over the enclosure surface, during windy conditions which intensi- \ufb01ed in the subsequent few days. A near-uniform distribution with depth was quickly established (see Fig. 2.20). The spores (d = 32.80 ± 3.18; \u3c1c = 1049 kg m\u22123; \u3d5r \u223c 2.2) had a measured sink- ing rate (ws) of 15.75 µm s\u22121 at 17 \u25e6C, which, adjusted for the density and viscosity of the water at the 4\u20135 \u25e6C obtaining in the \ufb01eld, predicted an in-situ intrinsic sinking rate of 0.96 m d\u22121. The theoretical time for spores to eliminate the enclo- sure (at the time, H \u223c 11.8 m) was thus calcu- lated to be (t\u2032 = ) 12.3 days. In fact, the elimi- nation proceeded smoothly, always from a near- uniformly distributed residual population at an average exponential rate of \u22120.10 m d\u22121, which value corresponds to a 95% removal in (te =) 30 days. The ratio te/t\u2032 is lower than predicted in Sec- tion 2.6.2 (2.44 against 3.0). This may be explained by probable violation of the initial assumption of full mixing of the water column throughout the experiment. Although no signi\ufb01cant density gra- dient developed, continuous and complete verti- cal mixing of the enclosure cannot be veri\ufb01ed. Nevertheless, the outcome is suf\ufb01ciently close to the model (Fig. 2.20) solution for us not to reject the hypothesis that entrained particles are lost from suspension at an exponential rate close to \u2212(ws/hm). In the second experiment, commenced in June, spores were dispersed at the top of the THE SPATIAL DISTRIBUTION OF PHYTOPLANKTON 77 Figure 2.20 Modelled (M) and actual (A) depth\u2013time distributions of preserved Lycopodium spores (of predetermined sinking characteristics) introduced at the water surface of one of the Blelham enclosures on each of three occasions (1, 9 January; 2, 3 June; 3, 9 September) during 1976, under sharply differing conditions of thermal stability. Lycopodium concentrations plotted as cylindrical curves; density gradients plotted as dashed lines. \u2217 \u2013 indicates no field data are available. Redrawn from Reynolds (1984a). strati\ufb01ed enclosure during relatively calm con- ditions. Sampling within 30 minutes showed a good dispersion but still restricted to the top 1 m only. However, 4 days later, spores were found at all depths but the bulk of the original addition was accounted for in a \u2018cloud\u2019 of spores located at a depth of 5\u20137 m. After a further 7 days, mea- surable concentrations were detected only in the bottom 2 m of the column, meaning that, effec- tively, the addition had cleared 10 m in 11 days, at a rate not less than 0.91 m d\u22121. Adjusted for the density and viscosity of the water at the top of the water column, the predicted sinking rate was 1.42 m d\u22121. Thus, overall, the value of t\u2032 for the \ufb01rst 10 m (= 7 days) was exceeded by the observed te (= 11 days) by a factor of only 1.57. Part of the explanation is that sinking spores would have sunk more slowly than 1.42 m d\u22121 in the colder hypolimnion. However, the model expla- nation envisages a daily export of the population from the upper mixed layer (varying between 0.5 and 4 m during the course of the experiment), calculated as N exp \u2212(ws/hm), whence it contin- ues to settle unentrained at the rate ws m d\u22121. To judge from Fig. 2.20, this is an oversimpli\ufb01cation but the prediction of the elimination is reason- able. The same model was applied to predict the distribution and settlement of Lycopodium spores in the third experiment, conducted during the autumnal period of weakening strati\ufb01cation and mixed-layer deepening. Variability in wind forc- ing was quite high and a certain degree of re- entrainment is known to have occurred but the time taken to achieve 95% elimination from the upper 9 m of the water column (te = 18.0 days) at the calculated in-situ sinking rate (ws) of 1.32 m d\u22121 exceeded the equivalent t\u2032 value (9/1.32 = 6.82) by a factor of 2.6. The three results are held to con\ufb01rm that the depth of entrainment by mixing is the major con- straint on elimination of non-motile plankters heavier than water, that the eventual elimina- tion is however delayed rather than avoided, and that prolongation of the period of suspension is in proportion to the depth of the mixed layer, wherein u\u2217 \u2265 15 (ws). 2.7 The spatial distribution of phytoplankton The focus of this chapter, the conditions of entrainment and embedding of phytoplankton in the constant movement of natural water masses, is now extended to the conditions where water movements are either insuf\ufb01ciently strong or insuf\ufb01ciently extensive to randomise the spa- tial distribution of phytoplankton. This section is concerned with the circumstances of plankters becoming disentrained and the consequences of 78 ENTRAINMENT AND DISTRIBUTION IN THE PELAGIC weakening entrainment for individuals, popula- tions and communities, as augured by spatial dif- ferentiation in the vertical and horizontal distri- butions of natural assemblages. Distributional variation is subject to issues of scaling which need to be clari\ufb01ed. It has already been made plain that aquatic environments are manifestly heterogenous, owing to spatial differ- ences in temperature, solute content, wind stress, etc., and that each of these drivers is itself sub- ject to almost continuous variation. However, while precise values are impossible to predict, the range of variability may be forecast with some con\ufb01dence, either on the basis of averag- ing or experience, or both. We may not be able to predict the intensity of wind mixing in a lake some three weeks or more into the future but we may estimate from the knowledge base the probability with which a given wind inten- sity will prevail. The changes in temperature, insolation, hydraulic exchanges and the delivery of essential nutrients affecting a given stretch of water also occur on simultaneously differ- ing scales of temporal oscillation \u2013 over minutes to hours, night\u2013day alternations, with changing season, interannually and over much broader scales of climatic change. The nesting of the smaller temporal scales within the larger scales holds consequences for phytoplankters in the other direction, too, towards the probabilities of being ingested by \ufb01lter-feeders, of the adequacy of light at the depths to which entrained cells may be circulated, even to the probability that the energy of the next photon hitting the photo- synthetic apparatus will be captured. The point is that the reactions of individual organelles, cells, populations and assemblages are now gen- erally predictable, but the impacts can only be judged at the relevant temporal scales. These responses and their outcomes are considered in later chapters in the context of the relevant pro- cesses (photosynthesis, assimilation, growth and population dynamics). However, the interrelation of scales makes for fascinating study (see, for instance, Reynolds, 1999a, 2002a): in the end, the distinction is determined by the reactivity of the response. This means that critical variations alter more rapidly than the process of interest can respond (for instance, light \ufb02uctuations are more frequent than cell division) or so much less rapidly that it is perceived as a constant (such as annual temperature \ufb02uctuations having a much lower frequency than cell division) which will be no more relevant to today\u2019s populations than is the onset of the next ice age to the mainte- nance of present-day forests (Reynolds, 1993b). In between, where driver and response scales are more closely matched, the interactions are rather more profound, as in the frequency with which new generations are recruited to a water col- umn mixed to a different extent on successive days. The variability in the instantaneous distribu- tion of phytoplankton may be considered in rela- tion to an analogous spatial scale. Consider \ufb01rst a randomised suspension of unicellular \ufb02agellates, such as Chlamydomonas or Dunaliella.