Performing permanova in R using adonis

I was tasked with performing a PERMANOVA test on my data, which consists of counts of specific genes found in different types of soils. The goal is to test if different types of lands differs in the number of total genes.
I am new to both the concept of PERMANOVA tests and the adonis function.

I have two files, one with the total genes in the sample

head(new_df)
              sum
L1   2107.2634619
L10  1916.4122739
L100 1129.1259035
L101   31.1241711
L102    4.3310406
L103    0.6941578

and another which associate the sample with the type of land

head(meta)
              land
L1        Woodland
L10      Grassland
L100     Grassland
L101     Grassland
L102     Grassland
L103      Cropland

is my understanding that the adonis function requires a dataframe and a factor in the formula, so I used this command:

results <- adonis2(formula = new_df ~ land, data = meta,  permutations = 999)

which gave me this output

Permutation test for adonis under reduced model
Terms added sequentially (first to last)
Permutation: free
Number of permutations: 999

adonis2(formula = new_df ~ LC_simpl_2018, data = meta, permutations = 999)
               Df SumOfSqs      R2     F Pr(>F)
LC_simpl_2018   3    0.521 0.00261 0.544  0.872
Residual      624  199.245 0.99739
Total         627  199.766 1.00000

My main question is if I correctly implemented the function.
Thanks in advance and sorry if is not clear.

No, you didn't correctly specify the test. For starters, sum is in mydatat (and that is all that mydatat contains) so it seems you tried to model sum using sum, which makes no sense. If there is more in mydatat than just sum, then you still over-inflated the variance explained because the same variable exists on both sides of the formula.

More generally, PERMANOVA is not a specific test but a "non parametric" version of a multivariate linear model. It is no more a test than regression is a test. Hence whether you did the "test" correctly or not depends entirely on what hypothesis you were trying to test or what features of the data you were trying to model. You do not provide such information, and regardless if you had, [so] is not the appropriate venue for discussion of statistical models.

If your data are actually a factor plus the sum variable, then I would just fit this as a Poisson GLM with glm(). If you want to fit this using PERMANOVA then you have the formula back to front. You want sum ~ group, data = mydatat where group is the factor for the soil types. I forget no whether or not adonis2() can work with a vector response - it really doesn't make sense to compute a dissimilarity from a single variable...