Wildlife Behaviour and Engineering PhD: Low-Power Sensors for Field Research

Wildlife Behaviour and Engineering PhD: Low-Power Sensors for Field Research

Quick Decision

This 2027 Exeter PhD addresses a practical field-biology problem: how to record individual animals and meaningful behaviour when cameras, thermal imagers and animal-borne microphones are too bulky, power-hungry or intrusive. The work would co-design sensing hardware and AI/data processing, benchmark candidate systems under controlled conditions, then validate them with established wild populations. It suits applicants who can contribute engineering, computer science/data science or zoology/biology skills and want to work across those fields.

Two strengths stand out: a supervisory team spanning behavioural validation, ecological sensing/AI and low-power electronics, and access to a long-running jackdaw field site. The project is not yet narrowed to a specific route. Visual, acoustic and thermal methods are examples, while the target species, behaviours, device form and field schedule remain open. International applicants can compete for full tuition within a cohort capped at 30% international students; visa, healthcare surcharge and relocation costs remain personal expenses.

Applications close at 12:00 noon on 8 December 2026. Before applying, check that the project-specific minimum degree and relevant-experience wording fits your profile, ask which technical route is most likely, and prepare a focused statement showing how you connect device constraints to a biological question. Confirm the interview schedule directly: the current DLA page gives a 2027 interview window in one section but retains a stale 2026 date in another.

Research Problem and Doctoral Work

Why field behaviour is difficult to measure

Behavioural responses shape whether animal populations persist or decline, but technologies designed for laboratories often fail in remote or sensitive field settings. Nest cameras may need mains electricity or may be too large to place without disturbing nesting birds. Thermal imaging can support non-invasive monitoring of stress-related responses, yet thermal cameras tend to be large, expensive and power-hungry. Animal-borne microphones can capture vocal communication and responses to human disturbance, but their battery life is limited and their mass is too high for many species. These are linked constraints: increasing sensor coverage can raise the device's size, weight, power, data volume and ecological footprint.

The project frames this as a joint engineering and biology question: how can hardware and intelligent processing be designed together so devices capture only what is needed to answer a biologically meaningful question, at the right spatial and temporal resolution? The advertised examples include recognising individual animals from visual or vocal characteristics and automatically quantifying behavioural or physiological responses. This emphasis on co-design matters. A model that is accurate after recording and uploading everything may not be useful if the device cannot run long enough, store or transmit the data, survive weather, or be used without altering the behaviour being measured.

Likely doctoral workflow

The official specification describes a sequence rather than a locked protocol:

1. Define a biological question and the minimum observations needed to address it.
2. Co-design one or more sensing systems and the data-processing pipeline around limits on mass, dimensions, power, storage and robustness.
3. Quantitatively benchmark candidate systems under controlled conditions.
4. Deploy and validate the selected approach with established wild-animal study populations.

The project lists visual, acoustic and thermal monitoring as possible approaches. It does not say that every modality will be built or used, nor specify a target species, a device attachment method, a model architecture, an on-device versus post-hoc processing split, or a named behavioural endpoint. The likely research contribution could therefore range from miniaturising a sensor and testing field reliability to developing an individual-recognition or behaviour-quantification method, but the balance must be agreed with the supervisory team. The candidate should ask how system performance will be measured beyond a laboratory accuracy score: battery life, missed events, false alarms, weather resilience, animal disturbance and the quality of downstream ecological inference may all matter.

The Cornish Jackdaw Project is identified as the primary early test setting. Established identity information, nest sites, observations and automated tracking offer a rare opportunity to compare device output with known individuals and biologically observed behaviour. The project says that later use at remote locations, including Hockings' West African study sites, is possible; it is not a promise of overseas fieldwork. Any site transfer also creates a domain-shift question: species, calls, lighting, vegetation, weather and field infrastructure may differ. Evidence from camera-trap AI shows that performance can drop on new locations and species, while acoustic workflows still need suitable annotated data and field validation. A robust thesis would make those limits explicit rather than equating a successful controlled benchmark with a reliable ecological tool.

Field direction and research opportunity

Current work in wildlife monitoring increasingly combines autonomous recording, computer vision or acoustic models, and long-duration field data. Recent methods show that transfer learning and active learning can reduce manual annotation effort in passive acoustic monitoring, while still requiring attention to precision, recall and sampling. Published gibbon-call workflows demonstrate how convolutional networks can move from call detection to multi-site ecological analysis, but they do not establish that a model transfers unchanged across taxa or landscapes. Reviews of animal-borne sensors likewise show that continuous individual-level monitoring can reveal behaviour in context, while the sensors themselves must be small enough and sufficiently robust for the animal and environment.

The project's distinctive opportunity is to close the gap between an ecological question and a deployable instrument. The strongest research design would specify what the sensor must resolve, why that resolution is biologically adequate, how the device affects the animal, and what uncertainty remains after automated processing. A useful extension could compare a low-power edge pipeline against full-resolution recording and later analysis, or test a semi-automated workflow that reserves human review for ambiguous events. These are research angles suggested by the field literature, not guaranteed project deliverables.

School, Programme and Field Context

The project PDF names a potential PhD in Biological Sciences, based at the University of Exeter's Centre for Ecology & Conservation on the Penryn Campus in Cornwall. Exeter's Biological Sciences research-degree environment at Penryn includes animal behaviour and cognition, behavioural ecology, conservation, movement ecology and sensory ecology. The university states that MPhil/PhD students typically start registered for an MPhil and undergo an upgrade to PhD during the first year, usually through a written progress report and internal viva. The programme page lists research training in areas such as quantitative biology and statistics, safety, research management, communication and career development. The particular training plan, registration title, full-time registration duration and campus/residency expectations for this award should be confirmed with admissions.

Exeter reports that, in REF 2021, 41% of its Biological Sciences submission (Unit of Assessment 5) was rated 4-star and 49% 3-star; the university programme page also describes Exeter as top 20 in the UK for world-leading Biological Sciences research, based on 4-star REF 2021 work. This is a subject-level research signal, not a guarantee of individual supervision or equipment access. Penryn is an Exeter campus shared with Falmouth University; the project identifies Exeter as the university and does not advertise a joint or double degree. Co-supervision from Engineering also does not itself create a second institutional award.

Supervisors and Research Environment

Complementary supervision

• Professor Alex Thornton leads the project and contributes behavioural ecology, field experiments and biological validation. His Wild Cognition group studies cognition and behaviour in wild jackdaws, as well as social learning and conservation applications.
• Dr Kim Hockings contributes passive acoustic and visual monitoring and ecological AI. Her conservation research centres on human–wildlife coexistence and includes field networks in Guinea and Guinea-Bissau; this supports a plausible route to remote validation, but the PhD PDF describes those sites only as possible later locations.
• Professor Monica Craciun contributes electronic-system and sensor design, particularly miniaturisation and low-power sensing. Her engineering research spans 2D materials, nanoelectronics and sensing.

The Cornish Jackdaw Project has operated since 2012 and reports more than 100 nest boxes across three colonies, with over 2,500 individually recognisable, colour-ringed and PIT-tagged jackdaws. This supplies a strong biological validation context. It does not by itself establish which equipment, datasets, technician support, calibration facilities or software the student will be able to use; ask for a concrete first-year plan and access arrangements.

Recent work that informs this project

Thornton and colleagues' 2025 study, Wild jackdaws recognise the contact calls of their mate, used playback trials in free-living, individually known birds and found that females responded faster to their partner's contact calls than to calls from a neighbour or an unfamiliar male. The work shows why vocal identity can be biologically meaningful and demonstrates an existing study population suitable for validation. It tested a defined behavioural response in jackdaws; it did not build a general-purpose vocal-recognition device.

In a 2023 Nature Communications study, Thornton and collaborators combined RFID-tagged wild jackdaws with an automated social-foraging task and social-network analysis. Birds adjusted associations to gain better rewards while preserving some valuable long-term relationships. This illustrates that individual recognition is not an endpoint by itself: the ecological value comes from linking identity to decisions and relationships over time.

Murray, Berger-Wolf, Schofield, Hockings and Hodgson's 2026 perspective on Conservation AI reviewed 611 biodiversity-sensor AI publications from 2000 to 2024. They report that only 16.9% shared new datasets and 18.0% shared code, and discuss ownership, data security and fragmentation. This is relevant if the PhD produces recordings or models from remote wildlife settings: publication, data access and local partner expectations should be considered alongside technical performance.

Craciun and collaborators' 2026 paper on spray-coated two-dimensional heterostructures presents flexible, approximately 1 g textile-integrated triboelectric sensors that harvest energy and sense humidity and physiological/environmental signals. It demonstrates adjacent expertise in lightweight, low-power sensing and device durability. The paper concerns human-health wearables, not a wildlife instrument, so it is evidence of relevant engineering capabilities rather than a ready-made solution for the advertised PhD.

Funding, Application and Living Cost

The project is attached to Exeter's September 2027 EPSRC Doctoral Landscape Award. The current scheme page offers an index-linked stipend for up to 3.5 years full time (currently £21,805 per year for 2026/27), payment of university tuition fees, and £5,000 total Research Training Support Grant (RTSG) over 3.5 years. These are scheme-level terms; the award letter should confirm the final amount, length, start date, research-cost rules and any fit with the thesis plan. International students can compete for a limited number of full studentships, up to 30% of the cohort. For successful international awardees, the scholarship covers the home-fee rate and Exeter waives the fee difference. Visa, healthcare surcharge and relocation costs are not covered.

Applications must use the link at the bottom of this project's PDF; the scheme states that other routes will not be considered for the studentship. Materials include an English personal statement naming the project, CV, full transcript, two references, and English-language evidence where required. All application documents must be in English; certified translations of academic qualifications are also required. The closing time is 12:00 noon, Tuesday 8 December 2026. Referees must send letters from institutional email addresses by 15 February 2027. The scheme overview expects interviews during a February 2027 window and outcomes in early March 2027, but a later paragraph on the same page still says February/March 2026. Treat the 2027 schedule as the apparent current cycle and confirm the actual interview date with the PGR team.

The project-specific minimum is a UK 2:1 degree or equivalent in Engineering, Computer Science/Data Science, or Zoology/Biology with relevant experience and an interest in working across disciplines. The general DLA page also describes a First or Upper Second in an appropriate science, engineering or technology area; its general lower-second exception should not be assumed to override the project-specific minimum. Confirm equivalency and how “relevant experience” applies to candidates whose first degree is outside the listed groups.

Exeter estimates basic living expenses in Cornwall at £1,300–£2,000 per month, using local spending data from April 2026. The 2026/27 Penryn accommodation schedule lists 40-week rooms between roughly £149.66 and £258.02 per week depending on type and catering. At the current stipend rate, the monthly average is about £1,817 before any personal or fieldwork costs. That falls inside Exeter's stated range, so the low-end budget depends on room choice, contract length and spending; visa, IHS, travel and initial setup costs require separate funds. The published accommodation contracts are 40 weeks, not a guarantee of year-round housing for a PhD student.

Applicant Fit and Preparation

Required conditions

• Minimum 2:1 (or equivalent) degree in Engineering, Computer Science/Data Science, or Zoology/Biology with relevant experience, plus an expressed interest in interdisciplinary work.
• Meet the EPSRC DLA's general academic and residency requirements. International funding is limited; applicants should verify fee-status eligibility rather than infer it from nationality alone.
• Submit the complete application through the project-specific route by noon on 8 December 2026 and ensure references reach the university by 15 February 2027.

Preferred experience and suggested preparation

The PDF does not name a required programming language, microcontroller, AI framework, prior field season or device-building portfolio. It does say that the project is especially suited to candidates interested in applying engineering, computer science or AI to biological problems. Before applying, prepare examples that show one side of the interface—electronics, embedded systems, power/thermal design, signal processing, computer vision, machine learning, quantitative ecology or animal behaviour—and explain how you would work with the complementary side. A small reproducible project or concise portfolio can make that bridge legible without claiming expertise in every modality.

Candidates focused only on software benchmarks may find the field constraints and biological validation demanding. Candidates focused only on field observation may need to show readiness to engage with hardware/data processing. Those are analytical fit considerations, not extra formal entry rules. For interviews, prepare a short testable question, a plan for controlled benchmarking, a biological ground-truth strategy, a way to assess disturbance, and a fallback if one sensor modality proves too heavy or power-hungry.

Career Directions and Main Risks

The training combination could support academic work in behavioural ecology, conservation technology or animal welfare; ecological monitoring roles; environmental data analysis; and engineering or AI roles in sensing and field instrumentation. UK Civil Service science and engineering careers include research, technical, engineering and AI-related work, while the Environment Agency lists scientists, engineers, data science and informatics among its occupations. These sources show adjacent career families, not a forecast of wildlife-AI vacancies, salaries or post-PhD placement rates. The project also mentions technology translation, stakeholder engagement and commercialisation opportunities, but no specific internship or industry partner is named.

Key risks to resolve are (1) which biological question and device modality anchor the thesis; (2) what constitutes acceptable device mass, power autonomy and environmental robustness for the chosen species; (3) how controlled benchmarks will connect to independent biological validation; (4) whether any fieldwork outside Cornwall is necessary and who pays for travel, permits and local collaboration; (5) how animal disturbance and ethics will be assessed; and (6) who can use resulting recordings/models and under what data-governance terms. The Association for the Study of Animal Behaviour stresses minimizing suffering and supporting conservation; non-invasive monitoring still requires ethical review and appropriate permissions.

Next Actions and Questions to Verify

Before applying

1. Ask Alex Thornton whether the team has a preferred starting question or modality and whether applicants may state an initial project direction.
2. Confirm that your degree, equivalency and relevant experience meet both the project-specific 2:1 condition and EPSRC residency rules; ask how hybrid profiles will be assessed.
3. Confirm the February 2027 interview date, reference procedure and any current application-form instructions; do not rely on the stale 2026 paragraph.
4. Build a statement around a single biological question and explain the matching sensing constraint, quantitative benchmark and validation pathway.

At interview

• Which species and behaviours are the leading candidates, and what makes the measurement biologically useful?
• Will the student design electronics, adapt existing equipment, develop models, or divide effort across these tasks? What skills can be learned during the PhD?
• What are the planned engineering metrics (mass, battery life, power, environmental tolerance, storage, latency), and what independent ground truth will be available?
• Which ethics approvals, animal-handling permissions, local partnerships and data-sharing arrangements may apply? Is any West African fieldwork optional, and what budget and time would it require?

Before accepting an offer

Request written confirmation of award length and current stipend, RTSG use, international fee-waiver coverage where relevant, thesis registration and upgrade requirements, fieldwork/travel resources, equipment access, and the supervisor team's expected division of work. Confirm that the advertised up-to-3.5-year funding aligns with the registration and field plan.

Evidence and Sources

Project, programme and application

• Official project specification
• EPSRC DLA project directory
• EPSRC DLA funding and application guidance
• Biological Sciences research-degree programme
• Exeter REF 2021 results
• Penryn Campus

Supervisors, research group and papers

• Alex Thornton: Wild Cognition Research Group
• Cornish Jackdaw Project
• Wild jackdaws recognise the contact calls of their mate
• Wild jackdaws can selectively adjust their social associations
• Kimberley Hockings: University of Exeter profile
• Cantanhez Chimpanzee Project team
• Open sharing in Conservation AI: Benefits and Barriers
• Monica Craciun: Henry Royce Institute profile
• Multifunctional, energy-autonomous textile sensors

Field methods, funding and applicant context

• Transfer and active learning for passive acoustic monitoring
• Automated gibbon-call detection from passive acoustic monitoring
• Evaluation of camera-trap AI processing platforms
• Onboard sensors and animal decision-making
• Exeter and Cornwall living-cost guidance
• Penryn accommodation prices for 2026/27
• Association for the Study of Animal Behaviour: ethics
• Civil Service science and engineering careers
• Environment Agency careers