OK you were in Av mode, which uses the cameras metering system to calculate the correct shutter speed, based on the currently selected ISO value. Although changing the metering mode might cause a small change in the systems assessment of the scene illumination, it will be small, as it is still essentially trying to achieve the same level of exposure for the current level of illumination. The smaller you make the area that is being used to measure the illumination level, the more chance you have of it being assessed wrongly due to the specific reflectivity of the subject differing from the presumed normal level. This is why when shooting in an automatic exposure mode I would usually try to use the largest metering area possible to find a good average value. Personally I prefer Center Weighted Average, over Evaluative, because in Evaluative the camera is trying to be clever, and using variable methods to weight that average, generally CWA is likely to be more repeatable in it's computation than Evaluative, which changes it's weightings based on other camera settings. So regardless of metering mode, the camera is always going to select a shutter speed that will result in what it has been programmed to consider the correct exposure for the current level of calculated illumination. Even if the level of illumination changes, or you change the selected aperture or ISO value, the metering system is still aiming for the same relative exposure based on it's programming, and the brightness of the resultant image will remain roughly constant.
Our cameras do though provide us with a control to vary the resulting level of brightness of the resulting image when using automatic exposure modes. That control is the Exposure Compensation, EC, control. All that the EC does is to say to the camera, I want you to make the image "this" much brighter or darker than your standard programming allows for. Thats it, that is all it does! The implications of that though are very significant, because there are so many things that can affect how the camera is calculating the level of illumination of the scene. The biggest problem is that the camera's metering system cannot actually directly measure the level of light falling on the subject, that is one great advantage of a hand held incident lightmeter, it CAN directly measure the level of illumination of a scene. Instead our camera systems measure the light reflected by the scene to the cameras metering system, through the cameras lens. As mentioned in the first paragraph, the meter is programmed to assume that the area of the scene that it is programmed to assess is of a fixed average level of reflectance (tone). If the area of the scene that is being measured is not overall of that standard average tone, then one has to start calculating the deviation from that norm, and so work out the correct level of EC to apply in that situation. In this case though it seems like the scene does average close to the nominal tone for metering, the issue is just that the programmed level of exposure results in a darker than wanted final image. This is actually a very common occurrence, as Canon seems to program it cameras to produce darker than the usually preferred level images in order to protect the camera from overexposing highlights. This is easily fixed, simply shoot with some added positive EC. I have owned Canon DSLR's now for 11 years, and I shot film for 40 years before that. Once I found out that Canon have the camera setup to shoot darker than I would normally expect, and that this was "normal" I had to simply revert to using the EC control to correct the camera's exposure to the level that I prefer, and so have always had my cameras set to +2/3 Stop of EC. There is nothing wrong with this, it is one of the reasons that EC is a control on a camera. Other than very basic point and shoot type cameras, an EC control has been included with all cameras since the introduction of TTL metered auto-exposure. So don't worry its quite normal to need to use EC.
Oh and the fact that the subject was in the shade is irrelevant in this situation, the camera is still underexposing subjects in full bright sunlight, the only difference is that in that situation, bringing the exposure down a bit from normal can produce nicer results from the in camera image processing.
Alan
