Early efforts to identify the world's remaining “frontier” forests highlighted the substantial abundance of pristine habitats in the tropics (Bryant et al., 1997). Since the identification of such Intact Forest Landscapes (IFL), which are forest/mosaics at least 500 km² (50,000 ha) lacking overt anthropogenic disturbance such as infrastructure (Potapov et al., 2008), there has been dramatic decline in such areas (Potapov et al., 2017). The startling loss of IFL is largely due to tropical nations' economies and infrastructural development being rooted in the exploitation of natural resources. Africa contains one of the three large blocks of the world's tropical forests and with the depletion of natural resources in Asia, multi-national companies have sought new outlets in timber-rich nations such as those in Western Equatorial Africa (Angola (Cabinda enclave), Cameroon, Central African Republic, mainland Equatorial Guinea, Gabon, and Republic of Congo). Selective logging is the primary extraction industry responsible for IFL loss in the region (Asner et al., 2010; Potapov et al., 2017) and Republic of Congo has been in the vanguard of this expansion. An accelerated rate of logging road construction has ensued particularly in the north of the country over the last two decades (Laporte et al., 2007; Kleinschroth and Healey, 2017). In the wake of such expansion follows considerable degradation of natural resources and increased human immigration (Geist and Lambin, 2002; Watson et al., 2018). If action is not taken to avert losses, it is estimated that all IFL outside of protected areas in Republic of Congo will have disappeared by 2050 (Potapov et al., 2017) and wildlife populations in the region will be reduced by 80% (Fa et al., 2003). These developments will assuredly have negative consequences on forest and biodiversity within protected areas including those of Natural World Heritage Sites as many are already under elevated levels of human pressure and forest conversion (DeFries et al., 2005; Laurance et al., 2012; Bailey et al., 2016; Lui and Coomes, 2016; Allan et al., 2017).

Identifying important environmental attributes of forest intactness and indicators of change in intactness are key to more informed conservation management. With 77.4% of critically endangered western lowland gorillas and 80.7% of endangered central chimpanzees existing outside of protected areas (Strindberg et al., 2018) there is a great need for management beyond protected areas to conserve these flagship species. Based on recent regional modeling, gorillas and chimpanzees occur at higher densities in IFL compared to non-IFL (Strindberg et al., 2018). The physical structure of these forests has considerable influence on great ape distributions with generally higher ape densities associated with increasing tree canopy height (Strindberg et al., 2018). The emergent and high canopy levels are the result of a few “biomass hyperdominant” tree species (Bastin et al., 2015) in a region overall typified by low tree stem densities (Lewis et al., 2013). Local-scale surveys and remote sensing indicate frequent disruption in canopy continuity even in an intact state (Devos et al., 2008). This is a product of the mixed species forest composition which is comprised of a mosaic of regenerating patches from natural disturbances with canopies varying in height and composition. The dynamic and complex nature of this habitat supports high densities of both chimpanzees and gorillas (Devos et al., 2008).

A key question remains as to how increases in canopy gaps associated with the loss of dominant canopy trees, as occurs in selective logging, affect forest composition and resources important to apes. In South America, highly disturbed forests have been shown to support elevated densities of climbers such as those of the genus Ficus, which include species bearing high-quality fruit found to predict populations of primates (Terborgh, 1986; Leighton, 1993; Wrangham et al., 1993; Marshall and Leighton, 2006). Removal of canopy trees also brings elevated light exposure to the lower understory strata which in turn bolsters the growth of non-arboreal pioneer species (Malcolm and Ray, 2000) that are likely of benefit to apes. Members of the families referred to as terrestrial herbaceous vegetation (THV) are important to gorillas and chimpanzees for both foraging and nesting (e.g., Wrangham, 1986; Rogers and Williamson, 1987; Fay, 1997). Chimpanzees are classically referred to as more of a dietary specialist with their resource use focused on fruit-bearing tree species whereas gorillas are typically considered to be more along the lines of a generalist with a diet focused mostly on herbaceous ground vegetation (Bourliere, 1985). This classification has proven to be a useful dichotomy when assessing factors shaping species responses to perturbation, with the former more often negatively impacted by forestry than the latter (Johns and Skorupa, 1987; Sodhi et al., 2010; Burivalova et al., 2014). To date, however, surveys of gorillas and chimpanzees in post-logged forests indicate increases as well as decreases in population numbers, which raises questions regarding the relationship between compositional changes in the environment and ape abundance. Overall, gorilla populations in this region are in decline (Strindberg et al., 2018) and negative impacts of anthropogenic disturbance on chimpanzee behaviors have become increasingly apparent (Kuehl et al., 2019). However, alteration in the structural complexity of IFL may not be the only or principal factor responsible for potential changes in ape abundance in logged habitats of Western Equatorial Africa.

The rise in unsustainable hunting of wildlife for meat (i.e., bushmeat) and body parts is the most severe and rapidly expanding threat facing species today (Ripple et al., 2017). Life-history traits are known to influence species-specific vulnerability to hunting pressure (Reynolds, 2003; Marshall and Leighton, 2006). Human settlements, consumption practices, and accessibility are also strong determining factors in the persistence or decline of wildlife in an area (Barnes and Lahm, 1997; Fa et al., 2000; Blake, 2002; Jerozolimski and Peres, 2003; Blake et al., 2008). We suggest that a putative pattern of hunting pressure is triggered by increasing access to IFL. Forests distant from human infrastructure have higher abundances of wildlife compared to forests with longer histories of human influence (e.g., Eves and Ruggiero, 2000; Fa et al., 2004; Dupain et al., 2012) and greater accessibility (Yackulic et al., 2011). It follows that larger and highly profitable animals are reportedly originating from more distant and less-accessible areas (Allebone-Webb et al., 2011). There are indications that declines in wildlife associated with opening of IFL are rapid. In a previously intact concession in northern Congo, (Wilkie et al., 1992) estimated that 3,140 km of primary roads, secondary roads, and transects were opened in a single year. This region has subsequently been shown to be a primary source of most ivory reaching markets, substantiated by the staggering 62% decline of forest elephants in the Congo Basin since the early 2000s (Wasser et al., 2004; Maisels et al., 2013). Understanding the temporal-spatial patterning of hunting pressure in relation to the decline of IFL is crucial to preventing species declines both in protected areas and neighboring forests through proactive conservation measures to address such threats.

As part of a regional monitoring effort, we partnered with local government officials, conservation NGOs, and the timber company working in the region to assess ape abundances in relation to habitat characteristics and anthropogenic disturbances and compare IFL and non-IFL in the Sangha Trinational landscape. Baseline estimates of great ape densities in an IFL are compared with post-logging densities to better understand population dynamics in relation to anthropogenic disturbance. We also document floristic differences between IFL and non-IFL and relate floral differences to great ape resource use needs. The study took place in and around the Goualougo Triangle which is located in the southern portion of the Nouabalé-Ndoki National Park. Initial surveys in this region conducted by the Wildlife Conservation Society which led to the creation of the Nouabalé-Ndoki National Park cited the intact nature of the Goualougo Triangle and its conservation as essential for maximizing protection of a key area of core habitat for the region (Fay et al., 1990; Fay, 1992; Blake, 1994). Among all long-term ape research sites in Africa, the Goualougo study area was found to be the least disturbed by anthropogenic disturbances (Wilson et al., 2014). As such, we provide an update on the expansion of timber harvesting in the forest surrounding this protected area and increasing anthropogenic pressures which have reached the most remote areas of the Goualougo Triangle. This provides a rare opportunity to observe the temporal patterning of increased accessibility to remote forests and how this relates to illegal poaching pressures. We use this information to provide recommendations for park management and forest certification policies, as well as to promote the potential for permanent research sites to contribute to conservation initiatives through monitoring and surveillance.

This study took place in the Nouabalé-Ndoki National Park (NNNP; 2°05′-3°03′ N; 16°51′-16°56′ E) and the adjacent Kabo Forestry Management Unit (FMU) which is an immediately neighboring logging concession in northern Republic of Congo. The National Park was established in 1993 and covers 5,000 km² of forest and is part of the Trinational de la Sangha (TNS), a designated UNESCO Natural World Heritage Site (NWHS) which spans the Republic of Congo, Cameroon, and Central African Republic. The landscape is primarily comprised of a vast stretch of lowland Guineo-Congolian forest (White, 1983) with altitudes ranging from 330 to 600 m. The major habitat types in this region include monodominant Gilbertiodendron forest, mixed species forest, transitional Gilbertiodendron to mixed species forest, and swamp forest. This semi-deciduous mixed species forest has a diverse flora and canopy that is not always continuous. Rainfall is bimodal with a main rainy season from August through November and a short rainy season in May. Annual rainfall averaged 1,728 ± 47 mm between 2010 and 2017.

The Wildlife Conservation Society has the mandate to manage the NNNP and the FMU bordering it. In 1995, the landscape encircling the NNNP was divided into four concessions. The Kabo FMU is comprised of 2,960 km² which surrounds the southern sector of the National Park. The western sector of the Kabo FMU was selectively logged between 1971 and 1972 by the Société Nouvelle des Bois de la Sangha (SNBS) and then harvested a second time from 2005 to 2009 by Congolaise Industrielle des Bois (CIB). The Kabo concession was among the first concessions in Western Equatorial Africa to achieve Forest Stewardship Council (FSC) certification in 2006. The majority of the volume extracted consisted of Entandrophragma species, Triplochiton scleroxylon, and Milicia excelsa (CIB 2014). The eastern sector of the Kabo FMU was an IFL until the mid-2000s (Figure 1).

The Goualougo Triangle Ape Project (GTAP) was established in 1999 with the aim of conducting applied conservation research on the behavioral ecology of gorillas and chimpanzees in northern Congo. The Goualougo Triangle is an enclave of lowland forest between the Ndoki and Goualougo Rivers, which forms the southernmost section of the NNNP. It is bordered to the south, west, and east by the Kabo FMU (Figure 1). The Djeke Triangle is located between the Ndoki National Park in Central African Republic and NNNP in Republic of Congo. The GTAP and Wildlife Conservation Society research teams maintain a year-round field presence in the Goualougo and Djeke Triangles with daily reconnaissance missions to follow great apes that have been habituated to human presence. Both the Goualougo and Djeke Triangles are represented as IFL.

Once considered a stronghold of pristine habitat, IFL loss in Western Equatorial Africa is occurring at an alarming rate. Comparisons between intact vs. disturbed habitats in the Kabo FMU confirmed that overall tree stem density and total basal area were lower in non-IFL than IFL forests. While ape food resources still persist in these forests, as selectively logged habitats contain fruit-bearing tree species and canopy gaps promote growth of high abundances of terrestrial herbs, the long-term ecological consequences of timber extraction on ape populations are not yet known. For example in this study, we documented species-specific changes in resource use in nesting between IFL and twice logged habitat. Chimpanzees nested significantly closer to the ground in logged forests suggesting that changes in forest structure associated with logging may impact resource use. In addition, nearly two decades of antipoaching surveillance in the region indicate concerning trends between proximity of new roads in IFL and increasing vulnerability of wildlife to illegal hunting in neighboring IFL including protected areas. Together, such biodiversity assessments and strategic aspects of long-term protection should be taken into account when identifying conservation set asides. For example, the majority of the remaining IFL in the Kabo concession is within the Djeke Triangle which is contiguous with the Ndoki National Park in Central African Republic and the Nouabalé-Ndoki National Park in Republic of Congo. Thus, this area is a strategic location for curbing future poaching incursions into both protected areas. It also comprises the home ranges of gorillas habituated to human presence for scientific study and tourism development. This information could be used to advocate for the formal protection of the Djeke Triangle, not only as remaining IFL but for its conservation value and role in serving as a buffer to neighboring protected areas. However, such initiatives will only be successful through collaboration of local stakeholders, government officials, conservation agencies, and industrial partners.

It has been previously asserted that chimpanzees prefer primary forest (Tutin and Fernandez, 1984; Furuichi et al., 1997), whereas primary lowland forests were thought to be insufficient in terrestrial herbaceous vegetation to support high gorilla numbers (Schaller, 1963; Groves, 1971). The intact Ndoki forest can be considered a climax forest with a high vertical canopy structure created by light-demanding pioneer species such as mahogany (Entandrophragma spp.) which were established centuries ago (Fay, 1997). Such large trees play critical ecological roles in forest dynamics (Lindenmayer and Laurance, 2017) and also influence neighboring floral communities. Natural canopy disturbance is consequential in terms of THV recruitment and can resemble structural changes similar to selectively logged forest. The impact is reflected in our botanical surveys, showing relatively high densities of THV in intact as well as logged forests. In sum, the diversity of tree species and variation in canopy coverage of IFL in northern Congo provide adequate resources to support relatively large numbers of both chimpanzees and gorillas.

Overall dietary flexibility and degree of folivory have been cited in primate species' abilities to cope with environmental disturbances (Johns, 1997; Meijaard and Sheil, 2008) and could explain high ape densities documented in non-IFL. However, it is also possible that these forests historically differed in their inherent suitability to support great apes which was not taken into account in this investigation. We found lower stem densities and total basal areas of larger trees in logged forest which has implications on carbon storage potentials and elevating environmental risks. Forest stand change was also accompanied by a successional shift to more numerous trees in the small and medium size classes. These findings are in accord with other studies on the effects of selective logging on forest composition and growth dynamics in semi-deciduous forests in Western Equatorial Africa (Gourlet-Fleury et al., 2013a,b) and support findings that logged tropical forests transition to shorter and more broken canopy stands (Felton et al., 2003). Evidence for changes in forest structure influencing ape resource use was provided by species-specific nest height selection. We found that chimpanzees preferentially nested in the middle and upper story tree stratums in IFL. Whereas, chimpanzees nested at significantly lower heights in logged forests than in IFL. Nesting options located higher in the canopy may have diminished with the loss of the large timber species. These findings are similar to reports that orangutans shift their nest locations lower in logged environments (Felton et al., 2003). Gorilla nests heights did not significantly differ between IFL and logged habitat. There was however notable variability in gorilla nest height location in logged vs. intact forests which may indicate this species is opportunistically responding to increased nesting options. Increased diversity in nest construction patterns of gorillas may be a result of the elevated availability of THV documented and associated growth in pioneer species in the exploitation zone.

Industrial logging is projected to continue at 30-year rotation cycles in most of Western Equatorial Africa and so the fate of many tree species is unknown. Repeated removal of timber even at low intensity levels can degrade the quality of habitat over time (Lindenmayer and Franklin, 2002), and could have negative consequences for apes as shown in the degraded forests of Asia (Rao and van Schaik, 1997; Felton et al., 2003; Wich et al., 2004; Husson et al., 2009). Structural changes in the logged forests in this study indicate some implications for floral climbers and epiphytes. Compared to the diverse representation of Ficus in the IFL, few figs were found in non-IFL which raises important questions about host specificity of strangler figs in relation to logging species and potential consequences to the frugivore community reliant on these resources. Ficus spp. are a critical component of the overall chimpanzee diet in IFL (Morgan and Sanz, 2006). Monitoring the direct impacts of logging on large fruit-bearing tree species preferred by chimpanzees and gorillas in future exploitation cycles will be important as such resources can influence reproduction and fitness in wild apes (Emery Thompson et al., 2007). Future studies of nesting resource use and distribution could also be informative for conservation planning, as the difference observed in chimpanzee nesting in IFL vs. logged habitat could be the result of indirect rather than direct disturbances associated with logging. Low intensity logging was practiced in the concession with off-take ranging between 0.5 and 3.0 trees/ha. Importantly, the top three marketable tree species exploited in the study area were rarely used for nest construction by chimpanzees or gorillas. Such insights on shifting resource use by apes should be considered in identification of High Conservation Value Forests (HCVF) which is an important environmental criterion of FSC certification.

Compared to IFL, apes in non-IFL are also at increased risk of synergistic interactions with other threats such as edge effects and emerging diseases. For example, areas of high disturbance and elevated undergrowth as documented in this study are likely to be more fire prone. Logging routes and their margins transform the local forest with elevated levels of disturbance (Brandt et al., 2016) which is followed by rapid growth of high densities of Marantaceae and Zingiberaceae families in the abandoned tracks and edges (Malcolm and Ray, 2000; Kleinschroth et al., 2015). Increased rates of tree mortality, leaf litter accumulation, and damage (Ferreira and Laurance, 1997; Laurance et al., 1998) likely combine with changes in microclimate conditions and combustibility of flora, elevating risks of severe fire (Leighton and Wirawan 1986; Campbell 1992; Dennis and Colfer, 2006). The recent fire that raged across IFL and non-IFL bordering the main national road accessing the north of Republic of Congo provides evidence that wildfire is now an agent of disturbance in this region (Potapov et al., 2017). Another threat that is exacerbated by reduction of IFL is considerable spatial overlap between apes and humans occurring in proximity to roads that may elevate risk for cross-species transmission of pathogens through handling of shared resources or even direct contact. Finally, dramatic reductions of great ape numbers in some regions of northern Congo have been attributed to excessive hunting levels (Bermejo et al., 2006).

Poaching remains the greatest threat to wildlife in this region, and the spatial-temporal relationship between expanding road networks and increased illegal hunting pressure within NNNP is both pervasive and instructive. Since the Park's creation, effective management of poaching pressure in the surrounding logging concessions has been a priority to protect the integrity of the core area (Elkan et al., 2006). Law enforcement patrols can lower threat levels to protected areas (Stokes et al., 2010; Tranquilli et al., 2014), but increased support for these activities is critical as expanding road networks are providing unprecedented access to previously remote areas. We documented the first instances of poacher incursions in Goualougo Triangle region of NNNP that coincided spatially and temporally with the arrival of roads and active logging in adjacent forest. The spatiotemporal patterning of high-value natural resource extraction in Africa has expanded across a gradient from high to low sourced areas (Ahrends et al., 2010). Currently, the NNNP is in the midst of experiencing the first wave of hunting pressure targeted at ivory. This supports assumptions that poachers preferentially target IFL. It is likely part of a larger spatial pattern that has typified hunting in Western Equatorial Africa since the onset of selective logging in the early 1970s, but is only now reaching the previously inaccessible forests which form the core protected areas of the Sangha Trinational NWHS. At the time of this study, the poaching profitability perimeter for ivory hunters remained spatially close to logging roads. Entry points of illegal raids into the southeastern sector of the NNNP were typically within 10 km of the nearest road which supports broader assumptions of distances traveled by hunters (Peres and Terborgh, 1995). We strongly urge research assessing whether the profitability of hunting diminishes with closure of logging roads. The arrival and sudden intensification of poaching in the Goualougo Triangle is particularly concerning given the incessant poaching pressure over the last decade in the IFL of the Dja Faunal Reserve NWHS which is another important landscape for great ape conservation (IUCN, 2014). Long-term studies in Malaysia also suggest survival prospects for species such as orangutans in production forests are mainly determined by hunting (Ancrenaz et al., 2004). If properly coordinated with regional antipoaching efforts, we assert that research outposts and field teams can serve as sentinels of such poaching pressure in remote areas and act as force multipliers in maintaining surveillance. In addition to improving coordination of ecological and behavioral data collection across research sites, technologically-enhanced monitoring with SMART has the potential to expedite reporting of regional antipoaching efforts. More importantly, this increased monitoring and collaboration can lead to the arrest of poachers, which proves that the relationship between law enforcement and research efforts can be mutually beneficial in safeguarding wildlife.

Beyond implementation of best practice guidelines in and outside production forests, the most effective way to protect the outstanding flora and fauna attributes of IFL is through policies that formally set aside such forests (Watson et al., 2016). Successful mandates to attain increased protected status require collaborative and holistic approaches (Peres, 2005; Haurez et al., 2017; Chazdon, 2018) such as those that led to the annexation of the Goualougo Triangle to the NNNP (Morgan et al., 2012). In these cases, increases in formal protected status were the result of coordination among local stakeholders, government officials, conservation agencies, and industry partners. Over the last several years the debate regarding the fate of IFL, particularly in certified logging concessions, has become a central focus of global conservation. In 2016, the IUCN World Conservation Congress adopted a motion (https://portals.iucn.org/congress/motion/048) encouraging the monitoring and sparing of IFL from degradation and loss. FSC has integrated the protection of IFL into their International Standards, urging governments and concessionaires to decrease the rate of IFL loss. The Djeke Triangle comprises the majority of remaining IFL in the Kabo concession, and is also the location of the longest-running gorilla research and tourism site in Western Equatorial Africa, which employs a large number of local people. Given its location along the international borders of two National Parks, increased protected status through annexation to the NNNP would further strengthen efforts to maintain the ecological integrity of the Sangha Trinational.

Our ground-based efforts to identify and verify IFL resulted in a more accurate depiction of both the amount and location of remaining intact forest in the Kabo FMU. We found that 69.7 km² of the terre firma forest in the concession had previously been erroneously classified as lacking anthropogenic disturbance, when it had actually been logged. Botanical surveys conducted in IFL revealed higher densities of trees across all size classes, relative to non-IFL. Non-IFL forests were composed of reduced numbers of trees, but elevated herb densities. Results from our longterm studies of chimpanzees and gorillas show that ape resource use may be affected by logging, as nesting patterns differed between intact and logged forests. The height of chimpanzee nests was significantly lower in non-IFL compared to IFL, indicating that nesting niche options may be reduced in modified habitats. In contrast, there is indication that the gorilla niche may have expanded in logged habitat as a result of greater nesting material options associated with secondary growth in logged habitats. Along with the ecological changes in logged forest, we also documented that road expansion was associated with increased poaching pressure in nearby IFL. Real time indicators of forest change and associated data on resource use by endangered species are critical to promoting long-term preservation of biodiversity across different landscapes. However, such opportunities to study and protect IFL areas are quickly vanishing. As of 2013, it was estimated that only 8% of forests within Central Africa remained intact (Potapov et al., 2017).

DM, SS, and CMS: designed the study and oversaw data collection. WW, WM, and FI: compiled data for this study. SS, WW, CRS, CT, and CMS: analyzed the data. CMS and DM: wrote the manuscript with input from all co-authors. DM, CA, CMS, SN, WM, DK, and FI: collected data in the field. DM, CA, and CMS provided logistical support and infrastructure for data collection.