Some current projects
Improved surveillance and detection of invading insects
Trap networks are essential components of many pest insect surveillance, delimitation, control, and eradication programs. With colleagues I developed a novel, spatially and demographically explicit model of attractant-based traps in a landscape context. We parameterized it through Mark-Release Recapture experiments in the field with three tephritid species in Hawaii, and I continue to seek ways to quantify trap attraction.
The model is avialble in a publicly available package called TrapGrid. In addition, there has been a reimplementation and extension called TGL Lambda. TrapGrid has been used by USDA-APHIS to generate population size estimates for systems approaches against Mediterranean fruit fly and inform trap density decisions in Florida, as well as for a risk analysis for cherry fruit fly in New York. Recently, TrapGrid has been used extensively to improve delimitation in fruit flies and other species including moths and beetles by USDA-APHIS. It has been adapted by CSIRO (Australia) researchers to determine pest prevalence at a farm level. A web-based implementation of the model was commissioned by the International Atomic Energy Agency (IAEA), which has also made it a central feature of an ongoing Research Coordination Project, to be used by researchers from Israel, Mauritius, Australia, South Africa, and Argentina.
Enhancing insect mass-rearing
Manipulation of the rearing environment and process for biological control agents or other released insects may enhance their fitness and capacity to attack target pests. Producing large numbers of fit insects are keys to successful biologically-based pest control and eradication programs, from SIT to classical biological control. Insects used is such release-based programs are usually from colonies in a mass-rearing facility or insectary, often conceived as “insect factories”: Production and quality control practices are adapted from manufacturing. Experiments to improve the yield or quality of insects based on critical environmental parameters like temperature often focus on testing multiple fixed temperatures and finding the one with the highest survival rate or fecundity. This temperature may then be considered optimal, and rearing conditions set accordingly.
However, biological organisms are not like manufactured goods in many ways; one is that there may not be an “optimum” temperature that produces consistently the “best” results. Rather, a diurnal oscillation in temperature might be superior, mimicking the regular patterns seen in nature. With various collaborators and in different species we have been testing the effect of oscillating temperatures on mass-reared fruit flies and biological control agents to see if these can be worth implementing in production facilities in the future.
Improving Male Annihilation Technique
Male Annihilation Technique (MAT) is a key technology for management and eradication of some Bactrocera fruit flies around the world. MAT is based on application of spots of a powerful male attractant plus an insecticide to attract and kill a large proportion of males thus reducing the growth rate of the population and even eradicating it. The prevailing notion has been that the higher the application density (spots per unit area), the higher the effectiveness against males. With colleagues I designed and led a series of field experiments that collectively indicated that the optimal application density for effective MAT against Bactrocera dorsalis fruit flies is much lower, under half, of that currently used in California and other parts of the world. The finding that a lower density of application points is more effective than a higher one is counter-intuitive and unexpected, though it can be explained by the principle of trap interference, known in the literature for decades, but never considered or tested in the context of MAT. As a result of this work, USDA-APHIS has revised their guidelines on application density to half the previous value as of January 2023, as have agencies overseas (e.g., an Australian areawide project in Malaysia). Reduction in application density will result in improved biosecurity, large cost savings for the material used and labor needed to apply over invasion areas, as well as reduced environmental consequences of the insecticide used in conjunction with the attractant in MAT. For a high-detection year in California such as 2015 the change is estimated to result in $380,000 in direct savings.
From 2024 - 2026 we extended this work to include MAT based on Cuelure with tests using Zeugodacus cucurbitae, the melon fly, in Hawaii. I am a collaborator on a similar project with trimedlure/Medfly in California. This work has increased the effectiveness of existing MAT programs and may extend the viability of the technique to other species.
Estimating the time to extirpation of invasive Medfly
Ceratitis capitata, Medfly, is a major pest of fruit crops around the world. In many areas where it is not established it is seen to recurrently invade, such as in S. California. When Medfly is found in these areas by monitoring programs intensive and costly quarantine and population elimination measures are put into place (in California the latter includes insecticide spraying, host fruit stripping, increased trapping and increased Sterile Insect Releases). One important question is how long to maintain the quarantine after Medfly is no longer detected. Currently, officials rely on traditional deterministic degree-day modeling to estimate how long, given historical temperature profiles, it should take for three generations of Medfly to pass. Depending on where the find is made, a quarantine can last 9 months or longer.
We developed an Agent Based Simulation (ABS) that allows for increased specificity, realism and uniform margins of safety when estimating quarantine lengths. In addition to the ABS, we are simulating outbreaks using wild Medflies in my lab. For this we are stepping through the exact hourly temperature, humidity and photoperiod as measured during several recent outbreaks in California. We aim to see how long the wild flies can live and how well that matches with our theoretical expectation.
Apps for coffee growers
Informatics are a critical driver of improvements in agriculture worldwide. With collaborators I was involved in the development and technology transfer of an app called “Best Beans” to help small to large growers produce the highest quality coffee. Based on results from the ARS Coffee Berry Borer (CBB) Areawide project, the app enables data collection, integration of sensor data, and decision support focusing on pest and disease control. The app includes functions to monitor CBB and Coffee Leaf Rust (CLR), trap catches, management practices and provides recommendations based on integrating these data with local weather data collected in real time.
The app was released to the public in late 2021 via the Google Play and Apple app stores, and has accumulated over 120 user accounts across around 70 individual farms. The last update was in October 2022. An offshoot app named "CoffeeMD", led by Dr. Johnson of TCCPRU, has focused on using machine learning/AI technologies to diagnose disease, pest, and nutritional issues with coffee plants based on images taken with a mobile device. This project is now in the late stages of development (as of Nov. 2026).
New approaches to quantifying insect behavior
Though a lot is known about pest fruit fly behavior, some aspects have remained stubbornly hard to measure. This includes the how and why they move over the landscape, and details on their attraction to semiochemicals. In my laboratory we are addressing both these questions using new approaches.
We have been successful in examining the time of attraction to cuelure by Bactrocera cucurbitae, the melon fly, by using a computer vision approach. We have used similar methods to examine the response of C. capitat to lures with and without insecticide, and also have run field trials of new lures to test the fine-scale behavioral response of individual flies to lures. We expect that these data will be important to bridging lure attraction to a trap capture. More recent work has focused on the oviposition behavior and on learning in braconid parasitoids that are useful biological control agents against fruit flies.
Selected past projects
Modeling coffee agroecosystems and the pest coffee berry borer
Coffee growers in Hawaii and Puerto Rico continue to be challenged by the introduction of the coffee berry borer (CBB) Hypothenemus hampei, first reported in Puerto Rico in 2007 and on Hawaii Island in 2010. The CBB spends the majority of its life cycle within the coffee berries where it is highly protected from control measures. Each attacking female may produce many progeny within a single coffee berry, increasing over 1-2 generations within a single bean before dozens of offspring emerge to attack more beans.
I served as lead scientist on a large ARS-funded areawide pest management program (“Areawide Integrated Pest Management (IPM) for Coffee Berry Borer in Hawaii and Puerto Rico”) from 2017 - 2022. This included a comprehensive monitoring system backed by a network of sensors and GIS data integrated with ground data collection on dozens of farms within four coffee growing areas on three islands. This monitoring program was the primary scaffold that enabled research, outreach/technology transfer and assessment components by 1) serving as a baseline against which to compare IPM variations in nearby farms; 2) serving to parameterize the models in development to produce projections of CBB for ecologically diverse locations; and 3) serving as a natural experiment to help us understand how environmental factors affect CBB population dynamics at a fundamental level.
The Areawide project substantially improved our understanding of CBB biology, ecology, and control in relation to variable climatic, geographic, and cultural factors, and has contributed to maintaining the economic viability of the US growers by enabling continued production of high-quality specialty coffee commanding a top price on the market. Over 5,000 people received CBB information including project research results during conferences, expos, and other public events. An economic analysis indicates that research results from this project resulted in a $130M benefit to coffee growers in Hawaii alone.
Transfer of the parasitoid Fopius arisanus to Brazil
I led the transfer, assisted with colony establishment, and participated in host specificity research on the fruit fly natural enemy Fopius arisanus in Brazil, where a member of the oriental fruit fly complex has invaded and is poised to cause serious damage. This effort is critical to the eventual release of F. arisanus in the North of Brazil for control of Bactrocera carambolae, the carambola fruit fly. In support of this accomplishment I produced and published a peer-reviewed video article documenting an ARS-developed rearing methodology for F. arisanus and was consulted by colleagues at The Brazilian Agricultural Research Corporation (EMBRAPA) on implementing this methodology in Brazil.
I led and completed experiments on host sex and its effects on F. arisanus to improve safety of parasitoid shipments. I also participated in team research with colleagues at EMBRAPA to examine the host specificity of this parasite when presented with native Brazilian hosts. As of 2023 this parasitoid wasp is established in quarantine in Brazil. It is on track to be released in the coming years.
The dance of An. gambiae in mating swarms
The reproductive behavior of male Anopheles gambiae was historically described in general terms: they form crepuscular swarms near markers of horizontal contrast, and mate recognition may be mediated by wing beat frequencies or through chemical cues. A more detailed view of this process and of differences between known subgroups (chromosomal/molecular forms) regarding male swarming behavior significantly improved our understanding of natural selection and mate specificity in the field. From early 2007 and for about the next eight years I worked to localize and track individual mosquitoes within swarms in the field using stereoscopic video together with Dr. Tovi Lehmann and Malian collaborators at MRTC. From 2009 I worked closely with the Paley Laboratory at the University of Mayland aerospace engineering department to create a semi-supervised 3D tracking system. As a result, we described the basic characteristics of mating swarms of A. gambiae and developed a second-order model of their motion.
Riparian mosquito ecology and malaria transmission in the Bancoumana region of Mali
The efficiency of interventions targeting malaria vectors might be improved by considering the population dynamics of the mosquito in a particular context. In the Bancoumana region, an area around the Niger River SW of Bamako, the dynamics of the mosquito population might be described as follows (from Baber et al 2010): "As rains diminish in November and December, larval habitats disappear and the environment becomes drier, so the population of mosquitoes in the larger inland village decreases to very few An. gambiae. At the same time, the Niger River retreats, leaving sandy pools in its river bed that sustain a population of mosquitoes in the thousands around the small fishing village. The riparian population persists throughout the dry season of December through April and is effectively contained to that village. Once the rains restart around May, movement of mosquitoes from the fishing village to the larger inland village allows the inland population to be seeded by a large number of migrants and grow quickly. That larger population can then be sustained by new local larval habitat created by the rains falling on mostly human-made ground cavities.." In some dynamical systems there may be a tipping point, which is a moment when a relatively small change might shift the system state into a different trajectory. The population of An. gambiae in and around Bancoumana might be such a dynamical system.
Chromosomal inversions and speciation in An. gambiae
Theoretical and empirical work suggests that chromosomal inversion polymorphism may lead to speciation though suppressed recombination. Factors leading to the establishment of inversion polymorphism are less clear. In Anophelines, adaptation to a peripheral ecological niche and cyclical changes in population size may establish and protect new paracentric inversions. We created a realistic, stochastic simulation model to test the likelihood of inversion polymorphism and which factors are most important to this process. Results indicate that inversion polymorphism occurs almost 8\% of the time using parameters within ranges reasonable for Anopheles gambiae and that the frequency of the inverted karyotype differs significantly between habitats. Migration and recombination rates were significant drivers of inversion polymorphism, in agreement with theoretical studies. However, our experiments suggest that they are less important than population size and the combined adaptive value of loci captured by the inversion.
Population structure movies
[from description on taylor0.biology.ucla.edu] We have a long standing interest in the genetic and population structure of An. gambiae. In this area you will find short movies we have created that allow visualization of some of the major features of this structure around Banambani, Mali. A video presentation describing this work has been published in JoVE (Journal of Visualized Experiments). You can view the presentation here.
Anopheles gambiae ecology and malaria transmission in Niono, Mali
With the growing population pressures on Africa, the continent has seen a large increase in the amount of irrigated crop land to feed its human population. Estimates of irrigated land area in Africa by 2020 are of about 15.9 million ha, a 30\% increase from 1990. This increase in irrigation may increase the frequency of vector-borne diseases such as malaria, making it an important public health question in sub-Saharan Africa. The effect of irrigation on malaria transmission is not simple, however. Our studies on the ecology of An. gambiae in Niono reveal that transmission can decrease with increasing vector density, contrary to expectation. Please refer to the publications section or papers describing the results of this work.