The Manú Gradient as a study system for bird pollination

Abstract Background This study establishes an altiudinal gradient, spanning from the highland Andes (2400 m) to lowland Amazon, as a productive region for the study of bird pollination in Southeastern Peru. The 'Manú Gradient' has a rich history of ornithological research, the published data and resources from which lay the groundwork for analyses of plant-bird interactions. In this preliminary expedition we documented 44 plants exhibting aspects of the bird pollination syndrome, and made field observations of hummingbird visits at three sites spanning the Manú Gradient: 2800 m (Wayqecha), 1400 m (San Pedro), and 400 m (Pantiacolla). Some of the documented plant taxa are underrepresented in the bird pollination literature and could be promising avenues for future analyses of their pollination biology. The Manú Gradient is currently the focus of a concerted, international effort to describe and study the birds in the region; we propose that this region of Southeastern Peru is a productive and perhaps underestimated system to gain insight into the ecology and evolution of bird pollination. New information Observations were made on 11, 19, and 14 putatively bird pollinated plant species found at the high-, mid- and low-elevation sites along the gradient, respectively. Hummingbirds visited 18 of these plant species, with some plant species being visited by multiple hummingbird species or the same hummingbird species on differing occasions. Morphometric data is presented for putatively bird-pollinated plants, along with bill measurements from hummingbirds captured at each of three sites. Voucher specimens from this study are deposited in the herbaria of the Universidad Nacional de Agraria de La Molina (MOL), Peru and the University of British Columbia (UBC), Canada. The specimens collected represent a ‘snapshot’ of the diversity of bird-pollinated flora as observed over 10 day sampling windows (per site) during the breeding season for hummingbirds of Manú .


Introduction
Manú National Park is a UNESCO Biosphere Reserve nested within the most biodiverse region in the world: the tropical Andes (Myers et al. 2000). Manú Park, and its surrounding forests encompass a remarkable elevational gradient (hereafter the 'Manú Gradient') of over 3000 m, reaching from the lowland Amazon rainforest to the Puna grasslands of the high Andes. The Manú Gradient has a rich history of ornithological research (discussed in Walker et al. 2006), and over the last decade the Manú Gradient has been the focus of numerous ornithological studies as part of the Manú Bird Project (e.g. Merkord 2010, Jankowski et al. 2012a, Jankowski et al. 2012b, Londoño et al. 2014, Londoño et al. 2016, Dehling et al. 2014, Munoz 2016. Along the gradient, tree composition and forest structure have also been described (e.g. Jankowski et al. 2012b, Hillyer and Silman 2010. The wide interest in the avian community of Manú make it an ideal system for studying hummingbird pollination: population structure, range limits, and locations of uncommon and understudied hummingbirds are described and published. For example, focused studies of the high elevation Shining Sunbeam (Aglaeactis cupripennis) have demonstrated the effectiveness of the Manú Gradient as a study system for bird pollination (Hazlehurst et al. 2016, Hazlehurst andKarubian 2016). Therefore, the objectives of this study were to, 1) document the occurrence of putatively bird pollinated plants with voucher specimens along the Manú Gradient, 2) describe the occurrence and diversity of hummingbirds using mist-net surveys, and 3) record hummingbird visitations to flowering plants.

Site Selection
We surveyed three field sites spanning an altitudinal gradient of 2400 m (400 m to 2800 m) in the southeastern Andes (Table 5): La Estación Biológica Wayqecha (Paucartambo Province, Cuzco Region, 2800 m), San Pedro (Paucartambo Province, Cuzco Region, 1400 m), and Pantiacolla (Manú Province, Madre de Dios Region, 400 m). This area is one of the most biologically rich regions in the world with an estimated species pool of nearly 1100 birds (Walker et al. 2006). To our knowledge, a comprehensive survey of the vascular plants of the region does not exist, although an increasing number of plant identification resources for this region are being made available by the Field Museum of Natural History ( http://fieldguides.fieldmuseum.org). Wayqecha is characterized as high elevation cloud forest, with a mosaic of mature forest and areas with shorter trees and woody shrubs that transitions into puna grassland above treeline. San Pedro is predominately mid-montane humid rainforest, but also includes the lower extent of the montane cloud forest. Pantiacolla is situated at the interface between the Andean foothills and the lowland Amazon. Detailed environmental characteristics for these sites have been summarized in . Sampling was carried out between September 4, 2016 and October 13, 2016, falling within the avian montane breeding season. A distinct rainy season occurs from November through April and a dry season from May through August. Annual precipitation for higher elevations (2700-3000 m) ranges from 1700-2000 mm ) and is generally >2000 mm for lowland (100-400 m elevation) sites (Rapp and Silman 2012). Time constraints afforded less than two weeks (10.3 ± 2.1 days) for botanical and avian sampling at each site.

Data Collection
Pre-cut singletrack trails were used to access sampling areas away from the Manú Road (main access road that runs along the southeastern border of Manú National Park). We sampled hummingbirds using standard (12 x 3 m, 34 mm mesh) mist-nets along trail systems only. Mist-netting sites were sampled during the primary breeding season (August-November) for two consecutive days from approximately 0600-1200 hrs during suitable weather conditions (i.e., no periods of extended heavy rain, high winds, or other situations that could compromise researcher or bird safety). Each site consisted of an array of ten to fifteen nets placed in forested and open habitat and spaced at intervals of 25-50 Table 5.
Summary information and site descriptions for three sampling points along the Manú Gradient.
m. Ten sites were sampled at Wayqecha and San Pedro, and 8 sites were sampled at Pantiacolla. Hummingbird bill length was measured from the bill tip to the nares. Bill width was measured from the anterior edge of the nares. All captured hummingbirds were marked by cutting the terminal 1-2 cm of one rectrix to avoid resampling of individuals.
Both trails and the Manú Road were used to opportunistically collect plants. Plants were considered putatively bird pollinated if they met criteria adhering to typical bird pollination 'syndromes'; namely, dilute nectar and long tubular flowers (Fenster et al. 2004), though we acknowledge the limitations of surveying by these critera (Ollerton et al. 2009). Plants of interest were photographed, their location marked using a hand-held Garmin 64s global positioning system, and a description of the immediate habitat recorded. We then
Diversity of plants exhibiting the bird pollination syndrome does not differ across the gradient in the time frame sampled (Table 4).  Table 1.  Along the gradient, putatively bird pollinated plants were generally characterized by long corollas and were predominantly coloured red, yellow, orange, or some combination thereof. Previous documentation of bird pollination exists for each of the 16 families collected (Cronk andOjeda 2008, Johnson andNicolson 2008), but undocumented species-level bird pollination systems may arise from focusing on lesser-stuided taxa (e.g. Thyrsacanthus, Pentagonia, Pachystachys). Many putatively bird pollinated plants contained too little nectar to effectively measure sugar concentration at the time of sampling. We suspect that early morning visitations by nectarivorous birds and insects (i.e., both pollinators and nectar robbers) influenced this outcome. Indeed, in some cases inspection of certain plants revealed that the flower had been recently robbed as indicated by punctures at the base of the corolla. In as little as bird pollination has been studied along the Manú Gradient, even less is known of the ecological and evolutionary dynamics of nectary robbery. As this survey was preliminary, time did not allow for multi-day sampling at one locale to isolate nectar. A focus on a specific plant taxon would allow familiarity for nectar phenology and hence, more effective collection of nectar.
We recorded 23 independent visits by hummingbirds to 19 different plant taxa over 33 days. These observations by no means represent a comprehensive list of the total diversity for hummingbirds (Walker et al. 2006), bird-pollinated plants, or the interactions between a b Figure 8.
a: Blue-fronted Lancebill (Doryfera johannae) b: White-necked Jacobin (Florisuga mellivora) these two groups. An estimate of total diversity will come only with an extended sampling effort at each site. Relatively few hummingbirds were captured or observed in the lowlands (Table 3) compared to the other two sites. It is likely that this resulted from differences in foraging behavior between hummingbird species, rather than local abundance. For example, in the lowlands, a higher proportion of hummingbirds (e.g. Phaethornis) exhibit traplining behaviour (i.e. repeated visits along a route of flowering locations) compared to territorial guarding of floral resources. In addition, because of the higher canopy, many of the trees, lianas, and epiphytes inhabit canopy heights that are logistically difficult to sample.
The number of plants exhibiting bird pollination syndrome and number of bird visits observed are comparable between sites. That is, at a coarse scale we did not find any indication that elevation affects the absolute diversity of bird pollinated plant taxa (as expected by Cruden 1972), although the Manú Gradient would be an ideal location to test the hypothesis that bird and insect pollinated plants occupy distinct ecological niches. Between species, corolla length and width encompasses a great amount of variation, but hummingbird bill morphology varies less Tables 1, 3. This may speak to the adaptability of flowers relative to bills. It may be that because flowers serve a singular purpose (attraction and exclusion of pollinators and robbers, respectively), whereas bills have many uses (feeding, aggression, preening, balance), that bill evolution is relatively constrained. Bill morphology data will be used to inform phylogenetic tests of bill-flower shape evolution in future studies.
Evaluating the extent to which plants and their pollinators contribute to maintaining local biodiversity, and identifying keystone species within these systems (Ebenman and Jonsson 2005) will be important to maintaining ecological and cultural heritage in the Manú region (Ministerio del Ambiente (Ministry of Environment) 2017). This study provides a baseline for future work in pollination ecology along the Manú Gradient. Any one of the 44 plant species highlighted here warrants closer investigation, and we anticipate that further studies will help clarify the roles of hummingbirds as pollinators for the plant taxa described herein..