I don’t know about your’s, but my mum gets worried when I don’t respond to her phone calls for a few hours. Once, I can’t remember what I was doing, but I didn’t hear the phone ringing. When I finally checked my phone I saw about 17483 missed calls, oops. I can only wonder what went through her mind when I wasn’t responding: she was probably picturing me skydiving, in an ambulance, or lost in the woods during a hike.

But what if she’d had a more statistical mindset and thought about why I hadn’t responded? Or even better: what if she’d thought about reasons why she could not detect me?

Ecologists and conservationists consider something similar when analysing data obtained from searching an area for a certain animal species. An animal could be present at a certain site, but still go undetected. First, they have to consider what ecological reasons might have determined where the species was present or absent (for instance, where is there suitable habitat within the considered area). Second, they have to take into account what factors might have influenced the likelihood of actually observing the species (such as the distance from the observer, or the fact that the surveyor may not be skilled enough to recognise the species). These are defined, respectively, as occupancy (which is the same as saying “presence”) and detection probabilities, and can be estimated by using statistical models.

Occupancy probability and detection probability are described by two different models and both of them will influence what will be observed during a survey. Taking into account that not all the animals will be observed is very important when attempting to accurately assess a species’ presence, which could otherwise be underestimated.

Peter Jahn, James Ross, Darryl MacKenzie and Laura Molles, in a study published in 2022, wanted to know how accurate acoustic surveys of roroa-great spotted kiwi (Apteryx maxima) were between 2011-2015. During this time, 18 birds were translocated from the Hawdon Valley, in Arthur’s Pass National Park, to the Nina Valley, in Lake Sumner Forest Park, representing one of the initial efforts of the Operation Nest Egg programme. The researchers also wanted to compare kiwi presence before and after 2015, and between the two areas.

They gathered data from a survey conducted in 2012-2013 by DOC in both the valleys and then repeated the methodology in 2017-2018. The technique they used was passive acoustic monitoring (PAM). PAM is effective when studying elusive species such as kiwi. Automatic recorders were deployed in the two study areas and left there for up to three weeks, activating just before sunset and switching off shortly after sunrise.

The team analysed the kiwi calls recorded in each of the valleys. The goal was to find a model that would best describe the obtained data, and use it as a base to estimate occupancy and detection probability. Peter Jahn and colleagues wanted to know which factors were important in detecting the kiwi and looked at the study area (Nina and Hawdon Valleys), year, length of the survey night, breeding/non-breeding season, precipitation, wind speed, night length, varying recorder battery capacity.

Similarly, my mum could have considered the fact that my phone may have been in silent mode, or had no service, or estimated the actual likelihood of me being in an ambulance. All of these factors could have influenced her imperfect detection of me.

In both the study areas, the detection probability was found to be higher during the breeding season, to increase with longer survey nights and to be influenced by wind speed, rain accumulation and recorder sensitivity. Also, as expected, kiwi presence in the Nina Valley increased after the translocation, as it did in the Hawdon Valley. Moreover, it was found that the number of sites where kiwi calls were recorded increased in 2017-2018 in both the areas and that, in total, many more calls were detected in the Hawdon Valley than in the Nina Valley.

Wait, the number of sites where calls were recorded and the presence of kiwi increased in the Hawdon Valley after kiwi were removed from there? How is that possible? Yeah, that was one surprising finding of the study. In fact, the researchers were expecting that occupancy would decrease after the birds’ removal, but what they found actually suggests that new pairs re-occupied the territories left inhabited by the translocated individuals.

This is a promising result, because it means that such conservation strategy doesn’t necessarily negatively influence the population from which the individuals are taken. Also, the ongoing pest mammal control in the Hawdon Valley could have balanced the negative effect of the translocation. I guess the only thing left to do now is find out what makes kiwi desire those territories so much that they can’t stay away: maybe they have the most delicious earthworms of New Zealand?

To conclude, these findings demonstrate that the species is reacting well to this reintroduction programme, considered that kiwi presence increased in the Nina Valley too. Furthermore, this study showed that combining occupancy estimates through statistical models with acoustic monitoring is very useful when studying the outcomes of kiwi’s translocations. However, if you, reader, can’t wait to know more about what happens to our dear kiwi when we move them around, sit back and read Peter Jahn’s PhD thesis: never stop learning.

Finally, going back to my mum trying to “detect” me: I suggest the probability would increase a lot if she learned to call outside of my usual napping times!

This article was prepared by Master of International Nature Conservation student Francisco Bini as part of the ECOL608 Research Methods in Ecology course.

Jahn, P., Ross, J. G., MacKenzie, D. I., & Molles, L. E. (2022). Acoustic monitoring and occupancy analysis: Cost-effective tools in reintroduction programmes for roroa-great spotted kiwi. New Zealand Journal of Ecology, 46(1), 3466.