Density of red-brocket deer (Mazama americana trinitatis cetartiodactyla: cervidae) on the continental island of Trinidad, WI

ABSTRACT The Trinidadian red-brocket deer (Mazama americana trinitatis) is the only species of deer in Trinidad and is classified as a unique subspecies of red-brocket deer. Despite the importance of the species in Trinidad, to date there have been no published studies on their densities. We report on the densities of this species at four different sites in Trinidad over a 3-year period, with the first 2 years sampled during a hunting moratorium. Diurnal surveys were conducted, with density estimates calculated using King’s Estimator. Surveys were repeated in both the dry and rainy seasons. A total of 312 deer on 1131.4 km of transect were observed over the study. Density estimates varied between different sites (1–16 individuals/km2), and between one and eight individuals/km2 island-wide. Overall, the estimated density of red-brocket deer increased in two of four sites during the study period, indicating that the institution of the moratorium may have had a positive impact on populations of deer. The fact that density estimates varied considerably across the island suggests that harvest levels may need to be tailored for different sites.


Introduction
Red-brocket deer (Mazama americana) are widely distributed in the Neotropics from Southern Mexico to Northern Argentina as well as the island of Trinidad [1].Although brocket deer (Mazama spp.) are highly sought after by subsistence hunters in the Neotropics [2], they are among the least studied species of deer [3,4], also being classified as Data Deficient by the IUCN Red List [1].Beyond their value as a source of meat for people, brocket deer also likely play a role in forest regeneration [5].While the taxonomy of Mazama americana is still debated, many authors consider the species in Trinidad to be a valid subspecies M.a.trinitatis [1].This is of conservation concern since M. americana is one of five game mammals hunted in the island of Trinidad [6] and may be unique to the island.The hunting fraternity is quite large in Trinidad and Tobago, estimated in 2013 at 13,000 persons [7], with a five-month hunting season open at the beginning of October and closed at the end of February [6], and with the sale and consumption of wild meat among the local population being especially popular during the Christmas season [8].At present, there is no bag limit with respect to mammals that can be hunted nor is there a limit placed on the number of licences granted for hunting that is site specific based on the populations of these hunted mammals.Coupled with this, to date there has been no published research on the population size, distribution and trends of this subspecies, with only one report estimating the density of this species from a single site on Trinidad [9].Estimating how the density of this species varies across the entire island of Trinidad is critically important to our understanding of how the hunting of this species might be affecting the viability of its populations.
In Trinidad and Tobago, a moratorium on hunting was established in 2013 after hunting return forms indicated that there were high levels of hunting pressure [8].Many in the hunting fraternity consider the data on these hunting return forms to be unreliable [10], and in order to get more reliable data, wildlife surveys were undertaken in 2014 to determine the densities of hunted mammals on the island of Trinidad.Since the hunting moratorium expired in October 2015, comparison of density estimates before the expiration of the moratorium (2014 and 2015) can be made to those subsequent to when hunting was resumed (2016).The objective of this study was to estimate densities of red brocket deer immediately after the hunting moratorium was established in 2013, and after its expiration in 2015 to evaluate its effects on population numbers of this species.

Materials and methods
To estimate deer density in Trinidad, we used linetransect surveys, which is a useful method where resources are limited, and logistics difficult [11].Transect surveys have been widely used in other studies to produce reliable estimates of the Neotropical ungulates [3,12].All study areas had a similar sampling effort independent of their size: Central Range (60 transects in about 161 km 2 ), Nariva Swamp (60 transects in about 65 km 2 ), Victoria Mayaro Forest Reserve (VMFR) (60 transects in about 305 km 2 ), and Southern Watershed (34 transects, sampled twice on each occasion to match sampling effort of other sites, in about 42 km 2 ).Rainfall varies across the different sites, with between 2250 and 3000 mm in the Central Range, 2000-2750 mm in Nariva, 1750-2750 mm in the VMFR, and between 1500 and 2000 mm in the Southern Watershed.The different rainfall levels contribute to the predominant vegetation at the sites, with the Central Range and Nariva Swamps having evergreen seasonal forest, while that in the VMFR was a mix of evergreen seasonal forest and semievergreen seasonal forest, and semi-evergreen seasonal forest in the Southern Watershed [13].
The majority of the transects were carried out within forest reserves (Figure 1) which permit hunting, and wildlife sanctuaries contained within these, where hunting is prohibited.An equal number of transects were set up in forest reserves and wildlife sanctuaries within these areas to gain an overall density estimate for a given area.Wildlife surveys consisted of the distance-based transect method [14], with surveys being undertaken from 2014 to 2016 in both the wet season (June to December) and dry season (January to May).Estimates for redbrocket deer density were therefore calculated for both the wet and dry seasons within a given year for specific locations, and overall estimates for the island were also calculated.In order to maintain consistency within a given study area, transects were placed in similar secondary or primary forest, which was informed by the vegetation classification of Beard [13].Transects sampled at the Central Range Wildlife Sanctuary were aligned with those of a previous study [9] to allow comparisons with this prior study.For all other study sites, the location of the first transect was selected at random, with other transects being placed relative to this one to allow for regular spacing, usually oriented in a north-south or east-west direction (Figure 1).As much as practicable, transects were set within similar vegetation types within sites using the aforementioned classification of Beard (1946).To ensure that transects were set in closed forest, the location of transects was compared to Ikonos imagery from 2007.To ensure that there would be no double counting of deer, which may be flushed from one transect onto another, teams of four persons on each transect line sampled at least 1 km from one another.Sampling was conducted during mornings (06:00-0:900) and afternoons (15:00-18:00), with the GPS location, distance to the animal (where it was first spotted), and angle relative to the transect line being recorded when a deer was encountered.
Due to the number of independent encounters in each study area not exceeding 30 individuals within a given season/year for most sites, King's Estimator was calculated for all sites.King's estimator has been found to be one of the more reliable estimators for animal density [15], has been used considerably in the Neotropics [16], and has also been shown to perform similarly to preferred estimates of density using distance models [17].The King's Estimator is parameterized as follows: with n being the number of animals observed, L being the length of the transect, and R the mean distance to the animal.The standard error associated with the estimates within a given sample session was also calculated.

Results
The results of the surveys on red-brocket deer in Trinidad are provided in (Table 1).Over the 3-year sample period, 312 deer were observed on 1131.4 km of transects sampled.Higher estimates of the density were found in the Central Range (mean 10/km 2 , range 3-16, s.d. 5) and Southern Watershed (mean 8/km 2 , range 2-13, s.d.4), while lower estimates of the density were found in the Nariva (mean 2/km 2 , range 1-3, s. d. 1) and VMFR (mean 3/km 2 , range 2-3, s.d. 1).When the data were pooled, the island-wide estimated mean value was five individuals/km 2 (range 1-8, s.d.3).We did not observe any pattern with respect to seasonality in the detection and thus estimates of deer density but did note an increase in deer density overall from 2014 to 2016 across sites.

Discussion
Estimates of red-brocket deer vary considerably across its range, from 1 to 2/km 2 in Peru [3,18], 4 to 7/km 2 in Mexico [12], 7 to 21/km 2 on Barro Colorado Island, Panama [19], and 15/km 2 in the Atlantic Forest of Brazil [20], with an average density of about 11/km 2 across its range [21].The estimates derived for Trinidadian forests are therefore within the range seen in other Neotropical sites, with most being below the average value.A prior estimate of four 2 [1] 3 [1] 8 [4] 5 [3] individuals/km 2 was derived by Nelson et al. [9] for the Central Range Wildlife Sanctuary.We acknowledge that the density estimates derived for other Neotropical sites may have used different and perhaps more modern methods of estimation, and that there are shortcomings of the King's Estimator technique [15], but due to insufficient detection at some of the sites we opted to use King's Estimator to allow for comparison between sites in Trinidad.There appears to be considerable variability in the densities of red-brocket deer in Trinidad, with lower values being found in Nariva and the VMFR.The higher densities estimated in the Central Range and Southern Watershed areas of Trinidad, may in part be due to differences in vegetation observed in these areas, with the species perhaps preferring these habitats.It has been found, for example, that red-brocket deer prefer thicker and moister forests compared to gray brocket deer (M.gouazoubira) [4].Further investigation into habitat preferences of this species in Trinidad would need to be undertaken to evaluate this.
Differences in vegetation preference by red-brocket deer in Trinidad would lead to differences being observed in populations within different areas and may also affect the detectability of deer in these habitats.We also acknowledge that there is variation in hunting pressure in the different areas where density estimates were undertaken.Unfortunately, data on hunting pressure in different parts of Trinidad are not available [22].A better estimate of the hunting pressure exerted throughout Trinidad would help determine whether this is also affecting the densities of redbrocket deer being observed.The variation in densities observed suggests that if hunting of this species is to continue, then this may need to be tailored to each site to ensure that deer are sustainably harvested, or bag limits need to be put in place for the species islandwide.Of note is the fact that the deer densities trended upwards from 2014 to 2016, which suggests that the institution of the moratorium may have had a positive impact on the populations of deer in Trinidad.The fact that the density estimates remained similar upon the resumption of hunting suggests that perhaps the 1 year of hunting activity in 2016 did not significantly impact the populations of deer, though this needs to be assessed with respect to the levels of hunting pressure within each of the areas.Red-brocket deer have also been categorized as species which may be less vulnerable to overhunting due to their being shortlived, having short generation times, and a high r max of 0.4 [23].
We recommend that continued monitoring of the populations of red-brocket deer in Trinidad be carried out in order to gain a more complete understanding of the effect harvesting of the species has on their populations, and to better estimate what harvesting levels are sustainable in each of these study sites.This would also include surveys of the levels of hunting pressure in these different areas to identify what if any effects these may be having on deer populations.The results of this study indicate that more site-specific management of the species be undertaken either in terms of bag limits or the number of hunting licenses which are granted.

Figure 1 .
Figure 1.Locations of 1 km transects surveyed in the island of Trinidad from 2014 to 2016.The outlines of associated forest reserves are provided.

Table 1 .
Number of individuals of M.a.trinitatis encountered on transect surveys, the length of transect surveyed, rate of encounter and density estimates calculated at different locations each season (with range calculated based on the standard error of the estimate).The mean density values and standard deviations are provided in the last row of the table.