet?s criterion (Wang and Willis3 1965: 192) and the fact that a Chi Square test on the remaining frequency distribution indi- cated that except for this datum the data are normally distributed. The adjusted values for 1972 became (RR - WR)i_y$ = -4.3 pg liter? and ( *; n = 25) (RR - WR)/(RR):yi&, = -.33 (*; n = 25). It should be noted that while the data exclu- sion criteria we have used permit a better estimate of the average value, the variance of the data remains unchanged and the denomi- nator in the Student?s t test is the same SZ as was calculated for the complete set of data. The paired sampling protocol applied to comparable systems makes it possible not only to treat the experimental design and the data by the powerful techniques of analysis of variance but to establish probability distribu- tions which justify the exclusion of ?bad? data points as we have demonstrated above. We can now say that in 1972 there was a 3 Note that there is a typographical error in this ref- erence. On page 192 the sentence should read ?...is equal to, or less than -&, .? instead of Yz N. 200 Ii. H. Seliger and M. E. Loftus TABLE 3. Summary of paired variate analyses for Rhode and West rivers for 1971 through 1972. There are three entries for each year and each class. The top entry is the calculated mean difference followed by symbols indicating the level of significance. The middle entry is the sample size. The bottom entry in brackets is the absolute value of the mean difference which would be significant at the P = 0.05 level, based on Sz of equation (4). Comparison Period Days 1970 1971 1972 Winter l-90 RR-WR RR Spring-Summer Fall 9 l-240 241-365 Whole Year l-365 Winter l-90 RR-WR [pg liter ?1 Spring-Summer Fall 91-240 241-365 Whole Year l-365 Winter l-90 Spring-Summer less Bloom 91-150 20 l-240 RR ~ CB Bloom RR Fall Whole Year 151-200 24 I-365 I-365 - - - - .08 NS n = 17 L.211 -.82 NS n=6 I.931 -.27 NS n = 23 L.281 - - - -.17NS n = 17 L3.91 -6.5 NS n=6 i9.11 - 1.8 NS n = 23 L3.61 - - - .50*** n= 11 [.I91 -.16NS n=6 L.391 ,jg*** n=5 [.I11 .34*** n = 22 I.181 -.I NS n=7 I.911 0.0 NS n= 19 L.151 -.I9 NS n=7 I.591 - .07 NS n = 33 L.201 2.7 NS n=7 l3.91 .07 NS n = 19 i2.11 I - 1.8 NS n=7 V.91 .24 NS n = 32 L2.11 .69** n=6 L.351 .09 NS n= 12 ~41 - .56 NS n=6 l.931 .15 NS n=6 L.381 .l NS n = 30 ~41 -.46 NS n=9 L.621 -.25* n = 14 L.221 p.28 ID n=2 - -.30* n = 26 L.221 -3.2 NS n=9 l3.71 -5.2* n = 14 L4.91 -2.9 ID n=2 - -4.3* n = 25 L3.91 .36 NS n=6 L.501 .34* n=7 L.281 .20 NS n=5 L.581 .38 ID n=2 - .32** n = 20 [.I81 TABLE 3. (con?). Growth and Dissipation of Phytoplankton in Chesapeake Bay 201 Comparison RR - CB [rg liter-?] Period Winter Spring-Summer less Bloom Bloom Fall Whole Year Days 1970 1971 1972 - 6.6* 3.6 NS l-90 - n=6 n=6 - 14.31 15.81 12.7** l.ONS 12.6* 91-150 n= 11 n= 12 n=7 20 l-240 17.41 13.21 [12.5] -1.3NS -6NS 17 NS 151-200 n=6 n=6 n=5 19.71 19.51 1301 7** 4.9 NS 5.1 ID 241-365 n=5 n=6 n=2 141 1101 - I.@* 1.4NS 10.3** l-365 n = 22 n = 30 n F 20 14.81 I31 16.81 1D:insufficient data for statistical analysis NS:not significant P > .05 *:significant .05 > P > .Ol TABLE 4. Frequency table of 1972 (RR - WR) chlorophyll a concentrations. Frequency (f-tj class of units of s, Observed f Expected P t? -2.5 0 .24 ,240 -2.0 1 .73 ,100 -1.5 1 1.70 ,288 ~ 1.0 2 3.15 ,420 -0.5 7 4.54 1.333 0 5 5.13 ,003 +0.5 8 4.54 2.637 +l.O 1 3.15 1.467 + 1.5 0 1.70 1.70 +2.0 0 .73 .73 +2.5 0 .24 .24 +3.0 0 .06 .06 +3.5 1 .Ol 98.0 Zf = 26 Summary: n = 26; Z = -3.0 p gram liter-? (NS); Sr - 1.90 p gram liter- I; n = 25; Hz = -4.3 p gram liter-? (*): xsz = 8.68. small but significant difference between the standing crops of phytoplankton in the Rhode River and the West River, a result not attainable from the baseline technique. If we used the very same yearly data and treated them as though the measurements had been **:very significant .Ol > P > ,001 ***:extremely significant ,001 > P made independently, the difference of the means remains the same, but the standard error of the means, the de- nominator of equation (2) becomes 6.0 as compared with 1.90 for the paired compari- son. Since the precision varies as l/G it follows that the application of the paired comparison sampling protocol can reduce the number of samples required to achieve a particular precisian; in this case by a factor of 10. Consider the following hypothetical case which will become apparent from observation of Fig. 13. Let us assume that some sewage or industrial effluent or other perturbation had been initiated in West River in late 1970. In 1971 we set about to measure, among other things, the spring pulse in phytoplankton standing crop in the West River and compare it to the 1970 data. Considering the before and after baseline our data for these two years give (&%)~~~~5,, - (WR)iy?i5,, = 17.4 pg liter-? (*; n + m = 16). Since (~)~~?~,,, = 25.6 pg liter-?, 202 H. H. Seliger and M. E. Loftus the West River has ?suffered? a 68% decrease in the spring phytoplankton standing crop compared with 1970. However from Table 3 the values of (RR - WR) and (RR - WR)/(RR) for spring-summer 1970 and spring-summer 197 1 show that, compared with the Rhode River, which presumably did not have the hypothetical effluent or pertur- bation, there was no significant difference in phytoplankton standing crops between RR and WR in spring-summer 1970 and defi- nitely not in spring-summer 1971; the 68% decrease observed in WR between 1970 and 1971 was reflected in RR as well. Therefore the hypothetical man-made perturbation in WR was not responsible for the observed changes in WR phytoplankton in 197 1. The advantages of the paired comparative sampling protocol are thus not only in reduc- ing the sample size, but in separating changes due to natural variability which relate to both systems from changes in a system due to specific perturbations in one system. b) The analyses tell us something about the natural variability of phytoplankton standing crops in these sections of the Rhode and West rivers. On the basis of the 3-year composite data for RR - WR we find that (RR - WR)1970-1g72 = - 1.7 pg liter? (*; n = 81). The fact that a small difference of only 1.7 pug liter- l of chlorophyll a is statistically signifi- cant is an indication of the precision of the comparative sampling protocol. More rele- vant, the standard deviation S, = 7.14 pg liter-?. Since SA, the instrumental and cali- bration error of a difference measurement of extractable chlorophyll a is equal to less than 2 pg liter-l it follows that the patchiness or irreducible natural variations still give rise to the major component of the total uncer- tainty, even when the total uncertainty is re- duced. It would therefore be unfruitful to at- tempt to improve the assay technique for chlorophyll a. Rather it should be possible to relax the precision requirements of the assay technique in favor of increasing the sampling frequency. The in vivo chlorophyll a fluorescence assay is, despite its uncertainties (Loftus et al. 1972) the method of choice for assays of phytoplankton standing crops. It can readily increase the sample size since it can be adapted to routine measurement of large areas by technical personnel and to unat- tended continuous operation in situ. c) The phytoplankton standing crops in Rhode River were generally higher than in the bay proper. However this was not the case in 1971. The mean differences RR - CB for the seasonally combined 1970, 1971 and 1972 data were 7.6 pg liter-? ** (n = 22); 1.4 pg liter-? NS (n = 30) and 10.3 pg liter-? ** (n = 20) respectively. d) The statistical analysis of the January- March, 1971 data, (??,-,CB) r;; sup- ports the insolation effect shown in Fig. 8. The effect occurred slightly later in 1972. If days l-150 are analyzed, ( RR - CB\1g72 RR / l-150 = 0.37* (n = 10). In 1970 our first data point was day 91. From day 91 through day 150, ( RRRIRCB)::::~o = 0.58** (n = 6). Conclusions The application of the paired comparison sampling protocol to a baseline study can represent an improvement in experimental design and in the statistical analysis of data. In the sample calculation contrasting the precision of the paired measurements with the separate system technique we arbitrarily cal- culated a standard deviation for the latter even though there was no a priori reason to assume a normal distribution for the annual data. This we justified only to compare the same data (see Sokal and Rohlf, 1969:333). It would be more reasonable to assume that the differences of parameters in comparable sec- tions are normally distributed since the ir- reducible experimental variance is the result of non-homogeneities in both sections. The relative differences (A - B)/A are not sym- metric. For example (A - B)/A + 1 for A > B and approaches -co for A << B. The distribution becomes important when A is quite different from B. The concentrations of chlorophyll a at any time are the result of previous production, Growth and Dissipation of Phytoplankton in Chesapeake Bay 203 predation and water exchange. The standing crop data in this paper are by themselves not sufficient for a comparative study of the Rhode River and West River phytoplankton communities. The parallel data on nutrients, primary production, species composition, succession and predators will be presented in a subsequent paper. The standing crop data have been used to demonstrate the degree of precision that can be obtained by the paired comparison technique. We have concluded that based on the combined data for 1970 through 1972 the mean annual phytoplankton standing crop in West River is higher than that in Rhode River; 1.7 pg liter-? at the 5% probability level. This is a small value and therefore the lack of exact comparability between the river sections could just as easily be responsible for the difference as some increased stress on the West River. It does not appear therefore that the relatively greater human population and boat use in the West River watershed have changed the phy- toplankton concentrations in the West River. It is possible that further upstream in West River there may be some small sections which have been disturbed by septic tank leakage or effluents from the boat marinas or silt runoff as the result of increased density of use of the watershed. The strong coupling and feedback in the phytoplankton-zooplankton-detritus- bacteria-protozoa food chain tends to damp out many of these stresses, and the effective- ness of this damping is a measure of the stability or assimilatory capacity of the sys- tem. If a local effect is extreme, if there are many small local effects or if a portion of the food web is interfered with, the recovery may not be complete. In this case the river section outside of the local areas may be affected. The present comparison was between river sections outside of the ?local effect? areas, i.e. outside of the creeks. However the extent of a ?local effect? depends on the size of the subestuary. In the Potomac River the local effects of blue-green algae due to eutrophica- tion can be extensive, as can the results of power plant predation by entrainment or inhibition by heat or chemical effluents. It is conceivable that in these extensive areas the comparable river sections chosen for the ?paired? comparison protocol could also in- clude adjacent sections of the river above and below the perturbation whose effect is to be measured. The standing crops of chlorophyll u in both rivers were higher than in the bay, except during the summer when there were Prorocentrum blooms in the bay (unpub- lished data). In 1971 however there was no spring pulse and therefore no significant difference between either river and the bay. Only in 1972 did the standing crops of chlorophyll a in the Rhode River differ significantly from those in the West River. We have no explanation for this. We know of no major man-introduced perturbations which might have changed conditions in either RR or WR from 1970 and 1971 when the phytoplankton populations were essen- tially the same. The precision demonstrated for the paired sampling protocol should make it feasible to study the response of the Rhode River sub- estuary to defined nutrient or biocide inputs, simulating concentrations observed or pro- posed for other, larger subestuaries. The effects should be measurable at a series of levels of additions which will not be exorbi- tantly expensive or impractical to handle physically, and from which the assimilatory capacity of the system can be estimated. LITERATURE CITED ANONYMOUS. 1973. The Chesapeake Bay: A pro- gram for research applied to national needs. Research proposal submitted to NSF RANN Feb. 1973. BEERS, J. R., and G. L. STEWART. 1969. The vertical distribution of microzooplankton and some ecological observations. J. Cons. Perma. Int. Explor. Mer. 33:30-44. BENSHNEIDER, K., and R. J. ROBINSON. 1952. A new spectrophotometric method for the determination of nitrate in seawater. J. Mar. Res. 11:87-92. BIGGLEY, W. H., E. SWIFT, R. J. BUCHANAN, and H. H. SELIGER. 1969. Stimulable and spontaneous bioluminescence in the marine dinoflagellates, Pyrodi- nium bahamense, Gonyaulax polyedra and Pyrocystis lunula. J. Gen. Phys. 54~96-122. BOWDEN, K. F. 1967. Circulation and diffusion, p. 15-36. In G. H. Lauff (ed.), Estuaries A.A.A.S. CARPENTER, J. H. 1965. The accuracy of the Winkler method for dissolved oxygen. Limnol. Oceanogr. 10:135-140. HUMPHREY, G. E. 1963. Chlorophylls a and c in cultures of marine algae. Australian J. Mar. Freshwa- ter Res. 14: 148-154. 204 H. H. Seliger and M. E. Loftus JERLOV, N. G. 1968. Optical oceanography. Elsevier Rhode River. Annual Report Ches. Res. Consortium, Pub. Co. 194 p. May 31, 1972 p. 446-457. LOFTUS, M. E., and J. H. CARPENTER. 1971. A fluorometric method for determining chlorophylls a, b, and c. J. Mar. Res. 29:319-338. -, D. V. SUBBA RAO, and H. H. SELIGER. 1972. Growth and dissipation of phytoplankton in Chesapeake Bay. I. Response to a large pulse of rainfall. Chesapeake Sci. 13:282-299. LORENZEN, C. J. 1966. A method for the continuous measurement of in vivo chlorophyll concentration. Deep Sea Res. 13x223-227. SELIGER, H. H. 1972. Annual report Chesapeake Research Consortium, May 31, 1972, p. 4588551. , and W. G. FASTIE. 1968. Studies at Oyster Bay in Jamaica, W. I. III. Measurements of underwater sunlight spectra. J. Mar. Res. 26:2733280. , and W. D. MCELROY. 1968. Studies at Oyster Bay in Jamaica, W. I. I. Intensity patterns of biolu- minescence in a natural environment. J. Mar. Res. 26~244-255. MCCARTHY, J. J., W. R. TAYLOR, and M. E. . LOFTUS. 1974. The significance of nannoplankton in the Chesapeake Bay estuary and problems associated with the measurement of nannoplankton productivity. Mar. Biol. 2417-16. MORRIS, A. W., and J. P. RILEY. 1963. Thedetermi- nation of nitrate in seawater. Anal. Chim. Acta 291272-279. MURPHY, J., and J. P. RILEY. 1962. A modified single solution method for the determination of phos- phate in natural waters. Anal. Chim. Acta 27:31-36. PRITCHARD, D. W. 1967a. What is an estuary: physical viewpoint, p. 3-5. In G. H. Lauff (ed.), Estuaries A.A.A.S. SOKAL, R. R., and F. J. ROHLF. 1969. Biometry, the principles and practice of statistics in biological re- seaich. W. H. Freeman and Co., 776 p. SOLORZANO, L. 1969. Determination of ammonia in natural waters by the phenol hypochlorite method. Limnol. Oceanogr. 14:799%801. STEEMANN-NIELSEN, E. 1952. The use of radioac- tive carbon (??C) for measuring organic production in the sea. J. Cons. Perma. Int. Explor. Mer. 18:117m140. STRICKLAND, J. D. H., and T. R. PARSONS. 1968. A practical handbook of seawater analysis. Fish. Res. Bd. Canada Bull. No. 167. 311 p. WANG, C. H., and D. L. WILLIS. 1965. Radiotracer methodology in biological science. Prentice Hall, New Jersey. 382 p, -. 1967b. Observations of circulation in Coastal plain estuaries, p. 37-44. In G. H. Lauff (ed.), Estuaries A.A.A.S. -, and G. HAN. 1972. Physical hydrography of the YENTSCH, C. S., and R. W. LEE. 1966. A study of photosynthetic light reactions, and a new interpreta- tion of sun and shade phytoplankton. J. Mar. Res. 24:319-337.