Nunney et al. previously proposed that IHR-generated genetic variation facilitated invasion of new hosts, based on the observation that all isolates from blueberry were recombinant-group X. fastidiosa subsp. multiplex. This hypothesis is further supported by the invasion of mulberry by the chimeric X. fastidiosa subsp. morus. These examples raise three additional points of support. First, given the long-term geographical association of the native X. fastidiosa subsp. multiplex with these 3 native host plants, the failure to infect them suggests that the genetic variation required for successful invasion had been absent from the native subspecies. Second, contact of these plant hosts with two newly introduced subspecies has failed to lead to infection of these plants; in all known cases of natural infection, these hosts were infected only by STs that had undergone large-scale IHR. Third, in each case, the STs found on these hosts show very little variation: blackberry, 1 ST; blueberry, 2 STs; and mulberry, 4 STs. This lack of within-host variation is consistent with host plants imposing strong host-specific selection on the bacterial genome. The data also suggest that host specificity is not determined by the lateral gene transfer of novel genetic material, since this would not impose the observed constraint on the genome. In addition, a similar pattern has been found in X. fastidiosa subsp. pauca in Brazil : evidence of large-scale IHR, combined with very limited genetic variation. From a sample of 55 citrus and 23 coffee isolates, only five STs were observed, with 85% of the citrus isolates having the same ST.

The data from X. fastidiosa show that massive recombination can occur between subspecies. We see this in the creation of X. fastidiosa subsp. morus,30 litre plant pots bulk and a similar event may have been involved in the genesis of the X. fastidiosa subsp. pauca strain that infects citrus and coffee in South America . But how did this happen? It has been established that conjugative plasmids can occur in X. fastidiosa , including a candidate found in the mulberry type . Furthermore, high rates of transformation have been observed in the laboratory . Which of these processes is involved in large-scale genomic exchange is not known. These data raise a second issue: how, given the clear potential for genetic exchange, X. fastidiosa subsp. morus and also the ancestral X. fastidiosa subsp. multiplex and X. fastidiosa subsp. fastidiosa strains have not introgressed into an ill-defined network of isolates. There are two main, nonexclusive hypotheses that might explain how these taxa have remained distinct: “opportunity” and “host selection.” The opportunity hypothesis is based on the distinct and almost completely non-overlapping range of plant hosts of the subspecies , which could severely limit contact between them and hence limit opportunity for IHR. This hypothesis is strengthened if it could be established that genetic exchange typically occurs in the plant host. On the other hand, the opportunity hypothesis would be weakened if genetic exchange typically occurs in the insect vector, since different subspecies can colonize the same insect . The host selection hypothesis proposes that different plant hosts impose strong host-specific selection such that, even if IHR occurs relatively frequently, most of the bacteria resulting from such exchange are maladapted and do not survive. Even moderate levels of recombination would be expected to generate high levels of genetic variability; however, very little genetic variability was observed within the mulberry type despite evidence of large-scale IHR and a broad geographical occurrence within the United States.

This near monomorphism of the mulberry-type isolates suggests that plant host specialization places severe constraints on the genome; i.e., the shift to the new host seems to have eliminated all but a narrowly defined set of genotypes. If the host shift had been due to some other genetic change, such as the acquisition of new extrachromosomal genes,then these genes would be expected to be seen in a number of different genetic backgrounds, which they are not. Thus, in summary, X. fastidiosa subsp. morus provides an important example for understanding the role of homologous recombination in bacterial adaptive evolution. We have been able to associate a clear ecological shift with a high level of recombination. But we are left with a puzzle. The data are consistent with X. fastidiosa subsp. morus and the recombinant-group X. fastidiosa subsp. multiplex originating from a single large-scale IHR, with no unambiguous evidence of any similar events involving the strains of X. fastidiosa subsp. fastidiosa currently found in the United States. Was this initial event a conjugation, followed by DNA fragmentation within the bacterial cell which resulted in large-scale recombination, or was it associated with a period during which conditions promoted a high rate of transformation, conditions that no longer prevail or occur only rarely? At present, it is far from clear if one or both of these possibilities could account for the pattern of evolution illustrated in Fig. 2.Understanding the relationship between species diversity and ecosystem functioning is a key issue given the global decline in biodiversity . Ecosystem functions such as nutrient cycling, soil formation, and pollination are crucial to environmental stability so an understanding of how and why these functions are related to species diversity will help to predict the broader consequences of species losses . Complementarity is niche differentiation by species/taxa which increases the efficiency of resource use. Large overlap between niches can indicate functional redundancy in a system, that is different species/taxa are doing similar things.

The functional redundancy and complementarity of species has been widely discussed, as it has implications for ecosystem functioning and prioritizing species conservation . There are several examples from studies of plants that show complementarity can contribute to a positive relationship between diversity and functioning . However, little is known about the role of complementarity in ecosystem functions mediated by organisms such as pollinators. Ecosystem functions can translate into short- or long term ecological or economic benefits to humans and in such cases are referred to as ecosystem services. Pollination is an ecosystem service crucial for wild plant reproduction , food production , and human nutrition , with bees being the main service provider . Complementarity is thought to play an important role in pollination service. With greater pollinator diversity and therefore, potentially greater complementarity, an increase in pollination service and therefore, fruit set may result. Pollination success in coffee was found to be positively correlated with pollinator functional group richness . In addition, pollinator functional diversity explained more of the variance in the seed set of pumpkin than species richness . However, as yet there are only a few studies on complementarity in pollination function and, to our knowledge, no data on how spatial complementarity of pollinator communities interacts with environmental change.Diversity in an ecosystem may appear redundant under a particular set of environmental conditions or at a given time. However, different species may not respond equally or in the same way to environmental changes. The diversity of what appear to be functionally similar species under one set of environmental conditions may buffer ecosystem function against fluctuations in these conditions, a condition known as response diversity . It has been observed that some non-Apis bees such as bumble bees and Osmia cornuta are more able to forage under inclement weather conditions than honey bees . For example,wholesale plant containers in apple orchards O. cornuta and muscoid flies were observed foraging under light rain when honey bees were not active and O. cornuta was the only pollinator species observed foraging in the orchards under high wind speeds.Such complementarity could be an extremely important mechanism for ensuring stable crop production. Agriculture has become increasingly pollinator dependent and recent findings of declines in both wild and managed bees have raised concerns about the potential impact on pollination services . For a large number of crop species, pollination is provided by honey bees , but there are many examples of crop species for which non-Apis pollinator species are more effective for fruit set on a per visit basis , coffee , and blueberry.Almond is a mass flowering crop, which requires biotic pollination and flowers early in the year when high wind speeds, low temperatures, and precipitation are common.In 23 almond orchards, the percentage fruit set was positively associated with the richness of flower visitors in the orchard and the species richness of wild bees .In this study, we investigated complementarity in almond, as a potential mechanism for this positive diversity-function relationship. Using the same 23 almond orchards, we investigated whether wild flower visitors showed spatial complementarity with honey bees and how spatial complementarity altered under changing environmental conditions . Our aims were to explore if different flower visitor taxa share or partition spatial niches at the tree scale; if flower visitor taxa show differential abilities to forage at high wind speeds and if the change in environmental conditions causes those taxa that forage at high wind speeds to change their spatial niche. Information from our study is important for predicting the consequences of functional pollinator diversity loss in a changing world.

The observations in the 23 orchards in 2008 under low wind speeds were used to investigate the foraging location of flower visitors within the trees, as all orchards had been sampled equally. The flower visitor community was divided into four functional taxa . The frequencies of visits by each of the taxa were analyzed in separate models. Due to a large number of zeros, data were summed for each observation day across trees at the edge and trees in the interior of each orchard. The explanatory variables were the location within the tree and the wind speed . The number of flowers observed was included as an offset as it was a covariate known to affect the flower visit counts. The random variables were the location , nested within the observation day, nested within the orchard. For the wild bees, hover flies, and all-others the error distribution was Poisson. For honey bees, the error distribution was log normal Poisson with a subject level random variable to account for over dispersion . For all models, stepwise deletion was carried out . After the removal of an explanatory variable, the models with and without the variable were compared by analysis of variance to test the loss of explanatory power from the removal of the variable . When there was no significant difference between the models, the explanatory variable was removed.Data were collected under high wind speeds in four orchards in 2008, 2009, and 2010 . These data were analyzed with the data collected at the orchard edge in the same four orchards under low wind speeds in 2008. To isolate the impact of wind speed from other environmental variables such as temperature, we conducted observations on days with high wind speeds and sunny conditions. High wind data were collected over 3 years as windy days were often also cooler and rainy; with windy, sunny days being rarer. In 2008, observations of flower visits were carried out over three separate days, per orchard and wind category. The data were summed across all five trees at the orchard edge observed in a day, in an orchard. The frequency of flower visits was the response variable in a mixed model with a log normal Poisson error distribution. The orchard’s pollinator diversity category, wind speed, and their interaction were included as explanatory variables. The wind speed was calculated as the average of the start and end wind speed of the observation period. Year was also included as an explanatory variable and the number of flowers observed as an offset . Only the observations in the two high pollinator diversity orchards were selected to analyze the effect of wind speed on the frequency of flower visits by each taxa. The high wind observations in 2008, 2009, and 2010 and the low wind observations from the same orchards in 2008, at the orchard edge only were analyzed. The data were summed across the five trees observed on each day, in each orchard. The number of flower visits recorded by each taxa was the response variable. The explanatory variables were the wind speed, the year and an offset of the number of flowers observed . Observation day nested within the orchard and a subject level random variable were included. A Poisson error distribution was selected.