1 This is a preprint version of the final version of this article. The 2 formatted article can be found at
6 Title: Assessing the long-term impact of Ranavirus infection in wild common frog (Rana
7 temporaria) populations.
1,2,3119 Amber G. F. Teacher, Andrew A. Cunningham, Trenton W. J. Garner
11 1. Institute of Zoology, Zoological Society of London, Regent’s Park, London NW1 4RY, 12 UK.
13 2. School of Biological and Chemical Sciences, Queen Mary University of London, Mile
14 End Road, London E1 4NS, UK.
15 3. Current address: Ecological Genetics Research Unit, Department of Biosciences, P.O.
16 Box 65, FIN-00014 University of Helsinki, Finland. 17
18 Corresponding author: Dr Amber G. F. Teacher
19 Address: Ecological Genetics Research Unit, Department of Biosciences, P.O. Box 65,
20 FIN-00014 University of Helsinki, Finland. 21 Email: email@example.com
23 Running head: Impact of Ranavirus on wild frog populations. 24
25 Keywords: Amphibian, disease, Ranavirus, common frog, population decline, Rana
3 Amphibians are declining worldwide, and one cause of this is infectious disease emergence. 4 Mass mortalities caused by a virus, or a group of viruses, belonging to the genus Ranavirus
5 have occurred in wild common frogs (Rana temporaria) in England since the 1980s, and
6 ranaviral disease is widespread in amphibians in North America and Canada, where it can 7 also cause mass die-offs. Although there have been numerous reports of Ranavirus-
8 associated mass mortality events, no study has yet evaluated the long-term impacts of this 9 disease. This study follows up archived records of English common frog mortalities likely 10 caused by Ranavirus. There is a preliminary indication that common frog populations can 11 respond differently to the emergence of disease: emergence may be transient, catastrophic, 12 or persistent with recurrent mortality events. We subsequently focused on populations that 13 had recurring mortality events (n=18), and we report median declines of 81% in the number 14 of adult frogs in these populations from 1996 to 2008. Comparable uninfected populations 15 (n=16) showed no change in population size over the same time period. Regressions show 16 that larger frog populations may be more likely to experience larger declines than smaller 17 populations, and linear models show that percentage population size change is significantly 18 correlated with disease status, but that habitat age (a possible proxy for environmental 19 quality) has no significant effect on population size change. Our results provide the first 20 evidence of long-term localised population declines of an amphibian species which appear 21 to be best explained by the presence of Ranavirus infection.
3 The Iridoviridae family consists of five genera; Iridovirus, Chloriridovirus, Ranavirus,
4 Lymphocystivirus and Megalocytivirus (Murphy et al., 1995). Of these, only ranaviruses are
5 known to infect amphibians. Unsolicited reports of unusual and often large-scale common 6 frog (Rana temporaria) mortalities have been received by various conservation
7 organisations in England since at least 1985 (Cunningham et al., 1996). The Frog Mortality
8 Project was jointly initiated in 1992 by the Institute of Zoology and Herpetofauna 9 Consultants International Ltd., and determined for the first time that these mortalities were 10 unequivocally caused by a virus belonging to the Ranavirus genus (Cunningham et al.,
11 2007a; Cunningham et al., 1996), specifically a variant, or variants, of frog virus 3 (FV3), 12 the type species for Ranavirus (Hyatt et al., 2000). Further records of amphibian mortalities
13 in Britain have been collected on an ad-hoc basis from 2002 onwards by the charity 14 Froglife. The mortalities originally occurred most often in the South-East of England, 15 however within a few years the range expanded to other parts of Britain (Cunningham, 16 2001). FV3 isolates from Rana temporaria in England are similar to North American
17 isolates, based on 586 base pairs of major capsid protein sequence data, where only a single 18 nucleotide differed between the isolates (Hyatt et al., 2000). This has led to speculation that
19 the virus was introduced to England from North America, potentially due to translocations 20 of the bullfrog (Rana catesbiana) or goldfish (Carassius auratus) for the pet trade
21 (Cunningham, Daszak & Rodriguez, 2003).
23 Infection of R. temporaria with Ranavirus can cause two disease syndromes; ulcerative
24 skin syndrome, characterised by dermal ulceration (Fig. 1) and haemorrhagic syndrome, 25 characterised by systemic haemorrhaging within the skeletal muscles and visceral organs 26 (Cunningham et al., 1996). Skin ulceration is easily visually recognizable and highly
1 diagnostic (Fig. 1), and is also associated with striking loss of muscle mass in limbs. Other 2 signs of ranaviral disease include lethargy and emaciation, although these alone are not 3 pathognomonic. Mortality levels can be high; Green et al. (2002) observed over 90% 4 mortality in their study of 25 Ranavirus–associated amphibian (9 species) mortality events.
5 However, mortalities are often restricted to a small geographical area, for example a single 6 pond (Carey, Cohen & Rollins-Smith, 1999). Although infection can lead to mass 7 mortalities within a population (Cunningham et al., 1996), some individuals are thought to
8 be able to survive infection, as scars characteristic of healed skin ulcers are sometimes 9 present on frogs in association with current signs of disease (Cunningham 2001, Fig.1). 10 Ranavirus isolates from the UK grow in vitro between 8-30?C, with slower replication
11 below 15?C and the fastest replication at 30?C (Cunningham, 2001). Likely reflecting this 12 temperature dependence, deaths caused by Ranavirus at all developmental stages (larval,
13 metamorph and adults) occur predominantly in the summer months (Chinchar, 2002; Green
14 et al., 2002). Findings from North American ranaviruses in other host species have, 15 however, shown that lower temperatures can cause increased mortality, perhaps due to 16 immunosuppression at low temperatures (Rojas et al., 2005).
18 Mass mortalities in wild populations that are caused by Ranavirus are known to have
19 occurred in many different amphibian species, including some that are thought to be in 20 decline (e.g. the Sonora tiger salamander) (Green et al., 2002). The most extensively
21 studied examples are mortalities of common frogs in England (Cunningham et al., 1996;
22 Teacher, Garner & Nichols, 2009a; Teacher, Garner & Nichols, 2009b) and tiger 23 salamanders (Ambystoma tigrinum) in North America and Canada (Bollinger et al., 1999;
24 Collins et al., 2004; Jancovich et al., 1997). Annual recurrence of the disease has been
25 noted in both of these species (Daszak et al., 1999). It is of note that the mortality events in
1 North America tend to involve larvae, whereas in the UK they are noted as yet only in 2 adults (Cunningham et al., 1996; Green et al., 2002). Despite causing high levels of 3 mortality, it is not yet known whether Ranavirus can cause long-term population declines
4 (Daszak, Cunningham & Hyatt, 2003), primarily due to a lack of tractable wild study 5 systems.
7 This study takes advantage of archived records from public reports of common frog 8 mortalities from the Frog Mortality Project and Froglife records. Stringent criteria were 9 used to identify archived records of populations that were most likely first infected with 10 Ranavirus (according to symptoms and timing of the mortalities) in 1996/7, and these 11 populations were resurveyed in 2008. The outcome of each population was noted, and those 12 that were still experiencing recurrent disease were investigated in more detail. Visits to 13 affected populations to look for signs of disease (skin ulceration), and laboratory detection 14 of Ranavirus from a subsample of carcasses were used to confirm that these populations 15 were indeed infected with Ranavirus, and to confirm the disease agent. The number of
16 frogs found in these populations before (1996/7) and after (2008) disease emergence was 17 then compared with population size changes in uninfected ponds over the same time frame. 18 Finally, linear models were employed to further investigate population size changes. 19
21 Identification of Ranavirus die-offs from archived records, and long-term outcome
22 Common frog populations in the UK are often found in relatively small, isolated and 23 artificial ponds in people’s back-gardens in urban and sub-urban settings (for an example,
24 see Fig. 2). Migration among ponds in such settings is expected to be lower than in rural
1 settings at similar geographic scales due to the presence of garden walls and fences, pet cats 2 and dogs, and the presence of barriers typically associated with urbanization (Fahrig et al.,
3 1995; Pellet, Hoehn & Perrin, 2004). Pond owners throughout the UK regularly have 4 submitted unsolicited records of amphibian mortalities to the Frog Mortality Project up 5 until 2002, and since then to the charity Froglife (UK registered charity no. 1093372).
6 Archives of records were searched for ponds where frogs had apparently been infected with 7 Ranavirus, using the following criteria:
8 a1. First mass mortality of common frogs in 1996 or 1997. As this study took place in 9 2005-2008, this allowed approximately a decade worth of impacts to be assessed. 10 a2. Mass mortality occurred between May and September. This excludes winterkill and 11 spawning-related deaths, which are the only other common causes of mass mortality 12 of frogs in the UK (Cunningham, 2001).
13 a3. Skin ulceration or signs of systemic haemorrhaging (bleeding through the mouth or 14 anus) recorded for the majority of frogs involved in the mass mortality. As stated 15 above, these are highly diagnostic signs of ranaviral disease and very unlikely to be 16 mistaken for any other etiology.
17 a4. Located in the South-East of England, within 80 miles of London. This region 18 received the greatest number of reports of mass mortalities in the mid-1990s. 19 a5. Urban or sub-urban location; this criterion, along with a4, was used to maintain 20 uniformity of habitat between the populations sampled.
21 In total, 70 such ponds were identified; the pond owners were contacted for further, up-to-22 date information in Autumn 2005. Of these ponds, some had since been filled in, others had 23 no frogs left, some had experienced no mortalities since, and others still experienced 24 recurrent disease (details in Results).
1 Identification of suitable study populations with recurrent disease
2 In addition to criteria a1-5, the following criteria were then used to identify suitable study 3 populations that had experienced recurrent disease:
4 b1. Frog population still present and breeding annually in the original pond. 5 b2. At least one diagnostic sign of disease in frogs noted by the pond owner during the 6 initial die-off (see a3).
7 b3. Continued presence of Ranavirus, as determined by at least one Ranavirus-
8 consistent death every two years (the cut-off was set low to ensure that we could be 9 sure that infection was still present, but to avoid artificially selecting only 10 populations with particularly high levels of mortality).
11 In total, 18 common frog populations were identified which satisfied all the criteria; given 12 that it is highly unlikely that any other cause produced this combination of observations, 13 these populations were termed Ranavirus-positive (Rv+) populations.
15 For the purpose of this study, control populations with no history of disease were also 16 required. An appeal was initiated for such ponds through local newspapers, Froglife, and
17 personal contacts; 16 Ranavirus-negative (Rv-) populations were identified based on the
18 fulfillment of the following criteria:
19 c1. Owners had lived at the same address since at least 1997.
20 c2. Owners had strong interest/awareness of their frog population. All populations were 21 urban or suburban and in gardens, ensuring the owners had easy and regular access 22 to highly visible populations.
23 c3. Owners had never identified any signs that might be indicative of Ranavirus
24 infection (see a2 and a3).
25 c4. Urban or sub-urban location the South-East of England, within 80 miles of London.
1 c5. Common frog population breeding annually in the same pond since at least 1997. 2
3 Confirmation of disease status of study populations
4 A sub-sample of Rv+ (n=8) and Rv- (n=9) populations were visited between May and July 5 2007. Between five and ten adult frogs (over 15 grams – weight criteria based on Home
6 Office requirements as blood samples were taken at the same time for another study; 7 (Teacher et al., 2009a)) from each population were caught by hand and observed for 8 evidence of Ranavirus infection; the number with active or healed skin ulceration was also 9 recorded. Those with both active and healed ulcers were recorded only as having active 10 ulcers to avoid duplication of results. Skin ulceration caused by Ranavirus typically results
11 in uniform rounded lesions, and the resulting scarring was identifiable primarily by the 12 characteristic shape (Fig. 1). Incidences of probable traumatic injury were also recorded, 13 including limb breakages and amputations, skin tears, penetrating injuries consistent with 14 bird predation, and septic or swollen digits. Scars consistent with probable traumatic injury 15 were identified as those with a non-rounded shape, such as would occur from skin tears or 16 piercing. The proportions of frogs with active or healed skin ulceration, and the proportions 17 with traumatic injuries, were compared between Rv+ and Rv- populations using Chi-18 squared contingency tests.
20 In addition, ten frog carcasses were obtained from seven Rv+ populations (submitted 21 during 2008). Cause of mortality was unknown in all cases. The liver was dissected from 22 each carcass ensuring no contamination occurred between animals. Liver samples were 23 used for DNA extraction using the Wizard SV96 Genomic Purification System (Promega, 24 UK). PCR was performed twice for each sample using published FV3-specific primers 25 which amplify a conserved region (420 base pairs) of the major capsid protein: forward: 5’-
1 GTCTCTGGAGAAGAAGAA-3’; reverse: 5’-GACTTGGCCACTTATGAC-3’ (Gantress et al., 2003;
2 Mao et al., 1996). The reaction mixture was: 2µl template DNA (approximately 0.3µg/µl), 3 4µl Qiagen Master Mix (Qiagen, UK), 1.8µl double-distilled and autoclaved water and 4 0.1µl of each primer (100pmol/µl). The PCR program used was: 35 cycles of denaturation 5 at 95ºC for 45 sec, annealing at 52ºC for 45 sec and extension at 72ºC for 45 sec (Pearman
6 et al., 2004). PCR controls consisted of two negative controls (no DNA) and two positive 7 controls (known infected individuals). Total PCR products were run on a 1.2% agarose gel 8 with 2；l loading buffer (Microzone, UK). Samples were recorded as positive for infection 9 with Ranavirus when both repeat PCR samples showed bands that matched the size of the 10 positive control bands. Two carcasses from a single pond had previously been screened for 11 Ranavirus using transmission electron microscopy (Cunningham, 2001).
13 The impact of disease on population size
14 We obtained estimated numbers of frogs killed in the initial mortality event for 53 of the 70 15 ponds from direct counts by owners (recorded in the original archived reports). Estimated 16 Rv+ adult population (n=18, see Results section on follow-ups) sizes were obtained from 17 direct approximations by owners (recorded in the archived reports) for the year prior to the 18 initial mass mortality event (1996/7). The majority (14/18) of the archived records had 19 population size estimates based on categories (<5, 5-10, 11-20, 21-40, 41-60, 61-80, 81-20 100, >100) – for the purposes of our analyses, a mid-point value was taken for these 21 records. In addition three records had an estimated number based on a direct owner count, 22 and one population had a direct count performed at the time by an author (A.A.C.). 23 Approximate adult population sizes based on direct owner counts in summer 1996/7 were 24 obtained for a subsample of Rv- population (n=10) ponds for which the owners had kept 25 records of this information. All pond owners (Rv+ and Rv-) were contacted in 2008 for a
1 contemporary adult population size approximation based on direct owner counts during the 2 summer months. The count methods could not be standardized as we relied on a mix of 3 archived data and public counts (raw data is available in Supplementary Material S1). 4 Detection probabilities should not differ consistently between Rv+ and Rv- populations, as 5 counts were performed in similar habitat types, around the same time of year. Although 6 ideally we would have used information on population sizes in the intermediate years, these 7 were unavailable. Instead our study was comparative and well-replicated within treatment 8 category, thus population responses as measured in 2008 were considered to be 9 representative of the factor compared amongst population treatment categories, namely 10 disease status.
12 Population size estimates were used to investigate change in population size. The number 13 of frogs in 2008 was regressed against the number in 1996/7 individually for each disease 14 category. A 1:1 regression would indicate no change in population size. We use methods 15 described by Verzani (2004) performed in R (R Development Core Team, 2008) to test 16 whether the Rv+ and Rv- slopes differed from one, and to test whether the Rv+ and Rv- 17 intercepts differed from zero (modified R code is available in Supplementary Material S2). 18
19 Percentage population size changes from 1996/7 to 2008 were calculated for each 20 population. The age of the pond was also included – there is a possibility older habitats may
21 provide better environments and potentially healthier animals (see discussion). Estimates of 22 the age of the ponds were obtained from pond owners, all of whom knew the year that their 23 ponds were built. To assess the variables influencing population size changes in the 24 sampled ponds, we used a linear model with normal error structure implemented in R (R 25 Development Core Team, 2008). We used percentage population size change as the