I was wondering if Gerundives, the verbal adjectives referred to as "passive future participles" by Latin grammarians, can appear as predicates of (true) Ablative Absolute (AA) constructions.
As is well-known, the predicates of AAs can be passive perfect participles (e.g., Cicerone occiso), present participles (e.g., Cicerone loquente), nouns (e.g., Cicerone consule), and adjectives (e.g., Cicerone vivo). Interestingly, active future participles are also found as predicates of AAs, as in the following example from Livy:
Carthaginienses, prima luce oppugnaturis hostibus castra, saxis undique congestis augent vallum. (Liv. 28, 15)
In contrast, Gerundives, which, as noted above, have sometimes been referred to as "passive future participles", seem to be prevented from appearing as predicates in (true) AAs. But perhaps such a prohibition is not well-grounded and examples similar to the following one are also found/possible. Are there examples of this use attested?
Carthaginienses, prima luce oppugnandis castris ab hostibus, saxis augent vallum.
NB 1: the absence of predicative Gerundives from (true) AAs can be inferred from statements like the following one by Vester (1990): "we may argue that the gerundive construction has a lower syntactic level, directly related to the nuclear predication, as opposed to the ablative absolute, which functions on sentence level" (Vester, E. (1990: 304). "Reflections on the gerund and gerundive"). Emphasis/bold mine: Mitomino.
NB 2: true AAs involve "functioning on sentence level". Accordingly, apparent AAs like Insectandis patribus tribunatum gessit (Livy 3, 65,4) 'He spent his tribunate in attacking the patricians' are not to be regarded as true AAs but rather as dominant participle-like constructions that involve lower syntactic embedding (e.g., in this case, Verbal Phrase (VP)-embedding rather than sentential one). As a result, the agent of insectari must be the same as that of the main verb. Importantly, such a restriction does not necessarily apply to "true" AAs. Accordingly, true AAs are those AAs that satisfy the following two conditions: (i) they function as adverbial subordinate clauses AND (ii) they do not involve any obligatory sharing of participants/arguments with the ones of the main clause. This said, it is worth pointing out that for other people only one condition (or even other conditions) must be satisfied.
NB 3: the present post is related to Joonas's insightful point on what a predicate of an AA can be: see his specific question on the (im)possibilitify of finding locative Prepositional Phrases (PPs) acting as predicates of AAs. As pointed out in my answer to his intriguing question, my intuition is that locative/PP phrases cannot be predicates in AAs either.