Hi depending on the ecosystem in question, I understand that acoustic trends are quite important but haven't been quite looked into,, rather than only using visual cues to determine the direction of change.
"Baseline" simply means a starting point, so the most appropriate baseline depends on what you are measuring and why. I suggest you look at Krebs "Ecological Methodology" (1999), and Quinn & Keough (2002) "Experimental Design and Data Analysis for Biologists" Look specifically at the chapters on repeated measures and power estimation (although power estimation has been criticised as being over-used). There are a variety of approaches to measuring change, such as Before-After-Control-Impact studies, or establishing a baseline using multivariate techniques based either on the variation within a number of initial sample sites, or the variation within the first few repeated samples. Anderson and Thompson 2004 describe Multivariate control charts for detecting the magnitude of change beyond the initial variation, although that doesn't describe the direction of change. I recommend you read Anderson's other work as well, as she addressed many of the statistical issues around detecting change.
For direction of change, it depends on whether you are tracking multiple or single variables. For multiple variables, the direction of change will be probably best shown by the shift of the sites along one or more axes of a multivariate plot over time. For univariate, it will be the slope (or shape) of the regression over time.
As with all studies, the selection of sites, the sample size and sampling strategy are all crucial. I've had to reject papers that have attempted to measure change over several years, simply because the initial sample design was so weak that the results were meaningless. Make sure that heterogeneity and other variables outside the scope of your study are accounted for.
See Mentis and Stalmans 1993 for an example of setting baselines, determining sample size and detecting change in grasslands.
Anderson MJ and Thompson AA. 2004. Multivariate control charts for ecological and environmental monitoring. Ecological Applications 14(6): 1921–1935.
Stalmans M and Mentis MT. 1993. Development of monitoring procedures for the herbaceous layer on the Northeastern Transvaal escarpment. African Journal of Range & Forage Science 10(3): 129–134.