The impact of chestnut blight, caused by the fungal pathogen Cryphonectria parasitica, has diminished in Europe presumably due to a natural biological control caused by hypovirus infection [1] [2] . Hypoviruses are virus- like particles of dsRNA found in the cytoplasm of debilitated strains of C. parasitica [3] [4] . Infected isolates display a significant reduction in fungal virulence (hypovirulence) coupled with reduced asexual spore production and sexual infertility. Cloning sequence analysis has allowed for closer examination of dsRNA, which often differs among Cryphonectria hypoviruses (CHV) with respect to size and number of coding regions [5] - [7] . Four types of dsRNA viruses (CHV1, CHV2, CHV3, and CHV4) have been described and placed in the genus Hypovirus in the family Hypoviridae [6] [8] - [10] .

CHV1 hypoviruses have been released in many areas of North America but are found naturally occurring in infected C. parasitica isolates in Europe and parts of Asia [11] - [13] . CHV1-Ep713 is a CHV1-type hypovirus isolate with a 12.7 kb genome comprised of two coding regions designated ORF A and ORF B [8] . When compared to CHV1-Euro7, another European CHV1-type isolate, the level of nucleotide identity for the entire 5’ non-coding domain is 93%; the CHV1-Euro7 genome is 11 nucleotides shorter than the CHV1-Ep713 genome [14] . Despite the similarity, these two isolates have phenotypic and morphological traits that differ considerably [5] . CHV1-Ep713 is a severely debilitating hypovirus isolate in vitro and produces small cankers and few asexual spores in vivo. In contrast, CHV1-Euro7 infection has a milder impact on fungal virulence; this isolate grows rapidly in vitro and produces larger cankers in vivo with moderate asexual sporulation [14] .

CHV3-type isolates are endemic to North America. Biological control with this hypovirus occurs naturally in chestnut stands in Michigan and has been recovered from experimental plots in West Virginia and Kentucky [15] [16] . CHV3-hypoviruses have a less effect on sporulation and pigmentation, however, significantly reduces the virulence of C. parasitica [17] . Typically, the vegetative growths of these isolates are severely debilitated, producing small, superficial cankers both in vivo and in vitro [16] . Previous studies have identified fundamental differences in the CHV3 dsRNAs when compared to CHV1 [18] . Sequencing results for hypovirulent isolate, CHV3-Grand Haven 2, have determined that this hypovirus is comprised of a single ORF and a 9.8 kb genome; considerably distinct from the two ORF, 12.7 kb genome organization of the CHV1 hypovirus [9] [17] .

Despite the variation among hypovirus-containing strains, they all have the potential to arrest canker expansion, as noted in the callousing cankers in European and in North American orchards [2] [19] [20] . Hypoviruses can also be transmitted into asexual conidia producing hypovirulent inoculum in natural populations [21] . In addition, each hypovirus-type is transmissible in nature via anastomosis among the same vegetative compatibility (vc) types in C. parasitica populations [14] [22] . Hypovirulence-mediated biological control has been far less successful in North America than in Europe, presumably due to barriers caused by diversity in vc types [16] [23] . However, HV transmission has also been reported to occur between different vc types, although to a much lesser extent [24] [25] . This indicates that hypovirulence conversion between incompatible isolates may provide successful transmission into natural populations of C. parasitica overtime [13] [25] [26] .

Field isolates that have the ability to produce non-lethal cankers, generate hypovirulent inoculum, and exhibit a greater ecological fitness are sought after to promote a successful biological control agent [12] [22] . The notion is that an isolate that has a greater tolerance to hypovirus-infection will produce a larger canker with the propensity to persist in the forest and produce hypovirus inoculum through viable asexual spores. Studies that analyze the interactions between the fungus and various hypoviruses have been conducted with the intention of selecting isolates that display a less debilitated growth habit after hypovirus acquisition [5] [14] [27] . In this present study, CHV1-Euro7, CHV1-Ep713, and CHV3-County Line hypoviruses were evaluated in five different strains of C. parasitica to determine the influences various hypoviruses have on different genomes of the fungus. Variables such as pathogenicity, HV persistence in the field, stroma production, and hypovirus transmission to conidia (in vivo and in vitro) were evaluated.

Characteristics thought to be a necessary for using hypoviruses as biological control agents include successful HV canker establishment, isolate persistence in the environment, and production of HV conidia. After 4 and 12 months in the field the influence of the hypovirus-infection on the growth of isolates was clearly demonstrated. The cankers initiated by virulent isolates produced significantly larger cankers than their HV counterparts. This finding was consistent with previous studies [27] [28] . With regard to CHV3-County Line hypovirus, isolates

infected with this hypovirus produced significantly smaller cankers than the CHV1-hypoviruses when assessed after either 4 or 12 months of growth (parental backgrounds pooled). Furthermore, all fungal backgrounds produced cankers similar in size when infected with CHV3-County Line. Isolates that produced larger cankers were severely affected by the CHV3-County Line hypovirus. The significant debilitation caused by the CHV3-type hypovirus has been observed in previous studies [9] [16] , presumably due to the significant down regulation of genes responsible for pathogencity.

Variability existed among the fungal strains, as in other studies [12] [29] . The C. parasitica fungal strain appeared to contribute to the differences in the size of cankers produced by the CHV1-Euro7 infected isolates. The variations in rankings observed among the isolates infected with CHV1-Euro7 were in most cases similar to the rankings of the isolates in their virulent form. That is, the most invasive strains, Bockenhauer and Schomberg, generally produced, with some exceptions, significantly larger cankers. This would indicate that the fungal genome plays a very important role in CHV1 hypovirus expression and therefore, how well suited particular field isolates are as biological control agents [12] . When comparing CHV1-Ep713 to CHV1-Euro7 isolates, there were generally no significant differences in size of cankers produced when cankers were compared per fungal strain. This was in contrast to previous studies that identified isolate Ep155 more debilitated when infected with CHV1-Ep713 with regard to canker size and sporulation than CHV1-Euro7 isolates growing on excised chestnut stems [14] [30] . Euro7ssv isolates produced cankers similar in size regardless of the infecting CHV1 hypovirus. Isolates JR10 and WR2, when infected with the CHV1-Euro7, produced smaller cankers than their CHV1- Ep713 counterparts.

As the cankers aged during this study, the percentage of hypovirulent isolates recovered increased. This occurred regardless of parental background or hypovirus. Cankers that enlarged the most (Schomberg, Bockenhauer, and Ep155) generally yielded the highest percentage of hypovirulent isolates. The recovery of the two CHV1-types was significantly higher than CHV3-County Line after 4 months in the field, consistent with the larger cankers yielding higher hypovirus recovery. However, after 12 months the HV recovery was similar among the three hypoviruses. Field studies have reported the persistence of CHV3 in West Virginia 12 years after initial establishment [15] . The natural establishment of CHV3 in Michigan may indicate a naturally suited hypovirus for the forests of North America [31] .

Stroma production on field cankers initiated by HV isolates was poor compared to virus-free isolates, a common consequence to the attenuation in virulence caused by hypovirus infection reported in other studies [20] [28] [32] [33] . If the genes responsible for enzymatic activity in the fungus are down regulated due to hypovirus-in- fection [9] an HV isolate will lack the ability to activate enzymes required for tissue degradation, nutrient acquisition, and the formation of stroma. Stroma production in vivo may have been further diminished by host response of the chestnut tree with the defensive production of polyphenolic compounds [34] , ethylene [35] , and callous tissue. The production of hypovirulent conidia in the forest ecosystem helps maintain the natural spread of hypoviruses and would seem to be a requirement for successful biological control. During this field study, isolates that also were the most pathogenic (Bockenhauer and Schomberg) were HV cankers that yielded HV inoculum.

Conidia formation as well as the percentage of hypovirus transmission to conidia was significantly higher in the laboratory than in the field. CHV3-County Line transmission percentages in vitro had a wider variation in rankings among parental backgrounds than those infected with CHV1-Euro7 or CHV1-Ep713. CHV1-infected isolates have been reported to transmit their hypoviruses at rates ranging from 90% - 100% [9] [16] . During this study CHV1-Ep713 and CHV1-Euro7 infected isolates had similar transmission rates averaging 80% - 87%, respectively. An increase in HV conidia from the CHV1-type hypoviruses may be a function of the protein p40 found on ORF A on the genomes of CHV1 hypoviruses. This protein is associated with CHV1 hypoviruses may increase ORF B expression that in turn enhances viral replication and dsRNA accumulation in asexual spores [9] . There were a few exceptions among the CHV1 infected isolates where certain backgrounds generated unusually low hypovirus transmission rates; a trait noted in previous single-conidial isolations indicating the fungal genome may have some regulatory effects on how well the hypovirus is able to replicate and be passed on during conidiogenesis [28] [32] .

In summary, this research provides additional evidence that each hypovirus interacts with its host differently. The results support previous research that reported varying morphological and pathogenic characteristics dependent on the genome of the isolate and hypovirus [14] [28] [32] . The most important message consequential to this study is that certain isolate/hypovirus combinations have better biological control potential than others. It would seem that those fungal isolates that are most pathogenic also have other traits such as increased ability to sporulate and high HV transmission rates that merit their use as biological control agents [22] . This, however, may not always be the case. For example, certain hypovirus/isolate combinations, JR10 and WR2 infected with CHV1-Euro7, performed very poorly in comparison to the other isolates infected with the same hypovirus. However, these isolates showed an improved biological control potential when harboring CHV1-Ep713. Therefore, hypovirus infection may not create strains that are uniform in all biological control characteristics. Rather, it appears to be a complex interaction between fungus and hypovirus genomes resulting in isolates with varying attributes.

Funding and research support for this project was provided by The American Chestnut Foundation and West Virginia University. Thank you to Dr. William MacDonald and Mark Double of West Virginia University for their guidance, contribution, and thoughtful feedback throughout this study. Author thanks Anthony Barker, Jessica Bright, and William Rittenour for their assistance in the field and laboratory procedures.