“Syllables are indicated sparingly, where necessary to avoid confusion, for example to break up sequences of vowels (Moai) or consonant clusters which an English speaker might misread as a digraph (Vancouveria, Windhoek).”
In other words, the dot helps the reader interpret the spelling in a way that avoids wrong interpretation of letter-sequences. The transcription of Moai is given a dot to indicate that there is a syllable division after ‘Mo’, and Windhoek has a dot apparently to stop the reader from interpreting ‘dh’ as a single consonant.
But these examples only show a need for a syllable division when the word is seen in its spelling form. As soon as one looks at the IPA symbols, the possibility of confusion disappears. When Moai is transcribed as /məʊaɪ/ the pronunciation (assuming it is being read by an English speaker) is quite predictable without the need for a dot. I can’t see how ‘Windhoek’ and ‘Vancouveria’, which are transcribed /ˈvɪnt.hʊk/ and /væn.kuːˈvɪəriə/ in this work, could be pronounced any differently if the dot were removed. The most puzzling is the example of Mikey/Myki, where the writer transcribes the former as /maɪki/ while the latter is given a syllable-boundary dot thus: /maɪ.kiː/. I can’t see what purpose the dot serves, as the boundary is bound to occur in the same place in both words.
NOTE: I have now changed the box marked 'Syllabification', and hope that this has made it better. I can't decide whether to go through WP now to eradicate redundant syllable-boundary dots where they have been put in.