(Continued from Part 1).

The kind of things plant ecologists often measure in the field – the height and percentage foliage cover of different plants, species lists – are difficult to obtain from looking at a photograph of the landscape because of distortions due to perspective. Or, in other words: “The geometry of the photographs inhibits quantification”.  Even mapping boundaries between vegetation types can be difficult because the process requires translating from an oblique 3-dimensional perspective to a 2-D aerial view. Further complicating matters, photos taken at different times may not be at exactly the same location, or in different seasons, or the lighting may be very different.

1980 (L) and 2014 (R) photographs of coastal terrace and escarpment near Langdon Point, Macquarie Island. Photographer: Jenny Scott.

While trying to quantify the magnitude of changes from repeat photographs is fraught, it is normally straightforward to determine the direction of change. This is the approach we are using with a time series of Macquarie Island landscape photographs. Three time periods have been chosen for analysis: 1980 (black and white film photographs digitized by scanning), 2009 and 2014 (colour digital images). These correspond with the end of a period of highest recorded rabbit population (1974-1980), the end of next highest population cycle (2004-2010), and the beginning of the post-rabbit phase (mid-2011 onward).

If there is a rabbit influence on the vegetation we would expect to see it strongly in 1980 (after several years of high rabbit numbers peaking at an estimated 350,000), less evident in 2009 (when the preceding rabbit population peak is estimated at 200,000), and some signs of vegetation recovery in 2014 (2.5 years after rabbit eradication). Of course, there may be other factors involved, notably climate.

The method involves viewing the photographs side by side on a screen, selecting distinct vegetation features which can be reliably identified in each of the three time series photographs and drawing these on a printed version. Key plant species or vegetation types such as feldmark, bare ground, moss, Pleurophyllum, tussock grass are scored as increase, decrease or no change.

Photographs of Whisky Creek from 1980, 2009, 2014 with annotations of visible vegetation change. Photos: Jenny Scott.

The centre-point of each of the zones identified from the photographs is then geo-located on a map of Macquarie Island. With all of the photograph locations recorded using GPS, these points can be uploaded to a GIS which allows the viewshed of each point to be determined. The viewshed shows which parts of the landscape are visible from a given point which helps to narrow down the features visible in the photographs taken from that point. A satellite image and contour lines provide enough information to confidently locate most of the zones in the photographs to a point on the map.

Satellite image of Macquarie Island overlaid with contour lines. Shading represents the viewshed from one of the photograph locations (red dots). Orange + symbols indicate locations of vegetation zones identified from photographs.

Combining observations of particular types of change (or, equally importantly, no change) with a geographic location allows us to look for spatial patterns in vegetation trends. Does aspect, elevation, landform, wind exposure or drainage influence changes in vegetation over a 25 year period? These questions will be answered after hundreds of observations are completed.