Use of fyke nets
A fyke net designed to capture platypus consists of a long, cylindrical netting chamber (divided into a series of internal compartments fitted with one-way funnels) flanked by rectangular netting wings (as shown at right). The deepest water in which a net can be set is limited by the height of its central chamber and wings (typically <0.6 metres). Mesh size is usually fairly coarse (>10 mm) to reduce the rate at which floating materials such as leaves accumulate on the upstream face of the wings. This can become very problematic in high flows, displacing rocks from the bottom margin and potentially dragging nets downstream.
Nets are normally set at 4-8 sites in the afternoon and removed from the water soon after the following dawn. The maximum number of sites is limited by the need to check nets at regular intervals after they’re set. To capture a high percentage of the animals occupying an area, the spacing between consecutive sites should ideally be a little less than the average distance that an animal travels overnight. This distance is typically longer in narrower and/or less productive water bodies as compared to wider and/or more productive water bodies. In practice, the ideal spacing between sites is likely to be around 500 to 1000 metres (Gust and Handasyde 1995; Serena and Williams 2012; Bino et al. 2018).
Two fyke nets are normally set per site, with one net facing upstream and the other facing downstream. To ensure that captured animals don’t drown, it’s imperative that a substantial air space is maintained in the final compartment of the central netting chamber (and ideally along its entire length) by securely attaching its end to a fixed anchor point, such as a metal stake hammered into the channel. Other platypus-related welfare issues associated with fyke net use potentially include hypothermia, hunger, stress, predation, exhaustion and fighting between adult males during the breeding season. To learn more about mitigating these concerns (and how to safeguard the welfare of non-target animals when fyke nets are deployed), see Platypus Fyke-netting Guidelines.
To work as effectively as possible, fyke nets must be set in a manner that reliably directs a platypus into the central netting chamber before it finds a way to scramble under, over or around the netting. This requires (1) weighting down the entire lower margin of each net (wings and frame) with rocks or the equivalent so all gaps between the net and channel substrate are eliminated, (2) ensuring that the entire top margin of each net is held well above the water, and (3) either extending wings up onto land for a distance of at least one metre (especially at sites where the banks are both flat and open) or pinning the wings securely against vertical banks so gaps are again eliminated. The manner in which fyke nets are set can have major consequences for platypus capture success: a statistically significant fourfold difference has been recorded in the number of animals captured by two different groups setting nets at the same sites in consecutive annual seasons near Melbourne. The less effective group often left gaps of 10 centimetres or more (easily large enough for a platypus to find and wriggle through) between the rocks used to weigh down a net (and sometimes used sand in place of rocks when these were in short supply); wings were often stretched less than 50 centimetres up a bank and sometimes didn’t extend beyond the water’s edge. This finding highlights that caution is warranted when combining fyke-netting results obtained by different researchers, e.g. to describe population trajectories.
Use of mesh (or gill) nets
Unweighted or minimally weighted rectangular mesh nets fitted with floats can be used to capture platypus in slow-flowing water bodies that are too deep for fyke nets to be used effectively. Nets are typically 15-50 metres long and 1.3-2.5 metres wide and have a mesh size of around 8 centimetres (Grant and Carrick 1974). Floats are painted white to improve their visibility at night, and the ends of nets are normally fastened to overhanging branches or emergent woody debris so nets are aligned with the direction of flow. This reduces the likelihood that nets are lifted in the water column by the current (which is presumed to contribute to an inverse relationship between high flow volume and mesh-netting capture success: Bino et al. 2015), and also reduces the amount of floating debris carried into nets.
To avoid drowning in a mesh net, a platypus must rise to the surface to breathe very soon after it becomes entangled. It is therefore essential that nets are not weighed down by entangled fish or become snagged on submerged logs or branches. In practice, the entire length of each net must be lifted by hand to the surface at intervals of one hour or less to release fish and ensure that it’s otherwise hanging freely (Grant and Carrick 1974). Nets should also be scanned with a spotlight after dark at intervals of 10 minutes or less to reveal the presence of an entangled platypus so it can be removed promptly (Grant and Carrick 1974). By adhering to these guidelines and proactively avoiding sites where large fish and/or bottom snags were abundant, Grant (2004) reported a netting-related platypus mortality rate of only 0.003 (2 deaths in 700 captures) in studies carried out along the upper Shoalhaven River in New South Wales.
Metrics used to assess abundance based on live-trapping data
To assess platypus numbers or population trends over time, the results of live-trapping studies can be summarised by using individual counts, by calculating population indices such as CPUE (catch per unit effort) or MNA (minimum number alive), or (if animals are marked and recaptured over time) by applying statistical capture-recapture models to estimate population size.
These approaches have the following features in common:
- The accuracy and precision of all population metrics are limited by the scope and effectiveness of fieldwork contributing to the analysis. Although self-evident, we believe this is worth repeating as it often seems to be overlooked in practice.
- Juveniles should not be included in metrics if the aim is to estimate the size and/or density of a resident platypus population. This reflects two facts: (1) Most juvenile platypus disappear from their birth population before maturity, with the rate of juvenile recruitment estimated to be as low as 9% (Grant 2004) or 18% (Serena et al. 2014). (2) Platypus reproductive success shows a strong positive relationship with the amount of stream or river flow in the months prior to breeding (Serena et al. 2014; Serena and Grant 2017). Lumping juveniles with established resident adults will therefore tend to bias population metrics upwards in relatively wet years and downwards in relatively dry years, regardless of whether the number of resident adults has changed.
- Caution is warranted before combining live-trapping results from different water bodies (e.g. to increase overall sample size), as demographic parameters and population trajectories may vary widely even among similar-sized streams found in close geographic proximity (Serena et al. 2014).
Counts (number of unique individuals captured)
Counts of the number of animals captured in a survey period can be usefully employed as an index to estimate abundance as long as enough replicated sampling is carried out (McKelvey and Pearson 2001). Counts have several useful features: they can be used to describe abundance at one location or over larger areas, they don’t require animals to be recaptured, and they don’t rely on animals being permanently marked (though some way is needed to identify animals that have already been encountered during a given netting session – or earlier in the same survey period – if netting is carried out on more than one occasion). In addition, counts necessarily take on a value between 0 and true population size, which is not always the case for the results of statistical models (McKelvey and Pearson 2001). After finding that field counts were highly correlated with population estimates derived by modelling mark-recapture data for five rodent species, Slade and Blair (2000) concluded that counts can provide a valid index to estimate and compare abundance for a given species as long as a consistent trapping protocol is employed.
MNA (minimum number alive)
Minimum Number Alive analysis generates a modified count, calculated as the number of animals captured in a given survey period plus the number that were not captured but were assumed to be present given that they were encountered on previous and subsequent occasions (Krebs 1966). This index has been found to perform very well when applied to long-term platypus mark-recapture data, yielding population estimates that are similar to (but more precise than) those generated by population modelling (e.g. Cormack-Jolly-Seber modelling: Serena and Grant 2017; modified Chao sparse estimator: compare Fig. 5 in Huggins et al. 2018 with Fig. 5 in Serena et al. 2014). However, use of MNA is known to underestimate population size in the initial and final stages of a longitudinal study (Pocock et al. 2004). It’s therefore most appropriately applied when data from the start and end of a long-term study can be sacrificed without compromising the goals of the analysis.
CPUE (catch per unit effort)
CPUE is calculated by dividing the total number of captures (not the number of unique captured individuals) by trapping effort. In the case of mesh nets (which hang freely in the channel and so can be assumed to contribute to trapping success in a more or less independent and additive manner), platypus trapping effort has previously been quantified in the scientific literature in units of net-hours (1 net-hour = 1 net of specified dimensions set in the water for 1 hour: Grant 1992). In the case of fyke nets (which are set for platypus in a configuration that blocks the channel so nets don’t contribute additively to trapping success if more than two are deployed), platypus trapping effort has previously been quantified in the scientific literature in units of site-nights (1 site-night = paired [or potentially more] nets set to block the channel overnight at a given location: Serena and Williams 2012) or site-hours (1 site-hour = paired [or potentially more] nets set to block the channel for one hour: Koch et al. 2006).
CPUE can be used to describe the status of platypus at either one location or across larger areas, does not require animals to be recaptured, and does not rely on animals being marked permanently (though it’s a very good idea to have some way of identifying animals that have previously been encountered in a given netting session – for example, due to their doubling back and again entering a net after having been released earlier in the night). Because this index is calculated as a decimal fraction, care needs to be taken to ensure that the number of significant figures used to report CPUE isn’t inappropriately inflated, conveying a false sense of precision (particularly when different values are being compared). As a good rule of thumb, CPUE should be rounded to the same number of decimal places as the least precise number used to calculate it. In other words, the level of precision implied by two significant figures (e.g. 1.2 or 0.38) shouldn’t be reported unless at least 10 captures were recorded based on at least 10 site-nights of live-trapping; three significant figures (e.g. 1.24 or 0.381) shouldn’t be used unless at least 100 captures were recorded based on at least 100 site-nights of live-trapping.
Statistical population models
Population models can provide abundance estimates that are relatively precise and unbiased as long as samples are large enough to support effective model selection (e.g. McKelvey and Pearson 2001 and references cited therein). They require mark-recapture data obtained through studies in which animals are uniquely marked. Both Jolly-Seber modelling (Grant and Carrick 1978) and Cormack-Jolly-Seber modelling (Bino et al. 2015) have been successfully applied to data obtained from a relatively large platypus population where annual netting occurred regularly for many years. However, use of the Cormack-Jolly-Seber model for sparser data sets may not be appropriate due to the need to estimate a large number of occasion-specific capture and survival parameters and – though Jolly-Seber modelling is more forgiving in this respect – its reliability is still limited by the frequency of recaptures (Huggins et al. 2018).
Alternatively, Huggins et al. (2018) extended Chao’s sparse data estimator to model open populations and then tested the method by applying it to mark-recapture data for a small (albeit closed) stream-dwelling platypus population. The new estimator was found to be prone to negative bias (so best interpreted as a lower limit to population size) and sensitive to population variation over a time scale corresponding to three consecutive sampling periods (so, to be appropriately applied, the population of interest should be relatively stable over such a time frame). In practice, the sparseness of the information available to the platypus model weighed on the precision of annual population estimates (with 95% confidence intervals typically ranging from about 7 to 35 individuals). As a matter of good practice, we therefore recommend that 95% confidence intervals should routinely be reported alongside mean estimates of platypus population size or abundance whenever these are generated through statistical modelling.
Recommended guidelines for designing live-trapping studies
Platypus population surveys
- Ideally conduct enough replicated sampling that live-trapping data can be used to generate reasonably reliable estimates of absolute population size or density for the area where nets are set. In practice, the amount of effort required to achieve this will depend on both the size of the live-trapping area and the spacing between netting sites. As a starting point, Serena et al. (2014) found that, on average, 58-65% of resident adults/subadults were captured annually in streams near Melbourne where pairs of fyke nets were set overnight on a mean 2.0-2.4 occasions annually, at mean spacings of 0.8-1.2 km. This suggests that at least 4 replicated surveys are likely to be needed to capture all or nearly all of the resident animals occupying a given length of stream sampled at this spacing. If nets need to be distributed across a really extensive area as an integral aspect of a survey, consider selecting some subsections for more intensive replicated sampling after the initial survey work has been carried out.
- Along with abundance, consider the implications of other demographic attributes when assessing population viability. For example, Serena et al. (2014) found that declining platypus populations were most clearly differentiated from stable populations by reproductive success. Similarly, Bino et al. (2015) found that the most important parameter to influence modelled platypus population survival was adult female survival, followed by survival rates of dispersing juveniles.
- Given the strong influence of net-setting technique on survey outcomes, photographic documentation of how nets are set at each survey site should routinely be provided as part of survey (or monitoring) reports.
- Similarly, given the strong influence of flow and season on live-trapping outcomes, platypus survey (or monitoring) reports should discuss how these variables may have affected the findings, and the amount of flow in the survey area should be documented if possible.
Long-term population monitoring
- To improve the reliability of findings, it’s recommended that nets should be set on several replicated sessions in each round of platypus monitoring, with consecutive sessions in a given monitoring round scheduled at intervals of a week or (ideally) more and consecutive rounds scheduled at intervals of two years or more. The decision to not conduct annual monitoring is appropriate given that the platypus is a long-lived species (with a life span of up to 21 years: Grant 2004) and little or no change in population size is therefore likely to be evident from one year to the next, particularly if nets are only set on one or two occasions. Allowing at least two years between consecutive monitoring efforts will also help to ensure that findings are not biased by trap-shyness, which is particularly likely to develop as an outcome of the fact that platypus nets are unbaited (Griffiths et al. 2013).
Fifth photo from top courtesy of Ken Mival, other photos: APC
Bino, G., Grant, T. R. and Kingsford, R. T. (2015). Life history and dynamics of a platypus (Ornithorhynchus anatinus) population: four decades of mark-recapture surveys. Scientific Reports 5, 16073.
Bino, G., Kingsford, R. T., Grant, T., Taylor, M. D., and Vogelnest, L. (2018). Use of implanted acoustic tags to assess platypus movement behaviour across spatial and temporal scales. Scientific Reports 8, 5117.
Grant, T. (2012). Environmental impact assessment: monitoring from a platypus perspective. Pp. 107-113 in Science under Siege: Zoology under Threat (P. Banks, D. Lunney and C. Dickman, eds). The Royal Zoological Society of New South Wales, Mosman.
Grant, T. R. (1992). Captures, movements and dispersal of platypuses, Ornithorhynchus anatinus, in the Shoalhaven River, New South Wales, with evaluation of capture and marking techniques. Pp. 255-262 in Playtpus and Echidnas (M. L, Augee, ed.). The Royal Zoological Society of New South Wales, Mosman.
Grant, T. R. (2004). Captures, capture mortality, age and sex ratios of platypuses, Ornithorhynchus anatinus, during studies over 30 years in the Upper Shoalhaven River in New South Wales. Proceedings of the Linnean Society of New South Wales 125, 217-226.
Grant, T. R. and Carrick, F. N. (1974). Capture and marking of the platypus, Ornithorhynchus anatinus, in the wild. The Australian Zoologist 18, 133-135.
Grant. T. R., and Carrick, F. N. (1978). Some aspects of the ecology of the platypus, Ornithorhynchus anatinus, in the Upper Shoalhaven River, New South Wales. Australian Zoologist 20, 181-199.
Griffiths, J., Kelly, T. and Weeks, A. (2013). Net-avoidance behaviour in platypuses. Australian Mammalogy 35, 245-247.
Gust, N., and Handasyde, K. (1995). Seasonal variation in the ranging behaviour of the platypus (Ornithorhynchus anatinus) on the Goulburn River, Victoria. Australian Journal of Zoology 43, 193-208.
Huggins, R., Stoklosa, J., Roach, C., and Yip, P. (2018). Estimating the size of an open population using sparse capture-recapture data. Biometrics 74, 280-288.
Koch, N., Munks, S. A., Utesch, M., Davies, P. E. and McIntosh, P. D. (2006). The platypus Ornithorhynchus anatinus in headwater streams, and effects of pre-Code forest clearfelling, in the South Esk River catchment, Tasmania, Australia. Australian Zoologist 33, 458-473.
Krebs, C. J. (1966). Demographic changes in fluctuating populations of Microtus californicus. Ecological Monographs 36, 239-273.
McKelvey, K. S., and Pearson, D. E. (2001). Population estimation with sparse data: the role of estimators versus indices revisited. Canadian Journal of Zoology 79, 1754-1765.
Pocock, M. J. O., Frantz, A. C., Cowan, D. P., White, P. C. L., and Searle, J. B. (2004). Tapering bias inherent in minimum number alive (MNA) population indices. Journal of Mammalogy 85, 959-962.
Serena, M., and Grant, T. R. (2017). Effect of flow on platypus (Ornithorhynchus anatinus) reproduction and related population processes in the upper Shoalhaven River. Australian Journal of Zoology 65, 130-139.
Serena, M. and Williams, G. A. (2012). Effect of sex and age on temporal variation in the frequency and direction of platypus (Ornithorhynchus anatinus) captures in fyke nets. Australian Mammalogy 34, 75-82.
Serena, M., Williams, G. A., Weeks, A. R., and Griffiths, J. (2014). Variation in platypus (Ornithorhynchus anatinus) life-history attributes and population trajectories in urban streams. Australian Journal of Zoology 62, 223-234.
Slade, N. A., and Blair, S. M. (2000). An empirical test of using counts of individuals captured as indices of population size. Journal of Mammalogy 81, 1035-1045.