You don't want to waste all those extra components, I'd think.
First of all, you can do what Malphas suggests: Add your in circuit programming header (most likely JTAG in this case) to your design for your development cycle. You can also use a system called "POGO Pins" to make a bed of nails, or buy a special pogo connector, so you only need a few tiny test pads for the JTAG, so you don't even need to waste space for the JTAG connector.
But second of all, your processor is BULKING with memories and what not, and I would not be surprised if Atmel actually included a ROM boot loader in it, but I'm too lazy to research that. You can use or write a boot loader, if you Google that, with your processor name or "AVR32" you should find plenty examples.
A boot loader will let you send new firmware to the device on the fly, even when it's already manufactured through any of the chip's on-board communication systems. I can't imagine your final purpose for this device doesn't include communication to somewhere else, seeing the huge amount of resources you are including.
If you want, though, you can also hook up an UART to an FTDI FT232 chip with a USB port, but that's throwing in many, many more items in your BOM that you should not need if you are going to put it into production.
Spending two days on firmware functions outweighs spending $10 extra per board if you will be making 100+ of them at some point.
Of course, I'm only inferring the production stuff from your wording and I may be wrong. In fact, from the "I'm new" bit, I'd advise you to first play around with a few designs before considering production, and maybe you were. Anyway. That ^^ is my advice.