forecasting model for insect pest

Forecasting Models for major Insect pests

1

2

Forecasting involves all the activities in ascertaining and notifying the growers of community that conditions are sufficiently favourable for certain insect pest, that application of control measures will result in economic gain or on the other hand and just as important that the amount expected is unlikely to be enough to justify the expenditure of time, energy and money for control.

Miller and O’Brien (1952)

Crop – Pests - Weather Relationship

Weather

Crop

Pests

Pre-requisites for developing a Forecast System

• The crop must be a cash crop(economic yield)

• The insect must have potential to cause damage(yield losses)

• The Insect pest should not be regular (uncertainty)

• Effective and economic control known (options to growers)

• Reliable means of communication with farmers

• Farmer should be adaptive and have purchase power

5

National Consultation on a Framework for Climate Services in Belize

Criteria for successful Insect pest forecasting system

Reliability -use of sound biological and environmental dataSimplicity - The simpler the system, the more likely it will be

applied and used by producersImportance -The insect pest is of economic importance to the crop, Usefulness -The forecasting model should be applied when the

insect can be detected reliably• Multipurpose applicability -monitoring and decision-making

tools for several diseases and pests should be available• Cost effectiveness -forecasting system should be cost affordable

relative to available insect pest management tactics.

7

Observed weather

Weather generator

Re-sampling approach

Climate change scenarios

Future hourly weather data

Pest and disease models

Future pest/disease scenarios

Experimental pest and disease data

Sensitivity studies

Schematic overview of the Forecast model for Insect and disease

calibration calibratio

n

8

Forecast Model- Types Between year models

◦ These models are developed using previous years’ data. ◦ The forecast for pests and diseases can be obtained by

substituting the current year data into a model developed upon the previous years.

Within year models◦ Sometimes, past data are not available but the pests status at

different points of time during the current crop season are available.

◦ In such situations, within years growth model can be used, provided there are 10-12 data points between time of first appearance of pests and maximum or most damaging stage.

(Amrender,2006)

Forecast Models Developed in the Past for insect pests Techniques used were essentially Statistical

(Correlation and Regression Analysis)◦T.P. Trivedi had proposed a regression model to predict the

pest attack. Model seems to work only for some years (1992-1994)

◦Correlation analysis was used by C.P. Srivastava to explore the relationship between the rainfall and pest abundance in different years. The technique is not effective as the attributes don’t follow normal

distribution

11

Model based on Regression method

Weather Related Forecasting Model

Observations-

Crop data: Phenological development, Growth, Leaf area and Variety

Insect pest : Pest population

Weather Data Required (hourly for ten or more years)PrecipitationTemperatureSunshine/cloudinessRelative humidityLeaf wetnessWind direction and speed

12National Consultation on a Framework for Climate Services in Belize

Degree-Day Models

Degree-days (DD) are used in models because they allow a

simple way of predicting development of cold-blooded

organisms (insects, mites, bacteria, fungi, plants). Degree-day models have long been used as part of decision

support systems to help growers predict spray timing or when

to begin pest scouting.1 degree-day (DD)-DD is way of measuring of Insect growth

and development in response to daily temperature

http://www.ipm.ucdavis.edu/MODELS

13

Calculating degree-days

• Degree days = (Maximum temperature + minimum temperature)/2 - Base Temperature

• Accumulated growing degree days was derived by using the formula

Where, Tmax = maximum temperature (°C) Tmin = minimum temperature (°C) Tb = base temperature (°C) (Iwata,1984)

14

n

ibT

TT

0

minmax )2

(

15

Phenology model

Deviation Model

Degree-Days Models types

Simulation model

Phenology Models

• Phenology models are driven by several weather parameters on hourly basis.

• Relationships between temperature and stage specific development rates of the insect life cycles are established in through laboratory experiments under controlled conditions.

• For validation, implemented model predictions are compared with independent field observations from several years.

16

(Stinner et al., 1974)

Two Methods to Manage Codling moth Larvae

Calendar Approach Treat 3 weeks after full

bloom

Degree Day Model Monitor adult flight with

pheromone traps Biofix = 1st consistent

catch of moths in traps Treat at 250 DD after

Biofix

17

18

19

Degree-day maps

Degree-day

calculator and

models

IPPC weather data homepage (http://pnwpest.org/wea)

Select location,forecast locationhistorical average location

Example on-line DD models:Fruit and Nut Crops:a) codling mothb) western cherry Fruit Flyc) oblique-banded leafrollerd) filbertworme) orange tortrixand 6 othersVegetable Crops:a) bertha armywormb) black cutwormc) cabbage looperd) corn earworme) sugarbeet root maggotPeppermint:5 speciesOther crops:4 species

IPPC weather data homepage (http://pnwpest.org/wea)

GDD approach

This method is based on the assumption that the pest becomes inactive below a certain temperature known as base temperature

GDD = (mean temperature – base temperature)

• Not much work on base temperature for various diseases and insect pest. Normally base temperature is taken as 50 C

• Under Indian conditions, mean temperature is seldom below 50 C• Need for work on base temperature and initial time of calculation

23

24

Limitations of Degree-Day Models

Insect response to temperature is not linear Lower Thresholds Temperature known for very few

species. Measured temperatures not the same as those

experienced by the pest.

Models developed at IASRI

• Mustard• Aphid

• Cotton• American boll worm• Pink boll worm• Spotted boll worm • Whitefly

• Groundnut• Spodoptera litura

• Onion • Thrips

• Sugarcane• Pyrilla• Early shoot borer & • Top borer

• Pigeon pea• Pod fly• Pod borer

• Rice• BPH• Gall midge

• Mango• hoppers• fruit-fly

S. No.

Parameters

Pest infestation in 10-14 standard week

High Medium Low

1 Sudden rise in the minimum temperature by >50C around 7-8 standard week

+ + + - + - -

2 Rainfall during 1-9 standard weeks

+ + - + - + -

3 Base adult moth population >15 per week during 5-7 standard week

+ - + + - - +26

Forecasting model for H. armigera

(Indian Institute of Pulses Research)

27

Forewarning model given by National research Center for Rapeseed Mustard

28

Potato aphid

Potato aphid (Myzus persicae) is an abundant potato pest and vector of potato leaf-roll virus, potato virus Y , etc.

Potato aphid population – Pantnagar (weekly models)

Data used: 1974-96 on MAXT, MINT and RH – [X1 to X3) lagged by 2 weeks

Model for December 3rd week

Y = 80.25 + 40.25 cos (2.70 X12 - 14.82) + 35.78 cos (6.81 X22 + 8.03)

Trivedi et al 1999

0

100

200

300

400

500

600

74-75

76-77

78-79

82-83

84-85

87-88

89-90

91-92

93-94

95-96

Year

Po

pu

lati

on Observed Predicted

Potato Aphid population in 3rd week of December at Pantnagar

Trivedi et al 1999

31

(Yadav and singh,2015)

Useful when only 5-6 year data available for different periods

Week-wise data not adequate for modelingCombined model considering complete data.Not used for disease forewarning but in pest forewarningAssumption : pest population in particular year at a given

point of time composed of two components.◦ Natural cycle of pest◦ Weather fluctuations

Deviation method

(Mehta et al,2001)

MangoMango fruit fly – Lucknow (weekly models)

Data used: 1993-94 to 1998-99 on MAXT, MINT and RH – [X1 to X3]

Model for natural pattern

2tt0067.0t16.01

t79.164.33Y

t = Week no.

Yt = Fruit fly population count at week t

Mehta et al.(2001)

0

50

100

150

200

250

300

April

I

II

III

IV

May

I

II

III

IV

June

I

II

III

IV

July

I

II

III

IV

Augus

t I

II

III

IV

Week

Pop

ulat

ion

Observed Expected

Mango fruit fly population prediction model based on deviation method in Lucknow

Mehta et al.(2001)

Pest simulation models

Pre –requisites for Simulation models –

• Mathematical descriptions of biological data.

• Computer programs or software to run these models.

• Application of these models in understanding population dynamics and dissemination of pest forecasts for timely pest management decisions.

(Coulson and Saunders, 1987 )13

EPIPRE

EPIPRE (EPidemics PREdiction and PREvention) is a system of supervised control of diseases and pests in wheat.The participating farmers do their own pest monitoring, simple and reliable observation and sampling techniques.Farmers send their field observations to the central team, which enters them in the data bank. Field data are updated daily by means of simplified simulation models. Expected damage and loss are calculated and used in a decision system, that leads to one of three major decisions :• treat • don't treat • make another field observationEx-Rhopalosiphum padi

37

Organization scheme for EPIPRE in relation to farmers and research organizations

38

Generic Pest Forecast System (GPFS)

◦Combination of multiple weather variables and biological processes into a single predictive model Temperature, relative humidity, leaf wetness (hourly)

◦Growth, mortality, infection

Predict timing and abundance of specific life stages Estimate damage to specific host plants based on pest

and host phenology

Hong et al,2013

GPFS Model Modules

• Development Response to temperature • Simple linear model (Tmin, Topt1, Topt2, Tmax) • Mortality factors: Heat, cold, aging, soil moisture,

food availability

39

Case Studies

Insects: • Bactrocera dorsalis

(oriental fruit fly) • Epiphyas postvittana

(light brown apple moth) • Helicoverpa armigera

(boll worm)

Hong et al,2013

Oriental Fruit Fly

Highly polyphagous and extremely invasive Development is temperature driven Available data sets of distribution records and

seasonal observations available for validation Model validation in three locations: India, USA (HI),

China Comparisons made with CLIMEX-compare locations

40

Case Study I: Bangalore, India

41(Jayanthi and Verghese, 2011)

Of

fru

it

fly

Ordinal logistic model – model for qualitative data

pest / disease outbreak can be taken even if the information on the extent of severity is not available but merely the epidemic status is accessible

models have added advantage that these could be obtained even if the detailed and exact information on pest count / disease severity is not available but only the qualitative status.

where z is a function of weather variables. Forecast / Prediction rule: If P ≥.5 more chance of occurrence of epidemic If P < .5 probability of occurrence of epidemic is minimum

42(Mehta et al. 2001; Mishra et al. 2004; Johnson et al. 1996; Agrawal, et al. 2004)

ez

EP

)exp(1

1)1(

43

Crop (Location)

Pest Important variables

Time of Workers

Damage Forecast

Cotton(south India)

Whitefly MAXT,MINT,RH I & RH II

Mid Dec. Mid Nov. Agrawal et al. (2004)

Sugarcane (Muzaffarnagar)

Pyrilla MAXT & RHM

Oct.-Nov.

May Mehta et al. (2001)

Mango (Lucknow)

Fruit fly MAXT,MINT& RH I

May-June

2nd week of March

Misra et al. (2004)

Forecasting outbreak of pest using Ordinal Logistic model

MAXT = maximum temperature, MINT =minimum temperature, RH 1= relative humidity (morning),RHII=relative humidity (evening) and RHM=mean relativehumidity

ANN provides an efficient alternative tool for forecasting.

Neural Networks don’t make any distributional assumption about the data.

It learns the patterns in the data, while statistical techniques try to do model fitting.ANNs can often correctly infer the unseen part of a population even if data contains noisy information.

This makes neural network modeling a powerful tool for exploring complex, nonlinear biological problems like pest incidence.

Uses of ANN(artificial neural network technique)

(Agrawal et al. 2004; Dewolf et al. 1997, 2000; Kumar, et al. 2010)

45(Wu et al., 2008)

Remote Sensing

• Remote sensing technologies are used to gather information about the surface of the earth from a distant platform, usually a satellite or airborne sensor

• The integration of remote sensing, GPS and GIS are valuable tools that can enable resource managers to develop maps showing the distribution of insect infestations over large areas

• Site-specific data, such as type of insect, level of damage, stage of attack, yield loss are collected from different sources, stored and managed in spatial database , either contained within the GIS or connected to the GIS from an external source.

46

Predicting the potential geographical distribution of sugarcane wooly aphid

Prediction of potential geographical distribution of sugarcane wooly aphid using its current distribution and data on range of environmental parameters.

Two approaches for the purpose: GPS (Global Positioning

System) GIS (Geographic

Information System)(Prabhakar et al., 2012)

Infrastructure to calculate risk maps

met.data

Geo.data

combine with GIS

Interpolation

Calculation of forecasting models with interpolated input parameters

Presentation of results

step1

step 2

step 3

step 4

The forecasting models database

The CIPRA (Computer Centre for Agricultural Pest Forecasting) software allows the user to visualise forecasts of insect development .

Within CIPRA, there is forecasting models for a total of 35 pests (25 insects and 10 diseases). In vegetables ( cabbage, carrot, onion, potato, tomato), fruits(apple, grape, strawberry), and cereals (wheat, barley, corn).

Each crop has its own independent computer programs file, known as DLL (Dynamic Link Library), which makes possible their integration in other specialized software.

(Bourgeois et al., 2008)

48

DEGREE-DAYS MODELS AVAILABLE IN CIPRA

Crop

• Apple• Crucifers• Potato• Tomato• Cotton

Insect

Codling moth Diamond back moth Colorado potato beetle Tomato fruit borer Spotted boll worm

• Mathematical models vary from simple degree-days approach based on air temperature to more detailed epidemiological system based on air temperature, relative humidity and duration of leaf wetness.

50

51

52

Computer Software Model for Prediction for Insect pest

• PEST-MAN is a computerized forecasting tool for apple and pear pests-Canada

• MORPH is predictive computer model for horticultural pest-UK

• SOPRA is applied as a decision support system for eight major insect pests of fruit orchards -Switzerland and southern Germany

• The SIMLEP decision support system for Colorado potato beetle (Leptinotarsa decemlineata)-Germany and Austria

53

mainstream technology of computer software for forecasting

54

55

Insect pest forecasting concept

56

57

58

success of a forecasting

The success of a forecasting system depends, among other things, on

• The commonness of epidemics (or need to intervene)• The accuracy of predictions of epidemic risk (based on

weather, for example)• The ability to deliver predictions in a timely fashion• The ability to implement a control tactic (Insecticide

application, for example)• The economic impact of using a predictive system

Uses of insect pest forecasts

Forewarning or assessment of insect pest important for crop production management

for timely plant protection measuresinformation whether the insect pest status is expected to be below or above the threshold level is enough, models based on qualitative data can be used – qualitative models

loss assessmentforewarning actual intensity is required - quantitative model

For making strategic decision-

• Prediction of the risks involved in planting a certain crop.

• Deciding about the need to apply strategic control measures (soil treatment, planting a resistant cultivar etc)

For making tactical decision-

Deciding about the need to implement insect pest management measure

Entomologist and meteorologists have often collaborated to develop insect pest forecasting or warning systems that attempt to help growers make economic decisions for managing insect.  

These types of warning systems may consist of supporting a producer’s decision making process for determining cost and benefits for applying pesticides, selecting seed or propagation materials, or whether to plant a crop in a particular area.

forecasting models in order to optimize timing of monitoring, management and control measures of insect pests

62

Limitations of forecasting Models-

Statistical models are simple in their usage and less parameter-intensive, but they are limited in the information they can provide outside the range of values for which the model is parameterized.

Crop simulation models are rather hard to use and parameterize. The need for calibration can be quite data extensive and not applicable to some developing countries.

The major limitation in use of satellite-borne(remote Sensing) data in pest forewarning is the timely availability of cloud-free data with the desired spatial and spectral resolution

Thank You