In this field study, we confirm the negative influence of Nosema infection on adult honey bee population size and brood production, and we show that when these microsporidia are not controlled, infection provokes a significant decrease in honey production. Since N. ceranae was the dominant species in the infected colonies throughout the study, the observed effects may principally be attributed to this microsporidium. However, although N. apis shows a small prevalence within the infected colonies and was not constantly present in them throughout the assay, a negative effect of this microsporidium over the colonies health cannot be discarded.
The number of adult bees, sealed brood and the amount of honey produced are valuable indicators of hive health . In our experimental conditions, the health status of the colonies was worse when they were highly infected with N. ceranae than when infection was controlled, suggesting that this microsporidium may have deleterious effects at the colony level, as suggested previously [17, 39, 40]. It is also worth mentioning that part of the non-treated colonies died leaving no bees in the hive at the moment of collapse, coinciding with signs described in a preceding study . Therefore, the parameters here presented regarding hive health status, taken together with other recently shown [36, 41, 42], may be considered as sub-clinical signs of nosemosis type C. However, as the number of colonies per group and sample sizes were limited, definitive conclusions cannot be made and the results presented here should be supported by more research.
The percentage of Nosema-infected forager bees is thought to be a potentially useful indicator of the extent of colony infection [12, 17, 43, 44]. The proportion of Nosema- infected forager bees in our colonies revealed significantly stronger infection than in the untreated colonies (groups CS and C) after the autumn and spring interventions, which may account for the observed differences in the fitness parameters like population size, brood production and honey production from those of the treated colonies (groups 1 T, 2 T and 4 T). Moreover, after the spring intervention we detected a significant negative correlation between N. ceranae infection and the size of the worker population, as seen previously [17, 39]. Indeed, there were significantly fewer bee combs in the spring and summer months in the colonies with the strongest N. ceranae infection (groups CS and C). Colony population size depends mainly on the worker bee life span  and as N. ceranae has a negative impact on this parameter in honey bees [21, 45, 46], it probably exerts a direct negative impact on colony size in accordance with our results.
Nosema infection was less prevalent in house bees than in forager bees, as reported previously [17, 35, 47]. Since the probability of acquiring infection is likely to increase with age and novel diseases are more often picked up by foragers, young workers should be less likely to be infected . A direct link has been reported between infection levels in forager and house bees [49, 50], but in the present assay we failed to establish such a correlation as the number of colonies in which the proportion of infected house bees was analysed was very limited. However, a subsequent study suggested that the interior bees may more accurately reflect the infection level within the colony . In our experimental conditions, forager bee infection rates of at least 45% were shown in all the collapsed colonies, while the percentage of infected house bees in collapsed colonies, only measured in colonies CS-7 and 4 T-6, was 20% and 35%, respectively. However, infection levels of 35% were also recorded in house bees in a colony that did not collapse (colony C-3) in July. This particular colony underwent natural queen supersedure in July, which may have subsequently decreased its infection rate and thus, the likelihood of collapse . Yet as the infection rate was not monitored after this time point, this hypothesis could not be corroborated. In colony C-6, which also underwent queen supersedure in June, the infection rate in house bees after one month was lower than that observed in the treated colonies (15%), while the forager bees exhibited a very high level of Nosema infection (75%). Thus, the influence of queen replacement on Nosema infection appears to primarily affect the house bee population, and the foragers to a lesser extent, although again this hypothesis could not be confirmed since infection rates were not subsequently measured.
Brood production was not correlated with infection levels, although less infected colonies had significantly more brood cells than those in highly infected groups at different time points throughout the study, consistent with earlier studies of Nosema- infected colonies [17, 29, 51]. Furthermore, brood production was slower in the untreated groups (groups CS and C), peaking in July, than in groups 1 T, 2 T and 4 T, which began growing earlier and reached maximum levels one month earlier (June).
Adult bee population size and brood production are correlated [39, 52, 53], and as mentioned above, increased bee mortality has been described in N. ceranae- infected bees. The smaller populations observed in strongly infected colonies may therefore reflect a combination of decreased brood rearing and increased forager death in these colonies. Furthermore, thermoregulation and brood food production may be dysregulated in infected colonies with small populations , resulting in reduced brood rearing efficiency . Nurse bees begin to forage precociously in response to high forager death rates [54, 56–58]. While this strategy restores the proportion of foragers in the population, it also shortens the overall lifespan of adult bees [59–61], as well as their effectiveness and resilience as foragers . This strategy also reduces the time each bee can dedicate to colony growth and brood production. Thus, when the rate of brood production is insufficient to replace the forager bees lost due to Nosema infection, the decline of the colony may accelerate [17, 58].
Colonies with high levels of N. ceranae infection throughout the study (groups CS and C) had significantly more brood cells relative to the adult bee population during the summer months (July and August). As the bee colony grows, a decrease in the ratio of brood to bees has been reported previously [53, 63, 64], as observed during the evolution of the colonies with a milder degree of infection here (groups 1 T, 2 T and 4 T). By contrast, strongly infected colonies (groups CS and C) did not grow at the same rate as treated colonies, and they did not experience a similar decrease in the proportion of brood to bees. The higher ratio of brood to bees in severely infected colonies may reflect the increased rearing effort in the colony in an attempt to replace the worker bees lost due to Nosema infection , although further studies will be necessary to confirm this hypothesis.
Since honey production is generally correlated with the number of worker brood cells and the worker population [27, 65], larger colonies tend to store more honey . In addition, nectar foragers from larger colonies visit more flowers on each trip, with a shorter handling time per flower when compared to those from small colonies . However, Nosema may negatively influence the flight ability of infected honey bee foragers [68, 69] and as such, the value of the nectar and pollen resources will depend on the status of the colony . According to our results, colonies with more severe N. ceranae infection, with smaller colonies and brood area, had lower yields of honey, as reported previously for N. apis- infected colonies [12, 51].
Although there was no correlation between N. ceranae infection and variations in colony strength or increased winter mortality in several recent studies [26, 71, 72], in our experimental conditions this microsporidium was very pathogenic to honey bee colonies, as shown by its impact on population size, the amount of brood and honey production. These conflicting findings may reflect differences in the experimental procedures used, as different methods and time points were used when measuring infection, colony strength and other parameters related to colony health. Accordingly, it may be necessary to standardise the procedures to evaluate Nosema infection in honey bee colonies in order to accurately compare the results from different studies and regions, and to better understand the differences in the epidemiology and pathology of this microsporidium worldwide.
The treatment used against Nosema in the present study (Fumidil B®) showed to be successful at temporarily reducing N. ceranae infection levels in the colonies, as demonstrated previously [17, 73], and allowed us to compare the health status of strongly and mildly infected colonies.
Collapsed colonies (2 in group 4 T, 3 in group CS and 3 in group C) exhibited severe N. ceranae infection prior to collapse, consistent with previous reports [17, 74]. The collapse of treated colonies in group 4 T may have occurred due to poor fumagillin consumption and hence, limited efficacy . Alternatively, in-hive conditions in this group may have been severely disturbed due to the administration of fumagillin in the winter time (unlike groups 1 T and 2 T), triggering a deadly infection by this microsporidium. Otherwise, since this group received the treatment more frequently, a negative impact of this molecule over colony health may have occurred , but all these suggestions should be studied more. One of the collapsed colonies of group 4 T was affected by chalkbrood disease, while another colony from the CS group that died exhibited symptoms of foulbrood disease. N. ceranae infection in these colonies may have provoked the outbreaks of stress-related diseases such as chalkbrood  and foulbrood disease. However, the surviving colonies in the CS and C control groups presented high levels of N. ceranae infection throughout the study, remaining alive but undergoing the effects of a chronic disease, as demonstrated by the reduced levels of fitness components such as colony size, brood rearing and honey production.
Despite the remarkable impact of Nosema infection on beekeeping economics [11, 12, 29], this effect is often underestimated by beekeepers . However, in our experimental conditions N. ceranae infection was highly pathogenic to honey bee colonies, as witnessed by sub-clinical signs such as significant decreases in colony size, brood rearing capacity, honey production and, as reported previously [17, 74], a decrease in the rate of survival. The study revealed that the control of Nosema in autumn and spring could be enough to mitigate these negative effects of nosemosis type C on colony health and productivity, which in turn may affect beekeeping profitability and have dramatic consequences on crop pollination and in natural ecosystems. As treatments for Nosema are currently unavailable in many countries, further studies of potential treatments or beekeeping techniques are urgently required to combat the rapid spread of this dangerous emerging disease.