INTRODUCTION

Understanding the relative abundance and habitat associations of a species, and the impacts that human activities have on them, is vital for wildlife conservation planning (Ferreguetti et al. 2018). Natural habitat structural characteristics, such as elevation and vegetative density, influence species presence, as do anthropogenic factors such as hunting, and proximity to roads (Ferreguetti et al. 2018), particularly in tropical forest ecosystems. Tropical forests are frequently the focal point of global biodiversity conservation; they support approximately 50% of all known and described species, and a much larger percentage that has not yet been described (Wright 2010). Many tropical forest mammals are of vital importance to tropical ecosystems for the role they play in ecosystem structure and function (Galetti et al. 2010).

The red-rumped agouti Dasyprocta leporina Linnaeus, 1758 is a medium-sized terrestrial rodent that is native to the southern Caribbean region of tropical South America, including French Guiana, Venezuela, Trinidad and Tobago, Guyana, Suriname, and Brazil (Jorge & Peres 2005). They typically weigh between 3-6 kg and can be found in high population densities ranging up to 63/ km2 (Jorge & Peres 2005). Agoutis are diurnal and exhibit bimodal activity patterns affected by both abiotic and biotic factors; i.e., they are usually highly active in the morning, rest during mid-day, and then are increasingly active again until sunset (Magalhães & Srbek-Araujo 2019). Agoutis have a broad diet of seeds, fallen fruits and nuts, and tend to bury excess food in holes for later consumption in times of scarcity (Rodrigues et al. 2006). Since they do not retrieve all hoarded seeds, they are one of the most effective seed dispersers in Neotropical forests (Kenup et al. 2017). This “scatter- hoarding” behavior positively enhances seed survival by increasing the probability that larger seeds will germinate; agoutis are therefore considered very important to the establishment and maintenance of forest structure (Galetti et al. 2010).

Although the red-rumped agouti is classified as ‘Least Concern’ (IUCN) primarily because of its wide distribution and common occurrence inside and outside of protected areas (Jax et al. 2015), knowledge of their ecology in areas where they haven’t been studied can benefit our collective understanding of forest ecology. In Trinidad, knowledge of the ecology and status of the red-rumped agouti is limited, despite it being an exploited species (Ali & Jones 2020). Knowledge of factors that may directly or indirectly impact agouti distribution and abundance have implications for understanding local ecosystem processes, and for the development of new hunting regulations and enforcement strategies to improve management outcomes. Aside from hunting, other factors that can potentially influence agouti habits and ecology include elevation and topography, and vegetation structure or density (Jax et al. 2015), as well as human activity or presence.

Our study used camera-traps to explore the relationship between the occurrence of red-rumped agoutis and hunting, elevation, proximity to paved roads, and vegetation density. We compared the importance of these factors on the relative abundance index (RAI) of agoutis across two sites of contrasting habitats, and the level of protection they are afforded from human hunting. We predicted that: agouti RAI would exhibit a negative relationship with anthropogenic disturbances, specifically with respect to hunting; 2) agouti RAI would be higher where the species is completely protected; 3) agouti would be detected less frequently closer to paved roads (hereafter "roads") due to expected higher levels of human activity; and 4) agouti RAI would be positively correlated to vegetative density, where we’d expect agoutis to find greater availability of food and suitable cover.

MATERIALS AND METHODS

Study Area

Our study was conducted at two study site locations: (1) the Arena Forest Reserve, and (2) the Spring Hill Estate. The Arena Forest Reserve (Fig. 1a) is a lowland, multi-use forest located 8 km south of the Borough of Arima and 1 km west of Cumuto; 10° 33’ 00.0" N 61° 13’ 00.0" W, comprising 30 km² of private and state land (Ministry Of The Environment And Water Resources 2016). From the early 1900’s through 1931, the Arena Forest was subject to intensive exploitation in the form of logging due to its valuable timber resource; most harvesting efforts however were based on selective logging, as one goal was to foster natural forest regeneration (Ministry Of The Environment And Water Resources 2016). In 1963, the harvest of timber in Arena Reserve ended (Ministry Of The Environment And Water Resources 2016).

Currently the Arena Forest Reserve accommodates diverse flora and fauna. Average annual rainfall ranges between 2000 - 3000 mm, with average temperature ranging from 19º to 28º C under forest canopy, and 16º to 35º C in the open (Ministry Of The Environment And Water Resources 2016). The topography of the reserve is relatively flat and varies from gently undulating to occasional short steep slopes; elevation varies from 23 m above sea level (minimum) to 87 m above sea level (maximum) (Ministry Of The Environment And Water Resources 2016). The main phytophysiognomy of the area is seasonal evergreen forest dominated by Eschweilera subglandulosa and Carapa guianensis. Additionally, non-timber forest products in the form of fruits, nuts, and ornamental and medicinal plants, are still important in Arena (Ministry Of The Environment And Water Resources 2016). Legal hunting has a defined season in Trinidad from the beginning of October to the end of February (five months). This seasonal hunting occurs annually in the area usually by locals.

The second site, Spring Hill Estate (Fig. 1b), once a former cocoa-coffee-citrus plantation, is also the location of the world-famous Asa Wright Nature Centre (AWNC). Established in 1967, the AWNC occurs in the mountains of the Northern Range approximately seven miles north of the town of Arima (10.717933°N 61.298267°W). It consists of approximately 6 km² of forested land now reclaimed largely by secondary forest; it is surrounded by older growth primary rainforest with steep slopes and canopy heights of greater than 30 meters (Asa Wright Nature Centre 2016). The area is very humid with temperatures varying between 18-30 ^oC annually due to its sub-montane location (Asa Wright Nature Centre 2016). The AWNC is an area internationally recognized for its protection of birds and other wildlife close to its natural state on Trinidad (Asa Wright Nature Centre 2016). Absolutely no hunting is permitted anywhere on Spring Hill Estate, even during the legal hunting season (Rutherford & Foon 2016).

Camera-Trapping

We conducted a camera-trap survey concurrently at both sites between 24 June and 6 November 2016. We deployed camera-trap (HCO Scoutguards) stations using the trail system at each study site; ten camera stations, each consisting of two opposite-facing cameras, were established in the Arena Forest Reserve, whereas five camera stations, also with each station consisting of two opposite-facing camera-traps, were established in the Spring Hill Estate. Camera-traps were set an average of about 500 m apart in a systematic grid or approximately uniform polygon pattern. All camera-trap station locations were chosen with no prior knowledge of wildlife abundance, presence, or diversity. Camera-traps were placed on trails and attached to the base of trees at about 50 cm above the ground (Tobler et al. 2008). All camera-traps were secured in steel boxes with locks and anchored to trees, and powered by eight lithium AA batteries; data was recorded on a 16GB SD card. Camera- trap settings were as follows: passive infrared (PIR) interval was set to five seconds; PIR sensor sensitivity was set to ‘high’; illumination was set to ‘short range’; and photo size was set to 8MP. Once a month the camera-traps were checked at all locations, and when necessary, batteries and SD cards were replaced. No attractants (i.e., sight, sound, scent) of any kind were used to lure animals to camera-trap locations. At the end of each sampling period, all memory cards at both study sites were recovered.

Data Preparation

Image files (photos) were renamed using the freeware program “ReNamer", then sorted and analyzed following Sanderson (2013) using an automated camera-trapping annotation software; this aided in the analysis of records to derive estimates of relative abundance for the agouti at each site. O’ Brien et al. (2003) recommended differentiating each species detection from a previous detection using some fixed interval to assign independence, unless an image clearly displayed a different individual. They noted that a 60-minute (1 hour) interval is commonly used and we followed their recommendations (i.e., we used a 1-hour interval) to declare the independence of sequential images of the same species. In those instances, where multiple photos were grouped together over an interval of <60 minutes, we used only the first such image obtained for subsequent analyses (Srbek-Araujo & Chiarello 2013). We then calculated the relative abundance index (RAI) of Dasyprocta leporina (number/100 trap-nights) using the formula:

Total number of independent pictures at each camera trap site x 100 Duration of camera trap-nights at each site

We also analysed occupancy data for agoutis using the software PRESENCE. Rovero et al. (2014) demonstrated that true occupancy estimation takes into consideration that a species may be present but not detected, with detection probability being p<1. During a visit when a species is detected it is denoted a value of "1" and when non-detection occurs it is indicated as "0" (Pease et al. 2016). Camera trapping data is very well linked to occupancy analysis in that detection data can be gathered over numerous sampling occasions (Rovero et al. 2014).

Covariate Analysis

For our analyses, we explored several coarse scale covariates relative to each camera-trap location, including: hunting, distance to roads, vegetation density, and elevation. For hunting, we calculated the relative abundance index (RAI) of hunters (people) and dogs per 100 trap-nights using the combined number of independent pictures of hunting dogs, people with weapons, and people with weapons accompanied by hunting dogs, at each camera trap station (15 sites) for both locations. Between March and September, it is illegal to hunt agoutis and other wildlife anywhere on Trinidad (Ministry of the Attorney General and Legal Affairs 2016). Over the course of our study (June-September) however, we obtained numerous photos of illegal hunters with their dogs. Because elevation can also impact the local ecology and environmental conditions for a species due to its relationships with forest type and composition, temperature, and other abiotic factors relevant to the species (Brodie et al. 2014), we included it as a covariate. All elevation readings were derived from a digital elevation model (DEM) of Trinidad we produced. We also used proximity of sampling units (i.e., camera-traps) to nearby roads as a covariate, which was measured via proximity analysis in ArcMap. This was to capture whether or not there might have been an effect of vehicular traffic. In the case of Spring Hill Estate, maintained human trails were all considered for the purposes of this study.

We also assessed the importance of vegetation density using the point-centred quarter method. This is a plot-less sampling method that involves the measurement of distances for a random section of trees, generally along a transect, and tends to be more efficient in achieving results (Silva et al. 2017). We used this method to establish 30 m transects at each camera station, and then to determine sampling “quarters”. Every 10 meters (three per transect), we established a sampling area divided into four quarters and labelled I, II, III, and IV (Silva et al. 2017) in a clockwise motion starting with 12-3 o’clock. The distance between the sampling point, and the nearest tree of three different size categories, was measured and recorded for each quarter. These categories included: the nearest sapling (<5cm dbh), trees of intermediate size (6-20cm dbh), and mature or larger trees (>20dbh); we also recorded the distance to the nearest palms, snags, and bamboo stands. This process was repeated at each camera station. After recording the data, we calculated vegetation density as:

D = A d²

where D is density, A is the specified area (1 m²), and d is the average of the four distances measured from the centre of the sampling point to the nearest tree in each quarter.

Statistical Analyses

All analyses were conducted in R 3.3.2. The relative abundance index (RAI) of D. leporina and the relative abundance index (RAI) of hunters and dogs were calculated as the number of independent pictures per 100 trap-nights. We performed simple linear regressions to test the hypotheses that each of our single predictor variables (i.e., our coarse scale covariates) exhibited a direct relationship with the relative abundance of agoutis (i.e., dependent variable).

The logs of our predictor variables were calculated to normalize the data, reducing the higher end range of values, expanding the lower end (Feng 2014), and ultimately log-transforming the data to establish more of a linear relationship. We also tested for multicollinearity among the variables by assessing for autocorrelation between independent variables using the Variance Inflation Factor (VIF), which measures the inflation of the variance estimation regression coefficient to test if there is a correlation among independent variables (Shrestha 2020). If VIF=1, this indicates that there is no correlation between factors of independent variables. A VIF between 5 and 10 can indicate a somewhat higher correlation between independent variables, which may be problematic. Multicollinearity can also occur among predictors in a multivariate model if the VIF ≥5 to 10. Furthermore, a VIF value of >10 could conflate the independence of model parameter estimates (Mason & Perreault 1991). To determine if significant differences existed in agouti’s relative abundance (RAI), or the “relative abundance” of hunters and dogs, between our two site locations (p<0.05; 95% confidence interval), we used a Mann-Whitney U test. To visualize the relative abundance of agouti, as well as hunters and dogs, at each study location, we used box plots to group and compare the data. We further used generalized linear models (Poisson regression) to test all potential combinations of predictor variables, and to determine their effects on agouti relative abundance at the two sites. Akaike’s Information Criterion (AIC) and R2 values were used to rank these effects from strongest to weakest, and to help select the best-fit model (Sellers & Shmueli 2010). Finally, we compared our relative abundance index (RAI) results to those of similar studies.

RESULTS

We accumulated 2016 camera trap-nights (CTN) over 15 camera-trap stations across both site locations: 680 CTNs for Spring Hill Estate (680 CTNs), and 1336 CTNs for the Arena Forest Reserve. Approximately 59.5% (n=1 554) of all photos contained independent images of Dasyprocta leporina, which were then used for analysis. Some inconsistencies in camera trap-nights among site cameras were due to malfunctions. Although we obtained photographs of other mammals, rodents, reptiles and birds, those records are beyond the scope of this project. We found that the mean RAI of agoutis across all camera-traps in Spring Hill Estate (120.94) was more than twice as high as the mean RAI for the Arena Forest Reserve (54.67/ 100 camera trap-nights) (Table 1). This higher relative abundance was also reflected in the range of RAI values for individual camera-traps in Spring Hill Estate (range: 52 to 245 number/100 trap-nights) as compared with the Arena Forest Reserve (range: 8 to 124 number/100 trap-nights). However, a Mann-Whitney-U test indicated that the difference in RAI of agoutis between both locations was not significant (p=0.129; U=38).

In testing for multicollinearity among our independent variables, we found all exhibited relatively low to no multicollinearity; VIF values for hunting (2.04) and elevation (2.06) were higher than values for proximity to roads (1.47) and vegetation density (1.55), but still relatively low overall. In addition, as analysis of agouti data demonstrated agouti were detected regularly at all camera-traps irrespective of survey interval size (i.e., Ψ =1 for all survey interval sizes), we further considered occupancy modelling as not appropriate for this study.

Of the four covariates we tested, only three appeared to affect the relative abundance of agouti: hunting, distance to roads, and elevation. The log of the relative abundance (RAI) of hunting, as measured using the occurrence of people and their dogs, had the strongest effect (or strongest relationship) on agouti relative abundance (RAI) with an R2=0.670 (Fig. 2a; p=0.000193). We also found a higher relative abundance of hunters and dogs in the Arena Forest Reserve compared to the Spring Hill Estate. Our results indicated a large disparity between the two study locations, as we found significantly more hunting in the Arena (per 100 trap-nights) than the Spring Hill Estate (Mann-Whitney U test; p=0.0118; U=46), not too surprising perhaps given hunting is never permitted legally in the latter. We also found the RAI of hunters and dogs to gradually increase from June to October (Fig. 3), a period which also saw a corresponding decrease in the relative abundance of red-rumped agouti at both locations (Fig. 4). We also found a significant positive relationship overall between elevation, and the RAI of agoutis (Fig. 2b; R2=0.298; p=0.0354); i.e.,as elevation increased, the relative abundance of agoutis increased. Finally, we found a strong inverse or negative relationship between the RAI of agouti and proximity to roads (Fig. 2c). This means that as distance to roads increased, the RAI of agoutis decreased ((p=0.00126) < 0.05; R2=0.564).

Finally, our overall generalized linear regression model (Poisson distribution; Table 2) indicated that Model 1—hunting and distance to roads combined—appeared to be the best model (6AIC = 0, R2 = 0.80) and also exhibited a greater goodness of fit for the data relative to competing models. Models 2, elevation + hunting + distance to roads (6AIC = 1.95, R2=0.80) and 3, hunting + distance to roads + vegetation density (6AIC=1.96, R2=0.80) were ranked second and third respectively, in their importance to agouti relative abundance (RAI) (6AIC < 2).

DISCUSSION

Although many studies have demonstrated the importance of terrestrial mammal fauna and their abundance to ecosystem functioning and services, few baseline studies of mammal abundance or ecology in Trinidad and Tobago have ever been carried out. The results from our study in northern Trinidad demonstrate the importance of several variables known to be important elsewhere, on two populations of Dasyprocta leporina, a locally important bushmeat species on Trinidad. Despite wide variation in agouti detection rates across individual cameras, which may have led to overlapping standard deviations and thus non-significant results in comparisons between study sites, hunting appeared to have the most significant negative impact on the RAI of agouti in both study areas. Unexpectedly, hunters were recorded on Spring Hill Estate—a strict protected area, for the first time; less unexpected was the documentation of hunting in the Arena Forest Reserve, a multi-use area where legal hunting is permitted seasonally. This may be due to greater pressure on the island’s natural resources. For example, over a 35-year span, Trinidad and Tobago has seen an almost threefold increase in hunters, from about 3500 in 1975 to approximately 10000 in 2010 (Hailey & Cazabon-Mannette 2011). Additionally, several factors may have influenced how hunters choose specific hunting locations (Lebel et al. 2012). Many images indicated that a common technique for hunting in our study locations was with the use of firearms and dogs. Evidence from the mainland suggests that dogs can be extremely effective at identifying terrestrial game species, such as agoutis (Koster 2008). In addition, the rate at which game species are detected when dogs are used in hunting activities may be significantly more than the encounter rate without them or using rifles alone (Koster 2008). Agoutis and many other tropical forest mammals worldwide are often a preferred source of wild meat wherever they occur, and are thus important to the diets of many rural people and communities (e.g., Brodie et al. 2014; Jones et al. 2019). Prey visibility by hunters is often a vital determinant of successful harvest (Lebel et al. 2012). For example, hunters using certain habitat types, such as those with low vegetative cover, may have greater success in seeing prey further away, thus increasing their ability to capture and kill prey (Lebel et al. 2012). Food availability is another predictor as to why hunters may prefer certain locations (Lebel et al. 2012). Knowledge and understanding of what foods certain wildlife consume, and the availability of those foods in a particular area, may allow for a greater likelihood that a wildlife species will be located in a specific area (Lebel et al. 2012) and thus influencing a hunter’s decision.

In October when the legal hunting season opened, we observed a significant increase in hunting as indicated by the relative number of photos of hunters and their dogs. The relative abundance of hunters and dogs in Arena Forest Reserve for example was approximately nine times higher than in Spring Hill Estate. This is so because, unlike Spring Hill, Arena lacks suitable monitoring and enforcement (Government of the Republic of Trinidad and Tobago 2011). The effects of road and trail access routes to destinations is yet another factor that may influence how a hunter chooses their location. The Arena Forest Reserve is a more accessible lowland forest area amidst adjacent urban areas criss-crossed by trails, characteristics which facilitate greater use by people (Margono et al. 2014). In contrast, Spring Hill Estate is well-known as a protected area; they have staff to tend to the area, and gates surrounding the front and restricting the entry. Furthermore, the steep, mountain location makes entry by other means more challenging. In fact, most people visit Spring Hill Estate and the Asa Wright Nature Center to enjoy recreational bird watching and nature in a secure environment. Regardless of where they are hunted on Trinidad, we know that the continuous hunting of agoutis elsewhere has caused them to become more vulnerable to overexploitation (Cullen et al. 2000).

Agoutis are also known to occur at higher abundances, or occur more frequently, in areas with tall lowland forest, and a diversity of palms and nuts. According to Ferreira De Pinho et al. (2017), more pronounced elevation gradient environments can make it difficult for medium to large-sized terrestrial animals to move locally. Elevation is also used as a proxy for forest productivity, i.e., forest productivity declines with increasing elevations. Elevation and temperature therefore can affect the occupancy and distribution of agouti species. In contrast to our findings, Ahumada et al. (2013) observed that agouti occupancy decreases with increasing elevation. They contended that because of the slower growth of palm species and other trees at higher elevations and lowered resilience to fluctuating temperatures, such habitats are unsuitable for agoutis. They also found agouti occupancy was positively correlated to taller forests and closed canopy relating to mature forests. These forest types occur more frequently in less disturbed, lowland forested areas (Ferreguetti et al. 2018), but this is the case for the Arena Forest. However, both of our study sites fell between 32-425 m in elevation and thus technically occurred within the 0-500 m range of “lowland forest” elevation as defined by Ahumada et al. (2013).

Roads may also compromise access to certain game species locally, even within protected areas (Espinosa et al. 2014). Roads can facilitate greater human access and vehicular traffic in areas presenting otherwise greater logistical challenges, and we expected that agouti’s relative abundance would be higher further away from roads. Instead, we found that agoutis were more abundant closer to roads, despite more evidence of human disturbance and foot traffic. According to Jax et al. (2015), agoutis were recorded more frequently on narrow foot or game roads than on wider roads and roads which permit greater traffic. Furthermore, Dias et al. (2019), stated that agoutis were more likely to frequent forest edges, which contradicts some other studies. It is important to note, that a specific species’ tendency to use trails and roads may vary depending on their ease of locomotion in a given habitat type, understory vegetation density, or the probability of predation and other human pressures (Di Bitetti et al. 2014). Ouboter & Kadosoe (2016) found that medium-sized predators, including ocelots, generally avoided areas with tourists and other disturbances. Interestingly, red-rumped agoutis may prefer disturbed areas not only because of potential resource availability, but also because predation pressure near tourist-frequented roads and trails may be less, and thus serve as a “refuge” from predation (Ouboter & Kadosoe 2016).

Another possible reason for higher relative abundances near roads and other areas of higher human activity may be due to the artificial provision of food to wildlife (Tortato & Izzo 2017). Regardless of the ethics which are debatable, the act of feeding wildlife has become extremely popular among visitors seeking contact with target species in nature (Tortato & Izzo 2017). Spring Hill Estate has such tourists moving across their grounds, and there is also a designated area near headquarters for bird-watching and bird feeding. This area contains bird feeders full of seeds, nuts, and fruit, which are replenished daily, and at which agoutis may frequently be observed. Agoutis in Spring Hill therefore are well adapted to and familiar with people, may be less wary of human activity, and may even associate humans with the possibility of food or greater food availability. We should also note that at both study sites, we did not place camera-traps on roads or trails that are open to motorized traffic. Although the setting of camera-traps closer to certain trails could also therefore help clarify the local relative abundance of red-rumped agoutis (Ouboter & Kadosoe 2016), unless sampling design is carefully considered, such camera placement could also prove more biased when study objectives include wider investigations of agouti distribution, ecology, or habitat preferences.

Interestingly, we found no significant relationship between the relative abundance of agouti and vegetation density. This was not unexpected, however, given that both areas encompassed forest with a high density of palm species and other potential nut, fruit, or seed-producing food sources for agoutis (Ahumada et al. 2013). According to Duquette et al. (2017), agouti density tends to be positively correlated to food density. Areas that are mainly enclosed and comprised of seed-bearing primary and secondary forests are generally prime habitats and contain optimal food resources for agoutis (Duquette et al. 2017). Our study areas were comprised of both these preferred forest types. A study in the Brazilian Atlantic forest concluded that one-third of all palm species documented by the authors likely germinated as a result of seed and nut dispersal by scatter-hoarding rodents, especially agoutis (Ferreguetti et al. 2018). This important activity carried out by red-rumped and other agouti species across Latin America helps to safeguard many palm seeds and nuts from predation, thereby improving seed germination success and sapling growth (Kuprewicz 2015). Through our own transect surveys, we found that the density of palms across both locations was relatively high, a potential reason as to why vegetative growth and density was not an influencing factor on agouti RAI at either site.

Lastly, although camera-traps are frequently efficient tools for studying medium-sized mammals, including red-rumped agoutis, we should contextualize some potential constraints stemming from our approach. We note for example that it is very likely we recorded captures of the same individual at camera-traps many times (Mccleery et al. 2014). This is despite the fact that we consistently used one-hour periods to distinguish “independent” photos of agoutis. One reason for this is that agouti home ranges are small, usually only a few hectares (Silvius & Fragoso 2017). Because it is important that themethods used to assess local population status fora species are unbiased (Jax et al. 2015), these are certainly potential challenges. Small sample sizesover shorter study periods can also lead to results not robust enough to adequately contrast study sites (Cullen et al. 2000); in the case of our study, however,it is also possible that agouti population carryingcapacity at both sites was similar, resulting in a lackof significant differences in agouti RAI between sites. And since some camera-traps malfunctioned due to water accumulation and moisture build-up, we knowthat heavy rainfall impacted our sampling eficacy; it may have also impacted camera-trap sensitivity,response time, or other camera-trap features that werelied on to collect accurate data.

IMPLICATIONS FOR CONSERVATION

We found that hunting activity had an inverse relationship on the relative abundance of red-rumped agouti in both of our study areas. Because agoutis are often hunted legally and illegally despite their small size or body weight (Kosydar et al. 2014), management or regulation of recreational and sport hunting is essential to ensure that local populations on Trinidad at least aren’t overexploited. Failure to effectively protect hunted game species can also have cascading ecological impacts, due to their functional ecological importance to forests as stated herein (e.g., Galetti et al. 2010). In the northern forests of Trinidad, safeguarding mammals from hunting requires properly designed and implemented management plans to address different types of hunters, their means and techniques, and their motivations, to ensure the long-term viability of local populations.

Despite the economic importance of agouti populations in Trinidad, the monitoring and protection of this species have not been a priority at the Arena Forest Reserve. Our data provides strong evidence that the disruptive human activities at the Arena Forest Reserve location may pose a regular threat to the ecosystem more generally, if not to agoutis directly. It is important therefore that trained game wardens or rangers conduct regular monitoring or patrols (Jenks 2011). Increased enforcement presence in this area, and even at Spring Hill Estate, may serve as an effective deterrent to illegal hunting.

Due to the shortened sampling period, we were unable to properly compare RAI and the impact of hunting and weather changes across the full extent of hunting and non-hunting seasons. We recommend that future surveys occur over at least one year and integrate more sampling units (i.e., camera-traps) to obtain accurate baseline data for the monitoring of agoutis and other wildlife. We expect this would be one way to detect how future changes in hunting intensity or the expansion of hunting might impact local wildlife populations. Sampling for longer periods or across more areas would also yield more insights as to the role that predation and competition play in affecting agouti abundance.

Finally, although we did not find a significant difference in the relative abundance index (RAI) of agoutis between our two study locations, possibly due to the wide variation we observed in mean RAI among individual camera-traps within each site, we do believe the overall differences are reflective of the level of protection afforded to agoutis at Spring Hill Estate versus the Arena Forest Reserve. This is in part because of the importance of hunter presence on agouti’s RAI, and the inverse association between RAI and proximity to trails. We also recommend that additional studies seek to corroborate our findings at these and additional study sites, as the implications of more detailed studies would benefit the management of both forests, as well as other ecosystems.

We also believe that our scientific findings, and those collected in the future, should be used to enhance awareness among the public, particularly the hunting community, about the potential negative impact of overhunting or hunting outside the legal season and the need to avoid strict protected areas which prohibit these activities.

Acknowledgments

We would like to thank the University of the West Indies, St. Augustine Campus Diploma/MSc Biodiversity Conservation and Sustainable Development in the Caribbean programme, S.P.E.C.I.E.S. The Society for the Preservation of Endangered Carnivores and their International Ecological Study and EarthWatch International, for providing funding, support, and guidance throughout this project. We also want to thank Dr. Nigel Noriega, Carrie Lezama, and countless EarthWatch volunteers for their support with logistics, field work, data retrieval, and their contribution to project management.

REFERENCES

Ali, A. J., & K. R. Jones. 2020. Nutritive Value and Physical Properties of Neo-Tropical Rodent Meat—with Emphasis on the Capybara (Hydrochoerus hydrochaeris ). Animals 10:2134. https://doi.org/10.3390/ani10112134

Asa Wright Nature Centre. 2016. About the Asa Wright Nature Centre. http://asawright.org/about-the-centre/. Accessed April 14, 2017.

Ahumada, J. A., J. Hurtado, & D. Lizcano. 2013. Monitoring the status and trends of tropical forest terrestrial vertebrate communities from camera trap data: A tool for conservation. PLoS ONE 8(9). https://doi.org/10.1371/journal.pone.0073707

Brodie, J. F., A. J. Giordano, E. F. Zipkin, H. Bernard, J. Mohd-Azlan, & L. Ambu. 2014. Correlation and persistence of hunting and logging impacts on tropical rainforest mammals. Conservation Biology 29:110-121. https://doi.org/10.1111/cobi.12389

Cervera, L. et al. 2016. A camera trap assessment of terrestrial mammals in Machalilla National Park, western Ecuador. Check List 12:1868. https://doi.org/10.15560/12.2.1868

Cid, B., L. G. Oliveira-Santos, & G. Mourão. 2015. The relationship between external temperature and daily activity in a large rodent (Dasyprocta azarae) in the Brazilian Pantanal. Journal of Tropical Ecology 31:469-472. https://doi.org/10.1017/s0266467415000309

Cullen, L., R. E. Bodmer, & C. V. Pádua. 2000. Effects of hunting in habitat fragments of the Atlantic forests, Brazil. Biological Conservation 95:49-56. https://doi.org/10.1016/s0006-3207(00)00011-2

Dias, D. M., M. O. Almeida, T. G. Araújo-Piovezan, & J. O. Dantas. 2019. Habitat selection by mammals in an isolated fragment of Brazilian Atlantic forest. International Journal of Tropical Ecology 21:37-46.

Di Bitetti, M. S., A. Paviolo, & C. De Angelo. 2014. Camera trap photographic rates on roads Vs. off roads: location does matter. Mastozoología Neotropical 21:37-46.

Dubost, G. 1988. Ecology and social life of the red acouchy, Myoprocta exilis; comparison with the orange-rumped agouti, Dasyprocta leporina. Journal of Zoology 214:107-123. https://doi.org/10.1111/j.1469-7998.1988.tb04990.x

Duquette, J. F., L. Ureña, J. Ortega, I. Cisneros, R. Moreno, & E. E. Flores. 2017. Coiban Agouti (Dasyprocta coibae) Density and Temporal Activity on Coiba Island, Veraguas, Panama. Mammal Study 42(3):153-160. https://doi.org/10.3106/041.042.0305

Espinosa, S., L. C. Branch, & R. Cueva. 2014. Road development and the geography of hunting by an Amazonian indigenous group: Consequences for wildlife conservation. PLoS ONE 9(12). https://doi.org/10.1371/journal.pone.0114916

Feng, C. et al. 2014. Log-transformation and its implications for data analysis. Shanghai Archives of Psychiatry 26:105-109.

Ferreguetti, A. C., W. M. Tomas, & H. G. Bergallo. 2018. Density, habitat use, and daily activity patterns of the Red-rumped Agouti (Dasyprocta leporina) in the Atlantic Forest, Brazil. Studies on Neotropical Fauna and Environment 53:143-151. https://doi.org/10.1080/01650521.2018.1434743

Ferreira De Pinho, F., G. B. Ferreira, & A. P. Paglia. 2017. Influence of vegetation physiognomy, elevation and fire frequency on medium and large mammals in two protected areas of the Espinhaço Range. ZOOLOGIA 34:e11921. https://doi.org/10.3897/zoologia.34.e11921

Galetti, M., C. I. Donatti, C. Steffler, J. Genini, R. S. Bovendorp, & M. Fleury. 2010. The role of seed mass on the caching decision by agoutis, Dasyprocta leporina (Rodentia: Agoutidae). Zoologia (Curitiba) 27:472-476. https://doi.org/10.1590/s1984-46702010000300022

Government of the Republic of Trinidad and Tobago. 2011. National protected areas policy. http://www.ema.co.tt/new/images/policies/protected areas.pdf. Accessed April 15, 2017.

Hailey, A., & M. Cazabon-Mannette. 2011. Conservation of herpetofauna in the Republic of Trinidad and Tobago. Conservation of Caribbean Island Herpetofaunas Volume 1: Conservation Biology and the Wider Caribbean 183-218. https://doi.org/10.1163/ej.9789004183957.i-228.64

Head, J. S., C. Boesch, M. M. Robbins, L. I. Rabanal, L. Makaga, & H. S. Kühl. 2013. Effective sociodemographic population assessment of elusive species in ecology and conservation management. Ecology and Evolution 3:2903-2916. https://doi.org/10.1002/ece3.670

Islebe, G. A., B. Schmook, S. Calmé, & J. L. León-Cortés. 2015. Introduction: Biodiversity and conservation of the Yucatán Peninsula, Mexico. Biodiversity and Conservation of the Yucatán Peninsula, 1-5. https://doi.org/10.1007/978-3-319-06529-81

Jax, E., S. Marin, A. Rodríguez-Ferraro, & E. Isasi-Catalá. 2015. Habitat use and relative abundance of the spotted paca Cuniculus paca (Linnaeus, 1766) (Rodentia: Cuniculidae) and the red-rumped agouti Dasyprocta leporina (Linnaeus, 1758) (Rodentia: Dasyproctidae) in Guatopo National Park, Venezuela. Journal of Threatened Taxa 7:6739-6749. https://doi.org/10.11609/jott.o3915.6739-49

Jenks, K. E. et al. 2011. Using relative abundance indices from camera-trapping to test wildlife conservation hypotheses – An example from Khao Yai National Park, Thailand. Tropical Conservation Science 4:113-131. https://doi.org/10.1177/194008291100400203

Jiménez, C. F., H. Quintana, V. Pacheco, D. Melton, J. Torrealva, & G. Tello. 2010. Camera trap survey of medium and large mammals in a montane rainforest of northern Peru. Revista Peruana de Biologia, 17: 191-196.

Jones, K. R., K. R. Lall, & G. W. Garcia. 2019. Omnivorous Behaviour of the Agouti (Dasyprocta leporina ): A Neotropical Rodent with the Potential for Domestication. Hindawi 5 https://doi.org/10.1155/2019/3759783

Jorge, M. S., & C. A. Peres. 2005. Population density and home range size of red-rumped agoutis (Dasyprocta leporina) within and outside a natural Brazil nut stand in southeastern Amazonia. Biotropica 37:317-321. https://doi.org/10.1111/j.1744-7429.2005.00041.x

Kenup, C. F., R. Sepulvida, C. Kreischer, & A. A. S. Fernandez. 2017. Walking on their own legs: unassisted population growth of the agouti Dasyprocta leporina, reintroduced to restore seed dispersal in an Atlantic Forest reserve. Oryx 52:571-578. https://doi.org/10.1017/s0030605316001149

Kluever, B. M., E. M. Gese, & S. J Dempsey. 2016. The influence of wildlife water developments and vegetation on rodent abundance in the Great Basin Desert. Journal of Mammalogy 97:1209- 1218. https://doi.org/10.1093/jmammal/gyw077

Koster, J. 2008. The impact of hunting with dogs on wildlife harvests in the Bosawas Reserve, Nicaragua. Environmental Conservation 35:211. https://doi.org/10.1017/s0376892908005055

Kosydar, A. J., D. I. Rumiz, L. L. Conquest, & J. J. Tewksbury.2014. Effects of hunting and fragmentation on terrestrial mammals in the Chiquitano Forests of Bolivia. Tropical Conservation Science 7:288-307. https://doi.org/10.1177/194008291400700209

Kuprewicz, E. K. 2015. Scatter hoarding of seeds confers survival advantages and disadvantages to large-seeded tropical plants at different life stages. PLOS ONE 10:e0124932. https://doi.org/10.1371/journal.pone.0124932

Lebel, F., C. Dussault, A. Massé, & S. D. Cóté. 2012. Influence of Habitat Features and Hunter Behavior on White-Tailed Deer Harvest. Journal of Wildlife Management 76:1431-1440. https://doi.org/10.1002/jwmg.377

Magalhães, L. M., & A. C. Srbek-Araujo. 2019. Plasticity in the timing of activity in the Red-rumped Agouti, Dasyprocta leporina (Mammalia: Rodentia), in the Atlantic Forest of southeastern Brazil. Biota Neotropica 19:e20180625. https://doi.org/10.1590/1676-0611-bn-2018-0625

Margono, B. A., P. V. Potapov, S. Turubanova, F. Stolle, & M. C. Hansen. 2014. Primary forest cover loss in Indonesia over 2000–2012. Nature Climate Change 4:730-735. https://doi.org/10.1038/nclimate2277

Mason, C. H., & W. D. Perreault. 1991. Collinearity, power, and interpretation of multiple regression analysis. Journal of Marketing Research 28:268. https://doi.org/10.2307/3172863

Mccleery, R. A., C. L. Zweig, M. A. Desa, R. Hunt, W. M. Kitchens, & H. F. Percival. 2014. A novel method for camera-trapping small mammals. Wildlife Society Bulletin 38:887-891. https://doi.org/10.1002/wsb.447

Ministry of the Attorney General And Legal Affairs. 2016. Conservation of Wild Life Act. https://rgd.legalaffairs.gov.tt/laws2/alphabeticallist/lawspdfs/67.01.pdf. Accessed May 10, 2021.

Ministry of the Environment And Water Resources. 2016. Arena Forest Reserve [Pamphlet]. Forestry Division.

Nelson, H. & E. Devenish-Nelson. 2008. Population density of small mammals in the Arima Valley, Trinidad, West Indies. Living World: Journal of the Trinidad and Tobago Field Naturalists Club 8-14.

O’ Brien, T. G., M. F. Kinnaird, & H. T. Wibisono. 2003. Crouching tigers, hidden prey: Sumatran tiger and prey populations in a tropical forest landscape. Animal Conservation 6:131-139. https://doi.org/10.1017/s1367943003003172

Ouboter, E. P., & V. S. Kadosoe. 2016. Three years of continuous monitoring of the large terrestrial mammals of Brownsberg Nature Park, Suriname. Academic Journal of Suriname 7:643-660. https://doi.org/10.1371/journal.pone.0250390

Palmer, M. S., A. Swanson, M. Kosmala, T. Arnold, & C. Packer. 2018. Evaluating relative abundance indices for terrestrial herbivores from large-scale camera trap surveys. African Journal of Ecology 56 (4):791-803. https://doi.org/10.1111/aje.12566

Pease, B. S., C. K. Nielsen, & E. J. Holzmueller. 2016. Single- Camera Trap Survey Designs Miss Detections: Impacts on Estimates of Occupancy and Community Metrics. PLOS ONE 11(11):e0166689. https://doi.org/10.1371/journal.pone.0166689

Peres, C., & J. Barlow. 2004. ECOLOGY | Human influences on tropical forest wildlife. Encyclopedia of Forest Sciences 90-95. https://doi.org/10.1016/b0-12-145160-7/00012-0

R Core Team. 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

Rodrigues, R. F., A. M. Carter, C. E. Ambrosio, T. C. Santos, & M. A. Miglino. 2006. The subplacenta of the red-rumped agouti (Dasyprocta leporina L). Reproductive Biology and Endocrinology 4-31. https://doi.org/10.1186/1477-7827-4-31

Rovero, F., F. Zimmermann, D. Berzi, & P. Meek. 2014. Which camera trap type and how many do I need? A review of camera features and study designs for a range of wildlife research applications. Hystrix, the Italian Journal of Mammalogy 24:148- 156.

Rutherford, M. G., & K. C. Foon. 2016. Activity patterns of terrestrial mammals at Springhill, Arima Valley, Trinidad. Living World, Journal of the Trinidad and Tobago Field Naturalists’Club. http://ttfnc.org/livingworld/index.php/lwj/article/view/404/572

Sanderson, J. & G. Harris. 2013. Automatic data organization, storage, and analysis of camera trap pictures. Journal of Indonesia Natural History 1.

Santos-Fita, D., E. J. Naranjo, & J. Rangel-Salazar. 2012. Wildlife uses and hunting patterns in rural communities of the Yucatan Peninsula, Mexico. Journal of Ethnobiology and Ethnomedicine 8:38. https://doi.org/10.1186/1746-4269-8-38

Sellers, K. F., & G. Shmueli. 2010. A flexible regression model for count data. The Annals of Applied Statistics 4:943-961.

Shrestha, N. 2020. Detecting Multicollinearity in Regression Analysis. American Journal of Applied Mathematics and Statistics 8:39-42.

Silva, L. B., M. Alves, R. B. Elias, & L. Silva. 2017. Comparison of T-square, point centered quarter, and n-tree sampling methods in Pittosporum undulatum invaded woodlands. International Journal of Forestry Research 1-13. https://doi.org/10.1155/2017/2818132

Silvius, K. M., & J. M.V. Fragoso. 2017. Red-rumped Agouti (Dasyprocta leporina) Home Range Use in an Amazonian Forest: Implications for the Aggregated Distribution of Forest Trees. Biotropica 35:74-83. https://doi.org/10.1111/j.1744-7429.2003.tb00264.x

Srbek-Araujo, A. C., & A. G. Chiarello. 2013. Influence of camera- trap sampling design on mammal species capture rates and community structures in southeastern Brazil. Biota Neotropica 13:51-62. https://doi.org/10.1590/s1676-06032013000200005

Tobler, M. W., S. E. Carrillo-Percastegui, R. L. Pitman, R. Mares, & G. Powell. 2008. An evaluation of camera traps for inventorying large- and medium-sized terrestrial rainforest mammals. Animal Conservation 11:169-178. https://doi.org/10.1111/j.1469-1795.2008.00169.x

Tortato, F. R., & T. J. Izzo. 2017. Advances and barriers to the development of jaguar-tourism in the Brazilian Pantanal. Perspectives in Ecology and Conservation 15:61-63. https://doi.org/10.1016/j.pecon.2017.02.003

Underhill, J., & P. Angold. 2000. Effects of roads on wildlife in an intensively modified landscape. Environmental Reviews 8:21-39. https://doi.org/10.1139/er-8-1-21

Wright, S. J. 2010. The future of tropical forests. Annals of the New York Academy of Sciences 1195:1-27. https://doi.org/10.1111/j.1749-6632.2010.05455.x

Información adicional

Editor asociado: Andrea Previtali

Cite as: Sharmila C. Ganpat, Anthony J. Giordano & Luke V. Rostant. 2021. Relative abundance of a geographically isolated population of red-rumped agouti (dasyprocta leporina): a first assessment for Ttrinidad, 28(2):e0617. https://doi.org/10.31687/saremMN.21.28.2.0.04.e0617

CONFLICT OF INTEREST: The authors have declared no conflicts of interest for this article.