Why aren't my services reachable in Marathon?

I’m deploying to Marathon using the reactive CLI. When I use the following:

$ rp generate-marathon-configuration "$image"  --env JAVA_OPTS="-Dplay.crypto.secret=$secret" --registry-use-local --instances 2 --external-service "kafka_native=$kafka_dns_entry" --external-service "cas_native=$cass_dns_entry" | dcos marathon app add

However, my services never become available because there are no seed nodes

16:37:42.412 [e[33mwarne[0m] application [] - system properties: play.crypto.secret is deprecated, use play.http.secret.key instead
16:37:45.644 [e[37minfoe[0m] akka.event.slf4j.Slf4jLogger [] - Slf4jLogger started
16:37:45.733 [e[37minfoe[0m] akka.remote.Remoting [sourceThread=main, akkaSource=akka.remote.Remoting, sourceActorSystem=application, akkaTimestamp=16:37:45.730UTC] - Starting remoting
16:37:46.232 [e[37minfoe[0m] akka.remote.Remoting [sourceThread=main, akkaTimestamp=16:37:46.229UTC, akkaSource=akka.remote.Remoting, sourceActorSystem=application] - Remoting started; listening on addresses :[akka.tcp://application@W.X.Y.Z:3531]
16:37:46.238 [e[37minfoe[0m] akka.remote.Remoting [sourceThread=main, akkaTimestamp=16:37:46.234UTC, akkaSource=akka.remote.Remoting, sourceActorSystem=application] - Remoting now listens on addresses: [akka.tcp://application@W.X.Y.Z:3531]
16:37:46.292 [e[37minfoe[0m] akka.cluster.Cluster(akka://application) [sourceThread=main, akkaTimestamp=16:37:46.292UTC, akkaSource=akka.cluster.Cluster(akka://application), sourceActorSystem=application] - Cluster Node [akka.tcp://application@W.X.Y.Z:3531] - Starting up, Akka version [2.5.16] ...
16:37:46.568 [e[37minfoe[0m] akka.cluster.Cluster(akka://application) [sourceThread=main, akkaTimestamp=16:37:46.567UTC, akkaSource=akka.cluster.Cluster(akka://application), sourceActorSystem=application] - Cluster Node [akka.tcp://application@W.X.Y.Z:3531] - Registered cluster JMX MBean [akka:type=Cluster]
16:37:46.569 [e[37minfoe[0m] akka.cluster.Cluster(akka://application) [sourceThread=main, akkaTimestamp=16:37:46.567UTC, akkaSource=akka.cluster.Cluster(akka://application), sourceActorSystem=application] - Cluster Node [akka.tcp://application@W.X.Y.Z:3531] - Started up successfully
16:37:46.923 [e[37minfoe[0m] akka.cluster.Cluster(akka://application) [sourceThread=application-akka.actor.default-dispatcher-16, akkaSource=akka.cluster.Cluster(akka://application), sourceActorSystem=application, akkaTimestamp=16:37:46.921UTC] - Cluster Node [akka.tcp://application@W.X.Y.Z:3531] - No seed-nodes configured, manual cluster join required
16:37:50.569 [e[37minfoe[0m] akka.management.cluster.bootstrap.ClusterBootstrap [sourceThread=main, akkaSource=ClusterBootstrap(akka://application), sourceActorSystem=application, akkaTimestamp=16:37:50.568UTC] - Bootstrap using default `akka.discovery` mechanism: MarathonApiSimpleServiceDiscovery
16:37:50.616 [e[37minfoe[0m] akka.management.cluster.bootstrap.ClusterBootstrap [sourceThread=main, akkaSource=ClusterBootstrap(akka://application), sourceActorSystem=application, akkaTimestamp=16:37:50.573UTC] - Initiating bootstrap procedure using akka.discovery method...
16:37:50.718 [e[37minfoe[0m] akka.management.cluster.bootstrap.internal.BootstrapCoordinator [sourceThread=application-akka.actor.default-dispatcher-20, akkaSource=akka.tcp://application@W.X.Y.Z:3531/system/bootstrapCoordinator, sourceActorSystem=application, akkaTimestamp=16:37:50.672UTC] - Locating service members. Using discovery [akka.discovery.marathon.MarathonApiSimpleServiceDiscovery], join decider [akka.management.cluster.bootstrap.LowestAddressJoinDecider]
16:37:51.613 [e[37minfoe[0m] akka.management.AkkaManagement [sourceThread=main, akkaSource=AkkaManagement(akka://application), sourceActorSystem=application, akkaTimestamp=16:37:51.545UTC] - Including HTTP management routes for ClusterBootstrap
16:37:51.632 [e[37minfoe[0m] akka.management.cluster.bootstrap.ClusterBootstrap [sourceThread=main, akkaTimestamp=16:37:51.629UTC, akkaSource=ClusterBootstrap(akka://application), sourceActorSystem=application] - Got self contact point address: http://W.X.Y.Z:3532
16:37:52.226 [e[37minfoe[0m] akka.management.AkkaManagement [sourceThread=main, akkaTimestamp=16:37:52.222UTC, akkaSource=AkkaManagement(akka://application), sourceActorSystem=application] - Including HTTP management routes for ApplicationStatus
16:37:52.226 [e[37minfoe[0m] akka.management.AkkaManagement [sourceThread=main, akkaTimestamp=16:37:52.225UTC, akkaSource=AkkaManagement(akka://application), sourceActorSystem=application] - Binding Akka Management (HTTP) endpoint to: 0.0.0.0:3532

I know lagom.cluster.join-self=on is a bad idea – is it expected that my service can’t find seed nodes?