I read your question quite a while ago, and recently read this in Carol Deppe’s book Breeding your own vegetable varieties. Bob had the gist of it quite right.
“The Cytoplasmic in Inheritance
More than 99 percent of the genes in plants are arranged on chromosomes located in the nuclei of cells, and they are inherited according to Mendel’s laws. However, some small fraction of genes are located on organelles in the cytoplasm, and they aren’t inherited in this way. Even though there aren’t very many extrachromosomal or cytoplasmic genes, they have a disproportionate importance in plant breeding. The inheritance of genes in cytoplasmic organelles is completely non-Mendelian. It is almost always exclusively maternal. The egg is a big cell with a lot of cytoplasm. The pollen is a very small cell with little cytoplasm, which is not normally transferred to the egg during fertilization. Nuclear DNA comes equally from both parents, but cytoplasm comes only from the maternal parent. Once in a while you can do a cross where one of the characteristics you are interested in is associated with a gene that is cytoplasmic instead of nuclear. Suppose you do a cross between a variety that has a mottled coloration pattern on the leaves and one that does not. If you cross a mottled female to a normal male, some or all of the F1 plants are likely to be mottled. But if you cross a mottled male with a normal female, none of the F1 plants will be mottled, nor will the offspring in any future generations. The cytoplasm is relevant to us in additional ways. The cytoplasm of a variety and its nuclear genome have evolved to be compatible. When distant relatives are crossed, sometimes the cytoplasm and the nuclear genome of the F1 or of some of the F2 progeny aren’t compatible. In extreme cases lethality results. In less extreme cases the plant may be sterile or pollen-sterile. Male-sterile cytoplasms have, in fact, been useful in the production of hybrids. Such cytoplasms aren’t absolutely male-sterile; they are sterile only in the presence of one or more nuclear genes. Anytime we are crossing two fairly distant relatives, we we need to realize that the reciprocal crosses might be genuinely different. Both will give rise to the same nuclear genotype, but they will set it down in the middle of different cytoplasms. This could have a dramatic effect on the phenotype of the F, and subsequent generations. Many modern crops trace their nuclear genomes to only a dozen or so sources. But it is common for vast fractions of a modern crop to have only one cytoplasmic genome. The Southern corn blight in the United States in the early 1970s was caused by the blight sensitivity of a particular cytoplasm – a male-sterile cytoplasm that was in virtually the entire commercial crop, whatever the nuclear genome involved. Whenever we do a cross using a particular variety as the maternal parent, we are preserving its cytoplasm and discarding that of the male. I think we should pay more attention to preserving and increasing the diversity of the cytoplasms in our food crops. All we have to do to preserve a particular cytoplasm is to make sure that we use it as the female parent when we do crosses.”