2. Methods

2.1 Study area 

The study was carried out in the Upper MacRitchie Basin in the Central Catchment Area of Singapore. 

  Fig 2. The location of the sampling sites in the Upper MacRitchie Basin in the Central Catchment Area (Source: Murphy, 1997)
 

The Upper MacRitchie Basin has topographic features that allow three main stream systems to flow through it. They are labeled stream systems G, H and J (Murphy, 1997). Access to the streams requires navigating through thick foliage, sometimes posing a difficulty for the researcher. Sites were selected at locations that are representative of the sub-streams being sampled.

2.2 Data collection

The data collection process followed the following standardised protocol: Sampling was usually done between 10 am and 2 pm on a sunny day to make sure that the water parameters and odonates behaviour were not affected by sunlight and temperature fluctuations. On arrival at the sampling site, GPS coordinates were taken to accurately map the sampling site. The habitat was then classified into ‘Shaded stream’ or ‘Open field’ and a photograph was taken of the site.

The water parameters were then collected using the YSI MDS650 and YSI MDS550 water parameter sensors. Specifically, the seven water parameters measured  were Dissolved Oxygen, Temperature, pH, Turbidity, Conductivity, Total Dissolved Solids and Salinity. These water parameters affect the ecosystem in different ways (Kwok, 2008). The table below spells out the water parameters in relation to the dragonfly and damselfly nymphs. 

Dissolved Oxygen (mg/L)	Dissolved Oxygen in surface water is usually acquired from dissolution of oxygen from the atmosphere or from photosynthesis of aquatic plants or algae. At 26°C, the maximum dissolved oxygen is 100% or 8.11 mg/L. Fish, invertebrates and other aquatic animals require oxygen for breathing. DO range from > 4.87 mg/L or 60 to 120% indicates that the water is healthy. Above and below these ranges, the waterways will become increasing stressed.
Water temperature (°C)	Surface water temperature is affected by the ambient temperature and solar radiation and is an important parameter to aquatic life. Thermal pollution or changes in the amount of solar radiation can affect the reproduction of macro-invertebrates and micro-organisms. Typical water temperature range is about 26°C for tropical Singapore.
pH	pH affects both chemical and biological processes, influencing aquatic life directly and indirectly. Unpolluted surface water should have a pH of between 6.5 to 8.5 which is ideal for the growth and well-being of most aquatic organisms. Typical soil in Singapore is acidic by nature.
Turbidity (NTU)	In water bodies, high turbidity levels can reduce the amount of light reaching lower depths, which can inhibit growth of submerged aquatic plants and consequently affect species which are dependent on them. Stream water ranges from 0 NTU for healthy water to 30 NTU for heavily turbid water. High turbidity may affect the predator-prey behaviour of fishes and dragonfly nymphs (Meutter, 2005).
Conductivity (mS/cm)	Electrical conductivity is commonly due to the presence of dissolved ions mostly caused by the erosion of rocks. Conductivity ranges from below 60 mS/cm for clean water and above 300 mS/cm for polluted water.
Total Dissolved Sollids (g/L)	The chemicals may be cations, anions, molecules or agglomerations on the order of one thousand or fewer molecules, so long as a soluble micro-granule is formed. Water containing a TDS level of over 0.50 g/L is unsuitable for most freshwater macro-invertebrates.
Salinity	Salinity is measured either in terms of Electrical Conductivity or Total dissolved solids. It is measured as a ratio of its salinity to that of the conductivity or TDS of KCI of 35 g/L.

  Table 1: Descriptions of the water parameters used in the survey of water (Hunter, 2011)

The sampling area was designated to be within a zone of 50 meters along the stream where the odonates were most likely to be found. Adult odonates were collected by hand-net within the zone. Because many forest streams were choked with hooked rattans or pandans, this was inefficient and some of the records had to be made by sighting alone. For those that were caught, photographs were taken at different angles before the odonates were released. In cases where there was difficulty in catching the dragonflies, a telescopic camera was used to take pictures from a distance. The identification of the odonates was done using an established taxonomy of odonates in Singapore (Tan, 2010).

The use of adults instead of the larvae or exuviae stages odonates has several advantages as highlighted by Simaika (2008). Firstly, sampling larvae would have been more difficult as they may hide in inaccessible parts of the stream. Secondly, it is not easy to identify the larvae until they become adults. Thirdly, most dragonfly larvae remain undescribed and their identification is difficult. Fourthly, we can assume that odonates mate and oviposit only in suitable freshwater habitats. Thus we can assume residency of most species collected.  

I have specifically chosen not to make an abundance count of the odonates nor to perform exhaustive species-richness estimation as presence-absence data sufficiently characterised community composition (Oertli, 2008). On the contrary, a sub-sampling was often enough to give an indicator that is representative of odonate diversity. This procedure was a much less labour intensive more efficient sampling method.

However, the use of estimated sub-sampling should be taken in the proper context of the time of sampling. In tropical Singapore, although there are no distinctive seasons, the variation of rainfall in different months affect the temperatures and hence the odonate populations in the Upper MacRitchie Basin. Furthermore, it would be difficult to conduct sampling during a rainy day. For the current research, almost all the sampling days were on sunny days in the first quarter of the year.  

2.3 Data analysis       

Based on the data collected, one can make a correlation between the species of odonates collected and the environmental indicators of habitat and physiochemical water parameters. The correlations were expressed in terms of a spatial distribution map of odonates to show how the various enclaves of odonates are correlated to the different habitats. The physiochemical parameters of the streams to be sampled were also correlated to the various odonates found. These two correlation maps could be used as a basis in which further investigation about the health of the water resources could be made.