Successful implementation of integrated pest management strategies requires accurate prediction of pest phenology. Because plants and insects exhibit temperature-dependent development, biological calendars based on plant phenological sequences can be user-friendly alternatives to degree-day models for predicting pest phenology. The primary objectives of this research were (1) to compare the accuracy of degree-day models and calendar date predictions of pest activity to determine if increased accuracy can be attained with customized species-specific models, or if one standardized model, such as that used by The Ohio State University Growing Degree-Day and Biological Calendar Website, can accurately predict large pest complexes; and (2) analyze data from The OSU Phenology Garden Network to assess consistency from location-to-location and year-to-year of phenological sequences for use as biological calendars for predicting pest activity.
A standardized degree-day model, customized degree-day model, and averaged calendar date model were developed for 43 arthropod pest species based on five years of data. ANOVA found no difference in the relative accuracy of the models based on the deviation of their predictions from actual date of occurrence in 2002. The standardized model most accurately predicted 28 of the 45 phenological events.
To further assess user-friendly prediction tools, The OSU Phenology Garden Network was established in 2004. The network consists of 34 replicate gardens across Ohio, each containing 16 clonal cultivars of woody ornamental plants. The phenological sequences constructed by ranking the chronological order of first and full bloom of each species were highly correlated from year-to-year with only one exception. Sequences were also highly correlated from location-to-location within a given year with only 124 of 1301 comparisons non-significant. The velocity (km/day) of the phenological wave of bloom as it progressed northward across Ohio over the growing season was quantified by regressing latitude against date of first bloom for each species. The velocity varied by species, year, and phenophase, which challenges the use of calendar days for predicting pest phenology.
The relationship between latitude and cumulative degree-days required for first bloom was analyzed via regression. This relationship was not significant for the majority of phenological events in four years, although there was a trend for most slopes to be negative. For all but one significant regression, the slope describing the location of garden and cumulative degree-days required for phenological events to occur was negative. This latitudinal gradient represents a previously undocumented source of variation in degree-day models for predicting plant phenology.
It is concluded that a standardized degree-day model (January 1, 10°C) is accurate for predicting the phenology of a pest complex consisting of multiple species. Overall, customized models did not improve accuracy. In The OSU Phenology Garden Network, the sequences in which phenological events occurred were consistent from year-to-year and location-to-location. These results indicate that biological calendars developed from phenological sequences can be accurate alternatives to complex degree-day models, and validate the utility of The OSU Growing Degree-Day and Biological Calendar Website, developed from phenological and temperature data at Wooster, Ohio, as a user-friendly tool for predicting pest phenology throughout Ohio
