Discussion

The results presented here indicate that the degree of relatedness of restriction patterns of higher plant ct-DNAs reflects their position in the taxonomic scheme.Tobacco and tomato，for example，are classified in the same family and share several similar EcoRI fragments whereas tobacco and corn.which are from different classes（dicot and monocot），show no common feature.This is essentially the same observation obtained by DNA-DNA hybridization studies in which 85%and 15%sequence homologies were estimated between tobacco and tomato and between tobacco and barley，respectively（Bisaro and Siegel，1980）.Similarly some common fragments are generated by HindⅢ（see Figure 1），EcoRI，BamHI，or Smal（Rhodes and Kung，1981）from ct-DNA of tobacco and spinach which share 60%—70%sequence homology（Bisaro and Siegel，1980）.If estimates of ct-DNA sequence homology based upon restriction fragment patterns parallel those based upon hybridization data then the restriction fragment evidence may be used for similar comparisons between Nicotiana species.

In theory，as long as the restriction enzyme recognition site represents a random distribution of nucleotides and only base substitution is considered as a mechanism for sequence alteration，the proportion of restriction fragments（S）shared by any two ct-DNAs can be used to estimate the fraction of nucleotides that are different（Po）between the two DNA sequences.Thus in practice，the number of restriction sites on the ct-DNA must be determined for each enzyme included in the study.Unlike SmaI，the observed EcoRI（see Figure 3），BamHI（see Figure 5），and XhoI（see Figure 2）recognition sites agree in number with those predicted from an assumption of random distribution of nucleotides（Upholt，1977）and thus are suitable for this type of analysis.In order to more accurately estimate the degree of homology，data from EcoRI，BamHI，and Xhol digests of ct-DNA were combined（Gotoh et al.，1979）.The resulting P。values presented in Table 4 for tobacco，spinach and tobacco，tomato agree quite well with sequence divergence estimates（40%and 15%respectively）obtained from DNA-DNA hybridization studies（Bisaro and Siegel，1980）.

Estimates of sequence divergence between N.glauca（Group Ⅱ）and members of ct-DNA groups Ⅰ，Ⅲ，or IVb（see Table 3）confirm that group Ⅱ ct-DNA is most closely related to group Ⅲ ct-DNA.This 0.21%estimate of divergence was similar to that obtained from comparisons among ct-DNA group Ⅰ species and represents a basal level of mutation for Nicotiana ct-DNA（0.63 mutations/300 base pairs）.The estimates of divergence and consequently of mutation（3 to 6 mutations/300 base pairs）were higher between ct-DNA groups I∶Ⅱ and Ⅱ∶IVb as well as among groups IVa，IVb，and Ⅳe（not shown）.It is worth noting，however，that even our lowest estimate of mutation rate for the entire chloroplast genome is higher than the mutation rate for the large subunit of RuBPCase estimated by Chen et al.（1976）to be 0.27 mutations/100 amino acids.Apparently certain regions of the chloroplast genome may experience fewer mutational events than others.

Table 3　Classification of Nicotiana species based on ct-DNA，RuBPCase，and morphology

Our grouping of Nicotiana species（see Table 3）and estimates of sequence divergence（see Table 4）based upon their ct-DNA restriction patterns are complemented by the taxonomic and biochemical organization of Nicotiana species outlined by Goodspeed（1954）and Chen et al.（1976）as presented in Table 4.In ct-DNA group I，for instance，N.excelsior，N.gossei，and N.megalosiphon have ct-DNA patterns similar to that of N.suaveolens and were postulated by Goodspeed to be derived from a cross between N.suaveolens and N.fragrans on the basis of morphological，geologic，and cytologic evidence.These four species have identical large subunits and multipte small subunits of RuBPCase which suggest that they were derived from one maternal and several paternal parents through multiple interspecific hybridizations.The EcoRI fragment pattern characteristic of these ct-DNAs supports this view.In a similar manner.the nearly identical EcoRI fragment patterns and identical isoelectric focusing patterns of RuBPCase large subunit for species in group Ⅲ（see Figure 5（b））suggest a single female progenitor for these species.The identity of ct-DNA of N.tabacum with N.sylvestris and other members in group Ⅲ was predictabte since N.sylvestris is the female progenitor of N.tabacum（Gray et al.，1974）.However，the low level of sequence divergence（see Table 4）between N.glauca and members of ct-DNA group Ⅲ was unexpected and implies a closer relationship than the assigned taxonomic categories would indicate.(https://www.daowen.com)

Table 4　Numbers of restriction enzyme cleavage sites which are common or different between pairs of chloroplast DNAs.Calculations of S（fraction of common recognition sites）and Po（fraction of nucleotide sites for which the two DNA sequences differ from each other）were based upon the equitations of Gotoh et al.（1979）

Also unexpected was the sequence diversity among members of ct-DNA groups IVa and IVb.According to Goodspeed these two groups have a common origin which is separate from that of ct-DNA group IVc.Yet analyses similar to those presented in Table 4 indicate a high degree of divergence（2.67%）between ct-DNA groups IVa and IVb which was nearly equal to that obtained between these groups and IVc.Some of these new fragments may have arisen from addition of a restriction recognition site as was demonstrated for a polymorphic form or N.debneyi（Scowcroft，1979）.Perhaps others arose with the loss of a restriction site or through deletion or inversion of DNA segments.But by whatever means，the changes in nucleotide sequence which occur in groups IVa，IVb，and Ⅳe are the most extensive of all Nicotiana species examined thus far.

Acknowledgement.This work was supported by NSF grant PCM7812126，NIH grant GM27746-01 and USDA Cooperative agreements 12-14-1001-967 and 12-14-1001-810.