"If connected to an unmanaged switch: It is possible that the unmanaged switch does implement spanning tree."
Personally, I'm unaware of any unmanaged switch supporting STP, but in theory, an unmanaged switch could be built supporting STP, although without management, does raise the issue of how you configure STP for operation in conjunction with other STP switches.
Without STP, though, switch will behave much the same as a hub, just as Rick notes although a switch doesn't flood all received frames to all its other ports, unconditionally, as does a hub. (As hubs are half duplex, looping frames will generally cause lots and lots of collisions, often effectively bringing useful traffic to a standstill.)
"If all the switch ports are in the same vlan then there will be elevated traffic levels on the switch ports."
Perhaps not, likely yes. Much depends on content of looped frame(s), as the upstream managed switch will still process the ingress traffic (from the hub or unmanaged/no-STP switch) as it would otherwise do. Unfortunately, just one broadcast frame, being looped, will cause the elevated traffic level (possibly to full bandwidth from looping hub/switch connection to upstream switch) noted by Rick and myself. Also, something like a looping broadcast frame, will be accepted for examination by all receiving hosts. (In such a case, all hosts within same L2 domain (like a VLAN, as noted by Rick) will be impacted to such a degree, generally those hosts will be network useless until you break the loop.)
"In this case if 2 ports are connected to each other then spanning tree will detect the loop and disable a port to prevent the loop."
Actually, it might be better to note the interface closing the loop will usually be "blocked" not really disabled (assuming the latter implies more of a permanent breakage). For example, if you remove the other end connection that also leads to the loop, the port that was blocked will be unblocked.
BTW, on Cisco switches that support portfast, I've read, it's possible for a looped connection to cause so much havoc, so quickly, that the switch won't be able to "block" the loop it allowed. (I.e. w/o portfast, STP should insure no loop will be created. With portfast, we hope the switch will detect an "active" loop and then block it.)
