My background is "research scientist": Specifically an analytical chemist. I grew up on the farm that has been in the family for 150 years. I feed my community with plant varieties that I have developed specifically for my farm without any formal training in biology, botany, genetics, or agriculture. The crop species that I grow were developed primarily by illiterate farmers. I am not a hobbyist. Farming and plant breeding are my full-time jobs.
I primarily use "Recurrent Mass Selection" among genetically diverse populations as my breeding strategy. That means I save seeds from whatever plants happen to be pleasing to me on the day that I make the selection or do the planting. I really like doing mass selection among F2 or F3 generations descended from natural or manual cross-pollinations.
I also do a lot of "Sibling Group Selection". By that I mean that I save the seeds from each mother plant separately, and then I plant the siblings together in a short row. Then when it is time to save seeds I save seeds primarily from the specific sibling groups that did best. This is much more powerful for me than random plantings from bulk seed, because it makes it easy to see which families are most suited for my garden. For example, in the corn patch this year, one sibling group lodged severely in a storm that barely affected the rest of the patch. It was trivial to detassel that family before it shed pollen into the rest of the patch. And the seed from those plants was fed to the chickens.
I don't plant duplicates of each sibling group. I don't randomize the planting locations. I tend to plant the sibling groups with the highest scores together on one end of the patch and work across the patch planting progressively lower scoring families. I usually don't record anything about the identity or pedigrees of the sibling groups. Some sibling groups are easy to identify from year to year because they carry a unique trait. I do best-guess evaluations regarding yield, productivity, agronomic suitability, taste, disease resistance, etc. For melons, date of harvest and weight are the two traits that I tend to measure and record. Every fruit undergoes taste testing. For popcorn, popping ability, ease of shelling, and harvest window (early/late) are the traits that I measure and record. I record fruit size (small/medium/large/XL) and evaluate taste for squash. Seeds are not saved from bad tasting squash, so there is no reason to record that something tasted good. I live in a very challenging environment for warm weather crops. There is no need to record if a variety failed in my garden. If it didn't produce seeds or pollen soon enough then it culls itself. No records needed about what failed. And very few about what succeeded. It's really easy to get so caught up in documentation and pedigrees that growing gets forgotten. If I had to keep meticulous records I wouldn't be doing plant breeding.
I have a refractometer that I could use for measuring brix in corn, tomatoes, melons, squash, etc. But I don't use it because my mouth is a much simpler and more accurate measurement tool for what tastes good.
With some crops I grow a "Control" or standard reference variety that I plant every year. I buy the seeds commercially in multi-year sized lots so that I can compare my crops to a consistent variety. In tomatoes the control is DX52-12. In sweet corn the reference variety is Ambrosia.
If I want to do a trial of a new variety, I plant it as if it were just another sibling group. If I like it I save seeds from it. If not then maybe it contributed some pollen to the patch and maybe it didn't. I do very little isolation, other than for gross traits like not planting sweet peppers and hot peppers in the same field. Promiscuous pollination is the name of the game in my garden. I do very little manual cross-pollination except that I might detassel a row of corn. When I find hybrids among predominately inbreeding crops they get a special place in the garden for a couple of years.
Here's some photos that demonstrate the huge differences that can show up when trialing varieties or comparing sibling groups. I observe this sort of pattern over and over again in just about every species that I work with. Weeding is more useful to me than formal record-keeping. Don't underestimate the power of trialing several varieties and saving seeds from what grows best in a particular location with a particular farmer's way of doing things.
Two different sibling groups of watermelons. Planted a few feet apart on the same day. Photographed on the same day.
Mal-adapted watermelon.
Well adapted watermelon.
My landrace spinach on the top vs a commercial variety being trialed (bottom). Planted and photographed at the same time a few feet away from each other. I didn't record the name of the commercial variety. It had it's chance and failed. It might get another chance some day and it might not.
I don't need to do any sort of measuring to be able to say definitively that my localized landrace did vastly better than the commercial variety that was being trialed. I am not chasing a few percentage points of higher yield. With most of the crops I am working on, it is all or nothing. Either I get a crop or I don't. It is common when I first start growing new species to have 75% to 99% failure rates. Failure being defined as not producing seeds. Runner beans and mixta squash completely failed 4-5 years in a row before I finally got a harvest. In the okra breeding project, it wasn't until the 3rd generation that there was enough excess that I could taste the crop.
A gardener without the ability to count chromosomes can still get a feel for whether or not a crop is diploid or polyploid. Phenotypes and growth patterns are often different. I attempted to do some diploid to tetraploid conversions on watermelon a couple of years ago using oryzalin. Based on phenotype differences I believe that they were successful:
Diploid Watermelon Control
Suspected Tetraploid Watermelon. Same variety but treated with a chromosome doubling agent.
Leaf size comparison between the smaller diploid and the suspected larger tetraploid.
Crosses could be made between diploids and the suspected polyploids to see if they are fertile. A diploid/tetrapoid cross is expected to be infertile. This summer I inter-planted the suspected tetraploid melons with a striped diploid melon. I'll be looking for an archetypical seedless watermelon. That will be another piece of evidence that the conversion was successful.