View Static Version
Loading

Maximising ranger-collected data on elephant poaching A 10 minute summary of a 3 year PhD

Globally, hundreds of thousands of wildlife rangers patrol wide areas within protected areas every day, observing diverse plants and animals as well as evidence of illegal activities like poaching. Data collection by rangers has enormous potential to track changes in biodiversity and threats to it - crucial information for both biodiversity science and policy.

This PhD investigated the reliability and conservation value of ranger-collected data on elephant poaching

Timothy Kuiper, supervised by Professor EJ Milner-Gulland

A ranger uses a handheld device to record an old elephant carcass while on patrol (Photo: Tim Kuiper).

Six headline messages

The results of this PhD point to six key ways to maximise the value of ranger-collected data for conservation:
  1. Define clear goals for how data can be used for management. The first step is for managers and scientists to identify clear examples of how ranger-collected data could inform specific management questions (such as identifying hotspots, tracking changes in poaching over time, or tracking how poaching responds to specific management strategies).
  2. Evaluate the power of ranger-collected data to meet these goals. Results suggest that only medium to large changes in poaching can reliably be detected under current patrolling at our site. Managers must critically assess whether ranger-collected data are suitable to address their specific questions, and understand the uncertainty involved.
  3. Be aware of and overcome data bias: Rangers patrol some areas more than others, so their data can be biased and managers should be careful when using raw data directly. Results show that basic statistical methods to correct bias can greatly improve conclusions drawn from ranger-collected data. Better recording of patrol routes is key.
  4. Engage rangers: Interviews with rangers point to the need for greater feedback to be given to rangers on the value of the data they collect and how it is used. We also suggest building on rangers’ strong sense of duty by rewarding excellence in data collection. Rangers also helped improve our statistical models based on their local knowledge. Their ideas and knowledge should be engaged by ZimParks and conservation NGOs.
  5. Engage park managers: Interviews revealed that managers did not buy-in to data-based adaptive management. They did not see how systematic analyses of ranger-collected data could help them and did not feel equipped to carry out such analyses. They preferred management based on intuition and patrol-to-patrol data use. When designing adaptive management, there is a need to listen to managers’ needs and priorities and identify specific ways that data-driven management and learning can help address these.
  6. Fostering interaction between scientific and management staff: Our results point to the need for ecologists/scientists to help managers analyse and interpret patrol data to address specific questions that managers identify as important.

Background

Ranger-based monitoring: evidence for conservation action

Biodiversity loss ranked 2nd on the World Economic Forum’s 2020 list of global risks. As society responds to this challenge, it is critical that policies to safeguard nature are well-grounded in science. Biodiversity conservation still lags behind healthcare and social policy in terms of a robust evidence base to guide action. What does and does not work to stem biodiversity loss? Do ivory trade bans reduce or stimulate elephant poaching? How effective are militarized anti-poaching patrols compared to community-based responses to illegal wildlife trade? This PhD evaluates a widespread and diverse, but underutilised, source of evidence on which to base conservation decisions: ranger-collected data.

Three rangers working in the Zambezi Valley, Zimbabwe (Photo: Tim Kuiper; rangers granted consent).

Research Questions

Understanding both people and data

People are at the centre of ranger-based monitoring and management. It is crucial that rangers are motivated and that park managers actively use data to inform management. Alongside this, data reliability and power to detect poaching trends must be carefully assessed.

Four key questions

1. What affects the reliability of ranger-collected data on poaching and how can reliability be improved?
2. What affects the extent to which park managers use these data to inform their decisions?
3. What affects the meaningful engagement of rangers with their monitoring efforts?
4. How can these insights help to maximise the contribution of ranger-collected data to anti-poaching and protected area management?
A young elephant calf and its mother in the Zambezi Valley, Zimbabwe (Photo: Tim Kuiper).

Study area and methods

The Mana-Chewore World Heritage Site

The site comprises three adjacent protected areas (Mana Pools, Chewore and Sapi) in the Zambezi Valley region of Zimbabwe, covering an area of 6 678 km2. The region’s elephant population declined by an estimated 42% between 2004 and 2014, due largely to poaching. Rangers at the site collect data on poached elephant carcasses as part of the global programme for Monitoring the Illegal Killing of Elephants (MIKE), which is administered alongside 65 additional African sites by the Convention on the International Trade in Endangered Species (CITES) - see Box 2. Two field trips to Zimbabwe were conducted in 2018 and 2019, respectively.

An Interdisciplinary approach

Advanced statistical and mathematical models were used to investigate the reliability of ranger-collected data and its effectiveness for detecting changes in elephant poaching (using 17 years of patrol data from the site). To complement this, semi-structured interviews were used to understand the motivations, values, and perspectives of rangers, park managers and senior-level stakeholders.

Key result 1: Identifying poaching hotspots using statistical models and rangers' knowledge

Headline: Simple statistical methods to correct biases in the data due to uneven patrols, and greater interaction between scientists, rangers and park managers, can greatly improve conclusions drawn from ranger-collected data.
Predictions based on raw ranger detections were very different from those that accounted for patrol bias.

We modelled spatial patterns of elephant poaching in Mana-Chewore based on ranger detections of 201 poached carcasses (2000–2017). Various model scenarios with different statistical methods were used to correct patrol bias - the fact that rangers patrol some areas more than others. We actively engaged rangers and park managers to help us evaluate our results and select the most robust scenario (see Box 3). Close proximity to water was the strongest predictor of poaching, reflecting both poacher and elephant behaviour. The scenario with no correction of patrol bias produced very different results from the corrected scenario. Our results also suggest that poaching hotspots change over time, making it difficult to predict where poachers might target next.

Rangers discuss model results with PhD research Tim Kuiper. Ranger insights led to critical reflection on model scenarios and more robust results (rangers consented to the use of this photo).

Key recommendations:

  • Be aware of biases in ranger-collected data and account for them (using raw patrol data directly can lead to poor conclusions). Consistent recording of patrol effort and routes will make this easier.
  • Ecologists and managers must work together to analyse patterns in ranger-collected data and identify management responses.

Key result 2: What do rangers think about data-collection?

Headline: Interviews with 26 individual rangers revealed that they had a strong sense of duty and deference to authority and saw data collection as a way to demonstrate a job well done to their supervisors. Rangers did not, however, always understand how their data were used. Greater feedback to them will likely motivate more engaged data collection.
Quotes from interviews with rangers.

Three elements of the occupational culture of rangers—a strong sense of duty, deference to authority and knowing their defined responsibilities within the organizational hierarchy—shaped their engagement with monitoring. Rangers saw biodiversity data collection as a routine duty that helped guide patrol strategy. Reporting data was a primary way that rangers demonstrated a job well done to their supervisors. Rangers did not, however, engage actively with data management and use. Ranger sentiment was evenly divided between those who said feedback on how the data they collected were used would motivate more engaged data collection, and those who said they would continue collecting data regardless, out of duty. Separation from family was the biggest challenge that rangers described facing, while experiencing and protecting wildlife was the part of their job they enjoyed most. More broadly, our work demonstrates the value of meaningfully engaging rangers in conceptualising and tackling conservation problems, rather than seeing them as passive nodes through which conservation strategies are enacted.

Key recommendations:

  • Foster greater awareness among rangers of the value of their data and how it ends up being used (managers and scientists should give rangers more feedback).
  • Build on rangers’ strong sense of duty and pride by recognizing and rewarding individuals for excellence in data collection.
  • Address key challenges around ranger well-being, such as separation from family and resource shortages.

Key result 3: Do ranger patrols have the power to detect changes in elephant poaching?

Headlines: Large yearly changes in poaching are detectable even with low patrol effort, but small changes are difficult to detect even with high effort. The reliability of detecting spatial hotspots of poaching is improved with higher patrol effort and wider coverage. Overall, the reliability of patrol data depends heavily on what managers hope to use the data for, and therefore setting clear monitoring goals is crucial.

Simulating poaching and patrols

There are no independent data on true poaching levels to compare to ranger observations, so we developed mathematical models to simulate realistic poaching trends and then ‘virtual rangers’ patrolling and detecting a portion of the poached elephant carcasses (Fig. 1). We could then test how different levels of patrol effort and different patrol strategies affected the power of ranger-collected data to detect different poaching trends. Models were parameterised using empirical poaching and patrol data from Mana-Chewore to ensure they were realistic.

Can patrols reliably detect increases/decreases in poaching?

Smaller changes in poaching (such as a 25% decrease occurring over 1-3 years) were almost impossible to detect, even with high patrol effort, whereas larger changes (such as a 75% decrease) were often detectable even with low patrol effort (Fig. 2). Therefore, increasing patrol effort was most important for detecting intermediate changes in poaching (e.g., 50% decrease). When the true baseline poaching rate was low (1% of the population or 30 elephants/year) it was difficult to detect even large changes in poaching as rangers typically detect only a small portion of poaching. Notably, increases in poaching were significantly harder to detect than decreases of the same magnitude (see the full report for an explanation).

Can patrol data reliably identify poaching hotspots?

Increasing patrol effort ensured that ranger observations captured poaching hotspots (spatial patterns) more accurately, but results were less accurate when patrol coverage was constrained to areas near the main ranger stations (Fig. 3). Surprisingly, patrols that were targeted to areas where rangers had previously found carcasses did not bias spatial pattern detection as expected. We discovered this was because poaching in Mana-Chewore is quite spread out, so rangers were less likely to repeatedly target just a few hotspots.

Key recommendations:

  • First define clear monitoring goals (such as identify poaching hotspots or annual poaching changes) and then critically assess whether ranger-collected data can meet these goals.
  • Be realistic about the likelihood of patrols being able detect change (results suggest that only medium to large changes in poaching can be reliably detected).
  • Be aware of uncertainty in ranger-collected data and make decisions that are robust to it.

Key result 4: Park managers do not see the value of systematic data-based management

Headline: Managers did not buy-in to adaptive management. They were unfamiliar with the technicalities of analysing ranger-collected data and did not appreciate how such analyses could help them. They preferred management based on intuition, experience, and basic data use.
Quotes are from parka managers unless otherwise indicated.

We sought to understand manager perspectives on adaptive management (which we define here as the analysis of trends in poaching data to improve anti-poaching strategies). Nine park managers, ten senior staff of the government wildlife authority, four local NGO leaders, and three senior staff of the MIKE programme were interviewed. Overall, there was a low level of ownership and adoption of adaptive management among park managers (reasons are shown in Fig. 1 below). Managers valued ranger-collected elephant poaching data and used them to guide one patrol to the next. They did not, however, systematically analyse longer-term trends in poaching data, nor did they adjust their anti-poaching strategies in response to these trends. Managers did not see how data-based management addressed their problems and were unfamiliar with the technicalities of data analysis. They felt that management based on intuition, experience and more reactive data-use was more familiar and dependable. Finally, the priorities and needs of park managers have not been adequately considered in the external programmes promoting adaptive management in Mana-Chewore.

Key recommendations

  • Focus on manager needs: understand their priorities and specific management problems, then identify ways that data-based management can help address these.
  • Help managers with data analysis and interpretation: foster greater interaction between scientists (such as ZimParks ecologists) and park managers.
  • Raise awareness among managers of how data-based management can help them.

Next steps

From January 2021, Oxford and ZimParks are working on a newly-funded project focussed on translating the above results into policy and action in Zimbabwe. We are excited by the opportunity to unlock the immense potential of ranger-collected data to inform better park management and anti-poaching, and ultimately help safeguard the country's beautiful biodiversity.

Thank you for reading!

This PhD research was made possible by generous funding from the Commonwealth Scholarship Commission in the UK. We would also like to thank Lynne Taylor of the Tashinga Initiative and Richard Maasdorp of the Zambezi Society for their excellent guidance and advice. Special thanks to Danica and Sally Kuiper for their help during field work.

Created By
Timothy Kuiper
Appreciate

Credits:

Timothy Kuiper

NextPrevious