Application of field sampling methodology

6.4.1 Spatial structure of study design

The study aimed to cover the geographical extent of the region: ranging from the more frequently flooded areas of the Coongie Lakes district to those least flooded areas of the far north and far south of the region. This also included the range of soil types and saline environments from north to south. This approach facilitated the collection of an appropriate data set that would be, “representative of and well distributed in the modelled geographical and

environmental ranges and that satisfy model assumption” (Bio et al., 2002) p2190. The heterogeneous spatial structure of the floodplain system informed the design and implementation of the sampling strategy.

The flood map classes identified by Costelloe (1998) (Chapter 2), were used in stratifying the region into sub-regions. These were selected to represent those areas of the floodplain where a maximum number of flood classes were shown to be present. This increased number of flood classes would be accompanied by an increased range of vegetation communities thus facilitating efficient sampling given financial, time and access constraints. To facilitate the finer scale location of sub-regions aerial photography was used in conjunction with the coarser flood class mapping.

Sixteen sub-regions were subsequently identified (see Figure 6-3). Within each of these sub-regions a number of sampling areas (patches) were identified for study. These were based upon the distinct vegetation communities derived from previous research (Gillen and Reid, 1988, Gillen and Drewien, 1993) and were deemed to be representative, homogeneous examples of the range of vegetation communities to be found within each sub-region (Mueller-Dombois and Ellenburg, 1974, Gauch, 1982). The number of vegetation communities encountered within each sub-region ranged from as few as one to as many as six.

Figure 6-3 Study area sub-regions

Twelve different vegetation communities and their variants were surveyed across the region. Not all communities were common to each sub-region. The vegetation communities included the following:

 Atriplex nummularia (Old-man saltbush) shrubland;

 Chenopodium auricomum (Queensland bluebush) shrubland;

 Chenopodium nitrariaceum (Nitre goosefoot) shrubland;

 Cressa cretica (Rosinweed) forbland;

 Cyperus gymnocaulos (Spiny sedge) sedgeland;

 Eucaluptus camaldulensis (River red gum) / Eucalyptus. coolabah (Coolibah) riverine woodland;

 E. coolabah (Coolibah) channel woodland;

 E. coolabah (Coolibah) floodout woodland;

 Muehlenbeckia florulenta (Lignum) shrubland;

 Sporobolus mitchellii (Rat‟s-tail couch) tussock grassland; and,

 Tecticornia indica (Samphire) shrubland.

Within each patch or unique homogeneous vegetation community, regarded as treatments for the design, two 100 metre line intercept transects, effectively pseudoreplications (Hurlbert, 1984, Dutilleul, 1993), were surveyed for the abundance of perennials present. The line intercept approach provides a suitable sampling procedure as vegetation communities of the region are often zonal in presentation, particularly around lake fringes and along and adjacent to channel levees. Transects ranged from 500 metres to 2 kilometres apart depending upon access restraints. Transects were placed to avoid ecotones and thus ensure homogenous representation of each community. In all 138 transects were positioned and assessed. In summary the spatial structure of the study design is multilevel or hierarchical with three levels of geographic stratification: 16 sub-regions within the Ramsar region; 1-6 patches (vegetation communities) within each of the sub-regions; and two transects within each of these (Figure 6-4).

Figure 6-4 Schematic of the heirarchical sampling design for the study area

An alpha-numeric nomenclature was adopted for site identification. Each of the 16 sub- regions was identified alphabetically from A to P (Figure 6-3). Each vegetation community type

sampled within a sub-region was sequentially numbered as encountered. The two transects located within each vegetation community were simply labelled 1 and 2. For example; in sub- region M there were five communities sampled. Thus the code, M 5-1, refers to transect 1 of the 5th community sampled, in this instance a Eucalyptus coolabah woodland.

The sampling structure then is an unbalanced, nested design which incorporates several magnitudes of spatial scale. The potential effects of spatial dependence and spatial autocorrelation are thus accommodated and hence the spatial random effects associated with this hierarchical structure can be incorporated within subsequent statistical modelling analyses (Latimer et al., 2006, Price et al., 2009). The fixed effect or treatment is the vegetation community associated with a particular patch. In effect the patch is the experimental unit and the transect is the observational unit. Patches were replicated across the region within the sub- regions whenever possible given the spatial heterogeneity of vegetation communities across the study region.

6.4.2 Field-work

6.4.2.1 Reconnaisance

An initial reconnaissance of the study region was made in mid 2004. Access was attempted from the Birdsville track. Unfortunately heavy local rains thwarted vehicular access.

A subsequent reconnaissance trip was conducted over a two week period in May 2005. Meetings were held with managers of Innamincka, Gidgealpa, and Clifton Hills pastoral properties. Intended survey methodology, acceptable access routes, stock movements and vegetation condition were discussed. Additionally, representatives of the Innamincka Regional Development Group were notified of the nature of the study and future research in the region and the thrust of the intended research.

A traverse was made from Innamincka to Coongie Lakes, across to Gidgealpa, up the Christmas Creek track to the Birdsville track and on to Clifton Hills Station. En route, potential access routes and river crossings were assessed, vegetation condition observed and sampling techniques for soil and vegetation survey trialled. The feasibility of conducting a topographic survey using a Differential Global Positioning System (DGPS) approach was assessed (and subsequently abandoned due to technical difficulties).

During the two phases of field work, for the sake of efficiency, base camps were established within a number of the sub-regions from which sorties were made to sites within each of the sub-regions and often beyond. During all periods in the field, contact was made and maintained with relevant regional pastoral stations (Innamincka, Gidgealpa and Clifton Hills). All stations were made aware of the research being conducted within region and the intended schedule. A regular daily schedule of satellite phone communications was maintained whist in

the field. As a further safety precaution an EPIRB or satellite safety beacon was carried at all times.

The first main period of field work took place during the period September–November 2005. Heavy rain disrupted field work twice during this period inhibiting movement on the heavy clay floodplains for up to a week at a time. This delay set back the time available for research. Rapidly increasing temperatures associated with the onset of summer rendered field conditions too dangerous to continue. A second field trip to the study region was required. Substantial progress was made during the first period of field work: 106 sites were assessed for soils and vegetation and over 100 plant voucher specimens were collected.

The second and final field trip was conducted from late July to late August 2006. An additional 32 sites were assessed from the more remote north-east area of the study region, providing the necessary representation of regional variation from Lake Hope in the south to Lakes Sir Richard and Lady Blanche in the far north-east.

6.4.2.2 Sampling procedure within transects

6.4.2.2.1 General

Upon arrival at each transect site, geographic coordinates were recorded using a GPS, the orientation of the 100 metre transect established, and a site photograph (Plate 6-1a) taken with compass bearing recorded. Brief notes recorded the structural characteristics of the vegetation community and basic local landform and soil features. Depending on the structural and floristic complexity of vegetation, sampling time ranged from 1.5 to 3 hours per transect.