Blastp Of Human Proteins Against A Combined Pig And Gorilla Protein Db, Why 10% Of Best Hits Are Against Pig?

wget ftp://ftp.ncbi.nih.gov/genomes/Homo_sapiens/protein/protein.fa.gz
wget ftp://ftp.ncbi.nih.gov/genomes/Sus_scrofa/protein/protein.fa.gz
wget ftp://ftp.ncbi.nih.gov/genomes/Gorilla_gorilla/protein/protein.fa.gz

sed 's/>/>pig_/' pig proteins > pig
sed 's/>/>gor_/' gorilla proteins > gor
cat pig gor > piggor.fa
makeblastdb -in piggor.fa -dbtype prot -hash_index -parse_seqids -out piggor

subsample 1,000 human proteins

blastp human subsample against piggor db with tabular output (outfmt 6)

sort for best hits with
sort -k1,1 -k12,12gr -k11,11g blast_output | sort -u -k1,1 --merge > best_hits

grep -c 'pig_gi' best_hits
grep -c 'gor_gi' best_hits

About 10% of best hits are against pig. Why?