Vegetation change observed from on-ground repeat photography, Macquarie Island - 1980-2021
Repeat photography from locations across Macquarie Island were visually compared to detect and categorise temporal changes in vegetation for each site. Photographs of the same scene in 1980, 2009 and 2014 were used. Terrain variables were used as predictor variables to investigate potential drivers of different types of vegetation change. Satellite-derived vegetation indices were compared with the on-ground photography for the latter period.
Dataset derived from visual analysis of a collection of repeat photography images from Macquarie Island. Three images from different years were compared for each location: 1980, 2009, 2014. Changes in vegetation were recorded for two change periods: 1980-2009 and 2009-2014.
MI Plateau 35year photo series_GPSdata and notes_Aug15_COPY.xlsx
Photograph metadata (location, dates, etc)
photochangesplateau.xlsx
Observed changes for each photopoint
photochanges_georef_3class.csv
Reclassified change data
WorldView_SVI_change_terrain.xlsx
Spectral vegetation indices (SVIs) derived from WorldView satellite sensor for photopoint locations. Calculated using ENVI software. NDVI - Normalized Difference Vegetation Index. MTVI - Modified Triangular Vegetation Index. SGI - Sum Green Index.
JScott photos 1980-2014.kmz
Location of photo points (i.e. location of photographer)
Shapefiles
ESRI shapefiles of point locations of centrepoint of 'photo zones', i.e. the subjectively defined areas where change was recorded within the viewshed of each photograph.
data
R files detailing data analysis plus csv files of data used for analysis.
Classes of vegetation/landscape:
Grass - short grassland (not Poa tussock) vegetation dominated by Agrostis, Luzula, Deschampsia, Festuca
Pleu - Pleurophyllum hookeri
Poa - Poa foliosa
Ace - Aceana magellanica and A. minor
Stilb - Stilbocarpa polaris (syn. Azorella polaris)
Bare - bare ground
Moss - bryophytes
Poly - Polystichum vestitum
Az - Azorella macquariensis
Az dieback - dieback of Azorella cushion plants (present/absent)
Change classes:
decrease
no change
increase
Data files use 0/1 values for binary data
0 = true/absent
1 = false/present
999 = NA/no data
e.g. GrassInc80 with a value of 0 means no increase in grass observed in the 1980-2009 period,
AceDec09 with a value of 1 means Acaena spp. did decrease in the 2009-2014 period,
landslide = 0 means site (polygon) not impacted by a landslide
Some datasets have been recoded to 4 categories:
0/1/2/3 value for decr/stable/incr/NA(unknown)
e.g. Poa80 = 0 means Poa foliosa decreased in the 1980-2009 period,
Bare09 = 1 means bare ground stable (no discernible change) over 2009-14
MICROINVERTEBRATE SAMPLING PROTOCOL
Macquarie Island
01 October 2001 - 28 February 2002
A.HABITATS SAMPLED
8 habitats representative of the following vegetation types were chosen:
1.Azorella macquariensis - Open cushion areas
2.Acaena (magellanica and minor) herbfield
3.Colobanthus muscoides (coastal cushion plants)
4.Mires - Upland
5.Pleurophyllum hookerii dominated areas
6.Poa foliosa Tall tussock
7.Short grassland (incl. Agrostis magellanica/ Festuca contracta/ Luzula)
8.Stilbocarpa polaris dominated coastal herbfield
B.HABITAT LOCALITIES
1.Range within which quadrats for a chosen habitat were located :
a) Altitudinal limits- Lowland (coast to +/- 300 - 350m)
b) Area- Spread over whole island
c) Distance- i) 500m min. distance from the perimeter of the Base/logistic zone Viz. none in the logistic zone.
- ii) 100m min. distance from an established hut
- iii) 50m min. distance from an established path
d) Aspect- East and west coasts
2.Types
a) Homogeneous areas
b) Least impacted areas (viz. Avoided heavily grazed Rabbit areas)
(viz. Avoided Alien dominated areas)
(viz. Avoided previously sampled or long term study sites)
C.GENERAL SAMPLING STRATEGY FOR EACH HABITAT
1.For each habitat Five 2m x 2m quadrats were located (similar in vegetation structure) and marked 1-5.
2.From each quadrat two random samples were taken with the O'Connor split corer (as per sampling protocol D below). Viz: 10 cores from each habitat.
3.Each sample was retained separately (in it's core-tube placed in a plastic bag) and marked accordingly. Viz: A and B from 1 through to 5 (e.g.: Poa1A-B, Poa2A-B, etc to Poa5A-B).
4.On return from the field samples were immediately stored the in a cool, safe (rodent free) place (lab refrigerator) for processing.
5.Invertebrate extraction followed as per protocol E below. Sample numbers were retained throughout the sampling period, together with sampling date.
6.Each habitat was sampled on an average of once every five - six weeks.
D.SAMPLING METHOD
1.Random numbers were obtained using a table of random numbers.
2.Numbers 1-100 are in top left quarter, progressing clockwise in the remaining three quarters for 101-200, 201-300 and 301-400.
3.If the position chosen for the first core had already been cored, the next random number and so on was used.
4.The core sample comprised a 70mm depth from ground level (viz. not including above ground vegetation growth and flowering parts).
5.Care was taken to disturb as little as possible of the vegetation in and around quadrat, as well as approach to site.
6.Sampling in or directly after heavy rain was avoided to prevent poor results (although it never rained hard or long enough for this situation to have occurred).
7.Samples were processed within 4 days (max) after return or safe / cool storage.
8.Before re-using any equipment (corer, cores, plastic bags, collecting jars and mesh cover etc), it was cleaned thoroughly to avoid contamination.
E.EXTRACTION AND SORTING
MESO-INVERTEBRATES : (These include all collembola and mites and enchytraeid earthworms).
1.In the collecting bottle of each sample placed in the HG extractor, an amount (+/- 2 cms high) of propylene * glycol was poured (*propylene glycol; CH3 CH(OH) CH2 OH = 76.10).
2.Core-samples were separated into litter-like top and about 5- 7 cm of soil.
3.Samples were retained in their respective core-rings, and where above ground vegetation biomass was more than could fit the depth of a ring, this was placed into additional rings. The veg (top)-side was covered with mesh or mutton cloth (approx. 1.5-2mm diam.) and secured with elastic bands (shock cord 3mm diam.).
4.The mesh covered side was placed facing down over the collection bottle in the HG extractor. The HG was left running for the first 2 days at 25 degrees C, and for the following two days (3rd and 4th days) at 30 degrees C. 5.Samples were transferred to 99% or 100% alcohol by draining off the propylene glycol through a 60 micron mesh, picking all the colembola and mites off it with a very fine paint-brush through the view of a good microscope, and placing these into labeled vials.
6.The filtered propylene glycol was re-used a couple of times.
7.Where time allowed, mites and colembola were separated for certain samples.
8.Sample details were noted in pencil on labels provided on the outside of each vial, and printed labels were inserted into each sample vial (see Macca Colembola and Mite labels 2001-02.doc).
F.DATA ACQUISITION AND ARCHIVAL
1.Field data were captured in pencil using one A6 hard-cover note-book.
2.Data was transferred to spreadsheet and document and stored on CD-R discs with a back-up copy.
This work was completed as part of the RiSCC project (Regional Sensitivity to Climate Change).
The fields in this dataset are:
Site name
Habitat
Location
Latitude
Longitude