Fourteenth IOCV Conference, 2000—Citrus Tristeza Virus The Effect of Aphid Vector Population Composition on Local and Background Components of Citrus Tristeza Virus Spread T. R. Gottwald, G. Gibson, S. M. Garnsey, and M. Irey ABSTRACT. Composition of aphid vector populations has been shown to affect the evolution of spatial patterns of citrus tristeza virus (CTV) by affecting transmission and spread of the virus. However, the spatial processes associated with various vector populations are not well described. In this study, the spatio-temporal dynamics of CTV were examined using research plots repre- senting two diverse pathosystems: i) where the melon or cotton aphid, Aphis gossypii, was the pre- dominant species and the brown citrus aphid, Toxoptera citricida, was absent, and ii) where T. citricida was the predominant vector. Data were analyzed using a spatio-temporal stochastic model for disease spread that was fitted using Markov-Chain Monte Carlo stochastic integration methods. For the CTV/Aphis gossypii pathosystem, the model parameter likelihood values sup- ported the theory that CTV was spread through a combination of random background transmis- sion (transmission originating from outside the plot) and a local interaction (transmission from sources within the plot) that operated over short distances. Conversely, for the CTV/Toxoptera cit- ricida pathosystem, results often suggested a local short range interaction that was not restricted to nearest-neighbor interactions, and that the presence of background infection was not necessary to explain the observed spread. Recent publications have demon- pathosystem is dominated by T. cit- strated that CTV pathosystems can ricida, but other aphid species, be separated into two general cate- including A. gossypii, are still gories (8, 9, 10). These two catego- present although overshadowed in ries are characterized by the species epidemiological significance (11). composition of aphid vectors that Citrus is the primary host for T. cit- exert the major influence on spread ricida, which can form large colonies of CTV. Historically, the most com- on citrus under favorable conditions mon pathosystem in the Western (1, 12, 14). In contrast to A. gossypii, Hemisphere has been the CTV/ citrus is the sole host for T. citricida. Aphis gossypii pathosystem. The transmission efficiency of T. cit- Although other aphid species are ricida, combined with the large pop- often present and at times more ulations it establishes in numerous than A. gossypii, the commercial citrus in relationship to aphid species assumed to most other aphid species, results in influence spread of CTV in this changes in the spatial and temporal pathosystem is A. gossypii. This is dynamics of CTV once the virus is due to its greater ability to transmit established. Following the introduc- CTV as compared to other species tion of T. citricida into Argentina in (2, 12, 13, 15). Citrus is not a pri- the 1930s, the aphid subsequently mary host for A. gossypii and is gen- spread throughout South America, erally not heavily colonized by this the Caribbean, and into Florida (9, aphid. Apparently, CTV is often vec- 14). As T. citricida became a compo- tored by A. gossypii as migrants of nent of the pathosystem, CTV this aphid species move through the increase and spread were elevated. orchards from surrounding areas or In South America this resulted in crops (10, 15). CTV-related tree and crop losses (9, Historically, the pathosystem 11, 14). most prevalent throughout Asia and Increase of virus infection is the Far East has been the CTV/Tox- strongly affected by pathosystem optera citricida pathosystem. This composition. Epidemics of CTV 88 Fourteenth IOCV Conference, 2000—Citrus Tristeza Virus 89 decline-inducing isolates pro- species present in the plots. The gressed from low to high CTV inci- first group consists of plots where A. dence (0.5 to 95%) in 8 to 15 yr gossypii was the predominant vector depending upon scion cultivar, etc. species = CTV/A. gossypii pathosys- in the presence of the CTV/A. gos- tem. In this group data were col- sypii pathosystem, whereas, the lected annually over a 7-yr period CTV/T. citricida pathosystem from 1989 to 1995 in nine plots resulted in the same increase in 2 to established within large commercial 6 yr (8, 9, 10). Spatial spread of CTV plantings of the U.S. Sugar Corpora- is also affected by pathosystem. In tion in south Florida. All plots con- the presence of the CTV/A. gossypii sisted of Rhode Red Valencia orange pathosystem, spread has been dem- grafted on sour orange rootstock onstrated to be either random (8) or planted in 1987. Each plot consisted a combination of local and back- of approximately 476 trees arranged ground interaction (5). In contrast, in 14 north-south oriented rows CTV spread associated with the each with 34 trees per row in a rect- CTV/T. citricida pathosystem has angular pattern. All data were col- been demonstrated to result from lected prior to the 1996 introduction primarily local interactions within a of T. citricida into Florida. For this defined area of about four to nine group, CTV incidence was low at the trees (9). beginning of the study in all cases. Gibson has recently demon- The second group consisted of strated the use of a spatio-temporal plots where T. citricida was the pre- stochastic model for disease spread dominant vector species = CTV/T. that was fitted using Markov-Chain citricida pathosystem. In this case Monte Carlo integration methods (5, data were collected from four plots, 6, 7). In the present study, this same each established within commercial methodology was applied to com- plantations in northwest Costa pare multiple assessments through Rica. All plots consisted of approxi- time. The purpose of this study was mately 20 rows of trees, each with to examine and compare the under- 20 trees per row in a rectangular lying mechanisms that affect CTV planting pattern within larger com- spread in relationship to the two mercial plantings that ranged from aforementioned pathosystems. Spe- 1- to 5-yr-old at the beginning of the cifically, we wanted to determine the study. Plots were designated: CR1 = contribution of local transmission a pineapple sweet orange planting (i.e., acquisition of the virus from on Cleopatra mandarin, CR2 = CTV-positive individuals in a Valencia sweet orange planting on defined host population followed by Cleopatra mandarin, CR3 = a Valen- transmission to other individuals cia sweet orange planting on local within that host population) versus grapefruit, and CR4 = a pineapple the contribution of background sweet orange planting on Carrizo transmission (i.e., virus transmis- rootstock. No aphid control proce- sion from vectors originating out- dures were applied in any of the side the host population being plots. T. citricida was present in all studied) on the spatial patterns of locations when the experiments CTV. were started. For this group CTV incidence varied from low to moder- MATERIALS AND METHODS ate at the beginning of the study. Sample collection and ELISA The experimental design was processing. The two pathosystems based on separating the research differed in rates of virus increase plots into two major groups accord- and thus different sampling inter- ing to the predominant aphid vector vals were used. The CTV/A. gossypii 90 Fourteenth IOCV Conference, 2000—Citrus Tristeza Virus pathosystem plots (in which CTV individual acquires the disease due incidence progressed slowly) were to primary infection from sources sampled once per year in the spring, outside the host population, whereas the CTV/T. citricida patho- whereas a2 quantifies the manner in system plots (where CTV incidence which the infective challenge pre- increased more rapidly) were sam- sented to a susceptible individual by pled in the spring and fall of each a diseased individual in the popula- year. Samples consisted of four leaf tion decreases with the distance petioles from young, nearly fully between them. As a2 increases the expanded leaves taken from the secondary transmissions occur over periphery of each tree. Every tree shorter ranges and, so long as b is was tested independently. For the not so large that primary infections Costa Rica plots, the four petioles dominate, disease maps generated from each tree were placed in a by the model exhibit aggregation. number-coded paper envelope and 20 individual envelopes correspond- RESULTS AND DISCUSSION ing to one row of trees were placed in sealable plastic bags to which was As previously reported, the tem- added ca. 50 g of a moisture-indicat- poral increase of CTV differed dra- ing silica gel. The silica gel was matically depending on the changed as needed and the dry sam- pathosystem, i.e., CTV increase is ples were then transported to the much more rapid for the CTV/T. cit- USDA-ARS laboratory in Orlando. ricida versus the CTV/A. gossypii For the Florida plots, fresh samples pathosystem (8, 9, 10). In the were transported to the U.S. Sugar present study we will examine the Corporation research labs in spatio-temporal dynamics of each Clewiston, Florida, for immediate pathosystem individually. Using processing. The four leaf petioles of the MCMC model, there was a each sample were placed in 5 ml of remarkable similarity among the PBS-Tween buffer and pulverized likelihood contours associated with for 30 sec in a Kleco tissue pulver- each pathosystem. Therefore only a izer. Extracts were assayed for pres- representative surface is shown for ence of CTV via double sandwich each pathosystem (Fig. 1a,b). For indirect (DAS-I) ELISA as previ- the CTV/A. gossypii pathosystem, ously reported (3, 4). the highest likelihood values corre- Spatio-temporal analysis. sponded to cases where a2 was posi- Data for the CTV epidemics were tioned towards the maximum of its analyzed using the spatio- temporal range and b was nonzero, usually in stochastic model for disease spread the range [0.25, 1.0] (Fig.