The reliability of GraVis as a new alternative for cuticle analysis and the determination of Svalbard climate using Salix polaris leaves.

Abstract

GraVis has proven to be a statistically reliable method to recognize and calculate UI, making it a promising alternative to ImageJ that is capable of producing and processing lots of data in a significantly shorter time span. However, this increased speed comes with the cost of reduced accuracy and higher standard deviations, as well as overall higher UI values. The main drawback is its struggle to recognize and distinguish cells in non-optimal images in which the leaf is in a higher state of decay. By training the image-recognition algorithm more extensively it should improve its ability to be able to recognize epidermal cell outlines more precisely and have less trouble with fossil leaves in a higher state of decay. Once the GraVis program has been improved upon, it will become possible to study inner leaf variation which can pick up deviations within the leaf in order to provide data with higher certainty and resolution. More research and data is needed to further improve the GraVis program, but it already provides promising results as of now. Using UI analysis of core SN22 and age depth correlation, the climate shifts during the last 600 years can be reconstructed. Core SN22 displays UI variation over time, with relatively high UI values at the start and end of the core. The UI values are the highest at the most recent part of the core, dated to the last 30 years. This is complimentary to the vast amount of studies done on the warming climate on Svalbard over the last couple of decades. The relatively low UI values between 1500 and 1800 coincide with the Little Ice Age. The strong increase in UI values during the last 30 decades of the core represent the recent warming of the surface temperatures at Svalbard in recent years, and it is predicted that the UI values will keep increasing in the following years. Further research can be done by coring in different locations of Svalbard while using the same GraVis analysis method to expand the dataset across Svalbard. Alternatively, more cores from the same locations could be studied to expand the age-range of the current dataset to further back into the Holocene and to improve resolution of the currently available data.