I think the best way to accomplish that is to use a gateway pool with your route destinations and create a priority group in the pool to establish the route's (effective) administrative distance. An example for you to get started (will likely require tweaking...not production ready). Note that priority-group activation is highest first, so as an administrative distance this is opposite a cisco route declaration. It's also dependent on monitors being accurate, so make sure however you decide to monitor your gateways, you're able to get successful responses during normal state.
ltm pool pool.tunnel_routes {
members {
172.16.2.2:any {
address 172.16.2.2
priority-group 10
session monitor-enabled
state down
}
172.16.2.3:any {
address 172.16.2.3
priority-group 5
session monitor-enabled
state down
}
}
min-active-members 1
monitor gateway_icmp
}
net route route.tunnel {
network 10.10.20.0/24
pool /Common/pool.tunnel_routes
}