Developing a Management Plan for Reducing Thrips-induced Damage on Timothy Hay

Final Report for GW06-030

Project Type: Graduate Student
Funds awarded in 2006: $10,000.00
Projected End Date: 12/31/2008
Grant Recipient: University of California, Davis
Region: Western
State: California
Graduate Student:
Principal Investigator:
Larry Godfrey
University of California, Davis
Principal Investigator:
Daniel Marcum
University of California
Expand All

Project Information


Timothy is one of the most important cool season grasses grown for hay. Thrips have recently been implicated in reducing hay quality. Producers have management options primarily limited to chemicals. Integrated pest management (IPM) issues such as sampling, basic ecology, economic threshold levels had not previously been studied in timothy. Sampling protocols and population dynamics were evaluated and a preliminary damage threshold was established in cooperation a local farm advisor and producers to establish an IPM program in California. This research, in combination with ongoing efforts, will improve the economic and environmental benefits of growing timothy in a sustainable system.


Timothy (Phleum pratense L.) is a cool-season grass grown as a forage crop in the United States. Most timothy grass in California is grown perennially for hay production in the Intermountain areas above 900 m elevation. Because it is a high-value forage, it is grown with agrichemical inputs and it must be irrigated during the dry summer in California. In 2007, California retail prices for alfalfa hay fluctuated between $160 to $225 per ton, while prices for timothy remained consistent at $300 to $350 per ton (USDA AMS 2007). This hay has stable export markets, with a high demand in Japan, and stable domestic markets, with demands in natural beef (fed only grains and grasses) and horse (race and hobby) production.
Timothy hay is largely purchased on aesthetic appearance. Visual appearance is considered the most important attribute for producers of timothy hay, followed by hay price (Curtis et al. 2007). “Brown leaf” is a condition that refers to dead leaves, usually in the lower canopy of timothy stands. These brown leaves are very obvious in a bale of hay, when compared to the rest of the green foliage, and cause a significant loss in marketability for the producer. Thrips have recently been implicated by producers as a major factor causing brown leaf and reducing yield in timothy hay and Anaphothrips obscurus Müller is the main thrips species found in California timothy (unpublished data). Other factors that may interact to cause brown leaf include mites (Tetranichidae and Eriophyidae), nutrient deficiency, especially nitrogen and potassium, seeding rates, plant senescence and disease.
The grass thrips, Anaphothrips obscurus Müller, was first documented infesting fescue range grass in California by Bailey (1948). The resulting injury was referred to as “silver-top” (silvertop). In timothy, silvertop refers to A. obscusus-caused damage that occurs in the growing points of the plant, which can include dead or abnormal inflorescences and white patches on the leaves (Hinds 1900, Kamm 1971). In addition to silvertop, undesirable frass is left by the thrips on the leaves. Silvertop damage was insignificant in California, until 1999, following a mild winter. It has also been hypothesized that stress induced by thrips could be interacting with other factors to effect brown leaf in timothy.
A. obscurus has a winged (macropterous) and non-winged (brachypterous) phenotype that has been studied in timothy and colonial bentgrass (Agrostis tennis Sibth. ‘Exeter’) and in both systems, macropterous phenotypes were more prevalent in the summer, when nights are shorter in the Northern Hemisphere (Köppä 1970, Kamm 1972). Kamm (1972) determined that photoperiod was one cue for phenotypic change in wing form, but other factors most likely influenced the dimorphism. Very limited pest management studies have been conducted in timothy. Hence, most basic arthropod ecology, including population dynamics, diversity, life histories, and biological control, is unknown. The need for such knowledge has increased with the demand for timothy hay, and the subsequent increase in acreage to meet this demand. For example, the number of timothy acres that were harvested in California increased by 46% from 1992 to 2002 (NASS 1997, 2002).
Economic thresholds are defined as the pest density at which the cost of taking a pest control action equals that of the value of the crop loss. Establishment of economic thresholds is an important part of IPM, with over 100 economic thresholds published for arthropod pests in 43 commodities, as of 1993 (Peterson 1996). They provide a standard by which economically feasible treatment decisions can be made based on scientific data. Economic thresholds depend heavily on monitoring for both their establishment and implementation. Monitoring insect pests requires routine sampling. Because pests may be difficult to identify, spatially disparate, or cryptic, sampling may be time consuming, expensive and physically challenging (Norris et al. 2003). In addition, sampling methods measure pest density in different ways. Absolute sampling methods describe methods in which the target organism is measured by unit-area, whereas relative methods do not (Pedigo 1994). Hence, a realistic sampling regime must be established that is economically, practically feasible, and correlated to realistic pest densities in the field.
There are no economic thresholds to indicate when thrips should be controlled. Although cultural controls such as field burning may control thrips (Hewitt 1914), it may decrease plant vigor if the timothy is not dormant when burned (Wasser 1982). Hence, the only options for managing arthropod pests in California timothy are the insecticides methidathion, cyfluthrin, and malathion. Long-term availability of methidathion is uncertain because it is registered under a special local needs label; moreover, both methidathion and malathion are organophosphate materials which are under regulatory scrutiny. Furthermore, in California, timothy treated with methidathion may only serve as horse feed. Because this thrips is multivoltine, with a short developmental time of 12-30 days (Hewitt 1914, Köppä 1970), malathion, which has a short residual period, is commonly applied 2 to 3 times more often than methidathion. This application frequency is economically and environmentally injurious and may lead to insecticide resistance.
A pyrethroid was registered for use on grasses in California in September 2006 for armyworm control. This chemical was mainly used by growers in 2007 as another option for thrips management, but it is restricted to one application per year. Unfortunately, pyrethroid insecticides can flare Tetranychid mite populations (Reisig and Godfrey 2006, Godfrey unpublished data), but without sustainable control options, producers are forced to accept these unwanted consequences.

Project Objectives:

Developing a management program to reduce brown leaf fits the goals of the Western SARE program. Our objectives included the development of sampling protocols for diagnostic evaluation of A. obscurus populations in timothy, correlating thrips numbers with damage, to establish an economic threshold, and studying some basic ecology (i.e. population dynamics) of this pest. We wanted to develop a sampling program that addressed Pedigo’s (1994) suggestions for choice of sampling technique and timing for thrips in timothy in California, USA. We restricted the sampling universe to timothy fields and did not directly address issues involving sample units, such as sample size, number, or collection pattern. In addition, we wanted to evaluate cost and precision in regards to sampling technique and timing. Finally, we wanted to study the impact of nutrient applications and plant density in regards to brown leaf, but we could not find a suitable location for this experiment. We hoped that studying thrips and correlating their levels to damage, that we could reveal one of the causes of brown leaf, enhancing grower knowledge. This will lead to more precise pest management methods and sustainable growing practices for timothy, increasing profits for growers, while protecting the environment.


Materials and methods:

Population dynamics. Three untreated timothy fields (Untreated Fields 1-3), with histories of thrips infestations, were donated by local producers in Shasta and Lassen counties, CA for these studies. Tiller (all the shoots extending from one vegetative node) samples were collected weekly during the growing season. Fifty randomly selected tillers were brought back to Davis, CA, and were washed for recovery of thrips (Leigh et. al 1984). In addition, fifty tillers were brought to Davis, CA where arthropod numbers were counted under a stereoscope and recorded. Furthermore, because thrips are attracted to sticky cards, they are commonly used to monitor thrips populations. Four individual blue sticky cards (10 cm x 18 cm) were placed in equal heights in the untreated fields and were replaced weekly, when the tillers were collected. Thrips numbers on the cards were compared to the numbers from the tiller samples.
Threshold trial. Small plots were established on 10 February 2006 and 4 July 2006, 6 April and 13 July 2007 in an untreated timothy field in the Fall River Valley, CA, near Fall River Mills (Field 1). This field was planted in 2000 and was the variety ‘Timfor.’ Plots were initially established on 6 April 2007 in a separate untreated timothy field in the Fall River Valley, CA, near Glenburn (Field 2); field age and variety were unknown. Plots were 12.2 x 12.2 m in 2006, with 3 replicates per treatment, and 6.1 x 6.1 m in 2007, with 6 replicates per treatment in a randomized complete block design. Treatments consisted of a malathion (Malathion 8 Aquamul at 917.8 g [AI]/ha; Agrium, Inc. Calgary, AB) application, a spinosad (Success at 28.7 g [AI]/ha; Dow Agrosciences Indianapolis, IN) application, and a cyfluthrin (Baythroid 2 at 5.4 g [AI]/ha; Bayer CropScience, Research Triangle Park, NC) application in 2006. Treatments were the same in 2007, but beta-cyfluthrin (Baythroid XL at 2.7 g [AI]/ha; Bayer CropScience) was substituted for cyfluthrin and an organic formulation of spinosad (Entrust at 18.4 g [AI]/ha; Dow AgroSciences) was used. Furthermore, a methidathion (Supracide 2E at 734.2 g [AI]/ha; Syngenta Crop Protection, Inc., Basel, Switzerland) treatment was added. All chemicals were combined with a silicone surfactant (Sylgard 309 at 0.25% total spray volume; Wilbur-Ellis Company, Fresno, CA) in 2007. These treatments were applied on 3, 10, 19 and 24 May 2006 and 26 April, 3, 10, and 17 May 2007, in the first cutting. In the second cutting, treatments were created with an application on 20 and 27 July, 3 and 10 August, 2006 and 19 and 26 July, 2 and 10 August 2007. Treatment dates were chosen to represent a range within the season when growers were most likely to treat for thrips.
Arthropods were sampled by weekly collecting 10 tillers per plot. Tillers were sampled by collecting at approximately equal distances from one another within the plots, but at least 1 m from the plot edge. In addition, tillers were not selected by sight, to avoid sample bias. All tillers were collected by carefully grasping the corm and gently pulling the tiller from the ground to avoid dislodging organisms. These tillers were stored in plastic bags and transported in a cooler to Davis, CA, where they were stored for 1-3 days in a refrigerator before processing. Tillers were washed according to the procedure described by Reisig and Godfrey (2006), but arthropods were backwashed into vials with alcohol storage and later quantification. Accumulated arthropod-days were calculated for the period after treatment (Ruppel 1983). Thrips and phytophagous mites from these experiments were not identified to species, but previous identification efforts indicate that the majority of thrips in California timothy are A. obscurus (unpublished data). Additionally, a sample of phytophagous mites was collected and indentified as Oligonychus pratensis Banks and Tetranychus spp. by Ronald Ochoa (personal communication). Predatory thrips and mites were not present in the samples in adequate numbers for analysis. Eriophyid mites were too small to be captured in our wash method, but were not observed in the field plots.
The timothy was harvested for yield analysis on 12 June, 7 September, 2006 and 21 June, 10 September, 2007 in Field 1 and on 14 June 2007 in Field 2. A sickle bar mower (Troy–Bilt Sickle Bar Mower Quantum Power 4HP; MTD Products, Cleveland, OH) was used to cut timothy in the following swath lengths: 3.56 m x 0.86 m (Field 1, 1st cutting, 2006), 4.27 m x 0.86 m (Field 1, 2nd cutting, 2006), and 2.97 m x 0.86 m (Field 2, 1st cutting, 2007). Cut hay was immediately raked onto tarps and weighed on a mechanical hanging scale (Chatillion 60 lb. x 1/10 lb. Milk Scale; AMETEK, Inc., Paoli, PA), tared to zero for tarp weight. A mechanical flail harvester (Carter Forage Harvester; Carter Manufacturing Co., Inc., Brookston, IN), equipped with an electronic scale (Weigh-tronix scale Model 615; Avery Weigh-Tronix, Fairmont, MN), was used to cut timothy in 0.91 m wide swaths in Field 1, 2007 for the second cutting. Lengths of the swaths varied among plots and the swaths length was measured after the cut was made; harvested timothy weight was recorded electronically. Lengths of the cuts ranged from approximately 8 to 11.6 m. These weights were converted into yields per hectare by extrapolation from the area harvested per plot. In addition, one sample was taken at random from each block immediately after harvest to determine the moisture of the hay. These samples were placed in paper sacks, sealed in plastic trash bags, and weighed. The paper sack was then removed from the plastic, and placed in a forced air dryer at 49° C for 7 days. The percent moisture was determined by dividing the weight of the hay after drying by the weight of the hay directly after harvest and multiplying by 100.
For damage analysis, samples were collected from a random location from within each plot on the same day that the timothy was harvested for yield analysis. We used hedge shears to cut timothy from a 0.5 m quadrat to the same height as the sickle bar mower and mechanical harvester. These samples were transported to Davis, CA, where a subsample of the hay leaves was separated based on 5 color categories: 0-20, 21-40, 41-60, 61-80, and 81-100 % brown. We assumed that hay leaves that were 100 % green was undamaged, whereas hay leaves that were 100 % brown was fully damaged. Each category was placed in a paper bag and dried at 49° C for 7 days and weighed. Weights were used to calculate average damage by calculating the total proportion of hay in each damage category for each treatment replication. This was then converted into a damage rating scale running continuously from 1-5. Damage category = 1 corresponds to hay that is 100 % green, whereas damage category = 5 corresponds to hay that is 100 % brown.
Each damage category compared to accumulated arthropod days for both thrips and phytophagous mites using multilinear regression (MLR; P=0.05; PROC REG; SAS Institute, Cary, NC, 2003). Blocks, accumulated thrips days and accumulated mite days were selected for inclusion or exclusion by Mallow’s Cp criterion (Mallows, 1966) and the lowest variance inflation factor. In one case, the regression was nonlinear, and a generalized linear mixed model approach was used for analysis (PROC GLIMMIX; SAS Institute, 2006).
Sampling study. Five sampling methods were tested in six locations in an untreated portion of Field 1. Sampling was replicated over three dates during the period before each harvest, of which there are typically two per year, one in mid to late June and one in mid September. Methods were tested during the period before the first yearly harvest on 13 April, 11 May, and 8 June 2007 and on 10 April, 8 May, and 5 June 2008. Methods were also tested before the second yearly harvest on 12 July, 9 August, and 6 September, 2006 and on 13 July, 10 August, and 7 September 2007. The methods used were direct observation, beat cup, tiller washing, sweep net and sticky card methods and they were tested at 08:00, 12:00 and 16:00 hours. All 5 methods were completed as a unit at each of the six locations until all six locations had been sampled; all locations were sampled within two hours of commencement. Tillers were pulled from a random location, within a 10 m radius of the sampling location, and tillers were not selected by sight, to avoid sample bias. All tillers were collected by carefully grasping the corm and gently pulling the tiller from the ground to avoid dislodging organisms. Ten tillers were sampled as replicates in each of the six locations using the direct observations, beat cup and tiller washing techniques.
For direct observation, the leaf blade on individual tillers was inspected, and the ligule of each leaf was disarticulated from the stem to reveal thrips that may have been concealed and so that the number could be counted. For the beat cup method, an individual tiller was rapped 10 times on the side of a 9.5 cm x 11 cm type 2 high density polyethylene cup (Berry Plastics Corporation, Evansville, IN). Ten tillers per location, date and time were placed in plastic bags and transported in a cooler to Davis, CA, where they were stored for 1-3 days in a refrigerator before processing. Tillers were washed according to the procedure described by Reisig and Godfrey (2006), but arthropods were backwashed into vials with alcohol storage and later quantification. Thrips from the washings were quantified and separated into either of two adult phenotypes- macropterous or brachypterous- or nymphs. To test more relative methods, a 25 sweep sample, consisting of 180º sweeps, was taken with a 36.8 cm sweep net at each of the six locations. Samples were stored in alcohol for later quantification. Finally yellow (10 cm x 18 cm) sticky cards were placed, at 10 cm above the canopy, in each of the six locations after the sweeping was completed. The sticky cards were removed within 4 hours of placement, and all the sticky cards were in place for ~2-4 hours.
A mixed model analysis (ANOVA; PROC MIXED; SAS Institute 2003) was used, with fixed effects specified as method, date, time, random effects specified as location, and repeated effects specified as method and time across date*location. The correlation structure was specified as unstructured among methods and compound symmetry between times. A similar mixed models analysis was used for the thrips forms obtained from the tiller washing and the fixed effect of method was replaced with thrips form. Heteroscedacity and normality were controlled using logarithmic or power transformations. Denominator degrees of freedom were calculated following the methods of Kenward-Roger (1997) and Tukey’s Honestly Significant Difference test was used for mean separation and a P value below 0.05 was assumed to represent an authentic rejection of the null hypothesis.

Research results and discussion:

Population dynamics. All thrips recovered from sticky cards were macropterous. In both 2006 and 2007, the number of thrips recovered from the sticky cards did not parallel those recovered with the hand searches and washes. Because thrips recovered from hand searches and washes were mixed phenotypes of adults (both macropterous and brachypterous) and nymphs, we posit that dispersal of macropterous individuals is not entirely dependent on thrips density, although it may be related. Why these thrips may disperse as macropterous individuals is not clear from these data.
In addition, the data from the hand searches paralleled those from the washes. We expected this, because both data sets were obtained from recovered thrips pulled from tillers in the same fields. Thrips numbers only increased throughout the season if thrips numbers passed a threshold of about one thrips per tiller. Sustained populations in fields of less than one thrips per tiller generally did not increase past this point and either decreased or remained constant.
Sampling study. In both 2006 and 2007, we found that there were similar numbers of thrips recovered among most of the more absolute sampling methods (direct observation, beat cup and tiller washes) across time of day and date sampled. Recorded thrips numbers varied slightly using more absolute sampling methods when sampling was done at different time of the day, but these differences were generally insignificant. Different numbers of thrips were found at different sampling dates using the more absolute methods, but the thrips numbers recorded were generally similar among these methods. As a result, we posit that any of these more absolute methods can be used to monitor thrips populations regardless of time of day or date that the sampling is done. Thrips recovered from the two more relative sampling methods were not correlated with one another (sweep net and sticky card) and they varied much more widely across time of day and the date that sampling was done.
These more relative methods are not useful for monitoring thrips populations within the field, because the numbers recovered vary so widely, but we gleaned other information from these data. For example, only macropterous (winged) thrips were recovered from the sticky cards. As are result, we hypothesize that the proportion of macropterous thrips is higher in the spring, summer, and leading into fall, consistent with Köppä (1970) and Kamm’s (1972) findings. Because we were able to look at the wing form of thrips recovered from the washes, we could tell when these macropterous individuals were dispersing. Although the number of macropterous individuals caught on sticky cards was higher in both the fall of 2006 and 2007 than in the summer, the number of macropterous individuals recovered from the tiller washes decreased or remained constant from summer to fall in both years. As a result, we hypothesize that the macropterous thrips disperse in the fall and overwinter as brachypterous individuals in the field. This is corroborated by observations and sampling done in the winter, when brachypterous individuals are almost exclusively found (Reisig, personal observation).
Threshold trial.
2006. Hay was ~70% moisture at each cutting. Although there were a range of thrips levels among treatments in the first cutting, accumulated thrips days were not correlated to damaged hay or to yields among treatments. Phytophagous mites were not abundant. Similarly, although there were a range of thrips levels among treatments in the second cutting, accumulated thrips days were not correlated to damaged hay or to yields among treatments. Phytophagous mites were abundant and accumulated mite days were significantly positively correlated with both increased damage and decreased yield.
Accumulated phytophagous mite days ranged from 70 to 5980 per tiller. Only 2 plots had mite days over 526 and these contributed significantly to the effect (3039 and 5980 accumulated mite days). Although these plots were outliers in the regression analysis and had a significant effect on leveraging the regression, the plots were visibly damaged by spider mites and the plants were stunted. As a result, they were included in the analysis, but because only 2 out of 12 plots had these high levels, we could not make an accurate prediction of how many mites are required to affect damage and yield.
2007. Hay was ~61% moisture in Field 1 and ~71% moisture in Field 2 during the first cutting. During the second cutting, the hay was ~68% moisture in Field 1. In Field 1 there was a range of both thrips and phytophagous mites among treatments before the first cutting. Both accumulated thrips and mite days were included in the regression analysis. There was no correlation among arthropod levels and yield. However, in contrast to 2006, accumulated thrips and mite days were negatively correlated to damaged hay among treatments. Thrips levels were lower than those in 2006 and treatments with low thrips levels generally had higher mite levels, which confounded the analysis. As a result, feeding by mites could have offset feeding by thrips to cause damage, but other factors probably caused the decreased damage in treatments with higher thrips and mite levels. In our opinion, neither thrips levels nor mite levels were high enough to have an effect on hay damage.
Additionally, in Field 1, before the second cutting, there was a range of thrips levels that was similar to those in the same field before the second cutting in 2006 and to those before the first cutting in 2007. Similar to our findings before the second cutting in 2006, accumulated thrips days were not correlated with damage levels or yield. Phytophagous mite days were not selected for inclusion in the regression analysis and were relatively low, ranging from 0 to 123 among treatments.
In Field 2, accumulated thrips numbers before the first cutting were similar to those in Field 1 before the first cutting, but the range among treatments was more pronounced. In addition, phytophagous mites were not present. Accumulated thrips levels were significantly positively correlated with damaged hay, but not yield. Populations built in all experiments, but the May treatment of all chemicals, except spinosad, kept populations below about 40 accumulated thrips days per tiller by the end of the experiment, which corresponded to 0.63 thrips per tiller per week over the experiment. Interestingly, when we monitored populations in untreated fields, sustained populations of less than 1 thrips per tiller decreased or remained the same over the growing season. This experiment is being duplicated in both fields in 2008 to establish a firmer threshold, but based on these regression data, we are setting a preliminary damage threshold at a single thrips per tiller in May before the first cutting. Furthermore, because there was a growing population of thrips in this field before May, this preliminary threshold should only be used if significant thrips populations, equivalent to 1 thrips per tiller per week, are found in early spring. Finally, in the future we will obtain economic data to correlate damage with economic loss. This will allow us to set an economic threshold level for management of thrips.

Participation Summary

Research Outcomes

No research outcomes

Education and Outreach

Participation Summary:

Education and outreach methods and analyses:

These data were used, in addition to others, for publications and outreach:
Reisig, D.D. In progress. Establishing an IPM program for thrips in California-grown
timothy and wing dimorphism in grass thrips (Anaphothrips obscurus Müller).
Ph.D. Thesis.

Reisig, D.D., L.D. Godfrey, and D.B. Marcum. 2007. Refinements in IPM programs
for arthropod pests of California cool-season hay crops, 9-12 December 2007, San Diego, CA. 55th Annual Meeting of the Entomological Society of America.

Reisig, D.D., L.D. Godfrey, and D.B. Marcum. 2007. Evaluation of sampling methods
for thrips in timothy and their bearing on population dynamics, 25-28 March 2007, Portland, OR. 91st Annual Meeting of the Pacific Branch, Entomological Society of America.

Reisig, D.D., L.D. Godfrey, and D.B. Marcum. 2006. Relationship of thrips population
levels to yield and damage in California-grown timothy, 10-13 December 2006, Indianapolis, IN. 54th Annual Meeting of the Entomological Society of America.

Reisig, D.D., L.D. Godfrey, and D.B. Marcum. 2008. IPM of thrips in timothy. 28
August 2008. Intermountain Ag Hay Meeting. McArthur, CA.

Reisig, D.D., L.D. Godfrey, and D.B. Marcum. 2007. Chemical control of thrips and its
relationship to yield and damage in timothy. 23 January 2007. Winter Ag Meeting. McArthur, CA.

Reisig, D.D., L.D. Godfrey, and D.B. Marcum. 2006. IPM of thrips and Tetranychid
mites in CA timothy. 24 October 2006. Diamond Valley Hay Grower Meeting. Eureka, NV.

In addition to these outreach efforts, I will present my research at the 2008 California Alfalfa and Forage Symposium, which is attended by producers and researchers. I will also present my research at the 2008 annual Entomological Society of America meeting, which is a meeting attended by entomological professionals nationwide. Moreover, I will publish many of these results in peer-reviewed journals. Finally, my research results will be disseminated through the UC IPM pest management guidelines, which provide guidelines for pest monitoring and pesticide and non-pesticide control alternatives.

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.