Pollination is important for ecological processes and worldwide agricultural productivity (Potts et al ., 2016). Several causes of global pollinator decline have been described, including climate change, pathogens, habitat loss and pesticides (Goulson et al ., 2015).

Among pollinators, bumblebees ( Bombus Latreille, 1802) provide pollination services not only to wildflowers but also to economically important crops; for this, some species are reared and commercialized for greenhouse production (Ortiz-Sánchez, 1992; Ornosa, 1996; Velthuis & van Doorn, 2006). Several pathogens ( Crithidia Léger, 1902: Trypanosomatidae, Apicystis (Lipa & Triggiani, 1996): Neogregarine, Nosema Nägueli, 1857: Microsporidia, and viruses) are related to the decline of bumblebee species, and their prevalence has been monitored elsewhere but not in the Iberian Peninsula (Williams & Osborne, 2009; Cameron et al ., 2011; Meeus et al ., 2011; Gallot-Lavallée et al ., 2016). To address this knowledge gap, samples of the most widely distributed species Bombus terrestris (Linnaeus, 1758) and Bombus pascuorum (Scopoli, 1763) were collected to map pathogen prevalence. We hypothesized that infection rates differ between northern and southern Spain, between species and between sexes.

Worker and male bumblebees were sampled from 19 locations in Spain: either from a southern mountain region (Sierra Nevada National Park, Granada) at altitudes between 1041 and 2752 m a.s.l., where conservation and protection measures of fauna and flora are implemented, or in northern human-altered landscapes (mainly edges of agricultural fields, river banks and urban gardens) between 63 and 1031 m a.s.l (Table 1).

Bombus terrestris and B. pascuorum species remain abundant in the study area, and are not considered threatened according to the International Union for Conservation of Nature (Rasmont et al ., 2015). Individuals were sampled with nets while foraging on flowers during August 2013 with dry weather (above 18 ºC) and clear weather conditions, and preserved in 100% ethanol at – 4 ºC.

Every sampled bumblebee was dissected to remove the genitalia and only the remaining abdominal tissues (gut and fat body) were used for DNA extraction following the Chelex method (Scriven et al ., 2013). RNA was isolated with RNase-Free DNase kit (Quiagen) and transformed in cDNA with QuantiTec reverse transcription kit (Quiagen) following manufacturer’s instructions.

Bumblebee species identification was confirmed through morphological and sequence analyses (Murray et al ., 2008).

To amplify the fragment of the 18S rDNA gene of Crithidia spp. and Apicystis spp., the reaction mixture consisted of a triplex PCR reaction including Apidae primers (as DNA extraction and amplification control for Bombus species) (Meeus et al ., 2010). To amplify Nosema bombi (Fantham & Porter, 1914) the reaction mixture consisted of a duplex PCR also including the Apidae primers. For Nosema apis (Zander 1909) and Nosema ceranae (Fries et al ., 1996) BioTools amplification plates were used (Martín-Hernández et al ., 2007), consisting of a prepared microplate with gelled enzymes and primers to which only water and the DNA sample need to be added. To amplify viruses, the reaction mixtures were performed following Ravoet et al . (2013) for Lake Sinai Virus (LSV), Francis & Kryger (2012) for AKI (Acute bee paralysis, Kashmir bee and Israeli acute paralysis viruses complex) and Chantawannakul et al . (2006) for Deformed Wing Virus (DWV) and Black Queen Cell Virus (BQCV). For detecting BQCV and DWV virus, cDNA amplification was performed with a RT-PCR, using as internal control Apidae primers from Meeus et al . (2010). Detailed information about primer and probe sequences are provided in Table 2.

PCR reactions were run for 10 min at 95 ºC and followed by 35 cycles of denaturing at 95 ºC for 30 s, specific annealing temperature for 30 s, extending at 72 ºC for 45 s, and final extension of 7 min at 72 ºC. Positive (DNA of the pathogen) and negative (water instead of DNA) controls were included in each PCR reaction.

To verify that amplicons (those from the bumblebees for species confirmation, and from the pathogens) corresponded to the target organisms, positive samples were sequenced and compared with sequences uploaded in the NCBI through the BLAST tool in MEGA v6 (Tamura et al ., 2013).

Prevalence of infection was estimated as a percentage of the number of infected divided by the total number of individuals. Proportions were compared using Yates-corrected chi-squared test, or when appropriate, Fisher´s exact test. Medians were compared with the non-parametric Kruskall-Wallis test. The relationship between infection with any pathogen or with Crithidia spp. and Apicystis spp. and explanatory variables was further explored using mixed logistic regression analysis to correct for spatial autocorrelation. Infection status (infected or non-infected) was the response variable. Explanatory variables including sex (female worker or male), environmental temperature, altitude, geographical distribution (north and south) and bumblebee species ( B. terrestris and B. pascuorum ) were fitted as fixed effects, and location was included as a random effect. A backward selection strategy was used to select explanatory variables, starting with a model that included all variables. Due to the high correlation between explanatory variables (for example altitude and sampling area), different combinations of models with two variables were finally compared using Akaike´s Information Criteria to select the model with the lowest value. Models were estimated using the maximum likelihood estimation method and significance was set at p<0.05 for a double-sided test.

A total of 115 bumblebees were sampled: 83 B. terrestris and 32 B. pascuorum . Bumblebee species abundance differed between regions: 75% B. terrestris came from the south, while 68% B. pascuorum were from the north ( p <0.05). The sex ratio was similar for B. terrestris individuals, and 72% of B. pascuorum individuals were females ( p <0.05).

Alignment of the amplicons with sequences FN546181 and FN546182 (Meeus et al ., 2010) confirmed the presence of Crithidia bombi (Lipa & Triggiani 1988) and Apicystis bombi (Liu et al ., 1974). Pathogens distribution is described in Table 3. Both protozoa were present in 37% B. terrestris and in none B. pascuorum . The estimated prevalence (95% CI) for B. terrestris was 15.7% (7.8-23.5%), for C. bombi and 24.1% (14.9-33.3%) for A. bombi , respectively. Two individuals were infected by both pathogens. Prevalence of both protozoans was significantly higher in males than in female workers, and A. bombi was more frequent in the south and C. bombi in lower altitude areas ( p <0.05). Logistic regression models corroborated the relationship between protozoa infection, sex and area (north and south) or altitude. The risk of infection was greatest in middle range altitude ( p <0.05).

Of the three Nosema species, only one B. terrestris male from the north was infected with N. bombi . Viruses were only found in B. pascuorum and with low prevalence: two workers were infected with DWV (one southern and one northern), and three were infected with BQCV (one male and two workers from the north). No bumblebees were infected with LSV or AKI.

Overall, pathogens (protozoa, N. bombi and bum­ble-bee viruses) were detected in 34% of sampled bumblebees; 16% of B. pascuorum and 37% of B. terrestris ( p <0.05). Pathogen prevalence was significantly greater in males than in females ( p <0.05), and did not differ according to environmental temperature or altitude, although it was numerically greatest in B. terrestris males, in low altitudes and warmer places ( p >0.05). There was no evidence for significant variation in infection risk according to location.

This study shows that pathogen prevalence in two bumblebee species in Spain varies depending on host and pathogen species, sex and area, although in general the level of pathogens was low. Prevalence of infection was higher in B. terrestris than in B. pascuorum , however, the former had protozoa and a microsporidium infection, and the latter only carried viruses.

A. bombi showed the highest prevalence and appeared exclusively in southern B. terrestris . This neogregarine has been recently detected in other Bombus species (Gamboa et al ., 2015), but its dispersion has been reported mainly due to managed bumblebees such as B. terrestris (Graystock et al ., 2014). In Spain, the extensive use of this species for greenhouse tomato pollination may have contributed to the spread of this pathogen.

The second most common parasite in B. terrestris was C. bombi as in UK (Goulson et al ., 2012). Trypanosomatids have rapid adult-adult transmission linearly related to host density (Goulson et al ., 2012), occurring either in sympatric or allopatric populations (Meeus et al ., 2011). They can be transmitted by direct contact or indirectly by dispersal onto flowers (Graystock et al ., 2015). B. terrestris has larger nests than B. pascuorum (Goulson et al ., 2012), which could facilitate faster transmission. In this study, males showed higher infection prevalence than workers, which may be explained by biological reasons: males live mainly outside the nests, and are more exposed to pathogen spillover from other pollinator species (Goulson, 2010).

Regarding microsporidia, N. bombi was the only species detected in B. terrestris . This pathogen has also been found in the UK (Goulson et al ., 2012), but in Spain has not yet been reported. Notably, N. ceranae was not present in our sample, despite its high prevalence in bumblebees from UK (Graystock et al ., 2013b) and in honeybees in Spain (Martín-Hernández et al ., 2007).

As in other studies (Singh et al ., 2010; Meeus et al ., 2011), LSV and AKI-complex viruses did not appear in Spanish bumblebees albeit BQCV and DWV have been found with a low prevalence only in B. pascuorum . DWV is highly widespread among bumblebees in Europe, and has been found in B. terrestris and B. pascuorum living nearby honeybee colonies in Germany (Genersch et al ., 2006). Given that we did not account for honeybee colony density in the sampling areas, a cause-effect of this low prevalence cannot be assessed.

Two individual bumblebees were infected by more than one pathogen, suggesting individual differences regarding susceptibility to parasitic infection, or that pathogens can act synergistically (Whitehorn et al ., 2011). The higher prevalence of pathogens in B. terrestris might reflect a different resistance and susceptibility to infections between species. Alternatively, protozoa virulence in B. terrestris may be lower compared to B. pascuorum (Goulson et al ., 2012). Moreover, B. terrestris emerges early in the spring, whereas B. pascuorum emerges later (Ornosa & Ortiz-Sánchez, 2004; Ploquin et al ., 2013). We conclude that this ecological difference between host species might affect the success of the parasite’s biological cycle and consequently infection prevalence.

We very much appreciate the English editing by Carolyn Daher.