Hello, Sonia!
The question here starts with 'what is independent assortment'. It's a fancy term for the claim that what happens in the distribution of one locus (a locus is the 'home' or location for a gene) is not related to what happens at another locus. So here, you are being asked to use a Punnett square in support of the proposition "What happens with the alleles A1 and A2 does not determine/is not influenced by what happens with alleles B1 and B2"
Note that a Punnett square is just a way of 'tracking' what happens when the parents are creating gametes (sperm and eggs). The row labels (top) are the genotypes of gametes (sperm, eggs) that one parent makes; the column labels are the gametes the other parent creates.
The way you create the Punnett square represents certain assumptions, and the generated genotypes are a form of prediction. So if the assortment of A and B is independent, then you would predict that no matter which A allele gets put into a gamete, there would be equal likelihood of getting B1 or B2. In other words, you predict that the 4 possible gamete genotypes generated from a heterozygote (genotype A1/A2, B1/B2 =>A1B1, A1B2, A2B1, and A2B2) would each be equally likely.
So you would create a Punnett square with those genotypes as the row and column 'header' contents. The rest of the Punnett square represents the possible outcomes if the gametes 'meet'--so if A1B1 'meets' A1B1, you generate an offspring A1/A1,B1/B1
IF independent assortment applies here, THEN the likelihood of each of the gametes is equal, and each resulting diploid square is equally likely. So your Punnett square reveals the PREDICTED OFFSPRING from such a cross.
You then compare the PREDICTIONs of the "random assortment" model in the table to a TEST (actual cross performed between individuals of genotype A1/A2,B1/B2 x A1/A2, B1/B2 and see if the predictions of the Punnett square are met by the outcome in the real world.
IF the loci are NOT independently assorting, then there will NOT be an equal distribution of gametes. If, for example, the parental chromosomes were
----------A1------B1------
and
----------A2------B2------
(i.e. A gene and B gene are 'linked') then there would be a disproportionately HIGH number of gametes that got A1B1 or A2B2, whereas gametes A1B2 and A2B1 would be relatively rare--because they require a crossing over event in the small space between the A and B genes.
Sonia B.
If the two genes assort independently the offspring of this cross will be:
1/4 A1A1 B1B1
1/4 A1A1 B1B2
1/4 A1A2 B1B1
1/4 A1A2 B1B2
If you find the four types of offspring in a ratio that is not 1:1:1:1, then the two loci are not independently assorting, and are linked.
10/11/17