ERRATUM:
p.276, second line of section 5: SHPSf S should be SHPf S.
UPDATES:
Question 28(ii) ("Does SHPS(X) = HSP(X) for every class X of lattice-ordered groups?") is answered in the negative in
N. Ya. Medvedev, HSP ≠ SHPS for metabelian representable l-groups, Algebra Universalis 31 (1994) 151-156. MR95c:06035.
For further work on semigroups of operators generated by these and related operators, see
Boa Tasić, Partially ordered monoids generated by operators H, S, P and by H, S, Pf are isomorphic, Semigroup Forum 62 (2001) 485-490. MR2002b:06026
and subsequent papers by the same author.