YES

TRS:

  a__zeros() -> cons(0(),zeros())
  a__tail(cons(X,XS)) -> mark(XS)
  mark(zeros()) -> a__zeros()
  mark(tail(X)) -> a__tail(mark(X))
  mark(cons(X1,X2)) -> cons(mark(X1),X2)
  mark(0()) -> 0()
  a__zeros() -> zeros()
  a__tail(X) -> tail(X)

max/plus interpretations on N:

  a__zeros_A = 4
  a__zeros#_A = 4
  cons_A(x1,x2) = max{3, 3 + x1, 3 + x2}
  cons#_A(x1,x2) = max{3, 3 + x1, 3 + x2}
  0_A = 0
  0#_A = 0
  zeros_A = 0
  zeros#_A = 0
  a__tail_A(x1) = max{10, 4 + x1}
  a__tail#_A(x1) = max{10, 4 + x1}
  mark_A(x1) = max{2, 5 + x1}
  mark#_A(x1) = max{2, 5 + x1}
  tail_A(x1) = max{5, 4 + x1}
  tail#_A(x1) = max{5, 4 + x1}

precedence:

  0 = zeros > a__zeros > mark > cons = a__tail > tail