Differences in the Vulnerability of Waterbird Species to Botulism Outbreaks in Mediterranean Wetlands: an Assessment of Ecological and Physiological Factors

Study area.

The study area consisted of 3 wetlands in Castilla-La Mancha, a flat region situated in the central Spanish plateau mostly devoted to agriculture: Tablas de Daimiel floodplain, Navaseca Lake, and Prado Lake. Tablas de Daimiel is a national park (TDNP) that protects 1,675 ha of a floodplain wetland; it is a special protection area for birds and is included in the Ramsar List. Navaseca Lake (24.3 ha) is located near TDNP (about 6.5 km). It is a permanently flooded wetland that is highly eutrophic, because it receives the effluent of a wastewater treatment plant. Prado Lake (50.5 ha) is situated near TDNP (about 25 km); it is a temporary, endorheic, saline wetland that also receives inputs of treated wastewater and is highly eutrophic. These 3 wetlands are winter refuge and/or breeding sites of a variety of waterbird species, including endangered species, such as the crested coot (Fulica cristata), white-headed duck, ferrugineous duck, or marbled teal (Marmaronetta angustirostris). Avian botulism is endemic in the region, where outbreaks have been attributed to the mosaic C. botulinum type C/D (2, 11). Botulism outbreaks occurred in summer to early autumn in Navaseca and Prado Lakes in 2011 and 2012 during the study period. Botulism type C/D was confirmed as the principal cause of death in both lakes by using the mouse bioassay test with plasma from diseased birds (17). The antitoxin type C used in the bioassay was purchased from the Centers for Disease Control (reference no. BS0611). The climate in this area is cold-temperate continental, with a pronounced dry season and annual rainfall of around 400 to 500 mm. All of the wetlands studied are between 603 and 670 m above sea level.

Estimates of living and dead waterbirds in Navaseca Lake.

Waterbird censuses were performed in Navaseca Lake during outbreak periods in 2011 and 2012 in order to estimate the numbers of birds at risk in each moment. In this wetland, a protocol for the collection of sick waterbirds was established, which allowed us to compare the proportion of sick or dead birds removed every day with the number of live birds at risk in the previous census for the different species. Navaseca Lake was too large to be observed from one point, and there were islands in it, so we divided it in 6 sections to facilitate the census. A census of live birds of different species was performed from a peripheral road just after sunrise by using binoculars and a telescope at 6 observation points (one for each section). A tape recorder was used for counting the observed individuals of each species, and the record was afterward transcribed to a database. The censuses were performed during outbreak periods: in 2011, 4 censuses were performed between July and August at intervals of 15 days; in 2012, 9 censuses were performed between the second half of June and August with weekly intervals (except one missing count at the end of July). As part of the botulism control protocol, and to obtain data about affected waterbirds, 1 or 2 persons surveyed on foot the shore of Navaseca Lake every morning, collecting dead and sick birds. Once a week, the survey was also taken by boat in order to collect birds from several small islands located in the center of the lake. After identification of the species by experts, all of the dead birds were buried, and the sick ones were sent to the wildlife rehabilitation center of the regional government (Joint of Commonalities of Castilla-La Mancha [JCCM]). The numbers and species of all dead (and sick) birds were recorded. Although these counts tend to underestimate the mortality, they were used as an estimation index to be compared with the census of live birds. Finally, 23 sick birds that died in the wildlife rehabilitation center were necropsied, and samples of cecum, intestines, and liver were analyzed for the presence of C. botulinum by real-time PCR.

Sampling of aquatic invertebrates.

During outbreak periods in Navaseca and Prado Lakes in 2011, we sampled aquatic invertebrates and small fishes to identify possible sources of intoxication for waterbirds. We captured them from the water surface of the shore using a 500-μm-mesh sieve. In 2012, we made a greater sampling effort of snails in Navaseca Lake, because we observed that these invertebrates were very abundant and that they fed on carcasses, so we suspected that they could be an important source of BoNT for birds. Captured invertebrates were kept alive in sterile plastic containers until analysis during the following 24 h. In the laboratory, they were identified and pooled according to the taxonomic family. Samples of invertebrates and fishes included specimens of nonbiting midge larvae (Chironomidae, n = 15), water boatmen (Corixidae, n = 24), back swimmers (Notonectidae, n = 4), crustaceans (Ostracoda, Copepoda, and Cladocera, n = 15), freshwater snails (Physidae, n = 53), beetle larvae (Coleoptera, n = 5), mayfly larvae (Ephemeroptera, n = 3), soldier fly larvae (Stratiomyidae, n = 2), and mosquitofish (Gambussia holbrooki, n = 8).

Sampling of waterbird feces and healthy bird cloacal swabs.

Waterbird fecal samples were collected during outbreak and nonoutbreak periods. Fecal sampling during outbreak periods was performed in 2011 between late July and October in Navaseca Lake (number of sampling visits [n_s] = 4) and Prado Lake (n_s = 1). Fecal sampling during nonoutbreak periods was performed between June and early July 2011 in Navaseca Lake (n_s = 3), between August and October 2011 in TDNP (n_s = 3), and between January and April 2012 in Navaseca Lake (n_s = 5) and in Tablas de Daimiel National Park (n_s = 1). In each sampling visit, around 30 fresh fecal samples from Anatidae (ducks), Rallidae (rails), and Charadriiformes (shorebirds and gulls) were collected with sterile swabs and kept at 4°C in plastic bags with zip closures until processing during the next 24 h. In order to ensure the identity of the fecal samples from these birds, samplings were performed in the shores where flocks of birds of a single family or order were observed resting for long periods.

Healthy bird cloacal sampling was done at Navaseca Lake once per season in 2012. We principally captured common moorhens (Gallinula chloropus) by using funnel traps located on the shore of the lake, where groups of them usually fed. In previous studies, monitoring of feces showed the presence of C. botulinum in 2.9% of droppings of rails, including common moorhens, which suggested that healthy birds of this species could be carriers of the bacteria (14). We also tried trapping moorhens in TDNP, but we were not successful. Traps were baited with grain and checked every hour. Captured birds were removed and held in bags before banding and sample collection. Age (juvenile or adult) was determined using plumage criteria (18). Birds were marked with a metal ring, and a cloacal sample was taken with a sterile swab and kept in a sterile transport container (Aptaca, Copan) until analysis. After sampling, the birds were released. Cloacal swabs were also taken from recaptured birds.

Fecal excretion of C. botulinum type C/D by sick birds.

In 2012, waterbirds showing botulism clinical signs were collected during the 2 outbreaks that occurred in Navaseca and Prado Lakes. They were monitored to study the chronology of fecal excretion of C. botulinum type C/D after admission at the regional wildlife rehabilitation center. Upon admission, birds were marked with color plastic bands, and a cloacal sample was taken from each with a sterile swab that was kept in a transport container (Aptaca, Copan). For the following 5 days, birds were maintained in nursing pens and received support treatment consisting of antibiotic therapy (marbofloxacin [Marbocyl 2%]; Vetoquinol), vitamin complex (Duphafrarl Multi; Pfizer), intravenous fluids (Ringer's lactate), and tube feeding. During this period, another 2 cloacal swabs were taken on days 3 and 5. After this time, if birds recovered, they were taken to a bigger enclosure before being released. During this period, cloacal swabs were taken when birds needed to be handled and before they were released. The intended sampling scheme was to collect cloacal samples at 2-day intervals; however, due to the different times of recovery of each bird and to avoid interfering with the center's staff work, the sampling scheme was varied.

C. botulinum type C/D detection.

Cloacal swabs, bird feces, necropsy samples, and aquatic invertebrates were processed for detection of C. botulinum type C/D, as previously described (19). Briefly, samples were cultured in 9 ml of commercial cooked meat broth supplemented with vitamin K₁, glucose, and hemin (BD BBL cooked meat medium with glucose, hemin, and vitamin K) using an anaerobe container system (BD GasPak 129 EZ) over a period of 3 days at 37°C. DNA was extracted by boiling the pellet obtained from 1 ml of the culture broth in 300 μl of distilled water. The solution obtained after centrifugation was used as the PCR template. Real-time PCR was performed as described by Sánchez-Hernández et al. (20) for the detection of the genes encoding both type C and type C/D mosaic toxins (2). This protocol permitted detection of spores, vegetative cells, or DNA, so we could identify the birds harboring or excreting the bacteria.

Statistical analysis.

The overall mortality rate during botulism outbreaks in Navaseca Lake was calculated as the percentage of dead and sick individuals of each species relative to the maximum number of individuals of the species recorded in that year. Mortality rates were compared among species in each study year with the χ² test. The differences in the vulnerabilities to botulism between species were studied using the recorded data of sick and dead birds of each species relative to the census of live individuals in the previous days (1 to 2 weeks). The relative vulnerability index of each species has been calculated as the means of the standardized residuals obtained after running a generalized linear model (GLM) with the number of sick or dead birds collected each day as the dependent variable and with the square root or the logarithm (x + 1) of the previous census of live birds and the Julian date (count of days since 1 June) and the year as explanatory variables. This approach permitted us to use all of the data set at a daily scale (of sick and dead bird counts) and minimized the bias produced by low counts of some species. Therefore, this vulnerability index could be more appropriate than the mortality rate described before to rank the vulnerability of the species to botulism. Comparisons of this vulnerability index among species were performed with a one-way analysis of variance test followed by a post hoc comparison with the Tukey test. Only the most abundant species were considered in the study of the vulnerability index, i.e., Anatidae (5 species), including mallard, gadwall (Anas strepera), northern shoveler (Anas clypeata), common pochard (Aythya ferina), and white-headed duck; Rallidae (2 species), including common moorhen and Eurasian coot; and little grebe (Tachybaptus ruficollis), black-winged stilt (Himantopus himantopus), black-headed gull (Chroicocephalus ridibundus), and greater flamingo (Phoenicopterus roseus). The Fisher exact probability test was used to compare the frequencies of detection of C. botulinum type C/D in bird fecal samples collected during outbreak and nonoutbreak periods among species and between juveniles and adults. It was also used to compare the frequencies of the pathogen in cloacal swabs of different bird species at the wildlife rehabilitation center. Significance for the statistical analyses was set at a P value of <0.05. The analyses were performed with IBM SPSS Statistics 22.

Vulnerability of waterbirds species to avian botulism.

During the census performed in Navaseca Lake between 2011 and 2012, the most commonly observed species were Eurasian coot, greater flamingo, northern shoveler, white-headed duck, black-necked grebe, mallard, and moorhen, but differences in abundance of some species (i.e., black-headed gull, coot, and northern shoveler) were observed between years (Table 1). During the same period, 646 dead or sick birds with clinical signs of botulism were collected: 357 in 2011 and 289 in 2012 (Table 1). The presence of BoNT type C/D was confirmed with the mouse bioassay in the plasma of 5 birds collected during the outbreaks. The BoNT type C/D gene was detected in at least 1 sample of cecum, intestines, and liver of 8 of the 23 (35%) necropsied waterbirds during the outbreak in Navaseca Lake in 2012. In 2011, the first waterbird carcasses were observed in the third week of July, and the number of dead birds increased until mid-August; the last dead birds were collected at the beginning of October. In 2012, the first dead birds appeared in mid-June, and the number of dead or sick birds reached a peak at the end of June and then decreased until the end of August. The numbers of affected birds from certain species varied between years: shovelers, gadwalls, white-headed ducks, moorhens, and black-winged stilts were more affected in 2011, while mallards, gulls, grebes, and coots were more affected in 2012 (Table 1). Coots represented 39% of all the affected individuals, followed by mallards (14%), shovelers (10%), gulls (8%), white-headed ducks and gadwalls (7%), black-winged stilts and moorhens (5%), pochards (6%), and, finally, flamingos and grebes (0% to 1%). Regarding the percentage of dead and sick birds relative to the maximum census of live birds counted each year (percentages in 2011 to 2012), significant differences were observed among species (χ² tests, P < 0.001) (Table 1).

GLM analysis performed with the individual pairs of census and mortality data at different times showed a better fit with the square root transformation of the census (coefficient of determination [R²] = 0.32) than with the log transformation (R² = 0.24). The best-fitted model showed a significant positive relationship between the census and the number of sick and dead birds for each species (Wald's χ² = 47.84; P < 0.001) (Fig. 1A), and it also showed that, in 2011, the number of affected birds was significantly higher than in 2012 (Wald's χ² = 22.09; P < 0.001) (Table 1). The effect of the Julian date was not significant. The vulnerability index was calculated from the standardized residuals of the best-fitted model, and the comparison among species revealed that coots and mallards were the most vulnerable species to botulism intoxication, while greater flamingos and little grebes were less vulnerable (F_10,132 = 14.108; P < 0.001) (Fig. 1B).

Potential determinants of C. botulinum carriage.

The presence of aquatic invertebrates with toxigenic bacteria was evaluated as a potential factor for species that include these invertebrates in their diet. Regarding this, the C. botulinum BoNT type C/D gene was detected in 16 out of 53 (30%) samples of freshwater snails (Physa acuta) and in 2 out of 2 soldier fly larvae (Stratiomyidae). None of the other samples of aquatic invertebrates or fishes yielded C. botulinum BoNT type C/D genes.

The retention/excretion of the toxigenic bacteria through the gastrointestinal tract of waterbirds was considered another significant factor for the vulnerability and carriage of C. botulinum in waterbird species and, especially, for their contribution to the epidemiology of the outbreaks. A total of 89 moorhens, 1 water rail (Rallus aquaticus), and 1 mallard were captured in the funnel traps; from these, 46 individuals were adults, 38 were subadults, and 7 were chicks. Nineteen moorhens were recaptured once, 2 were recaptured twice, and 1 was recaptured 3 times. None of the 117 cloacal swabs from these healthy birds captured in the funnel traps in Navaseca Lake showed the presence of the BoNT type C/D gene (Table 2).

During nonoutbreak periods, 0.8% (n = 365) of waterbird fecal samples yielded BoNT type C/D genes, while during outbreak periods, 16.3% (n = 153) of the samples yielded it, with this difference being significant (Fisher's exact test, P < 0.001) (Table 2). Regarding excretion by the type of bird, 23% of fecal samples from Rallidae (coots and moorhens), 16% of samples from Laridae (gulls), and 9% of samples from Anatidae (ducks) yielded BoNT C/D genes when collected during botulism outbreaks; these percentages were not statistically different. Out of outbreak periods, 1.4% of fecal samples from Rallidae and 1.9% of samples from Laridae yielded BoNT C/D genes.

As for the sick birds admitted to the wildlife rehabilitation center with botulism signs, 68 individuals from 4 families and 9 species were sampled: 28 Anatidae, 21 Rallidae, 18 Laridae, and 1 Podicipeadidae (grebe) (Table 3). In total, 32 of these individuals were finally released, and 37 died during the recovery process. Twenty-six (37.7%) of the birds excreted C. botulinum BoNT type C/D at some time during the period at the wildlife rehabilitation center. The excretion of BoNT type C/D decreased with the time at the center, from 30% positive cloacal swabs on day 1 to 0% on day 7; however, 2 individuals, a coot and a gadwall, excreted it again after the 7th day, just before being released. Anatidae and Rallidae excreted BoNT type C/D more often (P < 0.05) and for a longer time than gulls (Table 3).

Conclusions.

Vulnerability to botulism intoxication is different for each waterbird species, and it depends on feeding behavior, turnover, and time of arrival to the wetland; also, some phylogenetic component may exist. This implies that concentrations of highly vulnerable species during high-risk botulism periods may increase the likelihood of an outbreak and the mortality rate. In this study, the threatened white-headed duck was not among the most vulnerable species, but botulism killed around 17% and 7% of the individuals inhabiting the studied lake in 2011 and 2012, respectively, so this disease should be considered in future conservation programs. Among the sources of intoxication for some waterbird species, the invasive water snail Physa acuta may be particularly important, as it is very abundant in some wetlands and frequently harbors C. botulinum. Even though C. botulinum does not seem to form part of the normal microbiota of healthy waterbirds, the waterbirds may excrete and disperse it between wetlands after ingesting it from the environment. The persistence of C. botulinum in the intestinal tract of some birds may open the possibility of the development of toxicoinfectious botulism if the intestinal microbiota has been altered, but further research is necessary to confirm this hypothesis.