Data on vascular plants of herb-rich forests in Finland were used to compar
e the efficiency of reserve selection methods in representing three measure
s of biodiversity: species richness, phylogenetic diversity, and restricted
-range diversity. Comparisons of reserve selection methods were carried out
both with and without consideration of the existing reserve system. Our re
sults showed that the success of a reserve network of forests in representi
ng different measures of biodiversity depends on the selection procedure, s
election criteria, and data set used. Ad hoc selection was the worst option
. A scoring procedure was generally more efficient than maximum random sele
ction. Heuristic methods also appeared to be efficient. Each biodiversity m
easure can be used as a criterion for a selection algorithm and as a measur
e of efficiency of protection at the network level. The results indicate th
at different measures of biodiversity should be taken into account. For ins
tance, phylogenetic diversity of a network was maximized by a heuristic met
hod using phylogenetic diversity as the selection criterion. However, a heu
ristic method based on restricted-range diversity was more efficient than t
hat based on species richness in representing high species richness in a fo
rest network. Our results showed also that complementing the existing reser
ves is less efficient than selection starting with a blank slate. That is,
the total number of sites required to protect a given level of biodiversity
is higher when complementing the existing network, because some previously
protected sites contribute relatively little diversity to the network. Dat
a on species of a historic network should be available when new sites are s
elected for complementing the historic network.