DEOXYRIBONUCLEIC ACID CHARACTERIZATION OF THREE SPECIES OF DIATOMS WITHIN GENUS THALASSIOSIRA

A comparison was made of the lengths of the restriction fragments produced when total DNA from eight clones of three species of diatoms, Thalassjosjra rotula (Meunier), L grayjda (Cleve) and L nordenskjoeldjj (Cleve), was digested with twelve restriction endonucleases and probed with two heterologous chloroplast DNA probes from ~ rejnhardtii. DNA from the L rotula clones would not hybridize with the .C..... rejnhardtjj probes under high stringency conditions but the L grayjda and L nordenskjoeldjj DNA hybridized strongly to these same probes. The most likely reason for lack of hybridization with L rotu la DNA is its contamination with high amounts of polysaccharide material. There was a high degree of molecular similarity in the DNAs within a species for L nordenskjoeldjj and L grayjda (100 % and 91%, respectively), but much less between species (32%) implying a genetic congruence with the morphological species concept in diatom systematics. Genetic similarity between and within species is based on the probe detection of 44 restriction fragments, or 264 DNA bases which comprise 0.18% of the entire chloroplast genome, estimated at 147 kilobase pairs.


INTRODUCTION
The marine diatoms Thalassjosjra grayjda (Cleve) and L rotula (Meunier) exhibit a mutually exclusive distribution in the Pacific and Atlantic Oceans with some overlap. Lg rayjda is found in high latitudes of both hemispheres from about 70° to 80° down to approximately 30° S and N and L rotula occupies the low latitudes approximately between 30° S and N (Hasle 1976). Specific morphologies characterize each species in their respective biogeographical zones, but transitional morphotypes are seen at the latitudes where the two species overlap (Syvertsen 1977). These observations along with physiological studies of L rotula and L grayjda in culture, varying temperature and nutrient concentrations (Syvertsen 1977), have cast doubt on whether L rotula and L grayjda are indeed distinct species. To add to the biogeographical and morphological evidence, this research will attempt to determine if clones of each species are genetically distinguishable by restriction fragment analysis of their chloroplast DNA (cpDNA).
One method of genetically determining if there are distinct lineages for these species, and thereby prove (or disprove) that they are evolving along separate lines, is to compare the relative chloroplast DNA (cpDNA) nucleotide sequences. cpDNA is a very slowly evolving genome (Palmer 1985a;Palmer 1985b;Palmer 1987) and therefore is an ideal molecular marker for determining relative evolution of genomes between and within species. Comparative cpDNA sequence analysis has been used to determine taxonomic affinities between species of land plants such as the genera Clarkia (Sytsma and Gottlieb 1986), Lisjanthjus (Sytsma and Schaal 1985), Lycopersjcon (Palmer and Zamir 1982), to determine interspecies, intergenus and interfamilial taxonomic affinities in several kelp genera (Fain 1987), interspecies relationships within the centric diatom genus Cyclotella (Bourne et al. 1987), intraspecies and interspecies relationships in the diatom genera Coscjnodjscus and Odon te II a (Kowallik 1990), several red algae species (Goff and Coleman 1988) and the brown algae genera, Pylajella and Sphacelarja (Dalman et al. 1983), and to differentiate physiological races within the diatom genus Skeletonema (Stabile et al. 1990).
The · method of cpDNA analysis used in this study was restriction fragment length polymorphisms (RFLP), a technique of measuring genetic relatedness between organisms by comparing the size of DNA fragments produced upon digestion with restriction endonucleases. Restriction endonuclease enzymes recognize and cleave at specific double stranded DNA sequences of four to eight bases within the genome. The number of fragments produced depends on how many times that specific sequence appears in the DNA and the size of the fragments produced will depend on the relative positions of the numerous restriction sites. The variability of the fragment pattern produced when two or more DNA samples are digested with the same enzyme, will reflect the variability in the DNA sequence at those specific sites. When this analysis is done with many different enzymes, each specific for a different nucleotide sequence, a comparison of the resulting fragment patterns will reflect the degree of genetic relatedness.
In this study, cloned cpDNA gene sequences from highly conserved areas of the chloroplast genome of the chlorophyte Chlamydomonas rejnhardtii were used to probe restriction enzyme digests of total DNA from multiple clones of L. rotula and L grayjda. L nordenskjoeldjj, a species which is morphologically distinct from both L..grayjda and I..... rotula, were used for comparison. The restriction fragment length polymorphisms generated from these analyses indicates the degree of genetic relatedness on the intra-and the interspecies level within this genus, for these three species.

Taxonomy
The Thalassjosjra nordenskjoeldjj and L rotula clones used in this study were identified under the phase contrast microscope first as live chains of cells in natural assemblage taken from net tows from Narragansett Bay, Rhode Island, Hull Bay, Massachusetts and Friday Harbor, Washington. The L rotu la clone from Friday Harbor (clone F4) was kindly provided by Jan Rines. Once a pure culture of a suspected L rotula or L nordenskjoeldjj species was obtained, final identification was accomplished by observing the valve structure of hydrogen peroxide -cleaned frustules using high power, oil immersion, phase contrast microscopy. The L grayjda clones used were obtained as pure cultures from the Provasoli -Guillard Center for Culture of Marine Phytoplankton, West Boothbay Harbor, Maine. The species identification for Lg rayjda was also verified by observing valve morphology of cleaned material.
Thalassjosjra rotula Meunier, is identified in live mounts · as chains of disk-shaped cells with one or two thickened girdle bands, more noticeable in cells devoid of organic material (lebour 1930;Meunier 1910). The connecting strand is thicker than in most species of Thalassjosjra due to the numerous threads composing this strand. Seen in valves cleaned of organic material and mounted in resin are numerous central strutted processes, a single row of strutted processes at the margin and ribs of areolae, lacking tangential walls, radiating from the center. A single labiate process is located at the margin. In valve view, some isolated girdle bands appear "unevenly thickened" due to the presence of a septum on the internal side of the intercalary band. These bands also appear "unevenly broadened" in girdle view (Meunier 191 O;Syvertsen 1977).
L rotula belongs to the group of Thalassjosjra with the greater length of the processes on the external side of the valve and a single marginally located labiate process (Hasle 1968).
Cells of Thalassjosjra grayjda Cleve have been described as being less flattened in girdle view, and lacking the "unevenly thickened" girdle bands typical of L rotula (Cleve 1896;Meunier 191 O;Syvertsen 1977). In valve view of cleaned, mounted material, Lg rayjda is in all ways similar to L rotula except for the welldeveloped radial areola possessing tangential walls and thus eliminating the "ribbed" appearance (Hasle 1968 (Stabile, 1988 2). An absorbance value at 260 nm of 1.0 in a 1.0 cm cuvette is equal to a DNA concentration of 50 ug/ml or an RNA concentration of 40 ug/ml . Since absorbance at A280 is proportional to protein concentration (tryptophan and tyrosine amino acids absorb radiation at 280 nm) the A2sol A2ao ratio is then an indicator of relative purity of nucleic acid solutions (purification steps 3, 6, 9 and 12,  Table 2 lists the 15 DNA preps done on multiple clones of L nordenskjoeldji, L. grayjda, and L. rotula with some data from the significant steps in each prep. Although some of the purification steps and measurements were not done for some of the preps (dashed line in Table 2) there are ample data available to make some comparisons between and within preps. For example, centrifuging the trench press lysate (purification step 1) resulted in significantly less precipitated nucleic acid (A2 so units in purification step 5) in preps 1, 4 and 6 when compared to preps 9, 1 O and 13 where the centrifugation step was omitted. This is likely due to the large amount of cytoplasmic RNA which is not pelleted.

Results
The A2sol A2ao ratios in purification step 3 indicate relatively high amounts of protein in preps one and four which were eliminated in the subsequent digestions with proteinase K (purification step #4,  Table 2) the L rotula nucleic acid sample's A2so/ A2ao ratios remained at approximately 2.0. These DNA samples digested less consistently well than did the DNA with A2solA2ao ratios closer to 1.8 (L grayjda and L nordenskjoeldji) and exhibited less apparent homology to heterologous cpDNA cloned probes from ~ rejnhardtii than did the DNA from L nordenskjoeldii and L grayjda.
The total DNA samples from the eight diatom clones studied were assayed for their DNA, RNA, protein and polysaccharide content in order to determine a chemical dissimilarity with the L rotu la 20 DNA (Table 3). DNA concentrations, determined by absorbance at 260 nm, were always greater for the L rotu la DNA when compared to the concentrations as determined by Hoechst dye fluorescence.
Conversely, with the two clones of I... g rayjda and one of L nordenskjoeldjj, the OD 260 method DNA estimates were lower than the DNA estimates as determined by fluorescence. Protein concentrations were not significantly higher in the L rotula samples than in the others and this is also indicated by the fact that all the DNA solutions A260/A280 ratios were greater than 1.8.
Polysaccharide, as glucose equivalents, was detectable in all the diatom DNA assayed. When measured as microgram equivalents of glucose-D per microgram of DNA (determined fluorometrically) it was found in five to ten fold higher ratios in three clones of L rotula .DNA and in one clone of L nordenskjoeldjj DNA (clone HB3).
Although all the polysaccharide measurements were corrected for the known DNA content (nucleic acid is a ribose-polymer therefore detected by the assay) they were not corrected for RNA content.
When a qualitative RNA assay was done only the one L.
UV spectral analysis of all the DNA samples from the three species of Thalassjosjra also suggests a physical or chemical dissimilarity in the DNA isolated from L. rotula (figure 3). All the DNA samples, including purified salmon sperm DNA used as a standard, exhibited two absorbance peaks, one at approximately 260 nm and another at 213 to 224 nm. The 260 nm peak, the absorbance maximum for nucleic acid, broadens and flattens in the UV spectra for the two clones of L. rotu la while this peak in all other DNA spectra is sharp. The identity of the substance responsible for the second absorbance peak is not known, but is found in all the DNA samples including the lambda DNA standard and is therefore not considered anomalous.
CsCI isopycnic ultracentrifugation using the DNA binding dye Hoechst 33258 (Aldrich and Cattolico 1981;Aldrich et al. 1982;Goff and Coleman 1988)  and X1 and L grayjda clone TTG2 exhibited double broad diffuse bands with no sharp bands appearing at all. The L grayjda clone TTG4 DNA, however, exhibited a single lower broad diffuse band and · two upper sharp, well-defined bands which were so closely spaced, they had to be removed as one band, band TTG4-1. All the CsCl-Hoechst dye gradients were run identically, the . only difference being the quality of the total DNA loaded onto them. The three Qradients which produced broad diffuse bands, HB3, X1 and TTG2 were loaded with total DNA which had been initially purified in a CsCl-ethidium bromide gradient.
All other DNA loaded onto CsCl-Hoechst dye gradients had not been initially purified with a CsCl-ethidium bromide gradient (See table 2 That the upper band in the CsCl-Hoechst dye gradient contains the cpDNA fraction is consistent with the binding properties of Hoechst 33258. This dye preferentially binds to A-T rich regions of double-stranded DNA (Manuelidis 1977). cpDNA, containing a high proportion of A-T regions (Palmer 1987) binds Hoechst dye to a greater extent then mitochondrial or nuclear DNA. The intercalation of this dye within the rigid circular double helix structure of cpDNA causes some unwinding and results in a greater buoyancy in CsCI gradients. cpDNA from kelp (Fain et al. 1988), Olisthodiscus luteus (Aldrich and Cattolica 1981;Aldrich et al. 1982) and some red alga species (Goff and Coleman 1988) also are isolated from total cellular DNA as the uppermost fraction on CsCl-Hoechst 33258 isopycnic gradients .
Estjmates of chloroplast genome size for T. rotula.
To estimate the chloroplast genome size for L rotula , clone frequency within these genomes studied. Table 4 lists those, presumably chloroplast, DNA fragments which contain the sequences homologous to the rbc L and 235 rDNA genes of .Q... rejnhardtjj cpDNA. As these gene sequences are known to be highly conserved through phylogenetic lines (Palmer 1985a;Palmer 1987) and because high stringency hybridization conditions were employed, it is assumed that those fragments contain, in whole or in part, those same gene sequences in diatom cpDNA.
The rbc L probe and the 23S rDNA probe rarely hybridized to the same fragment (Table 4), with the exception that these two gene sequences are identified to the same 17 .5 kb Eco RV and 1.5 kb Ora I fragments in the three clones of L nordenskjoeldii (Table 5). More fragments were detected by the 23S rDNA probe than the rbc L probe (27 versus 19). This may be due to the higher concentration of the ribosomal DNA genes in cpDNA due to their typical occurrence on the inverted repeat region of the genome and the rbc L gene typically being a single copy gene (Palmer 1985a;Palmer 1987).
The fragment patterns detected by the two gene probes were identical for the three clones of L no rd ens k i o e Id j j for all ten restriction digests ( Figures: 8-12, Table: 5). There are limited data for clone HB3 due to lack of hybridization of probes to many of these digests, · probably due to small amounts of DNA loaded onto these gels. But for the data available, the two clones isolated from Hull Bay, Massachusetts, one year apart (clones HB3 and ZUD2) and the one clone isolated from Narragansett Bay (clone X1) show identical banding patterns. The two clones of Lg rayjda used in this study (clones TTG2 and TTG4), both isolated at the same time from Tromso, Norway but separated from one another in culture for more than ten years, also exhibit banding patterns identical to one another with the one exception of the Eco RI -238 rDNA fragments differing in size by 0.6 kilobases.
In none of the L grayjda DNA digests were the two probe gene sequences located on the same fragment, in contrast to th. e two L. nordenskjoeldjj fragments already mentioned. These results seemed to be highly reproducible for at least six of the enzymes and the P177 (238 rDNA) gene probe.
The fragments identified on the replicate gels (numbers in parenthesis, table 5) corresponded to one another very well, the calculated sizes differing, in most cases, by only a few kilobase pairs.
Data from table 5 can be used to build similarity matrices to quantify the size differences of those fragments detected by the two ~ rejnbardtjj cpDNA probes. Those data for each probe in table 5 which comprise the most complete data set are used (numbers in bold type). Because of the paucity of data for L nordenskjoeldii clone HB3, these data are omitted from the similarity matrix. Table   6 shows three matrices, one each for the two genes considered separately, and one considering both genes together. The equation  There are a number of possible reasons why the ~ rejnhardtjj cpDNA probes did not hybridize sufficiently well to L rotu la DNA as compared to the other diatom DNA, or why it sometimes resisted digestion with restriction enzymes. The most compelling reason is the quality of the DNA. The high amounts of polysaccharides found in the preparations of L rotu la DNA along with its anomalous UV spectra and UV absorbance for quantification, suggest that there may have been some kind of chemical interference responsible.
Polysaccharide contamination of DNA preparations is not unusual in algae (Edelman 1975;Edelman et al. 1967;Murray and Thompson 1980) and has been reported to affect DNA-DNA reassociation kinetics in plant studies (Murray and Thompson 1976). Edelman et al. (1967) found polysaccharides (from the blue-green algae, order Oscjllatoriales) to consistently contaminate their DNA preparations, even following repeated preparative CsCI gradients, phenol extractions and ethanol precipitations (Edelman et al. 1967).
The suspected polysaccharides all had buoyant densities in CsCI similar to that of DNA (DNA:1.75-1.68 g/cm3: polysac: 1.75-1.65 g/cm3) therefore they comigrated with the DNA through the CsCI gradient steps (Edelman 1975). Ethanol precipitation of the DNA may have also precipitated the polysaccharides as this is a customary procedure for isolating polysaccharides from algae (Beattie et al. 1961 ).
Also, although this possibility was not tested for the L rotula DNA, it is possible that the anomalous UV spectra could be a This scattered light, not transmitted to and counted by the photocollector, is "assumed" by the instrument to be absorbed, and is included in the optical density value (Kirk 1983).
The high relative amounts of polysaccharide to DNA in the L rotula DNA preparations (5.5:1 to 31.5:1; and hybridization reactions (Brooks 1987;Wahl et al. 1987).
Other possible contaminants which would have inhibited restriction and hybridization of L rotu la DNA are CsCI salts and phenol.
Although phenol is a known inhibitor of restriction endonucleases (Brooks 1987;Maniatis et al. 1983), it is unlikely that it was a problem here since several purification steps followed the phenol-chloroform extractions and would have rid the DNA of any trace of phenol. Cesium chloride contamination is an outside possibility, however. Several times following the cesium chloride step when the fluorescing DNA band, removed from the CsCI gradient, was precipitated with ethanol, crystalline salts also precipitated. These were assumed to be CsCI salts. This problem was remedied, however, by diluting the DNA/CsCI mixture many-fold with water before precipitating with ethanol.
Another possible reason for the L rotula DNA resistance to restriction digestion is the presence of methylated bases within the recognition sites for those enzymes. Modification of recognition site bases by methylation prevents restriction at that site by a particular enzyme (Raleigh 1987). There are two reasons, however, why this is not a viable reason for the lack of consistent restriction of L. re tu I a DNA.
Firstly, although plant nuclear DNA has consistently been found to contain methylated bases, to date, no cpDNA has been found to be methylated (Palmer 1986). Assuming The chemical dissimilarity of the isolated DNA product from the clones of Lg rayjda and L rotu la as indicated by the differences in their polysaccharide content (table 3) and UV spectra (figure 3), at best, only suggest ecophenotypic variation and not actual genetic differences between the species. The L rotula clones were cultured at higher temperatures (15° C) than both the L.grayjda and L nordenskjoeldjj clones (5° C) and also had a longer period in darkness just prior to harvesting. Because of these differences in culturing conditions, no conclusions can be made concerning genotypic differences between the L grayjda and L rotula clones on the basis of these results alone.
Autoecological studies with these two species to determine true genetic differences, either at the level of species or merely as physiological races, would have to be performed under identical growth and harvesting conditions to eliminate these environmental influences.
The differences in polysaccharide composition of the purified DNA's between the three species of Thalassjosjra probably reflects differences between species in production or utilization rates of intracellular storage polysaccharides. The most common storage polysaccharide found in the diatoms is chrysolaminarin (leucosin), a twelve-unit glucose polymer with B 1-3 and B 1-6 linkages, which has been isolated as a water soluble fraction from numerous diatoms (Allan et al. 1972) and other members of the Chrysophyta (Beattie et al. 1961 ). It is known that the cell concentration of storage polysaccharides in diatoms will decrease under conditions of darkness as the cells utilize this stored energy by respiration in the absence of lightdependent photosynthesis (Handa, 1969).
Although the diatoms cultures in this study were grown under continuous light, they were placed in darkness for one to several days just prior to harvesting to facilitate cell sinking. This cessation of photosynthesis undoubtedly resulted in a period of "dark respiration" and a reduction in cell polysaccharide in all cultures .
The L nordenskjoeldjj and Lgrayida cultures, however, were routinely kept in darkness several days longer than the L rotula cultures, due to their longer settling times, a function of slower sinking rates in the colder, more viscous seawater (5° C vs . 15°C,   (Hasle 1968;Hasle 1978) indicate a genetic divergence which should also be reflected in the relative sequence homologies of their cpDNA. Also, presuming from the strength of these hybridization signals, that there is a greater genetic distance between L. rotula and L grayjda than between L grayjda and .C.... rejnhardtii would seem unsound based on the vast morphological and biochemical differences separating these two classes (Dodge, 1973;Rothschild, 1987). These inferences tend to support the idea that the lack of strong hybridization of .Q.... rejnhardtii cpDNA probes to L rotula DNA does not indicate a difference in gene sequence but rather the presence of a chemical contaminant preventing strong DNA-DNA interactions.
The strong hybridization of .Q. reinhardtii chloroplast rbc L and 23S rDNA gene probes to the L grayjd~ and L nordenskjoeldjj DNA does, however, indicate a high sequence homology within these genes between these two classes of algae (Bacillariophyceae and Chlorophyceae) and supports the idea of the highly conserved nature of these genes across phylogenetic lines (Palmer 1985a;Palmer 1987;Palmer and Zamir 1982).

Members of the algae classes
Bacillariophyceae and Chlorophyceae are phylog~netically quite distant, and it has been proposed, within the framework of the endosymbiont theory, that they acquired their chloroplasts as separate endosymbiotic events either along a monophyletic track (Cavalier-Smith 1982;Cavalier-Smith 1986) or a polyphyletic one (Whatley et al. 1979).
There are numerous morphological, biochemical and ontogenetic characters which separate these two classes of algae including pigment types, storage products, cell enclosures, life cycle events, and chloroplast structure (Dodge, 1973;Rothschild, 1987). Indeed, even if the evolutionary relationships of the chloroplasts as endosymbionts alone are considered between these two classes there are ample differences here at the ultrastructural level to suggest either polyphyletic origins for plastids or extreme evolutionary changes. For instance, when representative members of the diatom and the green algae groups were examined for their cpDNA configuration within the chloroplast, it was discovered that in the diatoms, the cpDNA was arranged in a ring nucleoid within the girdle lamella and in the green algae as point nucleoid clusters surrounding pyrenoids (Coleman 1985).
Also, the chloroplasts of the green algae have two encircling membranes while the diatoms have four (Dodge, 1973;Rothschild, 1987). While most of these · features which delineate these two classes of algae are probably genetically controlled and there is undoubtedly vast heterogeneity within the genomes of these two classes of organisms, the primary sequences of some genes remain evolutionarily static, controlled by physiological and structural constraints. Evolution at these conservative sites proceeds very slowly and homologies extend across vast phylogenetic distances.
Estimates of the conservation of the primary nucleotide sequence of the 235 rDNA gene are 92% between tobacco and maize and 78% between tobacco and the cyanophyte Anacystjs (reviewed in Palmer, 1985b). Similarly for the rbc L gene (large subunit for RUBISCO), high sequence homology has been found between such divergent groups as Chlamydomonas and tobacco (86-95%) and chlorophytes and cyanobacteria (79-85%) (reviewed in Palmer, 1985b). No estimate on the sequence homology of these genes between the diatoms studies and .Q... rejnhardtjj can be made based on these thesis data other than that it was high enough to allow hybridization between DNAs. The restriction fragment data reported also does not indicate anything about the relative evolution of these specific genes between the classes, they are only used to identify those DNA fragments of chloroplast origin and are not indicative of restriction site homologies within the 235 rDNA and rbc L genes themselves.
Other ribosomal protein genes examined in the chloroplast genome have been shown to have much lower sequence homologies across similar phylogenetic lines (i.e ., angiosperms and Euglena: 39% for rps4 , 43% for rp/2, 55-58% for rps 19, reviewed in Palmer, 1985b). Indeed, when cpDNA ribosomal protein gene segments cloned from Njcotjana tabacum (obtained from Dr. Sigiura, Nagoya University, Japan; plasmids pTB28 and pTB13 containing ribosomal protein genes rpl 23, rpl 2, rps 19, rpl 2, rps 3 a~d rpl 16) were used as gene probes against L rotula and L nordenskjoeldjj total DNA digests, no hybridization occurred, even under low stringency conditions (data not shown). Tobacco rbc L and 23SrDNA probes, unfortunately, were not available for hybridization attempts to the diatom DNA.
The I... grayjda and I... nordenskjoeldii restriction fragment length polymorphism data (table 5 and 6) indicate inter-specific cpDNA variation greater than the variation between clones of the same species for the gene probes used. Roughly 32% of the restriction sites analyzed were found to be conserved between the two species whereas 91 % and 100% were found to be conserved within the species L grayjda and L nordenskjoeldjj respectively. This is only a nominal conclusion, however, because when only a portion of all the cpDNA restriction fragments are visualized, as is the case in these analyses, no conclusions can be made about the nature of the differences between genomes.
A length mutation (multiple base deletion or addition) occurring between, say, two Hind Ill restriction sites in one of two DNA samples being compared, can appear as a restriction site change unless the complete set of cpDNA fragments resulting from the restriction enzyme digestion of the complete genome can be visualized and compared. For example, consider two hypothetical species, denoted A and B, and species B is directly evolved from species A. If all of the restriction fragments that comprise the chloroplast genome are visible on an agarose gel for hypothetical species A and B, a gain of a restriction site within an A fragment due to a base deletion or addition, (in the evolution from species A to species B) will result in two B fragments whose summed sizes will equal the size of the ancestral A fragment. The error involved in quantifying fragment sizes on the agarose gels due to both the variability of fragment mobilities (as shown by the standard deviation of the lambda DNA standard digests, see methods) and the inability of 0.8% agarose gels to adequately separate DNA fragments greater than approximately ten kilobase pairs, puts a lower limit of approximately 7 kb on the fragment size variation which can be measured between clones. The extent of the error depends on the position to which the fragment migrates, being greatest near the top of the gel due to the non-linear nature of the separation (see figure 1 ). Therefore, this analysis, at worst, cannot detect base deletions or additions less than seven kilobase pairs when they occur within larger fragments migrated near the top of the gel. This would possibly exclude many of the diatom DNA length mutations from being detected as the majority of length mutations occurring in cpDNA are reported to be 1-10 kb in length (Palmer 1987).

Conversely, if a detectable length mutation occurs within an
The significance of these results for the two clones of L grayjda is tempered by the fact that the two cultures were isolated from the same area at the same time (see table 1) and may represent genetically fdentical asexual progeny of one another. It is interesting, though not surprising considering the slow rate of evolutionary change in the chloroplast genome, that their separation in culture for ten years did not produce significant observable cpDNA differences.
Two of the three clones of L nordenskjoeldjj (clones Zud2 and X1) which exhibited complete sequence homology for those areas of the chloroplast genome examined, were isolated from different water Massachusetts).

masses (north and south of Cape Cod,
Although it is conceivable that these water masses could mix and exchange biota via the Cape Cod Canal or around the peninsula, it is unlikely that X1 and Zud2 represent recent clonal siblings. It is not known, however, whether the extent of the intraspecific variation seen in this study is truly representative of the entire diatom populations in nature. To examine this problem would indeed be a Herculean task, considering the magnitude of geographically contiguous diatom populations and the · complexity and expense of the . DNA isolation and characterization methods. The chloroplast genome size estimate for L rotula (clone F4) of 14 7 kilo base pairs should be considered approximate. The value is the sum of the sizes of the fragments produced when the presumptive chloroplast DNA from the CsCl-Hoechst dye gradient is digested with a single restriction enzyme. All bands are assumed to be chloroplast DNA with no contribution from mitochondrial and nuclear. It is possible, though, that there is present some of this contaminating DNA. The presumptive cpDNA band from the CsCl-Hoechst dye gradient was removed as a very tight and distinct fluorescing band indicating that the DNA was of homogeneous physical and chemical character, but it is reasonable to · . assume that in removing the cpDNA band from the centrifuge tube, some contaminating, non-chloroplast DNA was removed also. This is seen in figure (F4 1-3 autorad) where the Chlamydomonas cpDNA gene probe hybridized to small amounts of cpDNA which contaminated the lower two CsCl-Hoechst gradient bands, F4-2 and F4-3. If cpDNA contaminated the lower two bands in the CsCl-Hoechst gradient, it is prudent to assume that the opposite occurred: DNA from these two lower bands may have contaminated the upper cpDNA band.
Estimates of DNA content per cell varied over a four-fold range for the three species of Thalassjosjra (table 3)  found . that cellular DNA content was one to three percent of the total carbon content, but no correlation statistics or linear regression equation was given for this relationship (Holm-Hansen 1969).
As an internal check on these values, the cellular DNA content for clones HB3 and TTG2 was also estimated using Holm-Hansen's empirical formulas relating the carbon content to DNA content of diatoms (Holm-Hansen 1969) (Table 7). Although carbon values were not determined for these diatom clones, estimates using the empirical formula of Smayda [Smayda, 1965] relating the plasma volume (to. tal volume minus the vacuole volume) to carbon content were used and considered reasonable for nutrient replete cell growth (Strathman 1967) but neglecting the effects of temperature (Eppley, 1972). The DNA content measured for L grayjda. clone TTG2 sequence homology and supports the idea that these genes are highly conserved across vast phylogenetic distances (Palmer 1987).
The species identification for the clones of L grayjda and L rotula, as described in methods, are assumed to differentiate the two taxa in this study based on the standard morphological criteria for diatoms. These taxonomic criteria, which recognize relatively "stable" characters (areola pattern, size and shape of girdle bands and overall size and shape of frustule), can to a great degree be controlled by external conditions, and, consequently, are likely to be of little taxonomic utility for distinguishing between L grayjda and L rotula. Many field observations of net tows taken in areas which are considered transition zones for the two species (Hasle 1976} reveal mixures of both L grayjda and L rotula cell types (Paasche 1961;Syvertsen 1977} sometimes present within the same colony (Hasle et al. 1971}. Indeed, the L rotula clones used in this study were observed to have valves with a mixture of areola pattern types, showing the "L rotula" type pattern near the valve center and the "L grayjda" type pattern near the valve margins (pers. observ.). These clones, however, were grown at 15° C, several degrees above what is considered the temperature of the transition zone ( -3 to 10° C).
Whatever the physiological reasons behind this morphological variability are, though, it is clear that these taxonomic criteria are not significantly rigorous to distinguish between L grayjda and L rotula under all ecological conditions.
The problems enountered in identifying and distinguishing between closely related species of diatoms are a result of a classification system based largely on the typological species concept, more specifically, based on the species defined by a specific set of morphologic characters (Makarova 1980;Mayr 1970).
This method of classification is necessary due to the relatively few or non-existant ecological, behavioral, ontogenetic and biochemical characters available to the diatom systematist to group this huge array of single celled eukaryotes. This is complicated by the fact that the diatom's primary mode of reproduction is by asexual fission, thus eliminating reproductive compatability and the biological species concept as a criteria by which to classify species.
All that is left to distinguish between species ·of diatoms is morphological characters.
This study, and others like it, has shown that molecular characterization of an organism's genome can be used to infer phylogenetic distances between and within most levels of taxa . If evolution in its strictest sense is considered to be the gradual accumulation of mutations within a population's gene pool through geologic time, then, by quantifying and characterizing these mutations, phylogenetic distances and relationships can be quantified, thus building a taxonomic system based on evolutionary relationships and trends. The concept of a species, as defined by Florkin (1966) is particularly salient here: ... a species [consists of] groups of individuals with more or less similar combinations of sequences of purine and pyrimidine bases in their macromolecules of DNA, and with a system of operators, controllers and repressors leading to the biosynthesis of similar sequences of amino acids, the integration of which, in one cell, or in a number of variably differentiated cells, leads to similar stuctural and functional characteristics, adapted to the ecological niche in which the species flourishes.
This definition can be considered as an addendum to the biological species concept as reviewed by Mayr (1970) which defines a species, -· in part, as a protected gene pool. By including the molecular functions of the genes and the gene products (proteins), Florkin implies in his definition that species can be defined at the molecular level. A system of taxonomic classification which uses molecular as well as other types of character data (morphological, ontogenetic, behavioral, biochemical, etc.) would benefit by incorporating information on the evolutionary trends and distances of the individuals within the lineages considered.
The method of total DNA purification used in this study was satisfactory for isolating large amounts of high molecular weight DNA.
It is fairly rapid and inexpensive, the only two pieces of expensive equipment required being an ultracentrifuge and a trench press. The effectiveness of the method, however, depends heavily on the nature and physiologic condition of the species of algae studied.
The interferences due to contaminating substances, such as polysaccharides, must be avoided by carefully controlling the initial growth conditions of the organism or by using suitable methods to eliminate the known contaminant. In this study, possibly longer periods in the dark for the L. rotula clones or further purifying the DNA by affinity chromatography, would have solved the problem of interfering substances.
The method of restriction fragment length polymorphism employed here has advantages over methods used previously for molecular systematic analysis such as DNA-DNA hybridization (Britten et al, 1974) and iso-enzyme electrophoresis (Murphy and Guillard, 1976  The RFLP method, however, is not without its problems, and will probably be rapidly supplanted in the future by more efficient means of analyzing genetic evolutionary change (Olsen 1990). Aside from the many technical problems already discussed for the RFLP method (purity, laborious methods, expense), there is the question of determining the nature and extent of the genetic change. The RFLP method used in this study, as already noted, does not describe the actual structural differences in the cpONA but merely shows that differences exist. In its most rigorous form, RFLP analysis compares restriction fragment patterns of the entire genome (requiring highly purified cpDNA) and identifies point and length mutations, thereby describing these differences with much less ambiguity. To detect rearrangements of the genome, a common evolutionary change in the algal lineages (Palmer 1985b), RFLP analysis and southern blot hybridizations can be enhanced with genome mapping to determine relative positions of specific genes (Palmer 1986). Also, using polymerase chain reaction (PCR) techniques, specific genes located on short portions of DNA can be amplified to obtain the required large amounts of material. In this way the nucleotide sequences of specific genes can be elucidated and their mode and tempo of evolution be determined. Using these methods, a clearer picture of the degree of genetic relatedness between extant algal species can be obtained from which lines of evolutionary descent can eventually be inferred.      Range 0.9-23 .0 kbp Mean size (+/-SD) 9.21 +/-5.96 Table 5 Data from autoradiographs in figures 7 and 8 showing a comparison of the hybridized restriction fragments for five clones of two species of diatoms. Bold print indicates those data used in calculating similarity matrix tables (  Table 6 Similarity matrix table calculated from equation S = 2 N xy/N x +Ny (see text for details) . The numbers above the diagonal indicated the total number of shared restriction fragments (N xy ), the numbers below the diagonal, the proportion of shared restriction fragments (S), and the numbers along the diagonal in parentheses are the total number of restriction fragments considered from each set (Nx and Ny).   3.5 4.5 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 Mobility (cm)

Figure 2.
Diatom Total DNA with and without RNAse digestion electrophoresed on 0.8% agarose gel in TAE buffer. All samples were incubated for 2 hr with and without 2 ug RNAse in lX Hind III reaction buffer in a total volume of 20 ul.
Optical density values increase upwards on y-axis. Centrifugation parameters are listed in methods section.
All CsCl -Hoechst dye gradients were run at 40,000 to 45 ,000 rpm for 48 hours in a 80Ti Beckman rotor and ultras peed centrifuge at 20°C. The CsCl density was 90% wt/vol. All samples except HB3, Xl and TTG2 were total nucleic acid without prior CsCl purification.
Remaining samples (F4, T3, TR4 and TTG4) were initially banded in 1.65 g/ml CsCl -ethidium bromide gradient run at 40,000 rpm for at least 12 hours in a 80Ti Beckman rotor at 20°C.