Frontier- Costa Rica Forest Programme

Transcription

1 Frontier- Costa Rica Forest Programme Phase 132 April June 2013 First Neotropical river otter (Lontra longicaudis) captured on camera trap as part of the first abundance estimation study in Costa Rica, and to our knowledge, in the world. Principal Author: Nathan Roberts (PI) Managing Director: Eibleis Fanning Co-authors: Helen Walters (ARO), Anik Levac (ARO), Kirsty Fraser (CA), Jessica Carpani (CA) Emma Gardner (RDI)

2 Frontier Science Staff Name Eleanor Keymer (EK) Nathan Roberts (NR) Helen Walters (HW) Anik Levac (AL) Kirsty Fraser (KF) Jessica Carpani (JC) Emma Gardner (RDI) Position Project Coordinator (PC) Principal Investigator (PI) Assistant Research Officer (ARO) Assistant Research Officer (ARO) Conservation Apprentice (CA) Conservation Apprentice (CA) Research and Development Intern (RDI)

3 Contents 1. Introduction Area overview and programme rationale Site and survey location with maps Frontier-Costa Rica Forest Research Programme 8 2. Aims Methodology Briefing Sessions Science lectures Field training BTEC Turtles Introduction Methods Results Discussion Primates: Population density and habitat associations of the four primate species present in Costa Rica: a case study on the Osa Peninsula Introduction Methods Results Discussion Otters Methods for monitoring the population status of the Neotropical river otter (Lontra longicaudis): calculating density from camera trap data and using indirect signs as an index of abundance Introduction Methodology Results Discussion Temporal variation in the spatial use of a river network on the Osa Peninsula, Costa Rica: latrine re-visitation and identifying factors affecting space use by the Neotropical river otter (Lontra longicaudis) Introduction Methodology 29

4 Results Discussion Is it possible to identify individual Neotropical river otters and apply capture-recapture methods to camera trap data to reliably estimate abundance? A controlled case study Introduction Methodology Results Discussion Birds: Species diversity and habitat specialisation of bird communities within primary and secondary forest and in a river course on the Osa Peninsula, Costa Rica Introduction Methods Results Discussion Butterflies: Species diversity of butterflies (Order: Lepidoptera) occupying different heights within the understorey of a secondary forest on the Osa Peninsula, Costa Rica Introduction Results Discussion Anurans: Diversity and abundance of leaf litter frogs on a forest trail in primary and secondary forest on the Osa Peninsula, Costa Rica: baseline data for a long-term monitoring programme of population responses to climate change Introduction Methodology Results Discussion Riverside Wrens: Documenting the vocal repertoire of the Riverside Wren (Cantorchilus semibadius) Introduction Ecology, life history, and conservation status of the Riverside Wren Bird songs and calls Duetting in Riverside Wrens Birds duets and conservation Vocal individuality 56

5 Population models and identifiable individuals Behavioural traits Ethical consideration Hypothesis Methodology Results Discussion 62 References 63 Appendices 79

6 1. Introduction 1.1. Area overview and programme rationale Costa Rica, a small Central American country covering approximately 51,000km², is situated between Panama and Nicaragua creating a land bridge connecting the North and South American continents (Figure 1.1). It lies about halfway between the Tropic of Cancer and the equator with an average annual temperature of approximately 27 C, which results in the average temperature of the hottest month of a year not exceeding the average temperature in the coldest month by more than 5 C (Leenders 2001). This small variance means that seasons in the tropics are not defined by temperature but by amount of rainfall. In Costa Rica the dry season generally starts in November until April or May where the rainy season begins. The geographical location combined with this tropical climate creates a biological mixing bowl of rich biodiversity where there are approximately 505,000 species of plants and animals, from jaguars, sloths, deer and tapirs to hummingbirds, armadillos, parrots and opossums (Henderson 2002). Figure 1.1 Location of Costa Rica and Osa Peninsula (marked A). This extremely biodiverse country was revolutionised by conservation and habitat preservation into one of the most successful nature tourism industries in the Western Hemisphere (Henderson 2002). One of the leading countries in environmental sustainability, ranking fifth in the Environmental Performance Index 2012, Costa Rica has taken a huge turn from the major effects of wildlife exploitation that occurred up until the 1960s. Tourists 6

7 would collect boa constrictor hides, caiman-skin briefcases, skins of spotted cats and sea turtle eggs as souvenirs from their travels in this tropical paradise (Henderson 2002). After the 1960s the process of wildlife conservation began through multiple fields of research, education, preservation, conservation and nature tourism where it remains today. Today Costa Rica contains 34 protected areas including 28 natural parks, a total of 1,415,000 acres, covering over 11% of the country s land. This huge area of protected land is justified when taking into account that an estimated 5% of all living species are found in this small tropical country even though it covers only 0.01% of the earth s landmass (Santchez-Azofeifa et al. 2002). However deforestation and habitat fragmentation outside of the country s protected areas and national parks is still a large problem because of the rising economic pressure; therefore, it remains an important area to survey and monitor Site and survey location with maps In a remote and lush corner of southern Costa Rica lies a realm of giant trees, potbellied spider monkeys, harpy eagles, prowling jaguars and herds of white-lipped peccary. This is the Osa Peninsula and there is no place in the world like it. (Larson 2010) This is how ecologist Dr Trond Larson chose to portray the biodiverse Osa Peninsula which is where our scientific research and monitoring programme is based. The Osa Peninsula covers an area of 1093km² on the South Pacific slopes of Costa Rica (Figure 1). It is a unique ecosystem in terms of endemic species and rich biodiversity making it an ideal location for scientific research. It consists of Tropical Wet, Premontane Wet and Tropical Moist forest types. Mean annual precipitations average around 5500mm and mean annual temperature around 27 C with elevations ranging between 200 and 760m (Santchez- Azofeifa et al. 2002). The peninsula separates the waters of the Golfo Dulce from those of the Pacific Ocean (Figure 1). The lack of mixing between these two bodies of water results in anoxic conditions in the depths of the Golfo Dulce basin (Hebbeln et al. 1996). This creates a tropical fjord which exceeds 200 metres in depth one of only four in the world like this (Hebbeln et al. 1996). The fjords are extremely rich in biodiversity and so unique that there are many endemic species solely inhabiting the areas around the gulf. Geologically, the peninsula is comprised of sedimentary material dating back to the Cretaceous period (Bundschuh & Alvarado 2007). Due to its diverse topography and having two major wind directions, the peninsula has a variable climate, with an average annual rainfall of 5500 mm, dropping to only 100 mm per month in the dry season (Cleveland et al. 2010). Humidity levels are high, almost never falling below 90%. The Osa Peninsula has a population of around 12,000 people mainly dispersed in small scattered settlements (Carrillo et al. 2000). The major sources of income in the region are agriculture (rice, bananas, beans and corn), livestock (cattle), gold mining, logging and, more recently, the expanding eco-tourism industry (Carrillo et al. 2000). However, with the Costa Rican population increasing at a rate of 2.6% annually, there are growing concerns for the 7

8 sustainability of the country s environmental resource demands (Sánchez-Azofeifa et al. 2001). Deforestation has been prevalent in Costa Rica, particularly during the 1970s and 1980s, resulting in high levels of forest fragmentation. Previous studies report that 21.5% of Costa Rica s forest cover was deforested between 1986 and 1991 (Sánchez-Azofeifa et al. 2001). This has had devastating environmental and economic impacts, and several reforestation programmes have since been implemented (Sánchez-Azofeifa et al. 2001). Costa Rica has a reputation for success in balancing conservation with economic improvement, with almost a quarter of Costa Rican land having been set aside by the government for conservation purposes (Sánchez-Azofeifa et al. 2001). Over the last two decades, there has been a dramatic rise in the number of hospitality businesses opening along the road from Puerto Jimenez to Carate as a result of the growing popularity of ecotourism since the 1990s (Minca & Linda 2002). Despite representing approximately 0.25% of the world s land mass, Costa Rica is exceptionally rich in faunal and floral biodiversity containing 4-5% of all plant and bird species (Sánchez-Azofeifa et al. 2001). There is a great scope for scientific research in the country as it is estimated that 84% of the 505,000 species present in Costa Rica are unknown to science (Sánchez-Azofeifa et al. 2001). Furthermore, Costa Rica has a wealth of endemic species which evolved largely as a consequence of the Talamanca Mountains which represent a geographical barrier, ultimately leading to genetic drift and speciation (Henderson 2010) Frontier-Costa Rica Forest Research Programme Frontier s forest research programme in Costa Rica began in July 2009, in collaboration with the local organisation Osa Conservation at the Osa Biodiversity Centre (OBC) (N , W ) (Figure 2). The long-term objective of the project is to investigate the effects of climate change on the terrestrial communities of Costa Rica and its impact on the country s network of protected areas. The current research focuses on very humid tropical forests and species that are particularly sensitive to climate change, such as turtles and amphibians. The site is located within a 1700 ha private forest reserve. The reserve is managed by the administrative unit of ACOSA (Osa Conservation Area) within the National System of Conservation Areas (SINAC). SINAC was established in 1995 with the mission to promote the management and conservation of natural resources, and has since established 11 Conservation Areas in Costa Rica (Sánchez-Azofeifa et al. 2003). The site is a prime location for carrying out both forest and shoreline surveys, as there is relatively easy access to both the primary and secondary forest as well as the pristine beach habitat. Furthermore, the site neighbours Corcovado National Park which is renowned for its high levels of biodiversity. 8

9 (i) (ii) Figure 1.2 i) Osa Peninsula, and; ii) the location of the Frontier Costa Rica Forest (CRF) camp and Osa Biodiversity Centre (OBC). 9

10 2. Aims The projects undertaken in phase 132 and their respective aims and objectives are: 1. Sea turtle monitoring programme continues as usual in collaboration with Osa Conservation with morning only patrols on both Playa Piro and Playa Pejeperro to make sure all the data possible is collected. The nesting peak of the Olive Ridley turtle and Pacific Green has ended though surveying continues to provide yearround data of turtle nesting activity on the local beaches. 2. Primate data is now being recorded on the trails within Osa Conservation s property in primary and secondary forest aiming to estimate species abundance of the populations of all four primate species. 3. The first of the otter projects is new for this phase, aiming to use camera traps and novel analytical techniques to provide the first abundance estimate for a population of this species in Costa Rica, if not the world. In addition, the study aims to provide a trial methodology for calculating an index of abundance by recording tracks and signs that may be used in the long term monitoring of the population. 4. The second of the Neotropical river otter project concerns the spatial ecology and the study aims to map the areas being used by the otters at different times of the year by non-invasive sampling. 5. The final otter project is an ex situ study of the species, aiming to test the reliability of identifying individuals of the species from camera trap images. The results of this study will aid in the standardisation of field techniques. 6. Bird surveys are being conducted as point counts within different habitat types including primary and secondary forest and within the course of a river. This allows us to investigate the diversity of the bird communities in these environments and to investigate and classify generalist and specialist species within the local area. 7. Canopy traps have been used to live-trap butterflies in an investigation of diversity within the understory of a secondary forest, as well as investigations of the heights at which butterflies utilise based on the capture success of traps set to different heights. 8. The diversity of leaf litter frogs is being investigated through hand capture techniques on a forest trail. The aim is to provide baseline data for a long term monitoring programme. 9. The final research project included in this report is a behavioural ecology study completed as part of HW s Master s thesis which aims to document the vocal repertoire of the Riverside Wren. 10

11 3. Methodology 3.1 Briefing Sessions In order to achieve the above aims, volunteer Research Assistants (RAs) received a number of briefing sessions (Table 3.1), science lectures (Table 3.2) and field training (Table 3.3). In addition, a number of RAs undertook a BTEC in Tropical Habitat Conservation this phase (Table 3.4). Table 3.1 RA Briefings LECTURES Welcome to Frontier Costa Rica Staff Introductions Health and Safety Talk and Tests LECTURER EK, NR EK, NR EK 3.2 Science lectures Table 3.2 RA Lectures LECTURES Introduction to the Science Programme Introduction to the BTEC / CoPE Introduction to Conservation Biology: What is it and what can I do for conservation? LECTURER NR NR NR 3.3 Field training Table 3.3 RA Training FIELD TRAINING Training in turtle monitoring and data collection Training in primate survey techniques Training in butterfly survey techniques Training in camera-trapping methods and setup Training in bird identification Training in anuran survey techniques BTEC Mentor Meetings TRAINER NR, HW NR, EK, HW, KF, AL NR, EK, HW, KF, AL NR AL, EK NR, EK, HW, KF, AL NR, HW, KF, AL 11

12 Training in field applications of GPS and GIS Training in statistical analyses using EstimateS and DIVERSITY Training in sound recording and statistical analyses using Syrinx NR NR HW 3.4 BTEC Table 3.4 BTEC projects conducted during CRF132 BTEC PROJECT TITLE John S. Web characteristics of the Golden orb-weaving spider (Nephila clavipes) and relationships with female body size. (10 weeks) Warbutton C. The diversity of birds at different canopy heights. (4 weeks) 12

13 4 Turtles 4.1 Introduction Sea turtles are both a flagship species for conservation due to their iconic nature, and an excellent indicator species for climate change due to their temperature-dependant sex determination (Janzen 1994). Their late maturation, in conjunction with anthropogenic threats such as beach development, long line fishing and pollution, mean that turtle populations are highly vulnerable and often unstable. Furthermore, climate change threatens to add to the stresses in several ways. Like many ectothermic reptiles, marine turtles exhibit temperature dependent sex determination whereby increased temperatures create a sex bias in favour of females, which could cause entire populations to collapse. Temperature induced changes in plant community composition together with rising sea levels may result in increased incidences of beach erosion and inundation of nests (Janzen 1994). Poaching and the illegal trade of turtle eggs cause further reductions to turtle populations, which may result in entire clutches being eradicated. Turtle hunting was at its worse from the 1950s until the 1970s, with half of the world s turtle catches being made in Mexico where as many as 350, 000 turtles were harvested annually (Marquez et al. 1996). Additional threats to sea turtles include fishing techniques using drift nets or long lines, predation by animals, and habitat loss which often results from increased tourism (Govan 1998). On the beaches of the Osa Peninsula, turtles are threatened predominantly by predation (dogs, coatis and humans) and land development (Drake 1996). Turtle conservation involves an array of strategies, such as patrolling of nesting beaches, preventing trade of turtle-based products, providing alternative fishing methods and sources of income for fishermen and poachers, planting shade trees, relocating nests, and educating locals. It was only relatively recently in 1963 that the license issued by the Costa Rican government permitting the collection of turtles and their eggs was ceased (Troëng & Rankin 2004). Despite this legal protection, illegal trade of turtle eggs and, to a lesser extent, meat trade, still continue to threaten turtle populations. The popularity for ecotourism in Costa Rica has, however, resulted in dramatic increases in turtles nesting on certain beaches, such as those at Tortuguero, which rose by 400% between 1971 and 2003 (WWF 2010). There are two species of marine turtle known to nest frequently on Playa Piro and Playa Pejeperro: the Olive Ridley turtle (Lepidochelys olivacea) and the Pacific Black turtle (Chelonia mydas agassizii). Leatherbacks (Dermochelys coriacea) and Hawksbill turtles (Eretmochelys imbricate) have been sighted here however nest less frequently on these beaches. The Olive Ridley turtle is the more commonly sighted of the two species, listed as Vulnerable, whereas the Black turtle is listed as Globally Endangered (IUCN Red List 2008). Pacific Blacks come ashore to nest between November until February, whereas Olive Ridley 13

14 turtles tend to nest from July until September; in some areas nesting together in their thousands in events known as arribadas (Gaos et al. 2006; Savage 2002). In 2003, Osa Conservation began patrolling the beaches of the Osa Peninsula, aiming to deter poachers and predators, and they are now collaborating with Frontier to assist in these patrols. These patrols cover approximately the 8km stretch of Playa Piro and Playa Pejeperro, and by monitoring the turtles they also created a valuable database. In addition, data was gathered relating to hatchling success, predation levels and methods of protection, location of the nesting sites and the size, health and species of turtle, along with flipper tagging individuals. Although they are the most abundant marine turtle species, L. olivacea are still of conservation concern and are currently listed as Vulnerable (Honarvar et al. 2008, IUCN Red List of Threatened Species, ). Olive Ridleys feed predominantly on fish, shrimp, crabs, squid, mussels, and clams (Spotila 2004). On average, a single adult female Olive Ridley lays around 105 eggs per nest, and will nest 2.5 times per season (IUCN ). Females reach sexual maturity between 10 and 18 years (IUCN ). Their eggs have an incubation period of days and adult turtles weigh on average 40kg (Sea Turtle Survival League). Survivorship rates beyond hatching are unknown for this species, but it is thought that, like other sea turtles, mortality rates are high during the early life stages (IUCN, ). Black turtles have experienced significant population declines worldwide and have been subsequently classed as globally Endangered (Trӧeng & Rankin 2005, IUCN, ). The numbers of females nesting annually, which are the only indicator of sea turtle population values, are believed to have declined by % worldwide over the past three generations (IUCN ). In Pre-Columbian times it is estimated that C. mydas agassizii numbered between million worldwide (Jackson 1997). Black turtles can take up to 26 years to reach maturity (Frazer & Ladner 1986). Green turtles are slightly larger than Olive Ridleys, with an average carapace length of cm and average weight of kg. Their diet is mainly herbivorous, mainly feeding on sea grasses, but they also eat fish and invertebrates. A number of conservation strategies have been established throughout Costa Rica. Some have used the approach of allowing limited egg collections, such as the Wildlife Conservation Law, 6919, whereby commercial egg harvesting was permitted on a nesting beach in Ostional, Costa Rica during the first 36 hours of wet season arribadas (Campbell 1998). Similarly, in Limόn, Cost Rica, fishermen were granted an annual catch of 1,800 green turtles (Trӧeng & Rankin 2005). In other regions, such as Tortuguero, the Costa Rican government has made egg poaching illegal, in addition to prohibiting the trade of calipee (the edible part of the shell) (Government of Costa Rica 1963 and 1969). Meanwhile, the growing ecotourism industry in Costa Rica has provided locals with an alternative source of income, and has promoted conservation throughout the country. It is imperative that turtle 14

15 conservation programmes are long-term since it can take decades for populations of species with late maturity to show a population response (Trӧeng & Rankin 2005; Bjorndal et al. 1999). 4.2 Methods Unlike in peak turtle season when night and morning patrols are conducted on both Playa Piro and Playa Pejeperro virtually on a daily basis managed between Frontier and Osa Conservation, during this phase it has only been appropriate for Frontier to conduct one morning patrol on each beach weekly due to the low nesting frequency. Morning patrols typically commence at approximately 0630hrs on Playa Piro and 0430hrs on Pejeperro in the interest of minimising surveyor exposure to unavoidable direct sunlight and high temperatures. As female turtles are very sensitive to lights, surveyors used red lights during early hours on Pejeperro and the survey team was a maximum of six persons to minimise disturbance. Each beach is divided into 100m sectors which are marked by signs displayed on trees. Two kilometres of Playa Piro is sampled and 4.3km of Playa Pejeperro. For each encounter of turtle tracks the following data are recorded: yesterday s date (i.e., the date the turtle probably nested) and the time of recording; name of data recorder; sector in which tracks are encountered; whether the tracks are associated with an in situ (IS) nest or a false crawl (FC; turtle returns to sea without nesting); characterising the tracks as symmetrical (S) or asymmetrical (AS); three measurements of the track width at different points, and; distance to vegetation. Symmetrical tracks are measured at the widest point of the front flippers, and asymmetrical tracks are measured at the widest point of the back flippers. The track is then crossed through with a deep heel drag in the sand to avoid the track being recorded again in following patrols. Track characteristics are used as an indicator of species where the turtle is absent, with Olive Ridley turtles having a smaller track width and leaving asymmetrical tracks. If the turtle is present, the species and the distance between the nest and tide is recorded, and if the turtle is in the process of laying eggs a health assessment is undertaken assessing the condition of the turtle in terms of mating scars, number of barnacles present and taking measurements of the carapace. An in situ nest can be suggested by large amounts of sand spray caused by the digging turtle, and in the case of the Olive Ridley turtle, the nest is often crater-like with a flattened surface and raised edges. However, to confirm the presence of an egg chamber it may be necessary for a stick to be inserted into the sand at the suspected site and where a chamber exists, the stick will enter a pocket of air surrounding the eggs, indicated by a marked change in resistance. If no egg chamber exists, the sand would be of the same consistency and resistance at all depths and be challenging for the surveyor to push into the sand. A false crawl is defined by the absence of a nest or where it is clear that the turtle returned to 15

16 sea without digging a nest. If the nest has been predated (typically evident by the presence of tracks, egg shells and signs that the nest has been dug up), the predator is identified by the tracks and recorded. Unlike in phase 131, during this phase no tagging or triangulation was performed as neither are required by the programme led by Osa Conservation at this time of year (low turtle season). 4.3 Results Between 25 March and 05 June 2013, 47 and 66 tracks were recorded on Playa Piro and Playa Pejeperro, respectively (Table 4.1). On both beaches, tracks of Chelonia mydas agassiz were encountered more frequently than those of Lepidochelys olivacea, and the former had larger average track width and nests were typically closer to the vegetation. No turtles were encountered whilst laying eggs and zero health assessments were therefore performed, resulting in no data of carapace length as well as distance from tide. Furthermore, no predated nests were encountered. Table 4.1 Summary of all turtle data collected between 25 March 05 June LO: Lepidochelys olivacea; CM: Chelonia mydas agassiz; S: symmetrical tracks; AS: asymmetrical tracks. Measurement Piro- LO (AS) Piro- CM (S) Peje- LO (AS) Total no. turtles seen Total no. tracks (AS&S) Total no. IS nests Total no. FC nests Average carapace length (cm.) Average track width (cm) Average distance from vegetation (m.) Average distance from tide (m) Predated nests Peje- CM (S) 16

17 4.4 Discussion The greater number of tracks of Chelonia mydas agassiz than Lepidochelys olivacea encountered is expected as the peak season for the latter ended in October as opposed to March for the former. Furthermore, the L. olivacea season begins in August so it is anticipated that this pattern observed here and in the last phase will change with respect to this. The smaller track width of L. olivacea and tendency to nest further away from the vegetation also agrees with data from previous phases. The absence of predated nests in this phase extends the period without predation noted in the last phase, suggesting that the abandonment of the house at sector 24 on Playa Pejeperro where high levels of dog-caused predation occurred has had lasting effect on the significant reduction of nest predation. The absence of turtles encountered whilst nesting during this phase is attributed to the sampling strategy of the phase, i.e. conducting morning patrols only. As in the previous phase, the comparatively high number of false crawls in unexpected. Turtles high sensitivity to choosing nesting sites and low level of tolerance to disturbance are known to perturb nesting success, and with a far lower presence on the beach by surveyors than during peak season, it would suggest that perhaps other environmental and abiotic factors are prevalent in affecting nesting success rather than human-induced disturbance. Known factors include digging impediments (e.g., roots, rocks and debris) and unsatisfactory thermal conditions (Wang & Cheng 1999). To support this, numerous tracks led to abandoned nesting attempts due to the presence of roots or debris. It is difficult to draw any extensive conclusions on the data collected within this sample period and year to date. However, it is possible that due to its greater length Playa Pejeperro has more suitable nesting sites available than Playa Piro which is considerably shorter. However Playa Piro is wider and therefore more suited to L. olivacea because of the larger area of sand, whereas the narrower Playa Pejeperro has much more vegetation where C.m.agassiz prefer to nest near to or within. Other possible factors influencing their behaviour include beach topography, temperature and light. The turtles that did not lay eggs during this phase either made no nesting attempts and crawled back to sea, or dug one or more false nests before returning to sea without laying their eggs. The leatherback turtle (Dermochelys coriacea) nest that was encountered on Playa Pejeperro in phase 131 has not been excavated as no evidence of hatchling emergence at this sector has been encountered. In the next phase we will be collaborating with Osa Conservation to manage virtually daily morning and night patrols of both beaches and it is anticipated that tagging and triangulation will again commence as per previous years. 17

18 5 Primates: Population density and habitat associations of the four primate species present in Costa Rica: a case study on the Osa Peninsula. 5.1 Introduction There are four species of primate that occur in Costa Rica: the Central American squirrel monkey (Saimiri oerstedii), mantled howler monkey (Alouatta palliata), Geoffroy s spider monkey (Ateles geoffroyi) and white-faced capuchin monkey (Cebus capuchinus). All four species are present within the Osa Conservation property at the study site in Piro. The spider monkey is listed as Endangered and the squirrel monkey as Vulnerable (IUCN 2013). Frontier Costa Rica has been surveying the presence and absence of primate troops of all four species for the past year. The overall aim of this research is to provide an insight into the habitat use by primates in the area to better aid management and policy decisions at a local level. With a recent change in methodology, data collection and analysis is at an early stage. Objectives of the project include estimating the population density of each of the four species and the production of a map detailing the locations of troops of each species. This can then be used in a GIS to model population dynamics and ecological interactions, including niche and resource use overlap between the species of primate and the interspecific competition and coexistence between different primate species. As many of the primates are predominantly frugivorous, they perform important roles as seed dispersers and are therefore crucial in maintaining plant biodiversity in the forests of the Osa Peninsula. Geoffroy s spider monkey is listed as an endangered species according to the most recent assessment by the IUCN, whilst the Central American Squirrel monkey is categorised as Vulnerable (IUCN 2012). Population declines exceeding 50% over the past 45 years have been reported for the Geoffroy s spider monkey, principally driven by stresses induced by habitat loss (Cuarón et al. 2008). The Central American squirrel monkey has a limited distribution, restricted to Panama and Costa Rica. The limited range of the squirrel monkey means populations are vulnerable to stochastic events. The extent of land conversion for agriculture and development as well as clear cut and selective logging threatens this squirrel monkey further, with vast areas of habitat lost. The Central American squirrel monkey is also exploited for the pet trade (Cropp & Boinski 2000). Consequently, this primate species has suffered drastic population declines over recent years and has been listed in Appendix I of CITES, meaning that there are strict regulations and controls over the illegal trade of this primate (CITES 2010). While the mantled howler monkey and white-faced capuchin have not suffered the same population declines as experienced by the spider and squirrel monkeys they are of interest for their intrinsic value and from an interspecific competition perspective. The white-faced capuchin is a significant scientific interest as an example of the independent evolution of traits thought to be restricted to the great apes and humans (Sechrest et al. 2002), while the 18

19 howler monkeys serve as important seed dispersers in the forest as well as constituting a prey base for arboreal predators (Glander 1980). 5.2 Methods The data is collected using the line-transect sampling protocol (Thomas et al. 2009; Buckland et al. 2010). Primary data are: perpendicular distance to the centre of each group of individuals observed; the number of individuals in the troop; and their location which is recorded on a GPS. Additional data include the number of males, females, juveniles and dependents in each troop. Transects are made up of the trails on Osa Conservation s land at Piro, totalling 9.1km, excluding Las Rocas and the Northern border. Primary forest: Ajo, Ocelote, Cerro Osa, Los Higerones, Chiricarno Alegre. Secondary forest: Piro, Terciopelo, Beach Trail, and New Trail. The road between the intersection with Beach Trail and the westernmost end of New Trail will also be surveyed. Surveys will be conducted between 0600 and 1100hrs, and each trail will be completed a minimum of twice per month. Following the distance-sampling method, trails will be walked at a consistent speed of between 1.5-2km/hr and all encounters with primates will be recorded. In order to determine population density a minimum of 40 independent records must be collected per species (Buckland et al. 2010). Species density was estimated in DISTANCE (Thomas et al. 2006), pooling data across all trails. 5.3 Results From 04 March 2013 to 13 June primate surveys were conducted. Of these there were 17 samplings days where no observations were made (Figure 5.1). As demonstrated in Figure 5.2, the species most often encountered was the spider monkey, with 27 independent encounters recorded during this project phase. Squirrel monkeys were the least encountered species during this phase with only three records of observations. Capuchins and howlers were encountered five and eight times respectively. Average troop sizes were calculated for each of the species (Figure 5.3). The howler monkeys and squirrel monkeys were found to have the largest troop sizes each averaging six individuals. Capuchins and spider monkey troops averaged five and four individuals respectively. 19

20 Number of encounters Number of encounters CRF 132 Science Report /4/2013 4/4/2013 5/4/2013 6/4/2013 Sampling days Figure 5.1Total number of independent observations for each sampling day C.capuchinus A.palliata A.geoffroyi S.oerstedii Species Figure 5.2 Number of observations of each species during the sampling period. As depicted in Figure 5.4, the mantled howler was encountered three times along the Ajo transect, and once on Ajo-Ocelot, Cerro Osa, Beach Trail, Beach Trail-Road and New Trail- Road over the course of the last project phase. The white-faced capuchins were encountered five times over the last project phase, once on Beach Trail, and four times along New Trail-Road. Spider monkeys were encountered on every transect, except Beach Trail and Beach Trail- Road. They were most often encountered on New Trail-Road (six observations), then Ajo and Cerros Osa (five encounters on each trail), then Ajo-Ocelot and Chiricano Alegre (four encounters per trail). Squirrel monkeys were recorded only three times during this project phase and only on New Trail-Road. 20

21 Number of encounters Number of individuals in troop CRF 132 Science Report Capuchin Howler Spider Squirrel Species Figure 5.3 Average troop size for each species based on observations from the sampling period Howler oberservations Capuchin Observations Spider Observations Squirrel observations 0 Transect Figure 5.4 The number of each individual primate species encountered per transect in the sample period. Spider monkey population density was estimated using DISTANCE 6.0 and was 43.7 individuals/km². Thirty three observations were analysed across an area totalling 20.4km² (total transect area walked); the greatest perpendicular distance was 41.6m from the observers to the centre of the troop. Detection probability was 47% and the encounter rate was 53%. 21

22 5.4 Discussion Primate population monitoring is vital throughout the tropical biodiversity hotspots (Mittermeier 1987; Sechrest et al. 2002). Primates are ecological engineers, their role as seed dispersal agents is well documented (Mittermeier 1987; Stevenson 2001; Estrada et al. 2005) and the study of their behaviour and social structures has provided great insight into human evolution (Fuentes 1999; Perry et al. 2002, 2003; Sechrest 2002). Before more detailed questions regarding the primate populations at Piro can be answered it is important to gather robust baseline data: troop numbers, troop size, population density and habitat use. The results discussed above are the beginning of this process. However there is still a long way to go. The results for the remaining three primate species highlight issues with the data collection. Troop size records for the squirrel monkeys and white-faced capuchins are much lower than expected. These two species are the smallest of the primate species found in Costa Rica and are known to live in large groups, numbering anything up to 25 individuals. Based on the numbers we are recording there is a substantial number of individuals that are not being recorded. There are several reasons why this could be the case although the number one reason is observer competence. Studying wild primates is notoriously difficult. It takes a long time to develop the field skills required. The more cryptic species, howlers, capuchins and squirrel, are not being recorded as often as the spider monkeys as surveyors do not have the skills to recognise acoustic cues such as vocalizations or canopy disturbances that signal the presence of primates. This is the focus of a determined training effort on the part of field staff and it is hoped that the results of this training will be evident in the next phase of the project. The troop sizes for spider monkeys and howlers are being accurately recorded; the former are noted for the fission/fusion nature of their troops and so records of one to four individuals are expected. The latter are easy to observe once they have been encountered due to the slow pace of their movements. In terms of habitat preference the mantled howler is known to occur in both primary and secondary forest (Anzures-Dadda & Manson 2006; Arroyo Rodriguez 2010) and this would appear to be supported by our data, although much more data must be gathered before any conclusions concerning habitat preference can be made. The white-faced capuchins have been incidentally noted in both primary and secondary habitat (H.M.Walters, pers. Obs.) The data displayed here suggest that this species prefers secondary habitat. However, once again there is not enough data to draw solid conclusions and as discussed above observer biases may have skewed the data in favour of the less dense habitat type. The spider monkeys do not demonstrate a habitat preference and are present in both secondary and primary forest. They were the only species of primate recorded on Tercioplo Trail-Piro Trail during this project phase. Whilst we have the greatest number of records for the spider monkey it will only improve the data to continue adding to it. The lack of squirrel monkey data collected during this phase of the project is of concern due primarily to the 22

23 conservation status of the species. The three records that were gathered cannot be used to make any inferences about habitat preference beyond the casual observation that all three encounters were in secondary forest that this species is known to prefer (Boinski et al. 1998). We were keen to examine population density for the spider monkey populations and so we ran the data through the DISTANCE 6.0 programme; it is likely that the figure of 43.7/km² is an overestimation due to the number of observations being less than the recommended number described in the literature. However this is a significant development in the study of spider monkeys here at Piro. It is the first time that the species population density has been examined in this area and with further data collection and/or the pooling of data with Osa Conservation there is an exciting opportunity to illuminate the state of the population of this endangered species here at Piro. Weghurst (2007) studied spider monkey population density at Sirena biological station, Corcovado National Park, Costa Rica. Their study found that spider monkeys in this area had one of the highest population densities ever recorded for the species. Our study of the spider monkey population density at Piro will continue and has potential to provide a continuation of the work carried out by Weghurst (2007). 23

24 6 Otters 6.1 Methods for monitoring the population status of the Neotropical river otter (Lontra longicaudis): calculating density from camera trap data and using indirect signs as an index of abundance Introduction In comparison to other otter species, the Neotropical river otter (Lontra longicaudis) is relatively understudied and knowledge is distinctly lacking (Kasper et al. 2008). Furthermore, as with most other otter species, the majority of studies concerning this species are concerned with diet (Rheigantz et al. 2011) and presently there are no systematic studies to evaluate the population size of this suspected rare and declining species, leading to its IUCN Red List classification as Data Deficient (Waldemarin & Alvarez 2008). Information on the population status is fundamental in assessing the species against the threatened categories and to inform appropriate conservation and management strategies. Furthermore, the IUCN notes that field surveys of current populations and identification of key habitats are primary conservation priorities for the Neotropical river otter (Waldemarin & Alvarez 2008). Our proposed study will provide the first population estimate for the species in Costa Rica using randomly located camera traps along the Rio Piro and Quebrada Coyunda and analysing data with novel methods which do not require the identification of individuals within the population. The study will also provide a standardized methodology for monitoring the status of the population, providing an index of abundance based on encounters of scat and tracks Methodology Two linear transects will be established along the course of the Rio Piro and Quebrada Coyunda, measuring 8.4km and 3.8km, respectively, extending beyond the boundaries of the Osa Conservation Property into public land. All encounters of scat which is not classified as a latrine (latrines are defined here by the presence of two or more scats of different age classes in the same location), and tracks of otters will be tallied to form an index of abundance. Specifically, each scat encountered will represent an independent record, and each individual track - or, where it is found, trail of tracks - will also be recorded as independent records. The index will simply be the sum of tracks (individual and trails) encountered, and separately, the sum of scat encounters. Commencing in April 2013 surveys will be conducted monthly at approximately equal intervals over the entire study period and surveyors will walk within the course of the river. 24

25 Seven camera traps (three Bushnell Trophy Cam HD, Bushnell Outdoor Products, Overland Park, KS, USA; and five Moultrie D55, Moultrie Feeders, Albaster, AL, USA) will be deployed for up to one year from May Cameras will be placed at computer-generated random locations along the full course of the Rio Piro and Quebrada Coyunda, Osa Peninsula, Costa Rica and relocated to new random locations approximately every 14 days. Set at a height of approximately 20cm, the cameras angle of view will be parallel to the overall slope of the ground and not be obstructed by vegetation or topography. Cameras will be operational 24 hours a day, assuming consistent functionality, and set to take the maximum number of photographs per trigger and minimal intervals between triggers to effectively produce short video clips of animal movements in the field of view and to maximize capture probability. Cameras will be secured to trees, roots or branches stuck into the ground to form posts where necessary at the edge of the river bank in such a way that it can be assumed that it is impossible for an otter to go behind the camera and still be considered to be within the area of the sampled river. Estimating density from camera trap data requires estimates of animal parameters v, day range (km per day) and g, average group size. These will be estimated from camera trap data collected within the study site. Day range will be estimated following Rowcliffe et al. (2012), examining a subset of all photographs in the field with a portable card viewer before retrieving the cameras, noting the position of the otters in each image and tracing the movement path across the field of view with a measuring tape. Speed will be calculated by dividing the distance moved by the time between the first and last capture as printed on the automatic time stamp on each image. All changes in direction of travel between consecutive images will be measured with a compass. Average group size will be a simple calculation based on all otter records. To determine the camera sensor characteristics required for density estimation, namely detection distance and angle, we will follow the method of Rowcliffe et al. (2011), noting the position of first detection in the field of view and measuring the distance and angle from the camera with measuring tape and compass. This process will be aided by the use of a portable card viewer to note the animal position in relation to landmarks such as rocks and logs, and will be performed prior to removal of the camera. The area of the rivers sampled will be quantified by measuring the width of the river from bank to bank with a measuring tape at the point of each camera trap station, inputting this width data into a Geographic Information System and creating a series of polygons along the rivers that connect the nearest neighbouring stations; total area will be calculated as the sum of all polygons. Otter density will be calculated from camera trap data using an adapted form of the methodology first applied by Rowcliffe et al. (2008), adjusting the latest 2-dimensional model to a single dimension, as otters effectively move in one dimension, i.e., up and down 25

26 the river (Rowcliffe, personal communication). This technique has been developed for calculating animal densities from camera trapping rates by modeling the detection process. Tallied frequencies of evidence encounters will be analyzed by simple summary statistics, quantifying the increase or decrease over time. Principally comparisons will be made between years rather than months, which may otherwise reflect changes in detectability over time in response to changes in the environment, typically rainfall fluctuations and exposed surfaces with impressionable substrate rather than changes in abundance Results During the April transect survey nine encounters of tracks and six encounters of scat were recorded on the Rio Piro and two tracks and one scat on the Quebrada Coyunda. Zero tracks were encountered in May on either river, and only the Rio Piro yielded any records of scat four encounters. Between 01 May and 16 June 2013, 17 camera stations (Figure 6.1) were sampled for an average of 13.1 trap nights per station, totalling 223 trap nights. In this time, one otter record was obtained on the Rio Piro (Figure 6.1.1; N W ), captured by a Moultrie unit at 2300hrs on 08 May 2013 at a distance and angle of 1.93m and 10 from the camera lens, respectively. The group size was one, and animal speed could not be calculated as the camera took only one photograph after being triggered Discussion During the transect surveys encounters of indirect evidence of otters were limited, particularly in May. This is expected to be a result of the increased rainfall around the sampling date; in line with the existing sampling protocol the day sampled could not be altered to the point of avoiding any effect of rainfall. It is therefore recommended that the sampling protocol is revised to conduct transect surveys every week to reduce these environmental effects on data and build a more robust dataset. Further, it should be noted that during the data collection of the simultaneous project Temporal variation in the spatial use of a river network on the Osa Peninsula, Costa Rica: latrine re-visitation and identifying factors affecting space use by the Neotropical river otter (Lontra longicaudis)., there were 19 records of latrine utilisation in April and six in May. Of the latter, four were recorded on the Quebrada Coyunda where there was otherwise no record of activity in this study. It is therefore advocated that the protocol is revised to include latrines and not exclusively individual scats. 26

27 Figure 6.1 Camera trap stations sampled between 01 May and 16 June 2013 within the Osa Conservation property, Osa Peninsula, Costa Rica. Insufficient data has been collected to estimate species abundance from the camera trap data collected to date. To ensure the analytical model has sufficient data to provide accurate estimates of abundance, the minimum requirements, according to developer Marcus Rowcliffe (personal communications) are: greater than 20 sampling sites, including camera relocations; 20 otter records within a year, though records would provide a more robust estimate; 40 records per camera model for calculating camera parameters (with no more than three otter records being used per station), and; 30 sequences for estimating animal speed. Based on this capture rate, an extremely cautious estimate of 8920 trap nights will be required to obtain the minimum sample size of 40 records, assuming all are obtained on the same camera model and no more than three records are captured at any one station. Within the sampling period limit of one year, this would require a minimum of 25 cameras in operation every day. Overall, the Moultrie units have performed comparatively poorly, with frequent flash failures when the camera is triggered at night, resulting in completely black images which are unusable and may have potentially otherwise recorded otters. In addition, the data from one of the Moultrie cameras was excluded from the reported dataset as the camera alerted low battery on retrieval and there was a gap in the photographs of 10 days without records 27

28 which is highly unexpected; the camera correctly functioned on the tenth day (final day of sampling at this station). It is recommended that the study progresses with only the Bushnell units which perform more reliably without any night-time failures and also record time of capture to the second, facilitating the calculation of animal speed. Despite the absence of otter records obtained by the Bushnell units in this study, during piloting of these units in February and March 2013 (Roberts unpublished data), two records were obtained on the Rio Piro, in one case capturing an otter in three consecutive photographs within two seconds, and in the other six consecutive photographs (i.e. two triggers recording three photographs each) within eight seconds. In both cases, it would have been possible to determine all animal and camera parameters required for analyses; these data were not included in the report here. To our knowledge there has not been a published estimate of abundance for the species for any population in the world, and the present study here, therefore, is of great significance not only on a local scale, but also internationally, providing a repeatable methodology that may be used to assess population status in other parts of the species range. Considering the experienced low capture rate, the success of the study is probably dependent on deploying a greater number of camera traps else our ultimate objective may not be achievable. However, assuming the additional cameras are deployed promptly and the required minimum records are obtained before the end of April 2014, the data contained within this report will be utilised in this pioneering study. 6.2 Temporal variation in the spatial use of a river network on the Osa Peninsula, Costa Rica: latrine re-visitation and identifying factors affecting space use by the Neotropical river otter (Lontra longicaudis) Introduction The Neotropical River Otter (Lontra longicaudis) is a solitary species, living in a variety of habitats, including swamps, streams and lagoons in Central and South America. They are an adaptive species in terms of tolerating disturbance and have been reported to inhabit irrigation ditches in rice and sugar cane plantations in Guyana (IOSF 2010). Listed in Appendix I of CITES, L. longicaudis populations were hunted, mainly for their pelt, to such an extent that they became extinct in parts of their former range (IUCN 2008; Parera 1993). Although listed as a protected species in many countries, including Costa Rica, the Neotropical River otter is still hunted illegally, although not to the same extent as in previous decades (IUCN 2008; Parera 1993). Other threats to L. longicaudis include agricultural activities, pollution of waterways, and deforestation (Quadros & Monteiro-Filho 2002). 28

29 This piscivorous mammal forages for a multitude of prey ranging from fish and basilisks to insects and small birds (Montero-Fillio 2001). These feeding habitats vary considerably depending on the area and prey availability. This has led to the debate on the species, it being a generalist or a specialist forager. Its main prey will include fish and crustaceans throughout the majority of the year but if prey availability is low it will shift to feeding on invertebrates, reptiles, fruit and other food items (Montero-Fillio, 2003). This information reiterates the theory that this species is more of an opportunistic feeder, choosing to feed on most abundant prey rather than preferred prey. As described above, the status of the Neotropical river otter is still highly debated with regards to its temporal and spatial habitat preferences. Resting sites and foraging areas are two of the key areas that define habitat use (Perrin & Carranza2000), in addition to the characteristic defecating in conspicuous sprainting sites which facilitate efficient means to determine spatial use within and among aquatic environments (Rheingantz et al. 2004; Kasper et al. 2008). The factors affecting site selections have been studied for the smoothcoated otter (Lutra perspicillata) in India (Anoop & Hussain 2004), and in Brazil Kasper et al. (2008) recorded shelters and latrines and monitored reutilization. Notably, previous studies have investigated the frequency of use (e.g., Kasper et al. 2008; Santos & Reiss 2012) and the factors that influence resting- and sprainting- site selection (e.g., Pardini & Trajano 1999). However, there is a gap in our knowledge as there have been no spatial analyses of latrines and resting sites, particularly with regard to seasonal changes in their distribution. Our proposed study will expand upon the work of Kasper et al. (2008), with a two-fold larger study area and new investigations of spatial ecology concerning site reutilization to identify and monitor core areas of habitat use on the Rio Piro and Quebrada Coyunda over time. This study will investigate holt-, resting site- and latrine reutilization rates and provide the first year-round study of the spatial ecology of the species. This study will be a forerunner in increasing our understanding of the movements of the Neotropical river otter, have the potential to form the basis of a standardized methodology to study spatial ecology through noninvasive sampling, and inform conservation strategies that effectively consider the otter s spatial requirements to ensure adequate long-term protection of this iconic species of conservation concern Methodology Two linear transects will be established along the course of the Rio Piro and Quebrada Coyunda, measuring 8.4km and 3.8km, respectively, extending beyond the boundaries of the Osa Conservation property into public land. All individual otters encountered and encounters of scat, latrines, potential holts and resting sites will be recorded on GPS and the sites (excluding visual encounters) photographed to verify against new records to build a catalogue and determine either site reutilisation or new spatial use. Latrines will be defined by the presence of a minimum of two scats of different age classes and resting sites will be 29

30 identified by the presence of a clear impression in the substrate, typically made by otters sliding or rolling. The substrate that the sign is on and the distance to river course will also be recorded. Surveys will be conducted monthly at approximately equal intervals over the entire study period and surveyors will walk within the course of the river starting in early morning, at approximately hrs. Using a Geographic Information System (GIS), the distribution of utilised sites will be mapped and analyzed descriptively Results In four sampling months (February May 2013), a total of 93 signs of evidence were encountered at 78 utilised sites summed across the two rivers; 37 sites on the Quebrada Coyunda (QC) and 41 sites on the Rio Piro (RP). Of all utilised sites there were 10 potential holts (RP: 3; QC: 7), 53 latrines (RP: 28; QC: 25), one visual encounter (RP), five resting sites (RP: 0; QC: 5) and nine scat locations (RP: 8; QC: 1) (Figure 6.2). Thirteen sites were utilised in at least two sampled months (RP: 6; QC: 7; Figure 6.2); LAT28 (RP) and REST2 (QC) were used the most during the sampling period, each utilised in three of the four sampling occasions Discussion The population of Neotropical river otters in the sample area have utilised the entire length of both the Rio Piro and Quebrada Coyunda. Of all of the sites, 16.7% (n = 13) were revisited at least a second time. Furthermore, despite the total number of encounters of evidence being marginally smaller on the QC, the river is significantly shorter, suggesting that the Quebrada Coyunda is more intensely utilised based on greater encounters per kilometre in addition to the higher frequency of reutilisation overall. Potentially, however it could be that the QC has a greater number of suitable sprainting sites than the RP; investigations of the factors affecting space use need investigation in order to validate conclusions. 30

31 Figure 6.2 Map of the distribution of Neotropical river otter evidence encountered (latrines, individual scats, resting sites, holts and individuals) and frequency of utilisation in monthly sampling between February and May It is advocated that to develop the study to further investigate the spatial ecology of the Neotropical river otter with a new focus on understanding the factors that influence space use, additional environmental variables are recorded. Specifically the key variables would be prey diversity and abundance, river width and depth, number and location of tributaries, rockiness and substrate. These variables could then be investigated with regards to their influence on site selection following the methods of Anoop & Hussain (2006) by Principal Component Analysis and logistic regression models. This extended study would further benefit from detailed scat analyses to determine the degree of dietary specialisation providing an indicator of the influence of certain prey items on otter distribution. Additionally, a detailed topographic map could further provide insight into the factors influencing space use, particularly the spatial configuration of the rivers in the study area in relation to other inland water. Sampling more frequently (i.e., at weekly intervals) would likely provide a larger and more robust dataset, increasing the number of encounters within the present sampling interval by accounting for the probability that scats will be washed away by heavy rainfall and increased water levels. Finally, with a larger dataset, it will be 31

32 possible to analyse patterns of core area distribution in relation to season, the ultimate objective of the study Is it possible to identify individual Neotropical river otters and apply capturerecapture methods to camera trap data to reliably estimate abundance? A controlled case study Introduction Basic information on population abundance is vital in conservation research to assess current status, monitor temporal trends and define conservation priorities (Mackenzie 2005; Oliveira-Santos et al. 2010). Without robust data on distribution and abundance, conservation strategies and management practices may be misguided, ineffective and inefficient (Langbein et al. 1999; O Brien 2008; Hájková et al. 2009; Obbard et al. 2010). The Neotropical river otter (Lontra longicaudis) is a Data Deficient species and there are no data available for population size (Waldemarin & Alvarez 2008). Though there is insofar no standardised method to estimate abundance of the species, it is thought that the species is rare and declining throughout its range from Northwestern Mexico to Peru and Uruguay, and field surveys of populations are thus one of the IUCN conservation priorities for the species (Waldermarin & Alvarez 2008). Furthermore, developing standardised methods is essential for reliably comparing data at multiple spatial and temporal scales. The difficulty associated with assessing the abundance of otters (Family: Mustelidae) by direct observations, obligates the implementation of alternative approaches (Hájková et al. 2009), such as recording distinct otter spraint to cost-effectively report species presence (Davison et al. 2002). However, estimating abundance - and therefore assessing population status - of elusive and rare species is often difficult (Balme et al. 2009; Hájková et al. 2009), and using sprainting intensity as an index of otter abundance is unreliable as it may rather reflect abundance of suitable, conspicuous sprainting sites and activity rather than true population size (Guter et al. 2008; Calzada et al. 2010). Camera-trapping is a relatively new methodological advancement (Pettorelli et al. 2009) that uses specialised equipment to detect and trap photographs of passing animals (Rowcliffe et al. 2008). Since its early applications in the 1920s, camera traps have become an increasingly popular tool in non-invasive wildlife research (Rowcliffe & Carbone 2008). It is often considered superior to human-observation-based survey methods due to its comparatively greater labour efficiency (O Brien 2008; Srbek-Araujo & Chiarello 2005; Pettorelli et al. 2009; Vine et al. 2009) and reliability of species identification (Vine et al. 2009; Roberts 2011). 32

33 The most common application of camera-trapping is the abundance estimation of wild cat species (Rowcliffe & Carbone 2008), which is most often calculated by simultaneously photographing the left and right flank of animals with paired camera traps, producing capture histories of individuals identified by unique stripe and spot patterns, and estimating population parameters by capture-recapture analysis (Karanth 1995). Camera trap studies of non-striped and non-spotted species however are less represented, and are generally limited to species inventories, rather than estimating abundance or density (Rowcliffe et al. 2008). Applying the camera-trapping methodology to these species will increase the value of this tool (Rowcliffe & Carbone 2008). Trolle et al. (2008) argue that species lacking conspicuous stripes and spots may be identified by subtle marks, coat colouration, scars, body structure and sex, and this approach has been applied in a number of studies concerning various taxa (e.g., coyotes (Canis latrans): Larrucea et al. 2007; foxes (Vulpes vulpes): Sarmento et al. 2009; maned wolf (Chrysocyon brachyurus): Trolle et al. 2007; tapirs (Tapirus spp.): Holden et al. 2003; Noss et al. 2003; Trolle et al. 2008; and pumas (Puma concolor): Kelly et al. 2008). Individuals however, may be misidentified if shallow scratches fade (Trolle et al. 2008), or photographs are captured in different light and humidity conditions, as well as varying approach path angles and distances relative to the camera (Oliveira-Santos et al. 2010). The inability of researchers to correctly identify individuals of non-striped and non-spotted animals was highlighted by Oliveira-Santos et al. (2010), where 55 photographs of a tapir population of known size were distributed to 12 participating researchers; errors in estimated population size ranged from an under-estimate of 50% to an over-estimate of 75%. It is believed that otters may be individually identified using the markings on the chin (Chanin 2003), however the reliability of this method has not been systematically investigated. Foster and Harmsen (2012) in their critique of camera trap studies, advocate that future studies validate these theories using data from captive individuals to test whether inter-individual variation at the population level is sufficient to reliably distinguish between individuals, and also state that for data to be analysed by capture-recapture methods, unique identifying markers must be visible in every photograph. The specific objectives of this study are to: i) investigate the ability of researchers to identify individual Neotropical river otters (Lontra longicaudis) in a set of photographs of a population with known size; ii) investigate which characteristics are being used by researchers to identify individuals within the population; iii) quantify the proportion of camera trap photographs which are discarded on the basis of inability to identify individuals; iv) assess the effect of any inter-researcher variability in individual identification on the reliability of species abundance estimates, and; v) provide informed recommendations towards developing standardised methodologies for estimating abundance of Neotropical river otters in the wild. 33

34 Methodology Camera-trapping will be conducted within the captive area(s) of an animal collection(s) holding at least one individual Neotropical river otter, including Barranquilla Zoo in Colombia. Camera(s) will be active 24hrd -1 and date and time stamps will be printed on each image. Resulting images will be grouped into three subsets: (1) right flank of animal captured; (2) left flank of animal captured; (3) animal facing camera. Cameras will be active until greater than 50 images within a single subset have been captured and it is agreed with animal collection staff that all individuals were photographed at least once. Animal collection staff will also be asked to identify the individual(s) in each photograph. The photographs will be sent to members of the IUCN Otter Specialist Group and other otter researchers who will be invited to respond to the following questions: (1) How many photos were discarded as inappropriate for identification?; (2) How many individuals were identified?; (3) What parts or characteristics of the animals were considered relevant in the identification of individuals? Respondents will also be asked to identify the individual(s) in each photograph, giving a unique numerical identifier to each individual. The number of individuals identified by each researcher will be expressed as a simple calculation of error proportion of over-estimation or under-estimation in relation to the known population size. Characteristics used by respondents to identify individuals will be assessed to investigate any trends in accuracy of estimate and characteristic(s) used. The proportion of photographs discarded will be expressed as a percentage of the total number of photographs collected. Capture-recapture methods will be applied to the capture histories created from researchers responses containing unique numerical identifiers for individuals in each photograph, and abundance estimates calculated in CAPTURE and compared with the estimate derived from the capture history based on responses from animal collection staff Results Data collection is still in progress at Barranquilla Zoo and no other zoos have yet confirmed participation in the project. No analyses have been undertaken Discussion The success of the project is dependent on additional animal collections assisting with the data collection; presently the sample size is too small to facilitate a robust study. 34

35 7. Birds: Species diversity and habitat specialisation of bird communities within primary and secondary forest and in a river course on the Osa Peninsula, Costa Rica Introduction Costa Rica is very rich in avifaunal diversity. Hosting approximately 850 species, Costa Rica is home to more bird species than the United States and Canada combined (Henderson 2010; Sánchez-Azofeifa et al. 2001). Moreover, 160 of these bird species are endemic to Costa Rica (Henderson 2010). The distribution of birds and the species richness of a given area can often be explained by the habitat type and characteristics of the environment (Wilme & Goodman 2003). From this information, it has been possible to classify species as habitat generalists or specialists, the former being more resilient to landscape change through deforestation, fragmentation and other changes (Pejchar et al. 2008). As birds are good indicators of the health of the habitats they occupy due to their sensitivity to change, recognising habitat specialists and monitoring trends in avifaunal diversity may help to identify species at greater risk of negative population responses to environmental change. Furthermore, declining bird populations will affect rates of pollination and seed dispersal, two fundamental functions of a healthy ecosystem facilitated, in part by birds (Pejchar et al. 2008). Considering this and the high levels of endemism in Costa Rica it is important that bird communities in this region are regularly monitored. The aim of this study is to identify and compare the bird communities found within primary and secondary forest and within the course of a river and further understand which species are habitat generalists and specialists based on presence/absence data at each sampling point Methods An even number of point count stations were made in three different habitat types; primary and secondary forest (stations set on trails) as well as a river course. Stations were set with a distance of 200m between all stations. Each station was visited a minimum of two times per month. In order to reduce the time of day effect, the order in which stations were visited was reversed each time. Surveys were performed in the morning between 06:00 and 08:00 am. Each point count lasted 10 minutes; before the next recording began a settling period of one minute was allowed. The following data was recorded: species seen or heard and the number of each species seen. Birds were recorded within a 50m radius of the station point. Sightings were done either using binoculars or by song recognition. This was done to identify birds to species level. When birds could not be identified to the species 35

36 level the bird group (e.g. Woodcreeper spp.) was still recorded however not included in the data analyses. Photographs were taken when birds were not identified in the field and needed further attention using guides. The main reference used was Garrigues & Dean (2007). Species diversity estimates were calculated for each point count station and an average value of diversity was then calculated for each habitat type. Species diversity was calculated using Simpson s Inverse Diversity Index (1-D). D = Σn(n-1) N(N-1) Where n is the total number of organisms of a particular species and N is the total number of organisms of all species. The value of the index ranges between 0 and 1; the greater the value the greater the sample diversity With the use of EstimateS (Colwell 2009) species accumulation curves for the bird species identified in each habitat type were obtained. Moreover, the estimated number of species present in each habitat type and their standard deviation were calculated by Chao Presence/Absence with the use of DIVERSITY (Henderson & Seaby 1998) Results A total of 699 individual birds representing 81 species and 13 groups have been identified up to the date of 14 June The complete list of species and groups identified are listed in Appendix A. Of the 81 species identified, 48 were found in the primary forest and 38 in both the secondary forest and river course. Species identified accounted for 34.59%, 41.8% and 57.96% of the number of species estimated by Chao Presence/Absence in the primary forest, secondary forest and river course habitats, respectively. Moreover, calculated averages for Simpson s Inverse Diversity Index (1-D) were , and for the same order of habitats. Seven bird groups were also recorded in both the primary forest and the river course, and eight were noted in the secondary forest habitat. It can be seen in Figure 7.1 that the average number of individuals per station is higher in the primary forest (6) than in both the river course (5) and secondary forest (4). Conversely, the river course showed a higher average number of species per station (5) than both of the forest habitats (3). 36

37 Average number CRF 132 Science Report Primary Secondary River course 1 0 Individuals per station Species per station Figure 7.1 Average number of individuals and species per station observed in the primary forest, secondary forest and river course surveyed. The following figures (7.2, 7.3 and 7.4) show the species accumulation curves produced for the three habitats surveyed. As seen in Figure 7.2, the primary forest habitat had the largest number of samples (24) and also the largest amount of species identified (48). The estimated species diversity from the Chao Presence/Absence was the highest of all the habitats; ± In Figure 7.3, the secondary forest habitat had a lower sample count (19) along with a lower number of species (38). The estimated species diversity from the Chao Presence/Absence was 90.9 ± Finally, the river course habitat had the lowest amount of samples (9) as seen in Figure 7.4. However, it had the same number of identified species as the secondary forest habitat (38). Moreover, the estimated species diversity from the Chao Presence/Absence was the lowest of all habitats ±

38 Number of species Number of species CRF 132 Science Report Number of samples Figure 7.2 Species accumulation curve for primary forest habitat and species diversity estimate as determined by Chao Presence/Absence (±SD) Number of samples Figure 7.3 Species accumulation curve for secondary forest habitat and species diversity estimate as determined by Chao Presence/Absence (±SD). 38

39 Number of species CRF 132 Science Report Number of samples Figure 7.4 Species accumulation curve for the river course habitat and species diversity estimate as determined by Chao Presence/Absence (±SD). The species most commonly encountered during surveys in decreasing order of abundance were the Black-mandibled Toucan, Scarlet Macaw, Riverside Wren, Chestnut-backed Antbird, White Ibis, Blue-crowned Manakin and Longbilled Hermit, which had individual sightings each. Species that have been sighted in all three habitat types include the Black-hooded Antshrike, Black-throated Trogon, Chestnut-backed Antbird, Blackmandibled Toucan, Longbilled Hermit, Riverside Wren, Rufous-tailed Hummingbird and Scarlet Macaw, whereas others were only present in one or two of the habitat types. Some of the bird groups identified, such as Green Parrots and Hummingbirds, had a high number of individual sightings (69 and 25, respectively) and were also present in all habitat types Discussion Summarized data obtained from surveys to date shows that there are a higher number of individuals in primary forest habitat, but a larger number of species on the river course. However, the species identified so far in primary forest habitat account for the smallest percentage of the estimated number of species present, while the river course has the highest percentage for this same estimate. The calculated Simpson s Inverse Diversity index is shown to be highest in secondary forest habitat, decreasing for primary forest and river course habitats. The estimated species diversity calculation using Chao Presence/Absence is shown to be highest in primary forest and lowest in the river course habitat, though standard deviations are relatively high in both forest habitats. 39

40 Further research on specialist and generalist bird species will need to be undertaken in the future. This will allow for further conclusions concerning the ecological significance of the species composition in the three habitats. The present results show a bias due to the varying amount of surveys conducted per habitat type, along with the varying amount of trails surveyed within each habitat. Moreover, identification of bird species can be limited by the expertise of surveyors and inaccuracy of identification, especially when considering the vast biodiversity of the Osa Peninsula. Recognition of key physical traits and bird songs and calls can prove difficult. Hence, species presence in surveys could vary in instances where some are more easily identifiable than others, which also depends on the experience of surveyors. Field guides and recordings of bird vocalizations can thus aid in the training of staff and volunteers in order to more easily identify birds during surveys. 40

41 8. Butterflies: Species diversity of butterflies (Order: Lepidoptera) occupying different heights within the understory of a secondary forest on the Osa Peninsula, Costa Rica Introduction Butterflies (Order: Lepidoptera) are a large and diverse group of invertebrates, with about three times the number of mammals, reptiles, dragonflies, mosquitoes, termites, and tiger beetles, and twice as many species as terrestrial birds (Robbins and Opler 1997), with at least 1,250 butterfly species being found in Costa Rica (Henderson 2002). Butterflies play a vital role within the ecosystem, as key pollinators of plants, effective indicators of biodiversity and ecosystem health due to their ability to flourish in many different climates and, being primary consumers, have a 98% mortality rate in the wild providing sources of food for a large number of predators (Kreman 1992; Koh 2007; Fleishman 2009; Bonebrake 2010). As stronger and more reliable indicator species than their temperate counterparts due to more stable population trends and year-round activity of adults, butterflies in the tropics have been used to monitor anthropogenic habitat degradation, land use change and climatic shifts, though their effectiveness has been contested. It is however understood that butterflies have very specific relationships with certain types of plants and habitats and greater understanding the ecology of butterflies may build our knowledge of these wildlife-habitat interactions that are a fundamental part in informing strategies to conserve both the butterflies and the environments they live in. However, presently, in comparison to their temperate counterparts, ecology of tropical butterflies is inadequately understood, despite approximately 90% of species being found in the tropics (Bonebrake et al. 2010). The present study aims to record baseline data on the diversity of butterflies in a secondary forest understory in the Osa Peninsula, Costa Rica and investigate their ecology with specific regard to their use of different heights within the forest structure Methods butterfly s fragility and irregular flying pattern make them exceptionally tricky to handle and capture, therefore we will use canopy traps to capture butterflies for identification. The traps that will be utilized during the study are commonly known as Blendon-Style canopy traps, comprising two parallel dinner-plate-sized plastic plates with approximately 1m of netting connecting the two to create - in effect - a cylinder (Figure 8.1). Fruit-feeding butterflies are lured to the traps to feed on fermenting bananas placed on the centre of the 41

42 plastic plate as bait, and become trapped once they enter the nets. The gap between the net and the plastic plate is crucial. If too large it will let butterflies escape; if too small it will prevent butterflies from entering the trap. The optimal width of the gap is roughly 5cm. The bottom of the canopy trap will be at a minimum of 2m off the ground and the maximum trap height will be 10m. Figure 8. 1 Blendon-style canopy trap used for phase 132 butterfly capture (Earth in Focus 2013). Butterflies are ectothermic and as such receive their energy from the sun and therefore will be most active in sunlit areas. The canopy traps will consequently be suspended by a rope in places where the midday sun would shine, although in dense forest this is not always possible. The difficulties encountered using these traps concern over- exposed areas where wind might blow the canopy traps into branches, thus either damaging the trap or making it difficult to lower easily, allowing butterflies to escape, along with blowing the plate and net out of line also, allowing the butterflies to escape. Trapping will be conducted in multiple series of five-week cycles whereby each of the five traps are raised to heights of two, four, six, eight and ten metres for three consecutive trapping days per week, adjusting the height between weeks to ensure that in each cycle, each trap is effective at each height, and in any given week, no two traps are set to the same height. Trap heights will be measured by temporarily fixing string marked at metre intervals to the bottom surface of the trap, and detaching the string by gently pulling on it once the trap is set to the respective height. A minimum of three complete cycles will be performed within the dry season, principally January to April where there are observed 42

43 peaks in butterfly abundance. The inter-trap distance will be approximately 70m on average, and traps will be above or within 5m of Piro Trail. This methodology will be form the protocol for a long-term monitoring programme, observing any fluctuations in diversity and abundance within the local population. Traps will be checked daily in the early afternoon rather than early morning to reduce the probability of butterflies remaining in the traps overnight. To identify butterflies in the field, they will be retrieved manually from the traps. They will then be held by the thorax between the forefinger and thumb so that the wing markings and shape can be seen clearly while referring to the field guide. To view the often more colourful dorsal side, the wings will either be blown apart or a pencil will be placed gently between the wings. The body will then be held with the thumb on the butterfly s back and the index finger on its belly so that the wings open for observation. Particular care will be taken not to crush the legs or damage the wings, whilst holding larger species more firmly to prevent them flying away (DeVries 1987). The individuals will then be released live. The analysis that will be undertaken will include inputting data collected into Microsoft Excel to construct tables from which species accumulation curves can be produced, plotting Sobs Mao Tau values computed in EstimateS ( 2013 Robert K. Colwell) with 100 randomisations without replacement to remove bias. A curve will be produced for each individual trap height, pooling data across traps and survey completeness will be investigated by calculating values of expected diversity by Chao Presence/Absence in DIVERSITY (Henderson & Seaby1998). Alpha diversity will also be computed by Shannon- Weiner test in DIVERSITY (Henderson & Seaby 1998), analysing each trapping height as an individual subset, in addition to analysis of the full dataset with data pooled across all heights. Species-specific analyses will investigate the relationships between species and canopy height utilisation. Graphs will be produced in Excel that visually represent the heights at which each species are captured. This will indicate which species are specialised to specific canopy heights, as well as identify those species which are more generalised to a broader range of heights used, as mentioned above. 43

44 Archaeoprepona amphicachus amphiktion Archaeoprepona camille camille Caligo atreus dionysos Caligo eurilochus sulanus Caligo eurilochus sulonus / Caligo telamonius Caligo illioneus oberon Caligo telamonius memnon Catonephele mexicana Catonephele orites Cissia agnata Colobura annulata Consul fabius (similar to cecrops) Eryphanis automedon lycomedon Hamadryas februa ferentina Hamadryas laodamia saurites Magneuptychia alcinoe Memphis ambrosia Memphis arginussa eubaena Memphis artacaena Memphis lyceus Memphis moruus boisduvali Morpho amathonte Morpho helenor marinita Nica flavilla canthara Opsiphanes bogotarnus alajuela Opsiphanes cassina chiriquensis Pareuptychia ocirrhoe ocirrhoe Pyrrhogyra otolais otolais Strathocles pulla Taygetis kerea Taygetis laches laches Taygetis xenana godami Temenis laothoe agatha Temenis pulchura Zaretis ellops Trap height (m) CRF 132 Science Report 8.3. Results Species Figure 8.2 Presence/absence of species diversity at each trap height within a secondary forest understory. The butterfly canopy trapping survey was initiated during the previous phase of the project, in the week beginning 18 February Since then five cycles of five-week surveys referred to in the methodology above have been completed. This has led to a total of 35 species consisting of 102 individuals being trapped. 44

45 Species (n) Species (n) Species (n) Species (n) Species (n) Species (n) CRF 132 Science Report Sample effort (days) Sample effort (days) (i) (ii) Sample effort (days) Sample effort (days) (iii) (iv) Sample effort (days) Sample effort (days) (v) (vi) Figure 8.3 Accumulation curves of species trapped at different heights within the secondary forest understorey, plotting Sobs (Mao Tau) values (blue) and estimated species richess (Chao Presence/Absence ± SD; red): i) Two meters; ii) Four meters; iii) Six meters; iv) Eight meters; v) Ten meters; vi) All data pooled across all trap heights. The most abundant species by far is Archaeoprepona amphicachus amphiktion (referred to from now as Archaeoprepona.a.a) of the Charaxinae family, with a total of 30 specimens spread across all five trap heights (Figure 8.2), of which the most abundant trapping heights 45

46 were six metres and ten metres, with nine and eight individuals respectively. The highest abundance trap capture for all species was at 10m with a total of 24 individuals covering 13 different species. Some species were trapped across various trap heights, whereas others have only been trapped at one specific height (Figure 8.2). Figure 8.3 shows the species accumulation curve for all specimens and species caught. Figures 8.3 show the species accumulation curves for trap heights 2-10m correspondingly. All data collected for the project so far has been entered into Microsoft Excel, EstimateS (Colwell 2009) and DIVERSITY (Henderson & Seaby 1998) to produce the species accumulation curves; in addition the estimated numbers of species still to trap and the standard deviation for each height have been plotted on each curve. Table 8.1 Percentage of survey completion for all trap heights. Trap Height (m) No. Spp. Trapped Estimate No. Spp. Survey Completeness (%) All Discussion Certain trends within the data are observed as to the differing heights that assorted species specifically fly, whereas other species are much more generalised and fly within a range of heights (Figure 8.2). We can see that the number of total species for all heights is considerably lower than that of the estimated number of species needed to complete this survey (±SD) (Figure 8.3), with the total number of species trapped so far being 35, and the estimated species number to complete this survey being anywhere between individual species. The percentage of completion for trapping at all five heights is currently just above 42% (Table 8.1). This would suggest that a large amount of surveying is still needed before this investigation is completed. As for individual trap heights, the level of survey completion varies greatly between specific heights. For instance, 15 weeks of trapping at 4 metres has produced the lowest completion percentage of 26.19%, whereas the same amount of trapping at 6 metres has produced the 46

47 highest completion percentage of 56.25% (Table 8.1). For a variance in height of only 2 metres, the difference in these percentages is possibly explained by the lack of multiple individuals of a single species being trapped at 4 metres, which can be seen by the considerable extent of the standard deviation for this trap height (±32.99) (Figure 8.3). The differences in completion percentages for the other trap heights are not as sizeable, generally being between 35-50% (Table 8.1). Standard deviation values for trap heights excluding 4 metres range from ±6.02-±11.75 (Figure 8.3), leading us to believe these particular height surveys are more complete than that of trap height 4m. However this could be explained again by the multiple individuals of a single species factor, as all these heights have had numerous multiples when compared to trap height 4m. These figures therefore lean towards the prospect of a substantial amount of trapping still yet to be completed. The canopy traps attracted the families of Nymphalidae, as they were baited with fruit, (expected by our initial hypothesis). The main families trapped were the Biblidinae, Brassolinae, Charaxinae, Morphinae, and Nymphalinae, with a few individuals of the Coeinae and Satryrinae families. The most abundant family present within the data is Charaxinae, mainly due to the high abundance of Archaeoprepona.a.a as mentioned above. The limiting factors for this project include the small primate populations in the area learning how to get into the canopy traps as they were attracted by the baits used, and possibly the butterflies trapped. Also, when the rains began towards the end of the cycles the likelihood of trapping butterflies decreased due to their ectothermic physiology. The rains also became a problem for elevating and lowering the traps as the ropes used to suspend the traps became saturated, leading to breakage. This caused data collection for specific traps to become delayed further into the rainy season, further reducing the likelihood of trapping butterflies. These factors could undoubtedly jeopardise the results and may require relocation of the traps to different locations, rotating them within the same habitat, and possibly changing the methodology. Due to the relatively low completion percentages and limiting factors mentioned above the methodology for the next stage of this project will be altered to include six consecutive days of canopy trapping, beginning at the end of the current wet season, at a practical time during November. This has been agreed between the Principal Investigator and relevant staff members as the best course of action for an improved continuation of the butterfly survey. 47

48 9. Anurans: Diversity and abundance of leaf litter frogs on a forest trail in primary and secondary forest on the Osa Peninsula, Costa Rica: baseline data for a long-term monitoring programme of population responses to climate change Introduction Since 1980, over 120 amphibian species have reportedly gone extinct, whilst more than a third of all remaining species are globally threatened (Whitfield et al. 2007). Primary threats include habitat loss and degradation, chemical pollution, and most significantly by climate change, for which amphibians are excellent indicators of resulting environmental stresses (Walther et al. 2002). Due to the porous nature of their skin, amphibians are vulnerable to desiccation in hot climates (Elinson et al. 1990). Reductions in the availability of their preferred microhabitat, generally small pools, and increased rates of leaf litter decomposition may lead to reduced breeding success in leaf litter frog species, which use leaves to build nests (Whitfield et al. 2007). A hotter climate also generates higher rates of disease transmission, therefore a higher incidence of death amongst amphibians (e.g., pathogenic cytrid fungus; Pounds et al. 2006). Increased UV-B radiation, which results from depletion of the ozone layer, has been reported to diminish hatching success of amphibian eggs (Blaustein & Kiesecker 2002). Already, there are reports that amphibian populations in Costa Rica have declined by as much as 75% over the past 35 years (Whitfield et al. 2007). This is a matter of serious concern to conservationists as amphibians constitute a major trophic level as prey and predators in many ecosystems and consequently their decline could generate a cascade of population declines in other taxa. Furthermore, mass extinction of these sensitive animals is of concern to human health, as amphibians largely control populations of insects, including those which transmit diseases. The aim of this study is to provide baseline data of the diversity and abundance of leaf litter frogs within the Osa Conservation property at Piro, Osa Peninsula, Costa Rica. The long-term aim is to determine the effects of climate change on leaf litter populations Methodology Due to the size and speed of the majority of leaf litter frogs and toads, the simplest way to identify these is by hand capture, followed by photographing for ex situ identification. The survey area is along New Trail which includes both primary and secondary forest. The forest 48

49 trail is divided into seven equal length sectors, each of 400m, including boardwalks, bridges and the Rio Piro crossing. Sectors are sampled in numerical order and re-sampled repeatedly until the species inventory is virtually complete. Surveys are conducted between 1530 and 1800hrs, sampling one sector per day only. During the survey, leaves are gently scraped from the trail with boots to uncover any anurans, often exposed when they jump. The individuals are then captured by hand, washing hands before and after handling each individual to reduce damage and minimise risk of disease spread. Where necessary, individuals are temporarily placed into a clean Ziploc bag to separate the individual from any leaves that may be in the hand too, and to gain a better, safer hold, restraining the individual by the hind leg between the surveyor s forefinger and thumb. Individuals are then photographed to record details of the proximal, lateral, dorsal and ventral sides to aid identification. Individuals are then released five paces backward along the trail from where they were caught to eliminate risk of pseudoreplication. Image numbers are recorded alongside the date and sector sampled, and the temperature and humidity readings from a handheld thermo-hydrometer. Species will be identified from the photographs using a dichotomous key and expert knowledge, referring to all major external characteristics, inter alia head form, pupil shape, tympanum, cranial crests, paratoid and other glands, vocal sacs, axillary web/membrane, ventral disk, nuptial pads, and colouration. It is anticipated that it may later be possible to identify individuals to species level in the field, eradicating the need for photographs. Data will be pooled across sectors and a species accumulation curve produced plotting Sobs Mao Tau values calculated in EstimateS (Colwell 2013). Survey completeness will also be expressed as a percentage of the expected species richness, as calculated by Chao Presence/Absence in DIVERSITY (Henderson & Seaby 1998). Diversity will be measured by the Shannon-Weiner and Margalef indices, calculated in DIVERSITY (Henderson & Seaby 1998) to report more finely on evenness and abundance, respectively. 49

50 9.3. Results Between 09 April and 13 June 2013 a total of 201 individuals have been captured and photographed during 40 surveys. Images are pending external identification, after which analyses may be performed Discussion The sampling technique performs well and will be continued as part of this baseline survey and the long-term monitoring programme, subject to successful identification from images. No results to discuss at present. 50

51 10. Riverside Wrens: Documenting the vocal repertoire of the Riverside Wren (Cantorchilus semibadius) Introduction The riverside wren (Cantorchilus semibadius) is a species of oscine passerine that has a limited distribution, restricted to south-western Costa Rica and north-western Panama. Pairbonded individuals perform a complex duet, with males and females using an extensive repertoire of song phrases (~40 N.I. Mann, C. Templeton & P.J.B. Slater, pers.comm.) with a high degree of temporal coordination. The purpose of this study was to document the vocal repertoires of these birds to allow for further investigation of duet function and the learning process of juvenile birds in a duetting species Ecology, life history, and conservation status of the Riverside Wren Around 13-14cm in length the riverside wren is easily recognisable by the striking black and white barred pattern on its chest. The species is monomorphic and distinguishing between males and females without handling the birds is impossible. The riverside wren occupies dense vegetation in forest edges and alongside rivers. It feeds entirely upon invertebrates and forage throughout the forest canopy and understory (Skutch 2001). They are usually found in pairs or family groups that generally do not exceed four in number. It is thought that they are monogamous, although divorce and mortality rates in partners has not been studied, and hold year round territories. The IUCN lists the riverside wren as Least Concern. Although this species may have a small range, it is not believed to approach the thresholds for vulnerable status under the range size criterion. The population trend is not known and the population size has not been quantified (BirdLife International 2012). However, the species is noted as being sparse and uncommon (BirdLife International 2012) and the areas of humid lowland Pacific forest where it occurs face significant threat in terms of land conversion and habitat degradation. Research into birdsong, and in particular duetting has a long history and over that time has provided significant insight into many fundamental biological functions (Catchpole & Slater 2008; Hall 2009), for these reasons, and the intrinsic value of the species it is important to increase the understanding of the riverside wrens duet and vocalization behaviour Bird songs and calls 51

52 Table 10.1 Terminology used in describing bird song and duetting throughout this paper. (Adapted from Hall, 2009) (i) Song Call Song Phrase Song initiating Phrase type Song type (ii) Duetting Duet Duet type Song answering Reaction time Answering rule Definition Short, simple vocalization, usually in particular contexts such as alarm, flight, begging Vocalization advertising for mates and territory ownership, usually longer and more complex than calls Unit within a song, may be an element (uninterrupted trace on a sonogram) or a syllable (set of elements occurring together in a particular pattern) Singing (non-duetting species), or producing the first phrase of a duet (duetting species) Version of a phrase the set of phrase types produced by one individual Version of a song the set of song types produced by one individual comprise its song type repertoire Coordinated singing by two individuals so that their phrases alternate or overlap Particular combination of the phrase types or song types of two individuals the set of duet types produced by a pair comprise their duet type repertoire Initiating a song in response to another individual to form a duet the individual-level behaviour resulting in duets Time interval between the start of an individual s song/phrase and the start/end of its partner s preceding song/phrase Consistent answering of a phrase type or song type in the partner s repertoire with a particular phrase type or song type from own repertoire the individual-level behaviour resulting in duet types Birds rely heavily on visual and acoustic communication. The importance of visual communication manifests itself through colourful plumage and display behaviour. In the poor light and dense vegetation of the tropical rainforest reliance on visual communication has disadvantages. In this type of habitat acoustic communication is more effective: sound can travel greater distances and can pass through and around objects. A song or call can be produced only as and when it is needed and large amounts of information can be transmitted quickly and efficiently (Catchpole & Slater 2008). The riverside wren relies heavily on acoustic communication. 52

53 Birds produce both songs and calls and the distinction is one of tradition and taxonomy (Catchpole & Slater 2008). Oscines and sub-oscines within the order of Passeriformes were originally separated based on the number and complexity of their syringeal muscles, with the oscines known commonly as the true songbirds due to the greater complexity of their songs (Catchpole & Slater 2008). Recent studies suggest that the separation of oscines from sub-oscines in fact has more to do with the birds underlying neurological structure and how they learn songs rather than simply the complexity of those songs (Catchpole & Slater 2008; Fortune et al. 2011). The classic definition of song is a long, complex vocalization with a clear structure and stereotypy, generally produced by males during breeding season, and seems to occur spontaneously with classic diurnal rhythms (Catchpole & Slater 2008). Whereas calls are often defined as being shorter, simpler, and produced by both sexes throughout the year, they are less spontaneous and generally occur in particular contexts related to specific functions such as flight, threat and alarm (Catchpole & Slater 2008). However, there are many exceptions to these broad definitions. In the tropics it is not uncommon for both sexes to produce song throughout the year regardless of breeding periods. It is also more common for tropical species to duet (Hall 2009). Songs are made up of a collection of phrases which contain elements, the elements are particular notes at different frequencies or amplitudes and they form the phrases which form the songs Duetting in Riverside Wrens The vocal repertoire of the Riverside Wren was first described by Mann et al. (2009) who found that they perform a cyclical, antiphonal duet (where each pair member repeats a phrase so that they alternate again and again within the duet, often with little or no overlap (Mann et al. 2009)). The duet of the Riverside Wren does not overlap, or if it did it did so only fractionally and generally only when pair members were far apart. This lack of overlap is even more notable as the latency between responses was generally seconds, and could regularly be as little as 0.02 seconds (Mann et al. 2009). They found that solos occurred in both sexes and the cyclical part of the duet was almost always initiated by the female, while the cycle itself was frequently, though not always, preceded by a separate phrase or call by the male (Mann et al. 2009). Both sexes contribute to duets by selecting from a sex-specific repertoire (phrase types), and these phrase types within duets are associated non-randomly to form a series of duet types (Mann et al. 2009). The presence of duet types almost certainly reflects the existence of a duet code (Logue 2006) between the members of each pair. A duet code is where one 53

54 of both members of a pair selects a particular phrase type depending on their partners last choice of phrase type (Logue 2006; Mann et al. 2009). These codes are thought to function in helping to identify an individual to its partner, or in identifying a pair to others that may be listening; alternatively, or in addition, they may play a role in improving duet coordination (Mann et al. 2009). The male Riverside Wrens introductory phrases are also linked non-randomly with the alternating male, female phrases that followed seeming to act as a cue for the format of the rest of the duet (Mann et al. 2009). The general form of the duet is, therefore: I (F M) n Where I is the male introductory phrase, M is the male song phrase, F is the female song phrase and n is the number of cycles in the duet Birds duets and conservation Identifying individuals within a population can generate information on life history parameters, generate input data for conservation models, and highlight behavioural traits that could affect management decisions and error or bias within census techniques (Terry et al. 2005). Vocal individuality is an emerging non-invasive method of identifying individuals within a population in order to count and monitor and count individuals over time (Terry et al. 2005). Where it is difficult to detect animals or they are sensitive to disturbance the vocalizations of animals can be used as an alternative marking technique (Terry et al. 2005). Knowledge of how individuals within a population communicate can be used to generate information from measures of habitat use to genetic fitness (McGregor et al. 2000; McGregor & Peake 1998) Vocal individuality The vocalizations used to identify individuals should demonstrate low within-individual variation and high between-individual variation (Falls 1982; Stoddard 1996). Individuality in vocalizations is likely a feature of all vocally active species and is a product of genetic, developmental and environmental factors (McGregor 1993; Suthers 1994). Vocal individuality distinguishes between two terms: discrimination and identification. Discrimination requires that individual's vocalizations differ enough at one point in time to be separated and is limited to census tasks. Identification requires that individual s vocal features remain constant enough to be associated with that individual for periods of time and allows individuals to be monitored over time thus yielding valuable life history data (Terry et al. 2005). Identification has greater conservation implications but is harder to demonstrate (McGregor et al. 2000; Peake 1998). 54

55 Population models and identifiable individuals Population Viability Analysis (PVA) is the most commonly used model in conservation biology. PVA uses demographic and environmental factors to predict the potential fate of a population (Lande 1993; Kendall 2002). These models include non-random individual variation in order to reduce the demographic stochasity, which can lead to extinction risk being exaggerated (Kendall 2002; Fox & Kendall 2002). By identifying individuals it is possible to generate more accurate predictions concerning the nature of individual variation in such parameters as fecundity and survival, therefore increasing the accuracy of PVA models (Sutherland 1996; Kendall 2002) Behavioural traits Identifying individuals can illuminate behavioural traits that may affect the conservation value of subsections of a population (McGregor & Peake 1998). An example of this is in breeding rates within populations. For example 17% of the population of common buzzard (Buteo buteo) accounted for 50% of the following years fledglings (Kruger 2001) and there are many more examples of the significant differences in reproductive success (Clutton- Brock 1988; Gompper et al. 1997; Kelly 2001) Ethical considerations Ethical concerns surround the capture and handling of study species (Nimon et al. 1995; Culok & Wilson 1991). It is safe to assume that capture and handling of animals can have deleterious effects, even if they are not immediately obvious. Therefore the default should be non-invasive techniques, such as vocal individuality (Terry et al. 2005). This non-invasive technique can be used to generate information about habitat use, survival, recruitment, immigration and emigration. This life-history data can then be used to test hypotheses concerning factors leading to population decline or the effects of management strategies (Greenwood 2000; Caughley & Gunn 1996). If vocalizations of individuals remain constant over time then this technique can be used as a long-term monitoring option (Gilbert et al. 2002; Galeotti & Sacchi 2001) Hypothesis Before more detailed questions regarding duets in C. semibadius can be investigated it is important to have a clear understanding of the Riverside Wren s full repertoire. The project 55

56 aims to document the full repertoire of six pairs in neighbouring territories, to determine the extent, if any, of phrase type sharing between the sexes in neighbouring territories and the extent, if any, of duet type sharing between neighbouring territories. The hypothesis is that there will be common phrase types shared between neighbouring males and females, while duet types will be unique to the pairs in each territory Methodology This research project was conducted on land owned by Osa Conservation at the Osa Biodiversity Centre, Piro on the Osa Peninsula, Costa Rica (N , W ). This private reserve consists of 1,700 ha, and is managed by the administrative unit of ACOSA (Osa Conservation Area) within the Costa Rican National System of Conservation Areas (SINAC) (Sánchez-Azofeifa et al. 2003). The study site consists of three kilometres of river edge Recording Prior to the commencing of recording, the territories of 26 pairs of Riverside Wrens were mapped along the Rio Piro. Pairs of Riverside Wrens were captured using mist nets (work carried out by N.I.Mann, P.J.B.Slater and C.N.Templeton); of the 26 pairs 22 have at least one member of each sex ringed using bright leg bands in unique colour combinations to facilitate individual recognition (N.I.Mann, P.J.B. Slater, C.N.Templeton pers.comm). Recordings of at least three hours of continuous song were taken in six territories (T29, T28, T30, T13, T12 and T10). Recordings were made between 05:00 and 08:30 using a Sennheiser K6/M66 directional microphone and Handy H2 solid recorder. Binoculars were used to identify the birds, to determine which individual was singing and what its contribution was to the song; this information was dictated into the recorder. Recordings were conducted four mornings per week for two months (May-June 2013). The recording required at least two researchers: one to record and narrate the recordings (territory, individual sighted, leg banded sighted, position of birds in vegetation, type of vocalisation i.e. duet, male/female solo) and another to locate and identify the individual birds using the binoculars describing the birds movements for the recorders. 56

57 Table 2 Ring codes for territories studied. The left leg ring is given before the right, & means one ring with narrow stripes of colours. Bk: black; DB: dark blue; DP: dark pink; LB: light blue; LG: light green; LP: light pink; M: mauve; O: orange; R: red; W: white; Y: yellow. PAIR Male Female T10 W&Bk Y T12 LP DB NR T13 LG O NR T30 R&Y R&Y M M T28 LB R DP LG T29 Bk&W Bk&W LP O Playback of conspecific song has been used to prompt birds to sing more of the song types in their repertoire (Langmore 1998; Hall 2006; Logue 2006; Mann et al. 2006; Catchpole & Slater 2008; Mann et al. 2009) Analysis Spectrograms of duets were generated using the acoustic analysis programme Syrinx (John Burt A single song was defined as temporally separated from others by 2 seconds either as a solo or as a duet (Mann et al. 2009). A single phrase was composed of a stereotyped sequence of notes (defined by a continuous trace on the spectrogram) produced by a single individual (Mann et al. 2009). A song could therefore be a solo performance comprising of one or more phrases from a single bird or a duet with phrases from two birds (Mann et al. 2009). Additional data collected included: the structure of the song and arrangement of phrase types present, including which bird had started the song and which bird terminated it, how many full cycles it involved, and whether the birds overlapped one another. Many duetting pairs combine specific phrases from their repertoires to form non-random associations (Logue 2006). I followed the methodology of Mann et al. (2003, 2009) to determine whether such associations are present, whereby a G test was used to compare the observed frequency of specific associations with random expectation assuming a Poisson distribution. The quantitative details of duet structure have been averaged over all recordings from each territory. The mean and standard deviation across territories were calculated with the knowledge that some song characteristics (i.e. relative frequency of duet 57

58 initiation by particular sex, frequency of replies to partner songs, and types and organisation of songs used from repertoire) can vary according to the time of day (Mann et al. 2009) a Bias The patterning of song types in a song bout introduces biases when attempting to document repertoires. Catchpole and Salter (2008) refer to two major patterns: eventual or immediate variety. If a bird sings with eventual variety they repeat one song type multiple times before moving on to another song type. If one spends too little time sampling then it may be that the bird s repertoire is underestimated. If a bird sings with immediate variety they cycle through their song types, sing one then the next with little or no repetition. If too little time is dedicated to sampling in these circumstances it would be easy to overestimate the repertoire of these birds (Catchpole & Slater 2008). The riverside wren sings with eventual variety which means it is important that sampling continues for enough time to ensure a robust estimate of repertoire size. To minimize this bias in my research project I have dedicated two months to data collection. By plotting the number of songs recorded against the time spent recording as described above I was able to estimate when I had a complete repertoire for each of the pairs in the six territories Results Recordings were transcribed into a spreadsheet following visual analysis in Syrinx. Male, female and introductory phrases were defined for the recordings so far collected. Recordings were conducted in territory 29 over the course of 2.5 hours. So far 23 male phrases, 19 introductory phrases, 13 female phrases and 23 duets have been recorded in this territory. Following two hours of recording in territory male phrases, 21 introductory phrases, 11 female phrases and 12 duets have documented. 58

59 Figure 10.1 Example of a duet from territory 29. The introductory phrase is denoted by intro, with male and female phrases denoted by M and F respectively. Figure 10.2 Example of a duet from territory 28. The introductory phrase is denoted my intro, with male and female phrases denoted by M and F respectively. In territory 28 two hours of recording have been completed. Six male and female phrases have been recorded so far, 16 introductory phrases, and three duet types have been documented. 59

60 Figure 10.3 Example of a duet from territory 30. The introductory phrase is denoted my intro, with male and female phrases denoted by M and F respectively. Figure 10.4 Example of a duet from territory 13. The introductory phrase is denoted my intro, with male and female phrases denoted by M and F respectively. In territory male phrases, 19 introductory phrases, 11 female phrases and 12 duets have been documented over two hours of recording. 60

61 Figure 10.5 Example of a duet from territory 12. The introductory phrase is missing, the female began the duet (F), with the male (M) joining in. Figure 10.6 Example for a duet from territory 10. The introductory phrase is denoted my intro, with male and female phrases denoted by M and F respectively. The final female phrase is curtailed - the final element of the phrase was not sung. Two hours of recording were carried out in territory 12 where seven male phrases, nine introductory phrases, five female phrases and five duets have been documented. Only 30 minutes of recordings have been conducted in territory 10. So far five male phrases and introductory phrases have been documented and four female phrases and four duets have been documented in this territory. 61