Introduction

A long-standing research focus in community ecology and biogeography is the coexistence among ecologically similar species within communities (MacArthur and Levins 1967; Hubbell 2005), a topic that has important implications for conservation (Dempster 1975). In animals, sympatric coexistence may be a result of niche partitioning and differentiation (Schoener 1974), and niche breadths (e.g., resource use, habitat type use) of coexisting species are expected to influence the mechanisms for niche differentiation (Büchi and Vuilleumier 2014). Among the variety of ways in which sympatric species can co-exist, and thus minimize competition, are temporal avoidance and differential resource use (Arlettaz 1999; Kronfeld-Schor and Dayan 2003). Within a particular guild, the potential for competition is increased, especially for species that are closely related taxonomically (MacArthur and Levins 1967). Thus, studies of carnivores in the same taxonomic family (i.e., “confamilial”), for example, can provide revealing insights into ecological co-existence (e.g., Powell and Zielinski 1983; Johnson et al. 1996; Santos et al. 2019), and on the potential for strong competitive interactions (Simberloff and Dayan 1991).

Two or more of the 12 species of skunks (Family Mephitidae) of the Americas (Hernández-Sánchez et al. 2022; McDonough et al. 2022) occur sympatrically throughout much of their range. Where more than one species has been studied, there appear to be some differences in activity peaks and patterns (Neiswenter et al. 2010; Farías-González and Hernández-Mendoza 2021), likely caused by interference competition (Hernández-Sánchez & Santos-Moreno 2022). In one recent study of sympatric skunks, a temporal avoidance strategy to decrease the potential for direct encounters with the dominant species, but activity patterns were also influenced by the activity of prey and native and exotic predators, as well as relative humidity, precipitation, cloud cover, and night length (Hernández-Sánchez and Santos-Moreno 2022). In addition, smaller spotted skunks (Spilogale spp.) are less generalist in habitat use (Doty and Dowler 2006; Neiswenter and Dowler 2007) and often differ in use from other skunk species (Cervantes et al. 2002).

Range overlap between southern spotted skunks (Spilogale angustifrons; 0.5–1 kg) and larger striped hog-nosed skunks (Conepatus semistriatus;1.4 to 3.5 kg) is reported to occur from the central and eastern Yucatan in Mexico in the north, and south through Belize, eastern Guatemala, central Honduras, eastern Nicaragua, to northwestern Costa Rica (Cuarón et al. 2016a; Helgen et al. 2016), but little is known of their co-existence. A third species, the hooded skunk (Mephitis macroura), is rarely reported as inhabiting northwestern Costa Rica, as well (Janzen 1983; Wainwright 2007, but see Cuarón et al. 2016b), and we obtained no photos of hooded skunks.

In this study we were interested in elucidating some factors that allow for co-existence between two sympatric skunk species in a dry forest of northwest Costa Rica. To that end, aspects of their space use, activity patterns, and photo rates were derived from camera trap data collected during long-term studies of jaguars (Montalvo et al. 2020). Our objective was to assess patterns of cover-type use and relative activity patterns in an area where both species co-exist. From these results, we also sought to identify if interspecific avoidance might be a means of co-existence.

Methods

Study area

We collected skunk photos in the Santa Rosa Sector of Guanacaste National Park located in northwest Costa Rica (10°53N 85°46′W; Boza 1992). The 387-km2 Santa Rosa Sector is dominated by one of the few seasonally dry forests remaining in Central America (Jimenez et al. 2016), and has undergone a large-scale restoration effort that was initiated in the 1980’s involving protected area status, the recovery of abandoned pastures by active fire suppression, protection from many human activities, and the recovering of large vertebrate populations (Janzen and Hallwachs 2016). The sector contains evergreen forests dominated by live oak (Quercus oleoides) but also including a large number of other species that co-occur in the adjacent mixed deciduous forest where oaks are rare, and more typical species from tropical dry forest (Powers et al. 2009). The forests on the hill slopes and in the lowlands are more diverse and with varying degrees of deciduousness. Several areas of Santa Rosa are covered with a mosaic of pasture and secondary growth in various stages of regeneration with areas that have different land use histories, past land use intensities as well as different occurrences of discrete anthropogenic events such as fire (Kalacska et al. 2004). Mangrove forests occur in lowland areas near the coast at stream and river outlets.

Mean annual rainfall in Santa Rosa totals ~ 1,600 mm but is highly seasonal; the wet season (months with an average > 100 mm of rain; average maximum temperatures ~ 29–31 °C) is May to November, and the dry season (with almost no rain and maximum temperatures > 35 °C) is December to April (Janzen 1993; Waylen et al. 1996). During 2011–2017, total precipitation during the rainy season varied from 617 to 2531 mm (mean = 1524). During the dry season, many forest patches lose their leaves and few evergreen-forest patches retain them. In addition, most of the rivers and streams in the study area run dry and the remaining waterholes become important providers of free water for wildlife (Campos and Fedigan 2009).

Photograph collection

To compare skunk cover type use and encounter rate, we used data collected continuously from 15 June 2016 to 13 June 2017 using 56 automatic trail cameras (Bushnell®, Trophy Cam models 119,436, 119,446, 119,456) in an area of 87 km2. Cameras were set in pairs throughout the area. For each pair, one of the cameras was set at a trail location that jaguars (Panthera onca) were likely to use; the other camera was set at an off-trail location an average of 0.59 km ± 0.25 SD away from the nearest trail camera (Fig. 1; Montalvo et al. 2023). Skunks have not been identified in jaguar diets (Hayward et al. 2016) and thus we did not consider that the results could have been biased by the presence/absence of jaguars. Also, a previous analysis (Montalvo et al. 2023) indicated no trail or season effects on striped hog-nosed skunk distribution (samples were too small to assess spotted skunks), and thus those factors were not considered in this analysis.

Fig. 1
figure 1

Detections of two skunk species during 15 June 2016 to 13 June 2017 at camera trap locations in Sector Santa Rosa, Guanacaste Conservation Area, Northwestern Costa Rica. There were no locations where both species were detected over the course of the study

Full size image

For assessment of skunk relative activity patterns, we used a larger camera trapping data set from 2011 to 2016 (including data from the survey noted above and restricted to cameras within that survey area). Cameras were deployed at a total of 101 different sites, including waterholes, on pathways (e.g., roads, human trails, and animal paths) and at random sites within the forest, for 34–365 days during each year. Camera placement often differed from one year to the next. The duration of continuous camera deployment at individual sites each year was affected by camera malfunctions, limited battery life in combination with logistics of camera checks, vandalism, and initial study design for deployment.

For all surveys, each camera was attached to a tree at a height of approximately 40 cm and set to be active for 24 h/day. Cameras were set either in video mode (30-s video, minimum 1 s between successive videos) or photograph mode (3 consecutive photos with a minimum delay of 1 s between consecutive triggers). Once deployed, cameras were checked on every month or so to replace batteries and change SD memory cards, if necessary. This research followed ASM guidelines (Sikes et al. 2016).

Photographs or videos were considered an independent photo events of a species if they were: (1) taken at least 30 min apart (e.g., a series of three photos of the same animal[s] taken in consecutive seconds = 1 photo event); (2) consecutive photos of the same species could be identified as different individuals (spots, scars, sex) and not part of the same group (e.g., > 15 min apart, going in opposite directions = two photo events); or (3) photos of the same species separated by photos of a different species (e.g., species 1, followed 2 min later by a species 2, followed 5 min later by species 1 = one species with two photo events and another species with one photo event).

Data analyses

Cover types at the 56 capture sites were pooled into five categories that included: Primary Forest; Secondary Forest, including early and later stage forests, some mixed with lesser amounts of primary forest; Mangrove and Beach Forest; Riparian Forest, sometimes with some secondary forest; and Grassland with trees, sometimes with some secondary forest. Chi-square tests were used to compare the number of trap sites in different cover types where a species was photographed, as well as photo rates (number of independent photos/100 trap nights) for each species among cover types. Photo rates did not differ for either species between seasons (P = 0.416–0.580), or by site (trail vs. off-trail; P = 0.138–0.263) so data were pooled across the year and sites. We also used Chi-square tests (Social Science Statistics—https://www.socscistatistics.com/tests/chisquare2/default2.aspx) to compare species specific-differences in the photo rates between species and among studies.

For relative activity comparisons, the time that the first photo of an independent photo event was taken was used in analyses. To quantify relative activity patterns, we used Ridout and Linkie’s (2009) approach with the R (https://www.r-project.org) software package activity 1.3.3 (Rowcliffe 2023), using a Wald test to contrast temporal distribution aggregation differences for circular data, smoothed with 10,000 bootstrap resamples to calculate confidence intervals (Rovero and Zimmermann 2016).

Results

Though during June 2016-June 2017 photo rates of spotted skunks (0.66; 135 independent photos/20,416 trap nights) in our study area were twice those of striped skunks (0.32; 65 independent photos/20,416 trap nights; \(\chi\)2 = 24.62, df = 1, P < 0.001), skunks of both species were photographed at camera stations throughout the study area (Fig. 1). However, at any given trap, only one of the species was ever photographed; that is, at no single camera were both species photographed over the course of the year.

We first compared the number of trap sites in different cover types where a species was photographed (Table 1). Each species was captured at traps in four or five cover types (no spotted skunks were photographed in Riparian Forest), and the number of sites in cover types where each skunk species were photographed was not significantly different (\(\chi\)2 = 6.61, df = 8, P = 0.579). We then compared photo rates of each species among cover types (Table 2), and found that for spotted skunks, the rate was significantly higher than expected in Secondary Forest (partial \(\chi\)2 = 4.70, df = 1, P = 0.030), and significantly lower in Riparian Forest (partial \(\chi\) 2 = 11.63, df = 1, P < 0.001; overall \(\chi\)2 = 22.02, df = 4, P < 0.001). For striped skunks, the rate was significantly higher than expected in Mangrove and Beach Forest (partial \(\chi\)2 = 21.73, df = 1, P < 0.001; overall \(\chi\)2 = 26.30, df = 4, P < 0.001).

Table 1 Number of trap sites (i.e., camera sets) in different cover types where a species was photographed in Santa Rosa National Park, northwestern Costa Rica during 15 June 2016 to 13 June 2017
Full size table
Table 2 Photo rates (independent photos/100 trap nights; total = 20,416 trap nights) of skunk species in different cover types in Santa Rosa National Park, northwestern Costa Rica during 15 June 2016 to 13 June 2017
Full size table

Relative activity patterns derived from photographs suggested that both species were clearly nocturnal (Fig. 2). We did not find statistical evidence that spotted skunks were more active than striped skunks before midnight, and striped skunks were more active after midnight. However, the only daytime photos we acquired were those of striped skunks.

Fig. 2
figure 2

Southern spotted skunk and striped hog-nose skunk activity overlap (gray shaded area = 77% overlap; Wald \(\chi\) 2 statistic = 0.10, df = 1, P = 0.917) as estimated from camera-trapping records collected during 2011–2017 in Sector Santa Rosa, Guanacaste Conservation Area, Northwestern Costa Rica

Full size image

Discussion

In our study, striped hog-nosed skunks photo rates were highest in mangrove and beach forest, though they were recorded in all other forest types. The striped hog-nosed skunk is fairly common in most parts of Costa Rica and reported to be found mostly in relatively open habitats such as pastures, secondary growth, and dry forest, but also in primary forest, where it feeds on small insects and vertebrates, and fruit (de la Rosa and Nocke 2000; Wainwright 2007). In general, striped hog-nosed skunks occur along the edges and in gaps of evergreen forests (and rare in areas with pronounced dry seasons; Reid 1997) and seem to be adaptable to some level of human disturbance such as agroforestry, forest edge, grasslands, plantations, etc., with a reported diet concentrated primarily on insects, but also small vertebrates such as lizards and birds, and some fruit (Reid 1997; Cuarón et al. 2016a).

Spotted skunks in our area seemed to be most common secondary forests, and were found in all other forest types except riparian forests. This skunk is restricted to the northwest and Central Valley of Costa Rica (but see Pacheco et al. 2006) and reported to be found in open areas with some cover, brush, and relatively dry forest (Wainwright 2007) and human habitations and farms (de la Rosa and Nocke 2000). Like the striped hog-nosed skunk it also is insectivorous, but also eats fruits and eggs. It also is considered more carnivorous than other skunks, preying or scavenging on rodents, lizards, snakes, and birds (Reid 1997; de la Rosa and Nocke 2000; Wainwright 2007; Helgen et al. 2016). Unlike other skunk species, southern spotted skunks can climb trees to escape predators, forage, or sleep/shelter.

Similar to our findings, where other species of spotted and hog-nosed skunks co-occur, there have been some habitat-specific differences in distribution reported, such as in the relative proportion of observations in grassland (less for spotted skunks) vs. scrubland for each species (Cervantes et al. 2002). Also, like our results, hog-nosed skunks were reported to be more active earlier during the night and spotted skunks more active later in the night (Farias-Gonzalez and Hernandez-Mendoza 2021). However, these two studies (Cervantes et al. 2002; Farías-González and Hernández-Mendoza 2021) also reported that the relative encounter rate and photo rates, respectively, of spotted skunks was lower than for hog-nosed skunks, in contrast to our findings. However, in the only other camera study we could find in which both species were photographed (in Costa Rica, just south of our study; Yaney-Keller et al. 2022), it appeared that, despite relatively small sample size (1498 trapnights), photo rates (number of independent photos/100 trap nights) of southern spotted skunks (2.20) were also significantly higher than for striped skunks (0.33).

Differences we observed in the distribution, activity, and photo rates of two skunk species were notable and seemed to point to local allopatry, or at least spatially interspecific avoidance. But the previously identified differences in species size and species-specific food habits, and perhaps the activity of prey and native and exotic predators, relative humidity, precipitation, cloud cover, and night length (Hernández-Sánchez and Santos-Moreno 2022), also could be factors that allow for co-occurrence in our area and other areas where the species are sympatric.

As a research technique, camera trapping certainly allows for site-specific identification of active (but not inactive) skunks. From these data, we can make preliminary inferences on the species’ occurrence, cover type use, and relative activity patterns, but cannot consider the influence of other factors. Clearly, site-specific studies on skunk foraging, food habits, predators, and other environmental factors, as have been conducted elsewhere (e.g., Jachowski and Edelman 2021; Dukes et al. 2022) will better clarify the means by which the species successfully co-occur.