Insecticide resistance

EMOD models insecticide resistance as a genome-dependent modification of insecticide effectiveness. When a vector with a resistant genotype encounters an insecticide-based intervention — a bednet, indoor residual spray, or larvicide — the insecticide's killing, blocking, and repelling effects are reduced according to resistance modifiers configured for that genotype. This allows simulations to capture the emergence and spread of resistance alleles in vector populations and their impact on intervention effectiveness.

Insecticide resistance operates through the vector genetics system described in vector-model-genetics. Resistance alleles are defined as ordinary alleles in the Genes configuration; what makes them confer resistance is the Resistances configuration on the insecticide that maps specific allele combinations to reduced effectiveness.

Please note, once you have defined Insecticides for your simulation, you will be required to specify the Insecticide_Name parameter for any intervention that has it. This is because the model needs to know which insecticide's resistance modifiers to apply when calculating the intervention's effectiveness against different vector genotypes. If you have interventions for which you do not wish to model resistance, you can create a dummy insecticide with no resistance entries and reference that insecticide in those interventions.

How resistance works

The insecticide resistance system has three components: insecticide definitions, resistance modifiers, and insecticide-based interventions. These are connected at runtime to produce genome-specific intervention effects.

Insecticide definitions in config.json name each insecticide and specify which allele combinations confer resistance and by how much.
Interventions in campaign.json reference an insecticide by name and define time-dependent effectiveness profiles (waning effects) for killing, blocking, and repelling.
At each time step, when a vector encounters an intervention, the base effectiveness from the waning effect is multiplied by the resistance modifier for that vector's genotype. A modifier of 1.0 means baseline susceptibility (no resistance); a modifier of 0.0 means complete resistance (insecticide has no effect); a modifier > 1.0 means increased susceptibility (insecticide is more effective than baseline).

For example, if a bednet has a killing effect of 0.8 and a vector's genotype has a killing modifier of 0.5, the effective killing probability for that vector is 0.8 × 0.5 = 0.4.

Resistance types

Each insecticide defines resistance modifiers for four effect types that correspond to the ways insecticide-based interventions affect vectors during the feeding cycle:

Killing, "Probability that a vector dies from contact with the insecticide. Applies to individual and node interventions that have both: Killing_Config and Insecticide_Name parameters.", "**Individual interventions**: IRSHousingModification, MultiInsecticideIRSHousingModification, IndoorIndividualEmanator, Ivermectin, ScreeningHousingModification, SimpleBednet, SimpleHousingModification, UsageDependentBednet,MultiInsecticideUsageDependentBednet; Node interventions: AnimalFeedKill, IndoorSpaceSpraying, MultiInsecticideIndoorSpaceSpraying, OutdoorNodeEmanator, OutdoorRestKill, SpaceSpraying, MultiInsecticideSpaceSpraying, SugarTrap;"
Blocking, "Probability that a vector is blocked from a blood meal. Applies to bednets blocking feeding attempts.", "**Individual interventions**: SimpleBednet, UsageDependentBednet, MultiInsecticideUsageDependentBednet; **Node interventions**: None;"
Repelling, "Probability that a vector is deterred from entering a house or approaching a host. Applies to individual and node interventions that have both: Repelling_Config and Insecticide_Name parameters.", "**Individual interventions**: IRSHousingModification, MultiInsecticideIRSHousingModification, IndoorIndividualEmanator, ScreeningHousingModification, SimpleBednet, SimpleHousingModification, SimpleIndividualRepellent, SpatialRepellentHousingModification, UsageDependentBednet, MultiInsecticideUsageDependentBednet; **Node interventions**: OutdoorNodeEmanator, SpatialRepellent;"
Larval_Killing, "Probability that larvae are killed by larvicide. Applies to larval habitat treatments.", "**Individual interventions**: None; **Node interventions**: Larvicides;"

A resistance modifier of 1.0 means the vector is fully susceptible — the insecticide works at full strength. A modifier less than 1.0 means the vector has partial resistance — the insecticide is less effective. A modifier of 0.0 means complete resistance. Modifiers greater than 1.0 are also possible, representing increased susceptibility (for example, a mutation that makes the insecticide more effective than baseline).

Configuring insecticide resistance parameters

Insecticides are defined in the Insecticides array in config.json, at the same level as Vector_Species_Params. Each insecticide has a unique Name and a Resistances array. Each entry in the Resistances array specifies a Species, the Allele_Combinations that confer resistance, and modifier values for each effect type.

Note

Parameters are case-sensitive. For Boolean parameters, set to 1 for true or 0 for false. Minimum, maximum, or default values of "NA" indicate that those values are not applicable for that parameter.

EMOD does not use true defaults; that is, if the dependency relationships indicate that a parameter is required, you must supply a value for it. However, many of the tools used to work with EMOD will use the default values provided below.

JSON format does not permit comments, but you can add "dummy" parameters to add contextual information to your files. Any keys that are not EMOD parameter names will be ignored by the model.

Parameter	Type	Min	Max	Default	Description
Name	string	NA	NA	UNINITIALIZED STRING	The name of the insecticide. This name is referenced by the Insecticide_Name parameter in campaign interventions to link an intervention's effects to the resistance modifiers defined here.
Resistances	array of json objects	NA	NA	[]	An array of resistance modifier objects that define how to modify the efficacy of the insecticide for particular vector genomes. Each object specifies a species, allele combination, and modifiers for killing, blocking, repelling, and larval killing effects. See example.
Allele_Combinations	array of string arrays	NA	NA	[]	The combination of alleles that vectors must have in order to be resistant. Each inner array represents one locus with two allele names. Use `` as a wildcard to match any allele at a position. The order of alleles within an inner array does not matter — `["a", ""]` and `["", "a"]` are treated identically. These alleles must be defined in the Genes* configuration for the associated species. Uses the same syntax as Gene_To_Trait_Modifiers allele combinations (see Vector genetics).
Blocking_Modifier	float	0	3.40282E+38	1.0	The value used to multiply the blocking effectiveness of an intervention for vectors matching the allele combination. A value of 1.0 means full susceptibility; less than 1.0 means partial resistance; 0.0 means complete resistance; greater than 1.0 means increased susceptibility.
Killing_Modifier	float	0	3.40282E+38	1.0	The value used to multiply the killing effectiveness of an intervention for vectors matching the allele combination. A value of 1.0 means full susceptibility; less than 1.0 means partial resistance; 0.0 means complete resistance; greater than 1.0 means increased susceptibility.
Larval_Killing_Modifier	float	0	3.40282E+38	1.0	The value used to multiply the larval killing effectiveness of an intervention for vectors matching the allele combination. A value of 1.0 means full susceptibility; less than 1.0 means partial resistance; 0.0 means complete resistance; greater than 1.0 means increased susceptibility.
Repelling_Modifier	float	0	3.40282E+38	1.0	The value used to multiply the repelling effectiveness of an intervention for vectors matching the allele combination. A value of 1.0 means full susceptibility; less than 1.0 means partial resistance; 0.0 means complete resistance; greater than 1.0 means increased susceptibility.
Species	string	NA	NA	UNINITIALIZED STRING	The name of the vector species for which this resistance entry applies. Must match a species name defined in Vector_Species_Params.

Example:

{
    "Insecticides": [
        {
            "Name": "pyrethroid",
            "Resistances": [
                {
                    "Species": "arabiensis",
                    "Allele_Combinations": [["kdr", "kdr"]],
                    "Killing_Modifier": 0.1,
                    "Blocking_Modifier": 0.8,
                    "Repelling_Modifier": 1.0,
                    "Larval_Killing_Modifier": 0.3
                },
                {
                    "Species": "funestus",
                    "Allele_Combinations": [["a1", "*"]],
                    "Killing_Modifier": 0.5,
                    "Blocking_Modifier": 1.0,
                    "Repelling_Modifier": 1.0,
                    "Larval_Killing_Modifier": 1.0
                }
            ]
        },
        {
            "Name": "organophosphate",
            "Resistances": []
        }
    ]
}

In this example, the insecticide "pyrethroid" has two resistance entries. "Arabiensis" vectors homozygous for the kdr allele (kdr/kdr) have strong resistance to killing of adults and larva (modifiers 0.1 and 0.3 respectively, meaning only 10% or 30% of the baseline killing effect applies), moderate resistance to blocking (modifier 0.8) and no resistance to repelling (modifier 1.0). Heterozygous "funestus" vectors carrying one a1 allele (matched by the wildcard * in the second entry) have intermediate resistance to killing (modifier 0.5). Please note, being homozygous for a1 (a1/a1) does not confer additional resistance beyond being heterozygous (a1/wt) because the second entry matches both genotypes with the same modifiers.

The insecticide "organophosphate" does not have any resistance entries, so all vectors are fully susceptible to it.

How effects are aggregated

When multiple insecticide-based interventions protect the same person — for example, a bednet and IRS in the same house — their effects are kept separate until they are resolved against a specific vector cohort. Each intervention may use a different insecticide, so each one contributes its own killing, blocking, and repelling probabilities independently. The effects are then cumulatively applied using independent probability combination:

\[P_{combined} = 1 - (1 - P_1)(1 - P_2) \cdots (1 - P_n)\]

where each \(P_i\) is the genome-specific probability from one intervention. Because each intervention's insecticide has its own resistance modifiers, the effective probability \(P_i\) depends on the vector's genotype — a vector resistant to pyrethroids but susceptible to organophosphates will experience reduced killing from a pyrethroid bednet but full killing from an organophosphate IRS. Two vectors with different genotypes encountering the same combination of interventions will therefore be affected differently.

The genome-specific probability is resolved at the final step, when the intervention effects are applied to a specific vector cohort. At that point, each intervention's probability is looked up for the vector's exact genotype: if it matches a resistant allele combination for that intervention's insecticide, the reduced modifier is used; otherwise the default (1.0, full susceptibility) applies. The per-intervention probabilities are then combined as shown above.

Effect on the vector population

Most insecticide-based interventions target only meal-seeking female vectors — those actively host-seeking or attempting to feed. This includes bednets (SimpleBednet, UsageDependentBednet), indoor residual spraying (IRSHousingModification), ivermectin (Ivermectin), and indoor emanators (IndoorIndividualEmanator).

However, some interventions affect vectors beyond the meal-seeking population, including non-meal-seeking females and males:

OutdoorNodeEmanator — can kill both male and female vectors in outdoor spaces, regardless of host-seeking status.
SugarTrap — targets both male and female vectors that feed on sugar sources, regardless of host-seeking status.
OutdoorRestKill — kills vectors resting outdoors, including males.
SpaceSpraying (and MultiInsecticideSpaceSpraying) — applies insecticide to all vectors present in the node, affecting males and females regardless of feeding status.

This distinction is important when modeling resistance dynamics: interventions that affect the entire vector population exert broader selection pressure on resistance alleles than those targeting only meal-seeking females.

Effect on the feeding cycle

Insecticide effects are applied during the vector feeding cycle described in vector-model-transmission. Each feeding attempt branches through a decision tree where the vector may be repelled, blocked, killed, or successfully feed. Insecticide resistance modifies the probabilities at each branch point.

For indoor-feeding vectors, the sequence is:

Repelling: The vector may be deterred from entering the house (IRS repelling, bednet repelling, spatial repellents). Resistant vectors are less likely to be repelled.
Blocking: If not repelled, the vector may be blocked from feeding (bednet blocking). Resistant vectors are more likely to penetrate the net.
Killing before feeding: If blocked, the vector then may be killed by insecticide (bednet killing).
Killing after feeding: The vector may die after feeding due to resting or Ivermectin.

At each step, the probability is genome-specific: a resistant vector faces lower killing and blocking probabilities than a susceptible vector encountering the same interventions.

For larvae, insecticide resistance modifies the larvicide killing probability. When a habitat is treated, the larval mortality rate is increased by the larvicide effect. Larvae with resistant genotypes experience a reduced mortality increase, proportional to their Larval_Killing_Modifier.

Modeling resistance dynamics

Insecticide resistance in EMOD depends on the genetics defined in the simulation. The mosquitoes with the resistant alleles are more likely to survive and feed than the other mosquitoes, therefore they will be more likely to lay eggs (if female) and more likely to mate (if male) spreading their genes to the next generation.

A typical workflow for modeling resistance dynamics involves:

Define resistance alleles as part of the species' Genes configuration, with initial frequencies reflecting baseline resistance prevalence.
Define fitness costs if any (optional). Sometimes the resistance alleles might have a fitness cost such as increased mortality or reduced fecundity, which creates a trade-off that prevents resistance from fixing in the population without selection pressure. To model that, you can use Gene_To_Trait_Modifiers to define fitness cost traits associated with the same genes that have insecticide resistance.
Configure insecticides with Resistances that map the resistance alleles to reduced intervention effectiveness.
Deploy interventions via campaign events. Selection pressure from the interventions favors resistant genotypes, increasing resistance allele frequency over time.
Track resistance using software-report-vector-genetics to monitor allele frequencies and genotype distributions across time and space.

For example, a simulation might define a kdr allele at 5% initial frequency with a 10% fitness cost (MORTALITY modifier of 1.1), then deploy pyrethroid-based bednets. Over time, the selection pressure from bednets increases the kdr allele frequency despite the fitness cost. Switching to a non-pyrethroid insecticide removes the selection pressure, and the fitness cost causes the kdr frequency to decline.

Configuration example

The following is a complete example showing the configuration of a resistance allele, an insecticide with resistance modifiers, a fitness cost, and a bednet intervention that uses the insecticide.

Note: When calculating resistance effects based on the alleles present in a vector, more specifically defined resistances are applied first, and each allele can contribute to a resistance effect only once. For example, if a vector has the genotype (a1, a1) and there are resistance effects defined for both (a1, *) and (a1, a1), the (a1, a1) effect will be applied because it is the more specific match. The (a1, *) effect will not be applied, since both alleles have already been accounted for.

config.json (relevant excerpts):

{
    "Vector_Species_Params": [
        {
            "Name": "gambiae",
            "Genes": [
                {
                    "Is_Gender_Gene": 0,
                    "Alleles": [
                        {"Name": "wt",  "Initial_Allele_Frequency": 0.95, "Is_Y_Chromosome": 0},
                        {"Name": "kdr", "Initial_Allele_Frequency": 0.05, "Is_Y_Chromosome": 0}
                    ]
                }
            ],
            "Gene_To_Trait_Modifiers": [
                {
                    "Allele_Combinations": [["kdr", "kdr"]],
                    "Trait_Modifiers": [{"Trait": "MORTALITY", "Modifier": 1.1}]
                }
            ]
        }
    ],
    "Insecticides": [
        {
            "Name": "pyrethroid",
            "Resistances": [
                {
                    "Species": "gambiae",
                    "Allele_Combinations": [["kdr", "kdr"]],
                    "Killing_Modifier": 0.1,
                    "Blocking_Modifier": 0.8,
                    "Repelling_Modifier": 1.0,
                    "Larval_Killing_Modifier": 0.3
                },
                {
                    "Species": "gambiae",
                    "Allele_Combinations": [["kdr", "*"]],
                    "Killing_Modifier": 0.5,
                    "Blocking_Modifier": 0.9,
                    "Repelling_Modifier": 1.0,
                    "Larval_Killing_Modifier": 0.6
                }
            ]
        }
    ]
}

campaign.json (relevant excerpt):

{
    "Events": [
        {
            "class": "CampaignEvent",
            "Start_Day": 365,
            "Event_Coordinator_Config": {
                "class": "StandardInterventionDistributionEventCoordinator",
                "Intervention_Config": {
                    "class": "SimpleBednet",
                    "Insecticide_Name": "pyrethroid",
                    "Killing_Config": {
                        "class": "WaningEffectExponential",
                        "Initial_Effect": 0.8,
                        "Decay_Time_Constant": 400
                    },
                    "Blocking_Config": {
                        "class": "WaningEffectExponential",
                        "Initial_Effect": 0.6,
                        "Decay_Time_Constant": 750
                    },
                    "Repelling_Config": {
                        "class": "WaningEffectExponential",
                        "Initial_Effect": 0.2,
                        "Decay_Time_Constant": 300
                    }
                }
            }
        }
    ]
}

With this configuration:

Wild-type vectors (wt/wt) experience full bednet effects (initially at killing = 0.8, blocking = 0.6, repelling = 0.2) and no fitness cost.
Heterozygous vectors (kdr/wt) experience reduced killing (0.8 × 0.5 = 0.4) and slightly reduced blocking (0.6 × 0.9 = 0.54), with no fitness cost.
Homozygous resistant vectors (kdr/kdr) experience strongly reduced killing (0.8 × 0.1 = 0.08) and reduced blocking (0.6 × 0.8 = 0.48), but have a 10% higher mortality rate from the fitness cost.

Output

Insecticide resistance dynamics can be tracked through the following reports:

software-report-vector-genetics — monitors allele frequencies and genotype counts over time, allowing direct observation of resistance allele spread.
software-report-inset-chart — tracks population-level metrics such as vector counts and human biting rates, which reflect the aggregate impact of resistance on intervention effectiveness.