Development of a Push-Pull System in Avocado Groves in South Florida

Project Overview

Project Type: On-Farm Research
Funds awarded in 2021: $19,923.00
Projected End Date: 03/31/2023
Grant Recipient: University of Florida
Region: Southern
State: Florida
Principal Investigator:
Dr. Xavier Martini
University of Florida


  • Fruits: avocados


  • Pest Management: integrated pest management

    Proposal abstract:

    The purpose of this project is to enhance the chemical and mechanical control of ambrosia beetles by developing a “push-pull” system that combines natural repellents and visually and chemically attractive insect traps. A push-pull system combines two forces: an attractant (the pull) and a repellent (the push). An initial push-pull system was tested in avocado by our team and produced encouraging results [4]. The field experiments demonstrated that the combined used of verbenone, an aggregation repellent, and the use of ethanol lures, a volatile highly attractive to the ambrosia beetles, significantly decreased the number of ambrosia beetles landing on traps attached to avocado trees (Riviera et al. 2020). Importantly, we did not capture a single ambrosia beetle in the traps placed on the trunk of the trees exposed to the push-pull treatment. However, the protection was only effective if verbenone was applied on each tree, as the active space around the repellents was less than 1 m. This limitation can be a severe drawback to the widespread use of this method by avocado growers.

    It is well known that ambrosia beetles use a combination of odors and visual cues to locate their host. These signals include the shape of the tree, the color of the trunk, and the volatiles emitted by the tree [7,8]. Typically, ambrosia beetles seek out large mature trees emitting some form of stress related volatiles such as ethanol [9]. We conceived mock trees made of cylindrical cardboard covered with a reflective material and associated with a slow-release ethanol lure. Preliminary experiments demonstrated that the addition of a reflective color dramatically increased the attractiveness of these mock trees. We propose to improve the current push-pull system presented in Rivera et al. [4] with the use of our enhanced mock-trees. We will investigate specific questions relevant for growers to apply this system in their grove. Specifically, we will investigate 1) to what distance mock-trees should be placed to attract the most beetles and 2) the density of mock-trees and repellent per block to obtain the best and most cost-effective protection of avocado trees.

    Project objectives from proposal:

    Objective 1: Mock-trees will be placed within, on the edge, and outside the avocado grove. White sticky traps will be attached to each mock-tree to trap adult beetles that land on them. White sticky traps will be placed on live asymptomatic avocado trees within the first and second row of the grove to serve as the treatment control. Mock-trees and white sticky traps will be spaced at a distance equivalent to four avocado rows in the grove. Beetles will be collected from the traps every 14 days and shipped to the NFREC in Quincy to be counted and identified at the species level. Beetles will be identified to species based on more recent identification keys for ambrosia beetles in Florida. The mock-trees will consist of a cylindrical cardboard tube with an 8-inch diameter, covered in a reflective mulch, emitting various spectrums of light. We obtained preliminary data showing that reflective material was the most attractive to ambrosia beetles present in avocado grove as compared to white or blue color. A slow-release lure of ethanol (Evergreen Growers Supply) will be placed on each mock-tree. Our preliminary data demonstrated that the addition of ethanol lures significantly increased the attractiveness of the mock-trees.

    Treatment 1: Control sticky trap on avocado tree.

    Treatment 2: mock-trees with ethanol lure and sticky trap within the interior of the avocado grove

    Treatment 3: mock-trees with ethanol lure and sticky trap on the edge (<5 m) of the avocado grove

    Treatment 4: mock-trees with ethanol lure and sticky trap outside the avocado grove, 10 meters from the first row of trees.

    Treatment 5: mock-trees with ethanol lure and sticky trap outside the avocado grove, 20 meters from the first row of trees.

    The avocado grove will be divided in 5 blocks of equal size, and the five treatments will be replicated in a generalized randomized block design. The number of beetles captured on each treatment will be compared with ANOVA for normally distributed data or a generalized linear model using Poisson distribution if the data are Poisson distributed. In addition, a principal component analysis will be used to determine the profile of the ambrosia beetle community attracted by the mock-trees disposed at the different distance from the avocado grove.

    Objective 2 will consist of a complete push/pull system combining the mock-trees + ethanol lures (‘pull’) from objective 1 with the application of verbenone on the trunks of avocado trees as the repellent (‘push’). Data from objective 1 will be used to determine the optimal placement of the mock-trees relative to the avocado border. Push treatments will consist of repellent formulations of Verbenone SPLAT® (ISCA technology, Riverside CA) provided in caulking tubes. Typically, Verbenone SPLAT is applied directly on the trunk around the circumference of the tree at 1 – 1.5 meters above ground level. For this experiment, two concentrations of verbenone will be tested 1) four 17.5 g dollops of Verbenone SPLAT, measuring 2 cm in diameter and totaling 70 g per tree on each tree in the plot (high rate) and 2) four 17.5 g dollops of Verbenone SPLAT, measuring 2 cm in diameter and totaling 70 g per tree on one tree out of two in the plot (low rate). Pull treatments will consist of visual targets + ethanol lures (cf. objective 1). Two densities of visual targets will be tested 1) a ratio of 1 visual target for 1 avocado row and 2) a ratio of 1 visual target for 2 avocado rows. Push-pull treatments will consist of both SPLAT application on the avocado trees and the visual targets on the border of the plot. The following treatments will be tested: 1) untreated control, 2) low verbenone rate, 3) high verbenone rate, 4) low verbenone rate + low trap density, 5) high verbenone rate + low trap density, 6) low verbenone rate + high trap density, and 7) high verbenone rate + high trap density.  

    The rate of 70g of verbenone per tree is based on our previous trial demonstrating that this rate significantly decreased the number of beetles in the push-pull treatment [4]. We hypothesize that the addition of visual targets will increase the efficacy of our system and will reduce the amount of verbenone needed to obtain a significant reduction of ambrosia beetles. Plots will be a 6 x 6 square of avocado trees separated with one buffer row. The field will be organized with a split-plot design with trap treatments as the whole plot factor and the verbenone as the split-plot treatment. Plots will be disposed on the border of the grove so the visual targets can be disposed in front of each plot. White 15 x 18 cm sticky cards will be attached to three trees per plot at approximately 1.5 meters from the base of the tree. Push-pull treatments will have sticky card traps attached at 1.5 meters on trees in between the splat; traps for untreated controls will be attached in approximately the same location. Traps will be placed on the central 16 trees within each plot. There will be four replicates per treatment. Ambrosia beetles will be collected from sticky card traps biweekly and sent to Quincy for identification. Collected ambrosia beetles will be examined visually and under a microscope for the presence of select species and other ambrosia beetle species. Ambrosia beetle species in the tribe Xyleborini will be identified using the key published in Gomez et al. (2018) [10]. The number of ambrosia beetles for each species will be compared across treatments with appropriate mixed model analyses using either an ANOVA for normally distributed data or a generalized linear model using Poisson distribution if the data are Poisson distributed.

    Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.