I have been saving seed from collard plants that overwinter well or are short lived perennials in my climate. Currently if I plant seed I've saved some portion of the plants will live until a particularly bad winter finally takes them out. Usually they produce seed in summer but also continue to produce leaves and thrive getting bigger each year usually falling over then respouting all along the stalk. This year my brassicas have become super infested with aphids. Some plants died, some survived, some were covered in aphids on the old growth but still growing vigorously  and some had hardly any aphids at all. This seems like a good opportunity to select for pest resistance. While it seems like I should save seed from the plants with no aphids I'm wondering if these plants produced some kind of aphid deterrent would that also be harmful to us?  If so would these genetics only produce this aphid toxin in response to heavy aphid pressure or always? I'm thinking it might be better to save seed from the plants covered in aphids but still growing good. These plants are thriving under heavy predation, seemingly better suited to heavy harvesting or pest pressure (normally slug and deer currently aphids) and possibly a healthier plant to eat but might also encourage a perpetual aphid problem. What do you think?

I think there is a reason why variation exists in a population for pest and disease resistance.

Take a land race population of wheat for example. This wheat might be coevolving with a rust. So in the landrace of wheat you might find some wild crop relatives like jointed goat grass and even several species of wheat at multiple ploidy levels. Hybridization within and between species and ploidy levels is a continuous ongoing process. Ripening time is variable. Plants vary in size, height, and productivity.

In regards to the rust some wheat plants may be totally immune, others totally susceptible, others having varying degrees of tolerance or resistance. Different genes drive these different strategies. When rust attacks a given plant it may not need genes that would allow it to attack another. The susceptible plants do great when rust has a bad year. The susceptible plants also foil the rusts attempts to evolve countermeasures.

Contrast this with a modern wheat. This wheat was developed using a process of doubled haploid anther culture or by extreme inbreeding. Its been selected for multiple rust resistancr gened but it is all genetically identical. If rust does manage to attack one plant- that same rust can attack the entire field even if it is 10,000 acres.

So if your collards have three aphid strategies susceptibility, resistance, and immunity. I  would try to save some seed from all three. The susceptible plants will give the aphids a home during bad aphid years. They will do great in years when aphids arent the problem. The resistant plants may provide an alternate strategy, and the immune plants may always do great in aphid years, but might not be the best stock in years when aphids are few.

This makes a lot of assumptions though. That the aphids are the disease and not just a symptom. That genes or at least one gene is at play in the collards. That it's resistance to aphids and not ants that tend the aphids. That the plants aren't being attacked randomly or based on conveneince to an ant colony or colonies that are tending the aphids.

Depending on your population size your collards may already be a highly inbred population. The three aphid s trategies you see could be controlled by a single gene. Some plants have no copies, some have one copy, and others have two copies. If you did want to follow the modern wheat strategy, and it was the case of a single gene, you should be able to eliminate the non resistant strain fairly fast by saving seed only from the mostly immune plants.

However if you have a copy of Carol Deppes book you might want to review it to see what she says about population sizes and inbreeding in brassicas. Also before you purge the population of non-resistance genes you might want to follow carols advice on drying some seed down and freezing it for a backup- in case it turns out later that those genes have some other function like say mildew resistance in years when aphids aren't the problem.

My seed saving strategy aligns with what William recommended. I feel like it's better to have more diversity in a population than to have more resistance...

Raul Robinson, who is one of my plant breeding mentors, recommends starting a plant breeding project by culling plants that are immune, and saving seeds from those that are susceptible.  The idea being that immunity due to single gene interactions is very susceptible to breaking down long term. And that tolerance that is due to dozens or hundreds of genes is much more resilient in the long term.

Even when growing clones, I see differences in growth between plants growing nearby to each other. Many highly localized variables can influence how each plant grows... Something in the soil. What weed is growing closest. The patterns of shading from other plants. Random interactions with animals or microbes. A bird dropping.

In my garden, most culling, and most selection is done by the interaction of the plants with the ecosystem. If a plant doesn't reproduce, it self-culls. Especially in small populations, I'm really inclined to save seeds from every plant that is capable of producing seeds. Diversity is more important to me than any particular trait.

Thank you for your input! There is huge variation in the genetics of my collards. They get white, green and purple and any combination of the three plus flat, wavy, frilly and waxy leaves. There is definitely some kale and brussel sprout genetics mixed in as well. I generally only cull super weak slow growing plants, ones that flower in their first year and ones that don't taste good. I don't water and provide very minimal care so the less resilient ones self cull. My garden is a mix of wild plants, "weeds", perennials, self seeding annuals and whatever new seed or starts I manage to fit in there. It's rather messy looking. This is the first time I've had any kind of infestation. I decided that instead of fighting it I would watch and see what happens. I hadnt considered the value of variation in pest resistance but that definitely makes sense. Maybe I'll save seed from all of them and leave seed on all of them to self seed then pay attention to what happens over the winter and next spring. Maybe they will just do their own thing and self cull the less resilient over time. Meanwhile I'll plant another patch of collards somewhere else far away from the little aphid farm I've got going to ensure I get plenty to eat too.

Joseph Lofthouse wrote:
... recommends starting a plant breeding project by culling plants that are immune, and saving seeds from those that are susceptible.  The idea being that immunity due to singe gene interactions is very susceptible to breaking down long term. And that tolerance that is due to dozens or hundreds of genes is much more resilient in the long term.

Just adding some additional food for thought.  I guess I would not cull plants that were immune, but I agree that one would want to keep a bit of everything that they thought might be valuable and maintain diversity within their stocks.  That some of the genes that confer immunity are indeed single genes and prone to break-down over time is well founded, yet there is some immunity (mostly in the realm of 'non-host immunity') for which increasing evidence points to a diversity of genes that underlie the trait.  Tolerance has been the subject of much debate, but I can agree with this discussion and definition:  

"A definition that would be limited to tolerance but include its broadest
scope should be available. Tolerance, like other resistance, is a relative concept
and may occur in varying amounts. It may also occur in combination
with other resistance. Tolerance may be defined as that capacity of a cultivar
resulting in less yield or quality loss relative to disease severity or pathogen
development when compared with other cultivars or crops. This definition encompasses
any degree of tolerance in combination with any degree of other
resistance or lack of resistance in respect to the comparison, thus fitting the
broad conceptual base. " --John Schafer "Tolerance to Plant Disease" (1971) Annual Review of Phytopathology

The good outcome of the high diversity Joseph is selecting for is that there is probably mixtures of different categories of resistance -- even if one of many plants was resistant/immune due to a single, more-precarious gene, the vast amount of resistance mechanism at work in the *population* would almost guarantee long term success as noted.

For a more in-depth discussion on the issue, an open-access paper entitled "Plant Innate Immune Response: Qualitative and Quantitative Resistance" -- https://www.tandfonline.com/doi/full/10.1080/07352689.2016.1148980

I actively cull for things like horrid taste, or poisons, etc...