“CHALLENGES OF MONITORING ANTARCTIC LICHEN COMMUNITIES USING PHOTO DIGITISATION”B y I s a b e l l a T o d d

The effects of global climate change are already being observed in the vegetation communities of the Windmill Islands, East Antarctica. East Antarctica has recently experienced unprecedented heatwaves, while some areas are also seeing regional drying with extreme consequences for the moisture-sensitive moss and lichen communities of this area. Previous studies have observed reduced moss growth, contraction of moss communities and the expansion of lichen-dominated communities.

Difficulties with long-term monitoring of slow growing lichen-dominated communities in harsh Antarctic conditions means there is very little empirical data on changes to lichen cover in these communities. Handheld digital photography is one of the most suitable, cost-effective methods for long-term monitoring of these communities.

Previous attempts to use semi-automatic object-based image analysis to classify vegetation types using RGB and HSI values were unsuccessful, as many Antarctic lichen species share similar spectral characteristics to rocks and dead mosses. Manual digitisation of digital photographs could be the solution to monitoring change in lichen communities - reducing the effect of observer bias from field estimates and the amount of time spent in the field, while also increasing the accuracy of cover estimates.

Investigate changes in the vegetation cover in lichen communities in East Antarctica using manual digitisation

Test the efficacy of using manual digitisation for future, long-term monitoring of these communities.

Image CollectionThe images used for this study were collected as part of a long-term monitoring program of vegetation communities at two sites in the Windmill Islands in East Antarctica: Antarctic Specially Protected Area and Robinson Ridge. Digital photographs of 20 lichen-dominated quadrats in this program were collected in 2003, 2008, 2012 and 2013. A subset of four of these quadrats were analysed in this study.

Manual Digitisation & AnalysisManual digitisation was undertaken in ArcGIS Pro. Each image was visually assessed, and the cover type (Fig 1) of all objects within the quadrat were classified and manually digitised (Fig 2). Mean % cover for each cover type was taken across all quadrats within each year and differences in mean % cover were compared across the four using analysis of variance.

A larger sample of quadrats is needed to verify if any trends in changes to percent cover are occurring. Further long-term monitoring is needed in order to understand the effects that regional and global climate change are having on the structure and dynamics of these communities.

Due to the time and labour-intensive nature of this method, it may not be scalable to long-term monitoring programs and may therefore be limited to studies that occur at smaller spatial and temporal scales. However, manual digitisation methods could be combined with semi-automated analysis to reduce labour times.

No significant differences in mean % cover were found for any cover type across the decade of measurements (Fig 3). There appears to be a slight increase in lichen cover over time for most of the quadrats studied. This could indicate a longer-term trend of increasing lichen cover due to drying. Due to the low productivity and slow growth rates of the species in these communities, any change will likewise occur at a relatively slow rate.

Low resolution of images made fine scale features, such as small crustose lichens, difficult to digitise. This method may therefore be insufficient to detect small scale changes in these small-scale species. Low contrast, low resolution and high shadow in some images made differentiation between lichens, moss and rocks of similar colours very difficult. This could be overcome by reference to and manipulation of raw images. However, this was a very time-consuming process with each quadrat taking anywhere from 2-4 hours to digitise.

INTRODUCTION

AIMS

METHODS

RESULTS & DISCUSSION

RECOMMENDATIONS

Figure 1: Images of cover types

References: 1. King, DH, Wasley, J, Ashcroft, MB, Ryan-Colton, E, Chisholm, LA, Robinson, SA & Lucieer, A 2020, ‘Semi-Automated Analysis of Digital Photographs for Monitoring East Antarctic Vegetation’, Frontiers in Plant Science, vol. 11, pp. 1-16.

0

10

20

30

40

50

60

70

Lichen Moss Rock Misc

Mea

n Pe

rcen

t cov

er

Cover type

2003

2008

2012

2013

1

2

Rock Macrolichens

Moribund moss Live moss

Macrolichen

Crutose lichens

Figure 3: Trends in Lichen community vegetation cover from 2003 to 2013 in the Windmill Islands, East Antarctica. Data are mean percent cover ±SE, n = 4.

Macrolichen

Crustose lichen

Moribund moss

Live moss

Rock

Shadow

Miscellaneous

Quadrat perimeter

1

2

1

2

Figure 2: Digitised quadrat