This program is aimed at solving problems related to increasing crop yields and preserving wild flora containing valuable genetic resources by introducing modern methods of monitoring, control and measures to protect plants from quarantine organisms that are limited in Kazakhstan. This work proposes the development of digital solutions based on artificial intelligence for fast and accurate identification of pests and weeds using remote sensing methods, highly sensitive test systems for detecting quarantine viral pathogens and pests. Identification of new entomophages with high activity in the fight against quarantine organisms. Testing of highly effective biological methods of protection against quarantine organisms. Determination of the mechanisms of interaction between the pest and its predator in controlled and natural conditions. Identification of heterogeneity of pest and weed populations and determination of the level of effectiveness of biological agents depending on the genetic profile of the population.

The goal of the program is science-based management of the spread and reduction of harmfulness of limitedly widespread quarantine pests that are adaptive in the conditions of Kazakhstan.

A digital technology for phytosanitary monitoring to identify quarantine pests has been developed. Digital system architectures have been created, including a computer program with a mobile application, as well as a set of neural network structures for identifying quarantine pests. Differential modules for administering the electronic system and a design for an electronic photo storage bank with an automatic sorting system have been developed. Large data sets were prepared, and the photographs were used to train a neural network focused on recognizing quarantine pests. The classification accuracy was 92.2% on the training data set and 89.1% on the validation data set. Artificial ecosystems that were as close as possible to the natural habitats of these organisms were created to conduct studies of quarantine objects, such as the brown marmorated stink bug (Halyomorpha halys), oriental codling moth (Grapholita molesta), South American tomato moth (Tuta absoluta), California scale (Quadraspidiotus perniciosus). This allowed us to study their biology and behavior in a controlled environment. For each type of quarantine object, control groups and four experimental ones were formed. A computer analysis of available genetic databases on gypsy moth was performed to determine specific gene regions for identifying the subspecies of the organism under study. Based on the analysis results, a set of primers for molecular identification and classification of the Lymantria dispar pest was developed. A total of 208 gypsy moth samples were collected from various regions of Kazakhstan, including Akmola, Almaty and East Kazakhstan regions. 47 samples of pure European race and 161 specimens with signs of Asian introgression were identified. Monitoring of the spread of quarantine objects in the territory of South-Eastern Kazakhstan, as well as border and transboundary territories, including China, Kyrgyzstan and Uzbekistan, was carried out. During the research, 43 gardens were surveyed, 66 trapping belts were installed and pheromone traps were placed, which made it possible to collect 154 samples of codling moth, of which 127 specimens were obtained using trapping belts.

A digital phytosanitary monitoring technology based on the EfficientNetB0 and YOLO neural networks was developed and tested. Training demonstrated a consistent reduction in the loss function and an increase in accuracy, confirming the effectiveness of the proposed architecture and overfitting control. The model demonstrated excellent results in classifying pest images in both laboratory and field conditions, achieving correct recognition in 74–92% of cases. The use of Grad-CAM and SHAP methods confirmed the biological interpretability of the network's solutions. Approximately 8,200 samples of quarantine and invasive organisms, including pests and weeds, were collected using pheromone traps, entomological nets, and manual selection. Geographically, the study covered the Almaty, Turkestan, Zhetysu, and Zhambyl regions, where key outbreaks were identified. Primary identification was conducted using morphological methods based on species-specific descriptors. Molecular genetic approaches based on sequencing of the COI, 16S rRNA, matK, and ITS marker loci were used for refinement and confirmation. The identification efficiency of molecular genetic methods was greater than 98%. The effectiveness of biological agents against major quarantine pests was assessed.
During the study, 1,500 specimens of brown marmorated stink bugs from various regions of southern Kazakhstan were collected and analyzed. A computer analysis of available genetic databases of the COI gene in the brown marmorated stink bug and other stink bugs common in southern Kazakhstan was conducted, and optimal gene regions were identified for the design of specific primers for detecting the brown marmorated stink bug. New primers for PCR identification of Halyomorpha halys were developed. The identification efficiency of the brown marmorated stink bug with the developed primers was greater than 98%. No false-positive results were identified as a result of cross-validation.
A genetic diversity assessment using the molecular markers SSR and RAPD for weeds and the markers Cytb and COI-COII for pests revealed that all studied species—common ragweed, common dodder, oriental codling moth, and California scale—are characterized by a high level of genetic diversity, reflecting their significant evolutionary and adaptive potential. Taken together, the obtained data confirm that both weed and pest species possess broad adaptive capabilities, high genetic plasticity, and the ability to rapidly evolve to environmental changes. Optimal herbicide application rates for the effective control of common ragweed and common dodder were established. For maximum weed control, it is recommended to apply herbicides during the budding and flowering phases of common ragweed, as well as during the active growth period before flowering of common ragweed and common dodder.