DATA ANALYSIS: DESERTS ➤
phosphorus) receives a lot of data analytic attention. With marginal vegetation comes marginal habitat, and studying wildlife populations in desert regions is a busy area for data analysis of all kinds. Progressive desertifi cation means loss of adjacent steppe and other ‘sub-desert’ habitats, producing either loss of indigenous species (and therefore biodiversity) or migrations that trigger wavefronts of environmental impact. Human populations, of course, are impacted by changes in both fl ora and fauna, not to mention water supplies themselves, and manifest modifi ed forms of the same processes. Water is becoming a strategic scarcity resource over which blood is already spilt and likely to be the cause of future wars as climates shift, populations continue to grow, and agricultural environments degrade. The experience of Abu Dhabi (see ‘Watering the future’) suggests that use of
‘Just the heterogeneity and lack of consensus in basic criteria so far agreed for analysis is enough to provide food for thought’
computer analytic methods could help to reduce the growth of such tensions. Reliable data is the starting point of meaningful analytic research, and gathering data on wildlife populations is not an easy business, even when they are large and visible. When they are small, and when they (as most do) make it their business to be inconspicuous, it becomes ever more diffi cult to assemble high-grade data sets. Collecting data on the individuals that make up those populations is more problematic still. As a young cog within the large machine of a small mammal survey in the former Yugoslavia, nearly three decades
Where the wild things are
Wildlife census studies, especially in hostile terrains, traditionally depend either on aggregate methods that do not allow detailed tracking or, alternatively, involve invasive procedures such as tagging or radio collaring – which can be both harmful and unreliable. WildTrack, an organisation cofounded by Sky Alibhai and Zoe Jewell, uses SAS and JMP software to ‘fi ngerprint’ an individual animal’s spoor using what they have named Footprint Identifi cation Technique, or FIT. In a 2001 paper, they described[18]
the
analytic roots of their process. Digital images were taken of left hind spoor for 15 known individual rhinos. Nearly 80 linear and angular measurements between manual markers on each print were chosen for comparison using discriminant and canonical analyses. Using a variety of techniques in combination, they established high-confi dence recognition of each animal from its footprint. Since then, both the methods and the software have been greatly refi ned. The original manual measurement methods have been replaced by computerised image analysis. JMP scripts apply recognition algorithms to the resulting data. FIT has moved on from rhino populations to other species – notably cheetah, tigers, lynx, polar bears and tapirs. Angie Nash, ecology and conservation lecturer at Cornwall College, is now developing FIT for use with brown bears in the Gobi Desert as part of a PhD programme at the University
14
ago, I spent much of my time sieving soil to extract, sort, measure, weigh and classify droppings, but scientifi c computing has moved things on somewhat since those days. FIT techniques (see ‘Where the wild things are’) were developed for rhinoceros tracking, but can apparently now be applied to something as small as the European dormouse, which should in principle mean that it could be used eventually to study almost anything that has feet. No look at arid region fauna in the early
21st century would be complete without mentioning the world’s love affair with its favourite viverrine, or the analytic software most associated with life sciences, and there is no shortage of data analytic stories that combine the two. My personal favourites are a series of GenStat analysed studies[14-16] published since 2008 by Alex Thornton (at Cambridge University’s department of zoology) using meerkat tribes in the Kalahari
As part of WildTrack’s noninvasive wildlife monitoring programme, a footprint (in this case, of a Namibian cheetah) is photographed for addition to the database and subsequent analysis. Photograph by, and courtesy of, Stephanie Hirn
of Exeter. ‘Using track data along with other non-invasive genetic data gathered through faeces and hair, I hope to estimate minimum number known alive for each region, and to work with local people to implement monitoring and custodianship of the species.’ That reference to local expertise is echoed by Florian Weise, research director of Na’an Ku Se Wildlife Sanctuary in Namibia, who comments on the WildTrack website that
‘San trackers ... have mastered this skill and technique over tens of thousands of years and FIT can build on and incorporate this precious knowledge...’ It also recurs in Alibhai’s expectation that, although most of the analysis is currently done by WildTrack, ‘when it’s evolved to the extent that it’s really user- friendly, the whole system can be turned over to the people in their own countries and they can become custodians of their own wildlife.’
SCIENTIFIC COMPUTING WORLD OCTOBER/NOVEMBER 2010
www.scientifi
c-computing.com
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56