Five steps for growers
Choose one pathway: use Regional activity for early warning without a field sample, or Forecast my orchard to anchor the weather trajectory with a recent orchard observation.
How to use the dashboard
- Nearest representative region: choose the named weather point nearest the orchard. “Near” identifies an approximate regional coordinate, not a private orchard or orchard-specific station.
- Weather information: the grower dashboard automatically uses recent observations and the available live regional weather forecast.
- Forecast through: leave the default date unchanged for the full available live outlook. Weather forecasts normally extend approximately two weeks; the dashboard clearly reports the last available date.
- Current macadamia crop stage: choose the closest observed BBCH group. This overlays crop exposure on the insect forecast; it does not alter the population curve.
- Click Show regional outlook. First read the coloured activity alert and direction, then the 7- and 14-day change, dominant stage and crop-stage overlap.
MIFE uses a fixed internal reference population of 100 adults solely for numerical scaling across the nine scenarios. It is not an estimate per orchard, tree or hectare, and it is not an economic threshold. Because the displayed pressure trajectory is normalized, this fixed reference does not determine its shape or peak timing.
Named orchard records used for research validation are retained in the scientific dataset but are not offered as grower forecast locations.
Which spotting bugs are covered?
The dashboard presents one combined Amblypelta forecast for fruitspotting bug (FSB, A. nitida) and banana spotting bug (BSB, A. lutescens lutescens). It does not currently produce separate FSB and BSB curves. Most field validation evidence is for FSB, so species-specific performance for BSB remains less certain. Read the output as regional spotting bug population timing and build-up, and use orchard observations to confirm the species present.
What the model does
MIFE is a deterministic, daily, stage-structured population model. It follows eggs, five nymphal instars and age-structured female and male adults. Each day it combines weather with measured biological schedules to update development, survival, adult ageing and reproduction.
| Component | Implementation | Meaning |
|---|---|---|
| Immature development | Temperature-dependent stage duration; linear interpolation between measured temperatures. Four points approximate a fluctuating daily temperature curve. | Warmer or cooler days alter physiological progress through Egg and N1–N5. |
| Immature survival | Stage- and temperature-specific conditional survival interpolated from the frozen parameter tables. | Only a proportion progresses successfully through each stage. |
| Adult survival | Daily age- and temperature-specific conditional survival for females and males, with daily age classes. | Adult persistence depends on both age and temperature. |
| Fecundity | Female-age and temperature-dependent egg production. Reproduction is zero at or below 15°C and at or above 35°C in the frozen model. | Female reproductive output changes through adult life and with temperature. |
| Seasonal activation | Three preregistered smooth day-length profiles: conservative, central and permissive. | Represents uncertainty in winter reproductive suppression without forcing reproduction to start on a calendar date. |
| Initialization | Overwintering adults distributed across young, mixed and older adult-age bands. | Represents uncertainty in the physiological age of adults entering the annual run. |
Temperatures outside measured immature-development and survival ranges are clamped to the nearest observed boundary rather than extrapolated. The seasonal activation curves are uncertainty scenarios, not fitted photoperiod parameters and not evidence of diapause.
Why nine scenarios?
The dashboard combines three seasonal-activation profiles with three overwintering-adult age bands, producing nine predefined scenarios. All nine are reported together; none is selected because it fits a particular orchard observation best.
| Adult-age band | Starting ages represented | Activation profiles |
|---|---|---|
| Young | 0–30 days | Conservative, central and permissive |
| Mixed | 0–90 days | |
| Older | 60–120 days |
The dark chart line is the median of the nine scenarios. The shaded area is their full range. It is a sensitivity envelope, not a statistical confidence interval.
How to interpret outputs
The dashboard separates the immediate regional outlook from the full-season trajectory. Current refers to today's date (or the latest available day when today is outside the selected period). The 7- and 14-day outlooks report change only where forecast weather is available.
| Output | Use | Do not interpret as |
|---|---|---|
| Population activity today and direction | Today's modelled regional activity and change over the preceding seven days. | A field observation or spray alert. |
| Predicted activity in 7 and 14 days | The predicted index on each future date, followed by its increase or decrease from today's index. A change of less than 2 index points is described as stable or little change. | A guarantee of local detection or damage. |
| Forecast weather available through | The actual last weather date returned and simulated. If earlier than the requested end date, the dashboard states the difference. | Evidence that weather is available beyond the reported date. |
| Population activity index (0–100) | A unitless index of modelled nymphs plus adults; eggs are excluded. In each model scenario, 100 is the highest predicted nymph-and-adult activity during the displayed period. Lower values show activity relative to that seasonal peak. | A field count or economic threshold. |
| Modelled seasonal population peak | Range of peak dates across the nine scenarios in the full-season trajectory. | A guaranteed outbreak date or automatically the period of greatest crop risk. |
| Stage composition | Modelled proportional composition of eggs, nymphs and adults. Component-wise medians across the nine scenarios are rescaled to total exactly 100% each day. | A direct field count, bugs per tree or bugs per hectare. |
| Seasonal activity calendar | Broad, ordered windows derived from a 15-day smoothed nine-scenario median nymph-and-adult activity trajectory. | Discrete generations, observed events, life-stage boundaries, damage thresholds or treatment triggers. |
| Phenology-informed monitoring priority | Combines the selected BBCH group with regional population activity and its direction to classify routine, enhanced, high or very high monitoring. | A second population forecast, a damage probability, an economic threshold or a treatment instruction. |
| Sampling-equivalent bugs/ha | Multiplies observed bugs per sampled tree by orchard trees/ha so results are expressed on a practical common area basis. | A census of every insect in the orchard or an economic threshold. |
| Field-adjusted 7/14-day outlook | Updates the regional count-rate relationship with today’s orchard sample and carries that local adjustment along the forecast trajectory. | A guaranteed future catch, damage forecast or treatment trigger. |
| Nut-damage check | Reports FSB/BSB-damaged nuts with the number examined and an approximate binomial interval. | Evidence that the insects causing the injury are still present today. |
Activity-level guide
| Level | Index | Plain-language meaning |
|---|---|---|
| Low | 0 to <10 | Well below the modelled seasonal peak. |
| Moderate | 10 to <30 | Noticeable regional activity. |
| High | 30 to <70 | Substantial regional activity. |
| Very high | 70–100 | Approaching the modelled seasonal peak. |
These are communication bands for the relative index. They are not economic-risk, crop-damage or pesticide-treatment thresholds. Always interpret the level together with its direction, crop stage and orchard observations.
Macadamia BBCH crop-stage guide
The dashboard uses a simplified reproductive macadamia BBCH grouping to describe crop tissue relevant to fruitspotting bug risk. BBCH 00–39 is intentionally omitted because vegetative and dormant stages are not a useful attack-risk selection for this grower forecast. Select the stage that best represents the orchard. If stages differ among cultivars, blocks or parts of a tree, choose Unknown or run separate interpretations.
| Dashboard group | BBCH | Field interpretation | MIFE use |
|---|---|---|---|
| Inflorescence emergence | 50–59 | Racemes developing and becoming visible. | Begin or increase standardized monitoring. |
| Flowering | 60–69 | Flowers opening through the end of flowering. | High flower exposure; prioritize direct flower/raceme observations. |
| Fruit set/very young nuts | 70–73 | Initial fruit set and very young nuts. | Early nut exposure; confirm fresh injury and field activity. |
| Nut development | 74–79 | Nuts expanding and developing. | Later nut overlap; interpret with local crop susceptibility. |
| Maturation/harvest | 80–89 | Nut and kernel maturation or harvest period. | Population overlap does not imply maximum crop damage. |
| Post-harvest/senescence | 90–99 | End of the reproductive cycle. | Seasonal carryover simulation only; not a direct treatment signal. |
| Unknown | — | Stage not confidently assessed. | Crop overlap is not assessed. |
How crop phenology is incorporated
MIFE now uses crop phenology in a monitoring decision layer. The regional insect-population curve is first calculated from weather and spotting-bug biology. The selected orchard BBCH group is then combined with the current, 7-day and 14-day activity values to produce a clearly separated phenology-adjusted crop-exposure outlook.
The short outlook assumes the selected crop stage remains representative for up to 14 days. Growers should update the selection when phenology advances, particularly around flowering, fruit set and early nut development. The crop-exposure categories describe monitoring priority—not the probability or expected amount of crop damage.
This separation is deliberate. The project archive contains 112 orchard/block-date sampling events with a recorded median BBCH value across 2024–25 and 2025–26. Those records confirm that FSB/BSB observations occurred across several reproductive stages, but BBCH also advanced with date, temperature, season, cultivar and region. The available observations therefore cannot yet isolate a causal crop-stage multiplier for regional insect abundance without double-counting seasonal effects.
| Layer | Inputs | Output |
|---|---|---|
| Regional population model | Temperature, rainfall, development, survival, reproduction and seasonal-activation scenarios. | Nymph-and-adult activity index and life-stage composition. |
| Crop phenology layer | Grower-observed BBCH group and regional activity today, in 7 days and in 14 days. | A visibly separate short-term crop-exposure and monitoring outlook. |
| Orchard field update | Current consistently collected FSB/BSB sample and orchard tree density. | Field-adjusted sampling-equivalent bugs/ha outlook. |
Scientific boundary: The monitoring priority is a transparent decision-support rule, not a fitted probability of crop damage or an economic threshold. A future abundance modifier should be estimated only after replicated sampling separates BBCH from date, weather, cultivar, orchard, sampling effort and management, and demonstrates improved out-of-sample prediction.
Integrated monitoring-risk model
MIFE v0.2 formalises the grower decision layer as a tested, versioned rule-based model. It does not relabel the population index as “damage risk”. Instead, it reports three separate lines of evidence and then assigns a monitoring priority.
| Risk layer | Question answered | Scientific boundary |
|---|---|---|
| Regional activity | Is weather-driven nymph-plus-adult activity low, moderate, high or very high, and is it changing? | Relative to the selected region’s modelled seasonal maximum; not bugs per hectare. |
| Crop exposure | Does current modelled activity overlap the grower-observed reproductive BBCH stage? | Monitoring interpretation, not probability or expected amount of damage. |
| Orchard evidence | Were live FSB/BSB or nut injury recorded in the current orchard sample? | A non-detection is not proof of absence; nut injury may pre-date today’s population. |
Current live-bug detection can increase monitoring priority because it confirms local presence. A zero sample does not cancel a high regional signal. Recorded nut injury is displayed as supporting crop evidence but does not rescale the current population forecast.
Model status: this is a transparent monitoring-risk framework supported by biological evidence and preliminary field validation. It is not a fitted economic-loss model, pesticide threshold or treatment recommendation. Converting it to a calibrated probability of crop damage requires linked, standardised insect, crop-stage and damage observations across additional orchards and seasons.
Field-adjusted sampling outlook
Combine FSB and BSB because field records are not always identified to species. Enter the number observed, the number of trees sampled and, where known, the number of trees with at least one detection. Growers may use a practical tree-based scouting method, but it should be applied consistently between visits.
Sampling-equivalent bugs/ha = (FSB/BSB observed ÷ trees sampled) × trees/ha
The field-adjusted result uses a preliminary Northern NSW count-rate calibration developed from research drop-sheet observations. The model is anchored to the activity index on the entered sampling date, then carries that orchard adjustment to today and the 7- and 14-day local-weather trajectory. The regional prior uses the equivalent of one four-tree research sample as an internal calibration weight; four trees are not a required grower sampling unit. A small zero sample does not force every future estimate to zero, while larger field samples have more influence.
Orchard trajectory graph: the submitted observation is shown as a point. The solid line uses available daily local gridded weather and is the operational forecast. Beyond that weather horizon, a dashed line and shaded range show a seasonal planning projection generated from five historical weather seasons and three predefined biological activation scenarios. The boundary is labelled on the graph. The seasonal section is a projection—not a daily weather forecast—and should be re-anchored whenever a new standardised orchard sample is collected.
Localisation boundary: orchard coordinates change the recent and forecast gridded weather and daylength used by the simulation. The nearest supported region still supplies the biological baseline and stored historical context. The orchard location and observation are used for the current calculation and are not presented as a validated orchard census.
Example: three FSB/BSB observed across 30 trees is 0.10 bugs per sampled tree. At 300 trees/ha, this is 30 sampling-equivalent bugs/ha. The field-adjusted future values may then rise or fall according to the regional weather trajectory.
These bugs/ha values are sampling-equivalent estimates, not a whole-orchard insect census. Sampling methods differ in detection efficiency, so comparisons are strongest when the same method and effort are used repeatedly. The calibration is based on sparse, overdispersed research observations and is not an economic or treatment threshold. A recent targeted spray is flagged because it can suppress the observed count; no unvalidated efficacy correction is applied.
Nut-damage check
Enter FSB/BSB-damaged nuts and total nuts examined. The dashboard reports the numerator, denominator, percentage and an approximate 95% Wilson interval. If sampled area is entered, it also reports damaged nuts/m². Damage is retained as a separate crop-impact observation because injury may have accumulated before the sampling date.
Seasonal activity calendar
The calendar translates the regional mobile-population trajectory into a small number of grower-readable activity periods: low-activity carryover, early seasonal increase, population build-up, seasonal maximum, seasonal decline, and, where reached within the displayed period, a return to low-activity carryover. It highlights the period containing today and provides a monitoring interpretation.
The classification uses a 15-day moving average of the nine-scenario median nymph-and-adult activity index to prevent short model fluctuations from creating numerous artificial transitions. The ordered periods are anchored to the principal seasonal maximum. Index crossings of 10, 30 and 70 define the broad communication bands before the maximum; after the maximum, the trajectory is described as declining until it returns below 10. These cut-offs are communication rules, not empirically estimated biological thresholds.
Life-stage composition is intentionally kept separate in the egg, nymph and adult graph. The calendar may mention today’s most common modelled stage for context, but a small change in stage proportions does not create a new activity period.
Scientific boundary: FSB/BSB cohorts overlap and weather can alter the timing and prominence of successive cohorts. The approximate windows are therefore a transparent summary of the model trajectory, not evidence of one generation, discrete generation or life-stage boundaries, observed events, crop-damage periods or treatment thresholds. Their timing should be checked against future field observations and refined if systematic disagreement is found.
Activity alert and live regional comparison
The coloured alert translates the selected region’s within-season index into Low, Moderate, High or Very high activity and reports whether the 14-day outlook is increasing, stable or decreasing. It is a population-activity alert, not a damage-risk category or pesticide trigger.
The nymph and adult comparison graphs use live weather at each distinct representative regional coordinate. Every location uses the same starting population and central biological scenario. Unlike the main graph, the comparison uses one shared maximum across both life stages and all displayed regions; it therefore retains modelled differences among regions and between nymphs and adults. Values remain a relative model comparison and are not bugs/ha.
Weather and location guidance
- Live regional weather merges recent and forecast daily minimum temperature, maximum temperature and precipitation from Open-Meteo at representative regional coordinates.
- Stored research records remain internal for reproducibility and model validation; they are not presented as a grower-selectable forecast source.
- The chart always labels its first and final displayed dates. A dashed line marks today and pale blue shading identifies the future-weather portion.
- The actual available end date is reported separately from the date requested by the user.
- The live coordinate is a regional environmental proxy, not an inferred private orchard location.
- Weather data alone do not represent immigration, insecticide effects, host movement, orchard-edge effects or local microclimate.
- Queensland regional application should be treated as exploratory where local species composition or ecology differs from the Northern NSW evidence base.
Validation status and limitations
MIFE v0.1 was initially compared with 21 field events: seven each from Malua, Knockrow and Dorey. The expanded evidence review now contains 156 de-identified orchard/block-date events spanning 2023–24 to 2025–26 across Northern NSW, southeast Queensland and central Queensland. The original 21 events are embedded in that archive and are not counted again as new evidence.
The 18 Northern NSW events from 2025–26 were reserved as an independent out-of-time test. Five events contained confirmed FSB/BSB, totalling seven confirmed individuals. Across the frozen nine-scenario ensemble, preliminary pooled discrimination between confirmed detection and non-detection was AUC 0.731. This is encouraging early evidence that the regional index can distinguish periods with and without confirmed detections, although the small number of positive events, management effects and the February end to sampling create uncertainty.
- The model forecasts weather-driven population timing, relative build-up and life-stage composition at regional scale.
- Crop phenology informs monitoring priority but does not currently multiply the regional population curve.
- Most field validation evidence is for FSB; BSB-specific performance requires further testing.
- The closed population engine has no active immigration/emigration term.
- No economic threshold or pesticide trigger is encoded.
- March–May standardized sampling is the priority for testing the modelled autumn peak.
No thermal-development, survival, fecundity, seasonal-activation or other frozen biological parameter was changed in response to either validation exercise.
Decision guidelines
- Confirm field presence and fresh damage; do not act on the forecast alone.
- Consider crop stage and tissue susceptibility separately from population pressure.
- Review recent insecticide applications, weather, orchard edges/hotspots and nearby alternative hosts.
- Use registered products strictly according to the current label or permit and seek qualified agronomic advice.
- Consider pollinators, parasitoids and predators before disruptive actions.
- Record observations so future versions can be independently validated.
Selected evidence
- NSW DPI: Fruit spotting bug in macadamia
- Ellis et al. (2023): Biology and ecology of insect pests in macadamia
- Australian Macadamia Society: Growing and processing macadamias
- Open-Meteo weather provider
- Roppolo et al. (2024): Macadamia phenological stages according to the BBCH scale
Manual version 0.4 · Prepared for the MIFE grower pilot · 20 August 2026.
Funding and acknowledgements
Funding: This project has been funded by Hort Innovation, using the Hort Innovation macadamia research and development levy and contributions from the Australian Government.
Field scouting: Christopher Themsen, Chris Fuller and Jarrah Coates are gratefully acknowledged for their field scouting contributions.
Project partner: Cervantes Agritech is gratefully acknowledged as a project partner.