Journal of Photochemistry and Photobiology B: Biology 62 (2001) 1?8 www.elsevier.com/ locate / jphotobiol Modeling the effects of ultraviolet radiation on estuarine phytoplankton production: impact of variations in exposure and sensitivity to inhibition Patrick J. Neale Smithsonian Environmental Research Center, Edgewater, MD 21037, USA Received 11 November 2000; accepted 7 June 2001 Abstract Spectral ultraviolet (UV) irradiance, water column attenuation and biological weighting functions for inhibition of phytoplankton photosynthesis have been measured for the Rhode River, a subestuary of the Chesapeake Bay. Together, these measurements can be used to estimate UV effects on water column production, but each factor shows a significant range of variability even just considering summer time conditions. A sensitivity analysis of UV inhibition is described which assesses the effect of this variation for different combinations of 28 irradiance spectra, 8 biological weighting functions (BWFs) and 16 water column irradiance profiles. Over all combinations, production averaged about 84% relative to potential production in the absence of UV effects. For a few combinations, relative production was as low as 67%, or as high as 97%, but for most combinations the range was 75?95%. Variations in the sensitivity of the phytoplankton assemblage, i.e. the BWF, and optical properties, represented by a transparency ratio of biologically effective UV to photosynthetically available radiation (PAR), had large effects on water column production. A simple relationship for UV inhibition of water column production is developed based on inhibition at the surface and the ratio of UV and PAR transparency. ? 2001 Elsevier Science B.V. All rights reserved. Keywords: Biological weighting functions; Photoinhibition; UV-B; UV-A 1. Introduction affected by variations in dissolved organic compounds and suspended particulates. Such variation is more important Ultraviolet radiation (UV, 280?400 nm) is increasingly than ozone depletion in determining UV exposure in the recognized as a potent influence on biological and chemi- water column of lakes [8?10]. Full band incident UV is cal processes in the aquatic environment [1?3]. In part, also affected by a number of variables that are vulnerable this recognitions stems from concerns about the effect of to anthropogenic and natural variation, such as cloud cover stratospheric ozone loss, which results in a wavelength- and atmospheric aerosols [11]. dependent increase in incident UV-B (280?320 nm, with The importance of variations in the UV climate to 290 nm the lower bound for solar irradiance) [4]. This chemical and biological processes in aquatic environments depletion is caused by the breakdown of anthropogenic can be assessed by application of spectral weighting chlorofluorocarbons (CFCs) in the stratosphere. The re- functions. Biological weighting functions (BWFs) describe lease of CFCs is now limited by the Montreal Protocol. the effectiveness of radiation of different wavelengths to Nevertheless ozone depletion continues, apparently due to produce a biological response, such as inhibition of the cooling of the stratosphere which accompanies ?green- photosynthesis. A wavelength-dependent description of house gas? induced surface warming [5]. UV effects on photosynthesis (BWF/P-I model) has been There are also strong ecological effects of the long- developed, which describes photosynthesis as a function of wavelength, solar UV-A (320?400 nm). Indeed, inhibition photosynthetically available radiation (PAR) and photo- of phytoplankton photosynthesis by solar UV-A is general- inhibition as a function of both PAR and UV [6,12]. The ly greater than inhibition by solar UV-B [6,7]. Many model is fit using laboratory measurements of photo- environmental changes affect UV exposure over the full synthesis under filtered solar-simulator (xenon arc) ir- waveband. Transmission of UV in aquatic environments is radiance [6,12]. The predictions of the model agree with the results of solar incubations in Antarctica [13] and in E-mail address: neale@serc.si.edu (P.J. Neale). situ productivity profiles in lakes [14]. Using this con- 1011-1344/01/$ ? see front matter ? 2001 Elsevier Science B.V. All rights reserved. PI I : S1011-1344( 01 )00159-2 2 P.J. Neale / Journal of Photochemistry and Photobiology B: Biology 62 (2001) 1 ?8 ceptual approach, BWFs have been measured for diverse irradiance (290?700 nm) were obtained using a radiative phytoplankton, including cultures [6,15] and natural as- transfer model [24] as implemented by the STAR software semblages in Antarctica [12,16] and a shallow subestuary package (H. Schwander, University of Munich). of the Chesapeake Bay [17]. These studies have revealed more than ten-fold variation in the sensitivity of phyto- 2.2. Sensitivity analysis plankton photosynthesis to UV inhibition. Among the issues that can be addressed using spectral The analysis focused on early summer conditions as UV weighting functions is the relative importance of variations exposure is highest during this period. Accordingly, a in environmental factors versus sensitivity to UV in 1?2-month period of observations was selected, with the enhancing or ameliorating the effects of UV in aquatic period centered around summer solstice. The selected data ecosystems. A modeling analysis of an Antarctic eco- included 28 days of irradiance measurements (June 7?July system (Weddell-Scotia Confluence, WSC) showed that 5, 1999), 8 BWFs (May 3?July 27, 1995 and 1996) and 16 both physiological variability (i.e. BWFs) and variation in profiles of downwelling spectral irradiance (June 6?August exposure, related to changes in the depth of vertical 6, 1999). The BWFs were measured in a different year to mixing, can profoundly affect predicted inhibition of water the irradiance data, however conditions were generally the column photosynthesis [18]. The effect of ozone depletion same between the two summers. Moreover, the BWFs was significant but secondary to these other factors. The showed no significant correlation with irradiance data and relative importance of variation in irradiance and sensitivi- proxy indicators of water column transparency measured ty has not been modeled in other aquatic environments. during 1995 and 1996 [17]. Variations in spectral trans- This report describes the results of an initial sensitivity parency of the Rhode River and incident solar irradiance analysis of factors affecting UV inhibition of water column also were not correlated during the period of study (data photosynthesis in the Rhode River, which is a turbid, not shown). Therefore, to a first approximation, variations eutrophic [19] subestuary on the western shore of the in each measurement over this period were considered Chesapeake Bay in Maryland, USA. The Smithsonian statistically independent. Environmental Research Center, located adjacent to the The general approach for the sensitivity analysis was to Rhode River, is a long-term monitoring site for solar calculate midday, integral water-column photosynthesis spectral UV-B [20]. The BWFs for phytoplankton assem- over surface to 1.6 m depth for each combination of blages in the Rhode River were measured on a monthly measured irradiance, BWF and spectral attenuation. The basis during 1995 and 1996 [17]. More recently, the Rhode BWF/P-I was used to calculate photosynthesis as a B 21 21River has been regularly sampled for spectral attenuation function of depth (P (z), mg C mg Chl h ) [21]. In this report, we make an initial evaluation of the 1B Brelative importance of variation in BWFs, solar UV and ]]]P (z) 5 P tanh (E (z) /E ) ? (1)S Ds PAR s *1 1 E (z)water transparency in modifying the predicted effect of UV inh on summertime water column production. B where P is the maximum attainable rate of photosynthesiss in the absence of photoinhibition; E (z) is PAR ex-PAR 22pressed as irradiance in energy units (W m ) and is determined by the attenuation coefficient for PAR (K ,PAR2. Methods 21 m ) using the expression E (0) exp(2K z); E is aPAR PAR s saturation parameter for photosynthesis. The inhibition2.1. Input data term is a function of UV spectral irradiance (E (l,z), mW 22 21 m nm ) expressed as biologically-weighted exposure,The primary sampling site is the Rhode River Station *E (z) (dimensionless), whereinh4B, which is located in an estuary segment with a mean 395 nmdepth of 1.6 m [22]. Sampling and experimental de- *E (z) 5 O e(l) ? E(l,z) ? Dl (2)termination of the BWFs using the photoinhibition ap- inh l5290nmproach has been previously described [17]. Profiles of 22 21 spectral irradiance (E (l,z), mW m nm ) were mea- e(l) is the wavelength-dependent biological effectivenessd sured using a Satlantic OCP 200 radiometer with filter (i.e. the BWF) for inhibition of photosynthesis by UV (mW 22 21 center wavelengths of 325, 340, 380 nm (2 nm bandwidth) m ) . Although a parameter for E dependent inhibi-PAR and ten wavelengths in the visible range (10 nm band- tion (i.e. e ) can be included in Eq. (2), it was notPAR 22 21 width). Incident spectral UV-B irradiance (mW m nm , needed [17]. The BWF/P-I model (Eq. (1)) is similar to 1 min averages) is continuously measured at SERC over that described by Cullen and Neale [25], except that the spectral range of 290?325 nm using a multifilter potential photosynthesis is described by a hyperbolic radiometer (SR18) with eighteen 2-nm bandwidth filters tangent P 2 E function [26]. For convenience of com- 22 21[23]. parison, water-column photosynthesis (P , mg C m h )z BModel estimates of clear sky, solar noon spectral was calculated for unit P and chlorophyll concentrations P.J. Neale / Journal of Photochemistry and Photobiology B: Biology 62 (2001) 1 ?8 3 23 B(1 mg Chl m ) by evaluating P (z) at 0.1 m intervals during the period: two-thirds of the measurements were from 0 to 1.6 m and integrating over depth. During the within 20% of modeled irradiance at 320 nm (Fig. 2a). B observation period, P actually varied from 3 to 18 mg C Cloudy conditions typically resulted in a reduction to abouts 21 21 23 mg Chl h and Chl varied from 15 to 45 mg Chl m . 30% of clear sky irradiance (Fig. 2a). For comparison, ?potential? water column photosynthesis Attenuation coefficients were calculated from depth 22 21in absence of UV effects (P , mg C m h ) was profiles of spectral irradiance and used to estimate in situzpot *calculated the same way except that E (z) was set to 0. spectral irradiance at 10-cm intervals between 0 and 1.6 m,inh Evaluation of the weighted irradiance requires full band after correcting for surface reflection (see [18]). Spectral spectral irradiance (290?700 nm) as function of depth. attenuation coefficients for wavelengths of the irradiance Incident spectral solar UV-B at the surface was measured data (SR18 wavelengths in UV-B and every nanometer in by the SR18, these measurements were averaged for the the UV-A) was obtained by fitting an exponential equation one period around solar noon (1300 h local daylight (i.e. linear regression on log transformed coefficients) to savings time). Complete evaluation of weighted irradiance attenuation coefficients estimated at 325, 340, 380, 412, required extension of this spectrum to the 325?400 nm 510, 532 and 555 nm. Attenuation coefficients for l , 325 spectral region not measured by the SR18 which was made nm were obtained by extrapolation of the fitted curve. by reference to model spectra calculated using the STAR Coefficients in this range may be underestimated if the program. Program inputs were as follows: measurement spectral slope increases in the UV-B, as has been reported date and time, i.e. solar noon (1700 h UTC at SERC in other waters [27]. In all 16 attenuation spectra, the 2location), ?continental polluted air? for aerosol type, sum- coefficient of determination (R ) for the exponential equa- mer ozone profile, total column ozone as estimated from TOMS (toms.gsfc.nasa.gov) and observed barometric pres- sure. The modeled spectra (smoothed using a 2-nm band- width gaussian filter) was very similar to the SR18 measurements under clear sky conditions (Fig. 1). The UV-A spectrum (325?400 nm, at 1-nm intervals) was then estimated by applying a constant factor to the model spectra for each day based on the proportion between observed and modeled spectra over 320?322 nm. Integral PAR (400?700 nm) irradiance was calculated using the same proportion. Clear sky conditions were prevalent Fig. 1. Example of incident UV-B irradiance spectrum for the Rhode River as used in the calculation of water column production. Shown are 22 21 spectra (mW m nm ) obtained for midday, clear-sky conditions on Fig. 2. Variation in factors affecting phytoplankton exposure to UV in the June 20, 1999. Spectra were measured with the SR18 multifilter UV-B Rhode River. (A) Distribution of incident spectral irradiance at 320 nm radiometer and averaged for the 1-h period centered around solar noon, for a 28-day period centered around June 21, 1999 as a proportion of the with irradiance plotted at the filter center wavelength (circles). Each filter clear sky 320 nm irradiance calculated for each day using the STAR has a nominal bandwidth of 2 nm (FWHM). Spectra were also calculated program (see Fig. 1). (B) Distribution of downwelling attenuation 21 using a radiative transfer model (line) as implemented by the STAR coefficient for spectral irradiance at 325 nm (K [325], m ) as calculatedd program using a 2-nm bandwidth, observed barometric pressure and total from 16 profiles of water-column irradiance in the Rhode River and column ozone from TOMS. Spectra are plotted on both a logarithmic (left) adjacent waters of the Chesapeake Bay measured during the period June and linear (right) scale. 6?August 6, 1999. 4 P.J. Neale / Journal of Photochemistry and Photobiology B: Biology 62 (2001) 1 ?8 26 21tion exceeded 0.98 (P,10 ). Spectral attenuation at 325 water column, actually K (325) was 8.7 m for thed 21 nm was usually between 4 and 8 m , but was occasion- minimum P* .z 21 ally as high as 15 m (Fig. 2b). An overall attenuation To further analyze how predicted P* is affected byz 21 variations in water column transparency, measures of UVcoefficient for PAR (K , m ) was calculated fromPAR (T ) and PAR (T ) transparency were calculated, bothattenuation coefficients in the 400?700 nm range. PIR PAR in units of meters. The transparency for inhibiting ir-The eight selected BWFs spanned the middle range of radiance (T ) was calculated using the equation ofvariability for the Rhode River [17]. The range of variation PIR Pienitz and Vincent [9] as modified [28]between the most and least sensitive (as indicated by weight at 325 nm) was about 3-fold, with other weights 395 nm 1 e(l) ? E(l,0)evenly distributed between these extremes. This compares ]] ]]]]T 5 O ? ? Dl (3)PIRto an approximate 10-fold variation between minimum and *K (l) E (0)l5290 nm d inh maximum sensitivity overall [17]. Average weight at 325 24 22 21 For the Rhode River and adjacent waters, calculated Tnm is 1.22310 (mW m ) . The coefficient of vari- PIR for the sensitivity analysis averaged 0.15 m (Table 1).ation (S.E. /mean) for the BWF coefficients is about 10%. Transparency for PAR (51/K [9],) was 0.55 m onPAR average. The ratio of these two transparencies (T /T ,PIR PAR average50.25) is an indicator of how water column 3. Results properties mediate the competing effects of solar irradiance as both a inhibitor and source of energy for photosynthesis. Calculations of water-column production and related Thus, the most severe inhibition of P* corresponded toz variables are summarized in Table 1. Midday productivity situations with the highest ratio, T /T . For severePIR PAR 22 21predicted by the model averages around 1 mg C m h inhibition this ratio was almost 0.5 compared to around 0.1 B(for unit Chl concentration and P ) when UV effects are for weak inhibition. While most combinations resulted in as included, compared to potential production of about 1.2 ratio near the mean, there were sets of conditions that 22 21 mg C m h in the absence of UV. In other words, under resulted in distinct groups near the upper and lower the chosen model conditions UV modeled production was extremes (Fig. 3c). about 84% overall of potential production and UV inhibi- Another indicator of the relative influence of exposure tion is about 16%. Over all combinations of variables, the and sensitivity factors on water-column production is the relative production (P* 5P /P ) ranged from 67 to variation resulting from each factor. This was estimated byz z zpot 97%, with most combinations lying in the range of 75? specifying a fixed factor level (e.g. each of the 8 BWFs) 95% (Fig. 3b). The lowest relative production (severest and calculating averages over all combinations of the other inhibition) occurred with the most sensitive assemblages factors. The calculation indicated that variations in BWFs (July 3, 1996 and July 12, 1995) under clear sky con- and water column transparency have the greatest influence ditions, whereas highest relative production (least inhibi- on the relative production (P* ), with the range ofz tion) occurred in combinations of the least sensitive averages about half of the overall range (Table 1). On the assemblage (June 17, 1995) under cloudy conditions other hand, absolute production (P ) is influenced more byz [E(l,0) , 30% of clear sky], as may be expected if incident irradiance and water column transparency. Vari- production is not strongly limited by PAR. Water trans- ation in the BWF is the strongest influence on relative parency to UV was also an important factor in determining production at the surface (P*(0)) and on T .PIR the extremes of production, but the relationship was not Column ozone ranged between 295 and 366 DU (Dob- simple. The greatest inhibition did not occur in the clearest son units) during the 28-day period for which irradiance Table 1 Variation in water-column production and related variables in the Rhode River P P P* P*(0) Tzpot z z PIR Average 1.17 0.98 0.84 0.39 0.15 Variation All (%) 662 665 618 688 661 BWF (%) 614 616 68.1 642 68.6 Incident irradiance (%) 632 628 64.9 650 2 Water column transparency (%) 632 633 67.5 2 653 22 21Data are the results of a sensitivity analysis of depth integrated potential production (P , mg C m h ), depth integrated production including thezpot 22 21 effect of inhibition by UV (P , mg C m h ), the ratio of UV-inhibited to potential production, for the water column (P* 5 P /P ) and at the surfacez z z zpot [P*(0)5P(0) /P (0)] and weighted transparency for biologically effective UV (T , m). Variation is quantified as the difference between maximum andpot PIR minimum values relative to the overall average, results are expressed as 6half this range. ?All? refers to the range over all 3584 combinations of conditions. The variation due to each factor was estimated as range resulting when means are computed for each of the 8 BWFs, 28 surface UV spectra or 16 water-column attenuation spectra, averaged over all combinations of the other factors. Cells are empty for factors that have minimal effect on the variable. P.J. Neale / Journal of Photochemistry and Photobiology B: Biology 62 (2001) 1 ?8 5 undertaken to determine the influence of ozone, per se. A series of model irradiance curves was generated for midday, clear-sky, summer solstice conditions using pa- rameters for June 21, 1999 as described previously, but varying total column ozone by 25 DU increments from 250 to 400 DU. A second series of production model calcula- tions was then performed to compute surface and water column production using these spectra over all combina- tions of BWFs and K . The results indicate that ozoned depletion, per se, has a small effect on production in the Rhode River, with an average of 9% drop in surface production and 1% decrease in water column production over the given range (Fig. 4). This reflects the limited transparency of the water column to UV-B [17]. Accord- ingly, excluding the UV-B contribution in model calcula- tions typically increases water column production by only 3% of P and decreases inhibition by approximatelyzpot 18% (53%/16%). The contribution of UV-B would be even less if attenuation coefficients in the UV-B range are actually higher than the extrapolated values used in the present analysis. Thus, the decrease associated with the 20% variation of column ozone in the observed irradiance series is probably obscured by concomitant variation caused by changes in other atmospheric conditions. The strong influence on predicted production of vari- ation in exposure and sensitivity factors suggested that a simple relationship may exist between these factors and Fig. 3. Distribution of model calculated water-column production and UV/PAR transparency ratio for conditions used in the sensitivity analysis. Each histogram shows the number of instances in each category over a 22total of 3584 combinations. (A) Water column production (P , mg C mz 21h ) for unit Chl and maximum rate of photosynthesis, (B) water column Fig. 4. Production in the Rhode River as a function of total column 23 21production as a proportion of the potential production in the absence of ozone. Surface production (P(0), mg C m h ) and water column 22 21UV inhibition (P* , dimensionless), (C) ratio of transparency for bio- production (P , mg C m h ) were calculated for all combinations ofz z logically effective UV (T , m) to PAR (T , m). BWFs and water column transparency using seven modeled irradiancePIR PAR spectra based on atmospheric conditions for June 21, 1999 but using seven levels of total column ozone ranging from 250 to 400 Dobson Units (DU), with an increment of 25 DU. Results are shown as percentage spectra were selected. This is about a 20% variation in decrease in production at each ozone level in proportion to production at ozone, but there was no significant correlation between 400 DU, left axis (points) water column production, right axis (line) ozone and modeled P . A separate analysis was therefore surface production.z 6 P.J. Neale / Journal of Photochemistry and Photobiology B: Biology 62 (2001) 1 ?8 averaged about 16% over all combinations, considerably less than the around 25?30% inhibition of daily production calculated for the Weddell-Scotia Confluence [18] and Antarctic Peninsula waters [29]. On the other hand, the predicted inhibition of integrated production for the Rhode River is at a similar level as calculated for Lake Michigan [30] (effect of ,370 nm only) and lakes in the Swiss Alps [14]. Interestingly, the ratio T /T in the Swiss lakesPIR PAR was in the same range as the Rhode River (0.2?0.4) despite the generally clearer water in the lakes. This reinforces the conclusion that the predicted effect of UV depends more on the relative penetration of UV and PAR rather than the UV transparency by itself. Thus, inhibition by UV can be significant even in a turbid estuary with high attenuation coefficient for biologically effective wave- lengths, especially under conditions of simultaneous low transparency for PAR. The Rhode River is much more dynamic, optically, than the WSC, so that changes in transparency emerge as a primary factor affecting exposure in this system, whereas vertical mixing was the primaryFig. 5. Relationship between inhibition of water column production (Inh ,z factor affecting exposure in the WSC. Indeed, fine tempo-dimensionless) calculated for sensitivity analysis conditions and inhibition of production as predicted by the product of surface inhibition (Inh ) and ral scale (1 h) monitoring of optical properties in the0 the ratio of UV to PAR transparency ratio (Inh ? T /T ); diagonal line0 PIR PAR Rhode River has revealed many-fold variations in trans- 2indicates a 1:1 relationship. The overall R between the two variables is parency on daily time scales [31].0.80. In the present analysis, vertical mixing was not consid- ered as factor since an irradiance-based model of UV inhibition by UV (i.e. less complex than a complete response was used, consistent with measured kinetics of numerical integration). A good agreement was found inhibition and recovery [17]. Strong vertical mixing could between inhibition of water column production (Inh 5 1 2 still affect production if residence times were comparablez P* ) and the product of surface inhibition (Inh 5 1 2 to the characteristic time scales of response to UV [18,32].z 0 P* ) and the ratio of UV to PAR transparency, i.e. However, an irradiance based BWF/P-I model was a good0 predictor of water column production in a Swiss lake, evenTPIR]]Inh 5 Inh ? (4) when incubations bottles were circulated during exposurez 0 TPAR [33]. The analysis also assumes that there is a homoge- This relationship reproduced model calculated inhibition neous vertical distribution of phytoplankton biomass. On 2 with an overall R of 0.80 (Fig. 5). The fidelity of the average, this is a good approximation to conditions in this relationship to a 1:1 ratio varied according to the water shallow estuary. However, localized and transient devia- optical properties. Eq. (4) slightly underestimated inhibi- tions from homogeneity certainly do occur. In particular, tion for the highest T values (0.8?1 m) and overesti- dinoflagellates appear to migrate away from the surfacePAR mated for the lowest T values (0.25?0.3 m). The under calm conditions. This would obviously affect thePAR standard error of the inhibition estimate was 2.6% of P . degree of UV inhibition of production.zpot Implementation of the BWF/P-I approach for predicting water column production requires full spectral irradiance at 4. Discussion nanometer scale resolution. Since only the UV-B spectrum is monitored at the site, UV-A was obtained by scaling the A modeling approach has been used to investigate the output of a radiative transfer model. A similar procedure relative importance of exposure and sensitivity factors in was used in the modeling of production in Lake Lucerne. determining UV inhibition of water column production At the Swiss site (EAWAG field station), SR18 measure- during early summer conditions in a temperate estuary. ments were conducted in parallel with operation of broad- The natural range of variations in sensitivity as measured band sensors (Macam) for UV-B, UV-A and PAR [34]. by the biological weighting function for inhibition of Model spectra were also calculated using the STAR program photosynthesis and variations in transparency both sig- and scaled to agree with the SR18 at 320 nm. The resultant nificantly influenced the calculation of water column UV-A and PAR irradiances were consistent with the production, with each factor making an approximate equal independent UV-A and PAR measurements, after account- contribution. ing for the respective spectral response of the sensors. Inhibition of midday production in the Rhode River Results for these two sites, together with similar efforts by P.J. Neale / Journal of Photochemistry and Photobiology B: Biology 62 (2001) 1 ?8 7 Environmental Intelligence Unit, R.G. Landes, Georgetown, TX,other groups [35], support the combined use of multifilter 1997.radiometers and radiative transfer models as an efficient [3] The effects of UV radiation on marine ecosystems, in: S.J. de Mora, method of obtaining spectral irradiance for use in quan- S. Demers, M. Vernet (Eds.), Environmental Chemistry Series, titating the effects of UV with weighting functions. Cambridge University Press, Cambridge, 2000. [4] D. Lubin, E.H. Jensen, Effects of clouds and stratospheric ozoneThe sensitivity analysis approach provides a detailed depletion on ultraviolet radiation trends, Nature 377 (1995) 710.picture of how multiple influences can combine to affect [5] S. Madronich, G.J.M. Velders, Halocarbon scenarios for the future ecosystem response to a primary stressor such as UV. The ozone layer and related consequences, in: C.A. Ennis (Ed.), Sci- results of this initial analysis motivate the extension of the entific Assessment of Ozone Depletion, World Meterological Or- ganization, Geneva, 1998, pp. 1111?1138.approach to other types of variation, such as vertical [6] J.J. Cullen, P.J. Neale, M.P. Lesser, Biological weighting functionmixing and to the quantitation of UV effects on other for the inhibition of phytoplankton photosynthesis by ultravioletprocesses, such as bacterial growth and survival. The main radiation, Science 258 (1992) 646?650. disadvantage of conducting a full sensitivity analysis is [7] H. Maske, Daylight ultraviolet radiation and the photoinhibition of that it is computationally intensive. However, the results phytoplankton carbon uptake, J. Plankton Res. 6 (1984) 351?357. [8] D.W. Schindler, P.J. Curtis, B. Parker, M.P. Stainton, Consequencesalso show that a simple expression involving surface of climate warming and lake acidification for UV-B penetration ininhibition, UV and PAR transparency gives a fairly accur- North American boreal lakes, Nature 379 (1996) 705?708. ate estimate of water column inhibition. This expression [9] R. Pienitz, W.F. Vincent, Effect of climate change relative to ozone could be used for general assessments of how variations in depletion on UV exposure in subarctic lakes, Nature 404 (2000) 484?487.some parameters translate to water column effects in [10] P.R. Leavitt, R.D. Vinebrooke, D.B. Donald, J.P. Smol, D.W.estuaries. For example, the expression could be used to Schindler, Past ultraviolet radiation environments in lakes derivedindicate how variations in nutrient availability (e.g. nitro- from fossil pigments, Nature 388 (1997) 457?459. gen) may influence UV effects on water column, since such [11] J. Herman, R.L. McKenzie, Ultraviolet radiation at the earth?s surface, in: C.A. Ennis (Ed.), Scientific Assessment of Ozoneavailability affects both phytoplankton BWFs [36] and Depletion: 1998, World Meteorological Organization, Geneva, 1999,water column optics (through controls on blooms [19]). pp. 91?944.Systematic deviations from the relationship are apparent [12] P.J. Neale, M.P. Lesser, J.J. Cullen, Effects of ultraviolet radiation suggesting that improvements in accuracy are possible. on the photosynthesis of phytoplankton in the vicinity of McMurdo The deviations are related to T (and thus K ), so a station (788S), in: C.S. Weiler, P.A. Penhale (Eds.), UltravioletPAR PAR Radiation in Antarctica: Measurements and Biological Effects, Am.more consistent relationship may be obtained if PAR is Geophysical Union, Washington, DC, 1994, pp. 125?142.estimated using spectral transmission (instead of an aver- [13] M.P. Lesser, P.J. Neale, J.J. Cullen, Acclimation of Antarctic age K) or by using photosynthetically utilizable irradiance phytoplankton to ultraviolet radiation: ultraviolet-absorbing com-(PUR) as a basis for the relationship. More complex pounds and carbon fixation, Molec. Mar. Biol. Biotech. 5 (1996) expressions utilizing PUR and other parameters have been 314?325. [14] P.J. Neale et al. Quantifying the response of phytoplankton photo-developed which accurately predict inhibition over a wide synthesis to ultraviolet radiation: Biological weighting functionsrange of optical conditions [28,37]. Future work will versus in situ measurements in two Swiss lakes, Aquatic Sci. 63 examine to what extent these simple approaches to asses- (2001) in press. sing UV effects apply to other temperate and polar [15] P.J. Neale, A.T. Banaszak, C.R. Jarriel, Ultraviolet sunscreens in environments. dinoflagellates: mycosporine-like amino acids protect against inhibi- tion of photosynthesis, J. Phycol. 34 (1998) 928?938. [16] P.J. Neale, J.J. Cullen, R.F. Davis, Inhibition of marine photo- synthesis by ultraviolet radiation: variable sensitivity of phyto- Acknowledgements plankton in the Weddell-Scotia Sea during the austral spring, Limnol. Oceanogr. 43 (1998) 433?448. [17] A.T. Banaszak, P.J. Neale, UV Sensitivity of photosynthesis inCharles Gallegos and Karen Yee are acknowledged for phytoplankton from an estuarine environment, Limnol. Oceanogr. making available optical data from the Rhode River, which 46 (2001) 592?600. were acquired with support from Environmental Protection [18] P.J. Neale, R.F. Davis, J.J. Cullen, Interactive effects of ozone Agency, CISNET grant number R826943. Additional depletion and vertical mixing on photosynthesis of Antarctic phyto- plankton, Nature 392 (1998) 585?589.support was provided by NSF Office of Polar Programs [19] C.L. Gallegos, T.E. Jordan, D.L. Correll, Event-scale response ofgrant OPP-9615342 and the Smithsonian Institution Scho- phytoplankton to watershed inputs in a subestuary: timing, mag-larly Studies Program. nitude and location of blooms, Limnol. Oceanogr. 37 (1992) 813? 828. [20] D.L. Correll, C.O. Clark, B. Goldberg, V.R. Goodrich, D.R. Hayes Jr., W.H. Klein, W.D. Schecher, Spectral ultraviolet-B radiationReferences fluxes at the earth?s surface: long-term variations at 398N, 778W, J. Geophys. Res. 97 (1992) 7579?7591. [1] O. Holm-Hansen, D. Lubin, E.W. Helbling, UVR and its effects on [21] C.L. Gallegos, K. Yee, D. Sparks, Unpublished spectral irradiance organisms in aquatic environments, in: A.R. Young et al. (Ed.), data for the Rhode River, 2000. Environmental UV Photobiology, Plenum, New York, 1993, pp. [22] T.E. Jordan, D.L. Correll, J. Miklas, D.E. Weller, Nutrients and 379?425. chlorophyll at the interface of a watershed and an estuary, Limnol. ?[2] D.-P. Hader, The effects of ozone depletion on aquatic ecosystems. Oceanogr. 36 (1991) 251?267. 8 P.J. Neale / Journal of Photochemistry and Photobiology B: Biology 62 (2001) 1 ?8 [23] E. Early et al., The 1995 North American interagency intercom- [31] C.L. Gallegos, P.J. Neale, unpublished data (2000). ?parison of ultraviolet monitoring spectroradiometers, J. Res. Natl. [32] E.W. Helbling, V. Villafane, O. Holm-Hansen, Effects of ultraviolet Inst. Stand. Technol. 103 (1998) 15?62. radiation on Antarctic marine phytoplankton photosynthesis with [24] A. Ruggaber, R. Dlugi, T. Nakajima, Modeling of radiation quan- particular attention to the influence of mixing, in: C.S. Weiler, P.A. tities and photolysis frequencies in the troposphere, J. Atmos. Chem. Penhale (Eds.), Ultraviolet Radiation and Biological Research in 18 (1994) 171?210. Antarctica, American Geophysical Union, Washington, DC, 1994, [25] J.J. Cullen, P.J. Neale, Biological weighting functions for describing pp. 207?227. ?the effects of ultraviolet radiation on aquatic systems, in: D.-P. [33] J. Kohler, M. Schmitt, H. Krumbeck, M. Kapfer, E. Litchman, P.J. ?Hader (Ed.), Effects of Ozone Depletion On Aquatic Ecosystems, Neale, Effects of UV on carbon assimilation of phytoplankton in a R.G. Landes, Austin, 1997, pp. 97?118. mixed water column, Aquatic Sci. 63 (2001) in press. [26] A.D. Jassby, T. Platt, Mathematical formulation of the relationship [34] P.J. Neale, P. Bossard, Y. Huot, Incident and in situ irradiance in between photosynthesis and light for phytoplankton, Limnol. Lakes Cadagno and Lucerne: a comparison of methods and models, Oceanogr. 21 (1976) 540?547. Aquatic Sci. 63 (2001) in press. [27] S. Markager, W.F. Vincent, Spectral light attenuation and the [35] J.T.O. Kirk et al., Measurements of UV-B radiation in two fresh- absorption of UV and blue light in natural waters, Limnol. water lakes: an instrument intercomparison, Arch. Hydrobiol. Beih Oceanogr. 45 (2000) 642?650. Ergebn. Limnol. 43 (1994) 71?99. [28] J.J. Cullen, R.F. Davis, Y. Huot, M.K. Lehmann, Quantifying effects [36] E. Litchman, P.J. Neale, A.T. Banaszak, Increased sensitivity to of ultraviolet radiation in surface waters, in: G.D. Gilbert, R.J. ultraviolet radiation in nitrogen-limited dinoflagellates: photoprotec- Frouin (Eds.), Ocean Optics: Remote Sensing and Underwater tion and repair, Limnol. Oceanogr. (2001) accepted for publication. Imaging, 2001. [37] M.K. Lehmann, R.F. Davis, Y. Huot, J.J. Cullen, Biologically ?[29] N.P. Boucher, B.B. Prezelin, Spectral modeling of UV inhibition of weighted transparency: a predictor for water column photosynthesis in situ Antarctic primary production using a field derived biological and its inhibition by ultraviolet radiation, Limnol. Oceanogr. (2001) weighting function, Photochem. Photobiol. 64 (1996) 407?418. submitted for publication. [30] W.R. Gala, J.P. Giesy, Effects of ultraviolet radiation on the primary production of natural phytoplankton assemblages in Lake Michigan, Ecotoxicol. Environ. Safety 22 (1991) 345?361.