id	siteCount	siteNestingDescription	verbatimSiteDescriptions	verbatimSiteNames	geospatialScopeAreaValue	geospatialScopeAreaUnit	totalAreaSampledValue	totalAreaSampledUnit	reportedWeather	reportedExtremeConditions	targetHabitatScope	excludedHabitatScope	eventDurationValue	eventDurationUnit	targetTaxonomicScope	excludedTaxonomicScope	taxonCompletenessReported	taxonCompletenessProtocols	isTaxonomicScopeFullyReported	isAbsenceReported	absentTaxa	hasNonTargetTaxa	nonTargetTaxa	areNonTargetTaxaFullyReported	targetLifeStageScope	excludedLifeStageScope	isLifeStageScopeFullyReported	targetDegreeOfEstablishmentScope	excludedDegreeOfEstablishmentScope	isDegreeOfEstablishmentScopeFullyReported	targetGrowthFormScope	excludedGrowthFormScope	isGrowthFormScopeFullyReported	hasNonTargetOrganisms	verbatimTargetScope	identifiedBy	identificationReferences	compilationTypes	compilationSourceTypes	inventoryTypes	protocolNames	protocolDescriptions	protocolReferences	isAbundanceReported	isAbundanceCapReported	abundanceCap	isVegetationCoverReported	isLeastSpecificTargetCategoryQuantityInclusive	hasVouchers	voucherInstitutions	hasMaterialSamples	materialSampleTypes	samplingPerformedBy	isSamplingEffortReported	samplingEffortProtocol	samplingEffortValue	samplingEffortUnit
1a8857bc-763a-4fa1-ad0b-d7a2ce5ae1d5test																																																									
4043f620-dd14-40fd-8c54-732b6fe9092dtest	3	3 sampling sites, each sampled via the following: visual and audio surveys along trail networks measuring 30, 31, and 18 km, respectively; large arrays of camera traps inside the forest; small arrays of audio recorders for birds and bats inside the forest; mistnets for bats inside the forest; fish sampling stations along small creeks, large creeks, and main rivers; riverine surveys along large rivers	mature upland forest | swamp forest dominated by Euterpe oleracea | floodplain forest along major rivers | vine-dominated secondary forest | rocky rapids in main river | stunted pole forest | swamp forest dominated by Mauritia flexuosa | sandy beaches in main river channels | main river channels | forest creeks, often with low flow or reduced to isolated pools	Acarai-Corentyne Corridor	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				39	days	Aves | Mammalia | Reptilia | Amphibia | Actinopterygii | Embryophyta		reported incomplete	based on comparison of lists of detected vs. expected species	TRUE	FALSE		FALSE		FALSE										FALSE	Mammals | Birds | Amphibians | Reptiles | Vascular plants | Freshwater fishes			sampling events only		open search | opportunistic search | restricted search			Pitman, Nigel, Cameron Rutt, Lesley S. de Souza, R. Elliott Oakley, Farah Carrasco-Rueda, Sophie Picq, and Jeremy M. Campbell, eds. 2025. Guyana: Acarai-Corentyne Corridor. Rapid Biological and Social Inventories Report 32. Field Museum, Chicago.	TRUE	FALSE		FALSE		TRUE	FMNH | CSBD	TRUE	whole organism | tissue sample | liver or muscle tissue sample | fertile branch | silica-dried leaf fragment | hair sample (bats only) | sample of the wing patagium using disposable biopsy punches of 2 mm (bats only) | liver tissue sample (bats only) | colon tissue sample (bats only) | DNA from water pumped through self-preserving polyethersulfone (PES) membrane filters of 0.8-µM pore size and 47-mm diameter					
6240d073-3fbf-4b38-adb5-a07b7398ce38test	3	3 sampling sites, each sampled via the following: visual and audio surveys along trail networks measuring 30, 31, and 18 km, respectively; large arrays of camera traps inside the forest; small arrays of audio recorders for birds and bats inside the forest; mistnets for bats inside the forest; fish sampling stations along small creeks, large creeks, and main rivers; riverine surveys along large rivers	mature upland forest | swamp forest dominated by Euterpe oleracea | floodplain forest along major rivers | vine-dominated secondary forest | rocky rapids in main river | stunted pole forest | swamp forest dominated by Mauritia flexuosa | sandy beaches in main river channels | main river channels | forest creeks, often with low flow or reduced to isolated pools	Acarai-Corentyne Corridor	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				39	days	Aves | Mammalia | Reptilia | Amphibia | Actinopterygii | Embryophyta		reported incomplete	based on comparison of lists of detected vs. expected species	TRUE	FALSE		FALSE		FALSE										FALSE	Mammals | Birds | Amphibians | Reptiles | Vascular plants | Freshwater fishes			sampling events only		open search | opportunistic search | restricted search			Pitman, Nigel, Cameron Rutt, Lesley S. de Souza, R. Elliott Oakley, Farah Carrasco-Rueda, Sophie Picq, and Jeremy M. Campbell, eds. 2025. Guyana: Acarai-Corentyne Corridor. Rapid Biological and Social Inventories Report 32. Field Museum, Chicago.	TRUE	FALSE		FALSE		TRUE	FMNH | CSBD	TRUE	whole organism | tissue sample | liver or muscle tissue sample | fertile branch | silica-dried leaf fragment | hair sample (bats only) | sample of the wing patagium using disposable biopsy punches of 2 mm (bats only) | liver tissue sample (bats only) | colon tissue sample (bats only) | DNA from water pumped through self-preserving polyethersulfone (PES) membrane filters of 0.8-µM pore size and 47-mm diameter					
4c287d7f-d7b4-4642-baee-935cd9a350b1test	3	3 sampling sites, each sampled via the following: visual and audio surveys along trail networks measuring 30, 31, and 18 km, respectively; large arrays of camera traps inside the forest; small arrays of audio recorders for birds and bats inside the forest; riverine surveys along large rivers	mature upland forest | swamp forest dominated by Euterpe oleracea | floodplain forest along major rivers | vine-dominated secondary forest | rocky rapids in main river | stunted pole forest | swamp forest dominated by Mauritia flexuosa | sandy beaches in main river channels | main river channels | forest creeks, often with low flow or reduced to isolated pools	Camp Amuku | Camp Monkey Jump | Camp New River	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				39	days	Aves		reported incomplete	based on comparison of lists of detected vs. expected species	TRUE	FALSE		FALSE		FALSE										FALSE	Birds			sampling events only		open search | opportunistic search	survey transects | audio recorders | opportunistic observations	We primarily sampled birds for 4–6.5 days at each camp: five days at Camp Amuku (6–10 November 2024), four days at Camp Monkey Jump (12–15 November 2024), and 6.5 days at Camp New River (18–24 November 2024) for a total of 15.5 sampling days. To this we added opportunistic observations during three days of river travel, avian bycatch from the mammal camera traps, taxa detected by environmental DNA (eDNA) and observations from the advance team. The bird sampling protocol mainly consisted of extended hiking transects, anchored to the trail system at each camp, but with the flexibility to pursue diverse mixed-species flocks off trail. Broadly, our objectives were to maximize the combined area surveyed each day, walk the entire trail system at every camp, and ensure that we covered all available habitats. Surveys typically began at dawn and continued through the early afternoon, resulting in ~9-hour days of fieldwork (roughly 04:45–13:45). Across three weeks of sampling at our three camps, sunrise shifted subtly between 05:35–05:38, with the first diurnal birds tuning up for the dawn chorus at ~05:10. To maximize coverage, Asaph Wilson and Cameron Rutt typically surveyed separate trails (or separate sections of the same trail) on any given day, taking care to sample the dawn chorus at unique locations. After the first morning at Camp Amuku, where the three of us birded together, Kwang Suse joined Asaph Wilson for all but the first days at Camp Monkey Jump and Camp New River. The tandem of Asaph and Kwang typically completed an entire trail in a single outing (5–8 km), while Cameron’s crawling pace meant that he often only covered ~1.5–2 km of trail (~3–4 km out and back). Weather did not influence any of our fieldwork, and we faced only a single morning with substantial tree drip from rainfall the previous evening. All bird observations, locations, and automated GPS tracks were recorded using eBird Mobile and a joint eBird account (username: Field Museum RI32), which was originally blocklisted from public output before being activated once the embargo period was lifted. These checklists primarily took the form of ‘traveling’ and ‘stationary’ counts and, at the end of each field day, Cameron compiled additional opportunistic observations from the camp into a single ‘incidental’ checklist, to ensure that all unique sightings from the day were logged. Wherever possible, we attempted to document unusual and charismatic species with either recordings on Merlin Bird ID (all members) or, for Cameron, a mirrorless camera (Canon EOS R5 Mark II) and telephoto lens (Canon RF100-500 mm) and/or a digital field recorder (Zoom F3) and shotgun microphone (Audio-Technica 8015). For many of the remaining species, we also archived photos and audio recordings at the Macaulay Library, which serve as publicly accessible, digital vouchers of these species within the corridor. Taxonomy and nomenclature follow the eBird/Clements Checklist of Birds of the World. For the first time on a Field Museum-led rapid inventory, the advance team deployed 15 autonomous recording units (ARUs; Song Meter Micro 2)—five at each camp—to both document and archive the soundscapes at these remote, undisturbed sites. All devices were deployed at least 500 m apart. These units began recording immediately after deployment, resulting in 12–28 days of sampling at each camp prior to our arrival (shortest at Camp Monkey Jump and most protracted at Camp New River). At the start of Camp Amuku and Camp Monkey Jump, Cameron collected the ARUs and relocated them elsewhere along the trail system (for three additional days of sampling, on average), enabling us to sample 24 unique soundscapes in all. Due to technical difficulties at Camp New River, we were unable to repeat this redeployment procedure there. Individual devices recorded a continuous three-hour time block every morning—one hour prior to sunrise until two hours after sunrise (with a sampling rate of 48 kHz, above the threshold for human hearing). Once properly formatted, all soundscapes will be archived in the Macaulay Library and run through Sound ID. At this time, Sound ID is able to identify 41% of birds in this region, but permanent archiving means that these recordings can be reanalyzed again and again as Sound ID becomes increasingly proficient with Neotropical birds. Two additional technologies also provided unique contributions of bird documentation. First, the camera trap work by the mammal team allowed us to detect shy terrestrial birds. This large-scale effort (2537 trap-nights at 131 locations) predominantly targeted medium- to large-bodied terrestrial mammals, but camera traps routinely capture photos and videos of game birds, in particular. Second, making its debut on a rapid inventory, eDNA collected from water samples was employed to detect vertebrates of all kinds, including birds. Lastly, other team members opportunistically contributed bird sightings. Álvaro del Campo was particularly helpful with his knowledge of macaw and amazon vocalizations. And thanks to the nocturnal nature of herpetological fieldwork, Leandro Moraes routinely surveyed for owls and potoos during the early hours of each night. No bird specimens were collected. For more information on the sampling protocol, including maps and figures, see the Rapid Inventory 32 report (Rutt, Cameron, et al., in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 129 - 137.)	Rutt, Cameron, et al., in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 129 - 137.	TRUE	FALSE		FALSE		FALSE		FALSE		Cameron Rutt | Asaph Wilson | Kwang Suse	TRUE	3 ornithologists performed visual and audio surveys | acoustic recorders	222 | 750	person hours | active hours
bdd2b3b2-6333-4ade-9512-d8d5f78e75fetest	3	3 sampling sites, each sampled via the following: surveys of upland creeks and pools along trail networks measuring 30, 31, and 18 km, respectively; surveys of larger creeks and main rivers	sandy beaches in main river channels | main river channels | rocky rapids in main rivers | forest creeks, often with low flow or reduced to isolated pools	Camp Amuku | Camp Monkey Jump | Camp New River	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				20	days	Actinopterygii		reported incomplete	based on comparison of lists of detected vs. expected species	TRUE	FALSE		FALSE		FALSE										FALSE	Freshwater fish			sampling events only		open search | opportunistic search | restricted search	gill nets | drag seines | baited hook | hand nets | cast nets | traditional indigenous fishing techniques | eDNA | opportunistic observations	Fieldwork was conducted over 21 days (5–24 November 2024) in the upper Essequibo and Corentyne river basins of southeastern Guyana. We sampled a total of 35 sites at three campsites: 17 sites at Camp Amuku; eight sites at Camp Monkey Jump; and 10 sites at Camp New River. Fishes were collected both during the day and night using several methods: a 72-foot experimental gill net with four 6-foot tall sections of different mesh sizes (0.5, 1, 1.5 and 2-inch mesh); an 8-foot diameter cast net with 0.5-inch mesh size; and small drag seines (6 x 6 feet with 0.25-inch mesh and 6 x 20 feet with 0.125-inch mesh). Baited hook and lines and fine-meshed hand nets were also used. At two sites, one in the Essequibo and one in the Corentyne River, we used traditional Indigenous fishing techniques. Juice from hairari vines (Deguelia chrysophylla cf.) was leached into slow currents of small water bodies to immobilize fishes via vasoconstriction properties similar to the action of the insecticide rotenone. This method produced the most complete and abundant fish samples obtained during this study. Sampling was conducted during the dry season with unusually high air and water temperatures according to local inhabitants. The four main fish habitats we sampled included sandy beaches in main river channels; the main river channels themselves (with hook and line); rocky rapids, frequently covered by aquatic plants (Podostemaceae); and forest creeks, often with low flow or reduced to isolated pools. Water type was mostly clear, sometimes a bit murky, and no sites had true black water. Water quality was mostly high, unpolluted, with acidic pH values and very low dissolved minerals (hardness and conductivity). We obtained geographic coordinates with a handheld GPS unit and took photographs of every site. Most of the fishes we collected were photographed in the f ield to capture live coloration. All specimens from which DNA tissue samples were taken were photographed after preservation to aid in their identification. Individuals were identified, sorted, and counted using current taxonomic keys and some were identified by experts using these photographs. Valid names were confirmed using the Catalog of Fishes. Fishes were preserved in 10% formalin, and larger specimens were injected with formalin to halt deterioration of the gut. All specimens were anesthetized with tricaine methanesulfonate (MS-222) and tissue samples were obtained from 642 select individuals (each photographed with the tissue tag number) and preserved in 95% ethanol for further genetic analyses. All whole fish specimens were processed and definitively identified at the Field Museum of Natural History in Chicago, Illinois, USA. Selected voucher specimens will be returned to the Centre for Study of Biological Diversity after identifications are verified. In addition to the methods described above, a number of fish species were recorded by a separate team using eDNA metabarcoding of 97 water samples collected during the inventory. For a detailed description of those methods, see the eDNA chapter on page 151. Below we present results from both the field surveys and the eDNA analyses. For more information on the sampling protocol, including maps and figures, see the Rapid Inventory 32 report (de Souza, Lesley S., et al. Fishes, in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 117 - 121.	de Souza, Lesley S., et al. Fishes, in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 117 - 121.	TRUE	FALSE		FALSE		TRUE	FMNH | CSBD	TRUE	whole organism | tissue sample	Lesley S. de Souza | Devya D. Hemraj-Naraine | Allister Henry | Onesiumus Rudolph | Syra Ayaw | Mark Suse | Tina Kenke | Carl Kenke | Ronnell Lewis | Albert Yaimo | Nehru Narine | Gideon Yaimo | Donald C. Taphorn	TRUE			
8af69ea4-0303-4227-a46f-953b0431ab06test	3	3 sampling sites, each sampled via the following: visual surveys along trail networks measuring 30, 31, and 18 km, respectively; large arrays of camera traps inside the forest; riverine surveys along large rivers	mature upland forest | swamp forest dominated by Euterpe oleracea | floodplain forest along major rivers | vine-dominated secondary forest | rocky rapids in main river | stunted pole forest | swamp forest dominated by Mauritia flexuosa | main river channels | forest creeks, often with low flow or reduced to isolated pools	Camp Amuku | Camp Monkey Jump | Camp New River	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				20	days	Amphibia | Reptilia		reported incomplete	based on comparison of lists of detected vs. expected species	TRUE	FALSE		FALSE		FALSE										FALSE	Amphibians | Reptiles			sampling events only		open search | opportunistic search | restricted search	survey transects | opportunistic observations | eDNA	Before undertaking fieldwork we generated a list of expected species for southern Guyana by compiling available data from previous inventories made in the direct vicinity of the Acarai-Corentyne, including much of the literature cited in the introduction. We surveyed amphibians and reptiles at three sites in the Acarai-Corentyne Corridor for 17 days (5–23 November 2024), totaling 5–7 sampling days at each sampling site. Two researchers visually or acoustically recorded individuals by diurnal and nocturnal active surveys along established trails for a total of 91 hours. To maximize the species recorded in a short period, we surveyed the most distinct microhabitats present in the area, incorporating local knowledge about the best sites to find different species. Surveys in water bodies were prioritized, as those habitats harbor a higher diversity of amphibians and support many reptile species, but also because the conditions were particularly dry. As many species, especially reptiles, are secretive and rarely encountered, we also recorded opportunistic encounters of amphibians and reptiles by all team members. All individuals recorded within the surveys were identified to the species level based on the literature. We also recorded their maturity, sex, microhabitats and activity patterns. We took photographs of at least one specimen of all the species collected, and identified specimens based on literature. Up to 10 specimens of each recorded species were collected. Collections adhered to Guyanese regulations, particularly regarding species listed in CITES Appendices I and II. All collected specimens were euthanized using standard methods (AVMA 2013). Amphibians were anesthetized by applying a small amount of benzocaine gel to the skin or by immersing them in a buffered solution of tricaine methanesulfonate (TMS/MS222), followed by a lethal intraperitoneal injection of TMS/MS222. Reptiles were euthanized using an intraperitoneal injection of TMS/MS222. Specimens were then f ixed in 10% formaldehyde and preserved in 70% ethanol (Heyer 1994). Prior to fixation, tissue samples (liver or muscle) were collected and stored in absolute ethanol for subsequent genetic analyses. For ecological analyses, rarefaction curves were generated using the R statistical software package iNEXT. In addition to the methods described above, a number of amphibian and reptile species were recorded by a separate team using eDNA metabarcoding of 97 water samples collected during the inventory and processed afterwards. For more information on the sampling protocol, including maps and figures, see the Rapid Inventory 32 report (Fouquet, Antoine & Moraes, Leandro. Amphibians and Reptiles, in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 121 - 129.)	Fouquet, Antoine & Moraes, Leandro. Amphibians and Reptiles, in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 121 - 129.	TRUE	FALSE		FALSE		TRUE	FMNH | CSBD	TRUE	whole organism | liver or muscle tissue sample	Antoine Fouquet | Leandro Moraes	TRUE	2 herpetologists performed diurnal and nocturnal active survyes	91	person hours
104f06e7-2601-489c-998d-8bf6dc5ebea7test	3	3 sampling sites, each sampled via the following: visual surveys along trail networks measuring 30, 31, and 18 km, respectively; large arrays of camera traps inside the forest; small arrays of audio recorders for birds and bats inside the forest; mistnets for bats inside the forest; riverine surveys along large rivers	mature upland forest | swamp forest dominated by Euterpe oleracea | floodplain forest along major rivers | vine-dominated secondary forest | rocky rapids in main river | stunted pole forest | swamp forest dominated by Mauritia flexuosa | sandy beaches in main river channels | main river channels | forest creeks, often with low flow or reduced to isolated pools	Camp Amuku | Camp Monkey Jump | Camp New River	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				39	days	Mammalia		reported incomplete	based on comparison of lists of detected vs. expected species	TRUE	FALSE		TRUE	Aves | Reptilia	TRUE										FALSE	Mammals			sampling events only		open search | opportunistic search	mist nets (bats only) | camera traps | survey transects | audio recorders | opportunistic observations | eDNA	Prior to field work, we obtained a list of mammal species expected for the Acarai-Corentyne Corridor from Map of Life. We validated this list based on the literature and personal knowledge, removing taxa we consider extremely unlikely to occur in the study area. The taxonomic reference we used for this study was mainly based on species names for Northern Neotropical mammal species. However, since the taxonomy of many groups has changed recently, we accepted some changes in primates, rodents, ungulates and other groups. We surveyed mammals over 14 days (6–21 November 2024) at three sites in the Acarai-Corentyne Corridor. This included f ive days at Camp Amuku (6–10 November), four days at Camp Monkey Jump (12–15 November), and five days at Camp New River (17–21 November). During the rapid inventory, the mammal team focused on bats, and medium- and large-bodied mammals. We used different methods that complement each other to record different groups of species and obtain different types of evidence, as no single method is effective for detecting all species in every context.  Trail surveys and opportunistic sightings  At each camp we walked all trails during the day between 06:30 and 17:30, including the periods when we collected the camera traps. We walked 14.4–69.6 km at each camp, for a total of 122 km. In addition to direct observations, we also recorded any other signs of mammal presence (i.e., prints, tracks, vocalizations, burrows, feces, feeding signs, scratches). Unlike other methods, this one allows us to make a record of arboreal species, especially primates. We complemented our observations with sightings made by other members of the biological team and by the advance team during the camp construction period. To estimate sampling success for this method, we divided the number of records made by the mammal team by the distance we walked.  Camera trapping  Camera trap photos and videos were obtained following well-established methods for camera trap research. With the support of local experts, 42–45 camera traps (Bushnell Core Prime #119932CB, Bushnell©, KS, USA and Reconyx HF2X HyperFire 2, Reconyx©, WI, USA) were set at each camp during the advance work and collected at the end of the field work. Cameras were left in the field for 6–36 trap-nights (mean = 19) between 17 October and 21 November 2024. The sampling effort totaled 2537 trap-nights in 131 locations across the three camps. Cameras were placed along the trail system, ~500 m apart, with a single camera at each trap location, set 30–60 cm from the ground in proximity to observed animal signs (tracks, paths, marks on trees, scat, animal trails), no farther than 30 m from the trails. Additional cameras were set opportunistically farther away, outside of the trail system (primarily Reconyx cameras set to take videos) in areas where signs of habitat or resource specialists were detected. Cameras were active 24 hours per day, with a 1- to 3-second delay between each 3- or 5-image photo sequence, or a 10-second video without a quiet period between triggers. In an effort to reduce wariness around cameras and avoid biased capture rates, no scents or lures were used, and all cameras employed were equipped with infrared flash. All camera traps recorded the date and time with each photo or video. For each, we recorded coordinates in decimal degrees (Datum: WGS 84), camera information, installation date and time, and physical description of the station. Camera trap data was uploaded in the project ‘RI-32 Guyana: Acarai-Corentyne Corridor’ in Wildlife Insights.  Mist-net sampling  Mist-net surveys are generally considered more effective for phyllostomid species that feed on the understory, closer to the forest floor. At each camp we installed 10 understory mist-nets (12 x 2.5 m) in pairs or triplets that remained active from dusk until 22:00, for an average sampling time of four hours per night. At each camp, we captured bats for at least two nights, weather permitting (no mist-netting was done in poor conditions). We totaled a sampling effort of 60 mist-net-nights, where one mist-net-night is equivalent to one mist-net of 12 m active for four hours per night. Nets were checked every 30 minutes. Captured individuals were placed in cloth bags until processing time. For each individual, we recorded weight, age, sex, and forearm length. We used experience and field identification keys to determine the species of the captured individuals. Following identification, we took hair samples from the dorsal region of all captured bats, using steel scissors that were disinfected with alcohol between every collection. This allowed recaptures to be recognized. Hair samples were stored in 1.5-ml Eppendorf plastic vials or coin envelopes to prevent mold. These samples will be used in the future to measure potential mercury levels in individuals through specialized analyses in laboratories. For all captured individuals, we took a sample of the wing patagium using disposable biopsy punches of 2 mm that we kept in cryovials with DNA/RNA Shield reagent. At each camp, we collected individuals of each species as museum specimens. We used standard euthanasia methods (AVMA 2020), by an overdose of inhaled anesthetics (isoflurane), and in some cases we administered an additional lethal dose of pentobarbital intraperitoneally. Once the specimen was confirmed to be euthanized via cessation of breathing (confirmed by sight) and stoppage of the heart (confirmed by touch), we opened a cavity in the ventral area of the body and removed samples of tissue for the liver and colon (5–7 mm). We kept the tissue in 2-ml cryovials with ~600 uL of DNA/RNA Shield reagent. Lastly, to preserve the specimens we injected 10% formalin into the chest and head muscle as well as the viscera and stored them in a container with 10% formalin. Immediately following field work, all collected samples were transported to the Centre for Study of Biological Diversity (CSBD) for processing and verification for exportation to the Field Museum in Chicago. Collections were done under permit provided by the Guyana government (EPA Permit to conduct Biodiversity Research # 20240918 BR 010, GWCMC research export permit FI/LD/25/2024) and Field Museum IACUC Study #2024-2.  Bat acoustic recordings  At all campsites and during the camp installation period, we placed ultrasonic recorders along the trails, near streams and tree gaps. We installed three Audiomoth recorders per camp, in forest clearings along trails, and configured them to obtain a 1-minute recording every 5 minutes every night (17:30 to 06:00) with a sampling rate set at 384 kHz. They were installed during the advance period and collected at the end of the inventory field work in each camp. In addition, we placed an Anabat recorder opportunistically in the surroundings of camps, creeks, and the Essequibo River at Camp Amuku and Camp Monkey Jump, based on sightings of bats or high likelihood of recording species associated with water. The Anabat was set to be active from 17:30 to 06:00 at Camp Amuku and Camp Monkey Jump during the advance and inventory period. We configured it with a division ratio of eight, a high sensitivity, and continuous recording mode. All recorders were set in Guyana Time (GMT-4) with an output format in .wav for Audiomoths and zero crossing (.zc) for Anabat. For more information on the sampling protocol, including maps and figures, see the Rapid Inventory 32 report (Carrasco-Rueda, Farah, et al., in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 138 - 151.)	Carrasco-Rueda, Farah, et al., in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 138 - 151.	TRUE	FALSE		FALSE		TRUE	FMNH | CSBD	TRUE	whole organism | fertile branch | silica-dried leaf fragment	Farah Carrasco-Rueda | Matthew T. Hallett | Arianne Harris | Huichang Yang | Alex Stewart | Octavious Hendricks | Wenceslaus Washington | Phillip Suse | Samson Yaimo	TRUE	131 camera trap locations | trail surveys | bat captures | acoustic recordings	2537 | 122 | 60 | 295.8	trap nights | km walked | mist net nights | active hours
c2765a6a-d105-4948-818e-bb95823cca55test	3	3 sampling sites, each sampled via the following: visual surveys along trail networks measuring 30, 31, and 18 km, respectively; riverine surveys along large rivers	mature upland forest | swamp forest dominated by Euterpe oleracea | floodplain forest along major rivers | vine-dominated secondary forest | rocky rapids in main river | stunted pole forest | swamp forest dominated by Mauritia flexuosa |	Camp Amuku | Camp Monkey Jump | Camp New River	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				20	days	Embryophyta	Bryophyta | Marchantiophyta | Anthoceratophyta	reported incomplete	based on comparison of lists of detected vs. expected species	TRUE	TRUE	Chlorocardium rodiei | Manilkara bidentata	FALSE		FALSE										FALSE	Plants			sampling events only		open search | opportunistic search | restricted search	floristic inventory | herbarium specimens | tree plots | opportunistic observations	We sampled plants for 4–5 days at each of three campsites. For each vegetation type accessible via the trail system, we recorded as many plant species as possible, documented especially common or notable plants in those vegetation types, and assessed the importance of each vegetation type for the regional flora. We collected and photographed a representative sample of every fruiting and flowering plant encountered, making 1–4 duplicates of each herbarium specimen. Collected plants were pressed in newspaper, preserved in 70% ethyl alcohol, and stored in plastic bags during the fieldwork. All specimens were recorded under the numbers of Kaslyn Holder-Collins, series KH468–652. For every herbarium collection, we preserved one 4 x 4-cm piece of fresh leaf for posterior genetic analyses in coin envelopes dried with silica gel. Specimen data were digitized in the field and all photographs uploaded to iNaturalist within a few hours of capture. Waiwai names for many plants were provided by Cemci ‘James’ Suse, Stephen Suse, Tikkil Kenke, Matias Waiwai, Reuben Yamochi, and Jo Yacipa. We established two tree plots in upland forest, one at Camp Monkey Jump and one at Camp New River. For both plots, all free-standing trees ≥10 cm diameter at breast height were surveyed in a 1-ha transect following one of our trails through upland forest (5 x 2000 m). Each tree in the plots was measured for circumference, identified, and/or collected (sterile vouchers under the numbers of Nigel Pitman, series NCAP11310–11706). While two 1-ha plots are not sufficient to characterize these upland forests, they help put it in context via comparisons with thousands of other such plots established across South America (ter Steege et al. 2023). In both tree plots, we used an auger to collect soil samples at 0–80 cm depth for posterior analysis of charcoal and pollen by Nina Witteveen and Crystal McMichael at the University of Amsterdam. This method allows a glimpse into the deeper history of these forests, by documenting fires, agriculture, and other landscape features over the last ~5,000 years. We sampled five other soil cores at the first two campsites and at old farm sites nearby. These analyses take time and we do not report results in this chapter. For many of the vegetation types and plant species we encountered, local experts provided targeted information on regional extent and abundance, habitat preferences, cultural importance, seasonal patterns, and other crucial insights that our fieldwork would not have otherwise revealed. For more information on the sampling protocol, including maps and figures, see the Rapid Inventory 32 report (Pitman, Nigel C., et al. Vegetation and Flora, in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 109 - 117.)	Pitman, Nigel C., et al. Vegetation and Flora, in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 109 - 117.	FALSE	FALSE		FALSE		TRUE	BRG | F	TRUE	whole organism (bats only) | hair sample (bats only) | sample of the wing patagium using disposable biopsy punches of 2 mm (bats only) | liver tissue sample (bats only) | colon tissue sample (bats only)	N. Pitman | Kaslyn Holder-Collins | M. A. Ríos Paredes | Cemci Suse | Stephen Suse | Matias Waiwai | Tikkil Kenke | Jo Yacipa | Reuben Yamochi | Ronnell Lewis | Christopher Bhola | Nehru Narine | Zachary R. Kachian	TRUE	3 botanists sampling 8 hours a day for 15 days	360	person hours
0df0970e-bee7-403d-85e7-0ccc2449ffd0test	3	3 sampling sites, each sampled with large arrays of camera traps inside the forest	mature upland forest | swamp forest dominated by Euterpe oleracea | floodplain forest along major rivers | vine-dominated secondary forest | stunted pole forest | swamp forest dominated by Mauritia flexuosa | sandy beaches in main river channels | forest creeks, often with low flow or reduced to isolated pools	Camp Amuku | Camp Monkey Jump | Camp New River	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				36	days	Mammalia		reported incomplete	based on comparison of lists of detected vs. expected species	TRUE	FALSE		TRUE	Aves | Reptilia	TRUE										FALSE	Mammals			sampling events only		open search	camera traps	Camera trap photos and videos were obtained following well-established methods for camera trap research. With the support of local experts, 42–45 camera traps (Bushnell Core Prime #119932CB, Bushnell©, KS, USA and Reconyx HF2X HyperFire 2, Reconyx©, WI, USA) were set at each camp during the advance work and collected at the end of the field work. Cameras were left in the field for 6–36 trap-nights (mean = 19) between 17 October and 21 November 2024. The sampling effort totaled 2537 trap-nights in 131 locations across the three camps. Cameras were placed along the trail system, ~500 m apart, with a single camera at each trap location, set 30–60 cm from the ground in proximity to observed animal signs (tracks, paths, marks on trees, scat, animal trails), no farther than 30 m from the trails. Additional cameras were set opportunistically farther away, outside of the trail system (primarily Reconyx cameras set to take videos) in areas where signs of habitat or resource specialists were detected. Cameras were active 24 hours per day, with a 1- to 3-second delay between each 3- or 5-image photo sequence, or a 10-second video without a quiet period between triggers. In an effort to reduce wariness around cameras and avoid biased capture rates, no scents or lures were used, and all cameras employed were equipped with infrared flash. All camera traps recorded the date and time with each photo or video. For each, we recorded coordinates in decimal degrees (Datum: WGS 84), camera information, installation date and time, and physical description of the station. Camera trap data was uploaded in the project ‘RI-32 Guyana: Acarai-Corentyne Corridor’ in Wildlife Insights. For more information on the sampling protocol, including maps and figures, see the Rapid Inventory 32 report (Carrasco-Rueda, Farah, et al., in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 138 - 151.)	Carrasco-Rueda, Farah, et al., in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 138 - 151.	TRUE	FALSE		FALSE		FALSE		FALSE		Farah Carrasco-Rueda | Matthew T. Hallett | Arianne Harris | Huichang Yang | Alex Stewart | Octavious Hendricks | Wenceslaus Washington | Phillip Suse | Samson Yaimo	TRUE	131 camera trap locations	2537	trap nights
e40978cd-d0db-47fb-9987-6d8c7d7fe0cetest	3	3 sampling sites, at each of which ~30 water samples were filtered from upland creeks and pools, larger creeks, and main rivers	mature upland forest | swamp forest dominated by Euterpe oleracea | floodplain forest along major rivers | vine-dominated secondary forest | rocky rapids in main river | stunted pole forest | swamp forest dominated by Mauritia flexuosa | sandy beaches in main river channels | main river channels | forest creeks, often with low flow or reduced to isolated pools	Camp Amuku | Camp Monkey Jump | Camp New River	15017.75	km2	153	km2	Extremely dry weather, very low river levels, many creeks dried out, only one short rainstorm during sampling				20	days	Aves | Mammalia | Reptilia | Amphibia | Actinopterygii		reported incomplete	with eDNA we were only able to detect the subset of vertebrate species for which we had sequences in our reference library; Sophie and Lesley did not report this number in their chapter but should be able to generate it	TRUE	FALSE		FALSE		FALSE										FALSE	Mammals | Birds | Amphibians | Reptiles | Freshwater fishes			sampling events only		open search	eDNA water samples	Building a DNA reference database for local vertebrate species In an effort to optimize detection of vertebrate species through eDNA, we created a DNA database composed only of known DNA barcodes for vertebrate species of the region, as the accuracy of taxonomic assignment in eDNA metabarcoding studies can be improved if prior expectations of species occurrence are included. This process typically involves using external information about species abundance or presence, ‘geographic filtering’, or excluding from consideration species that are not reasonably expected to occur at a site. To do this, we used the Map of Life Rapid Assessment dashboard to create a list of expected vertebrate species in a 100-km radius area around our sampling zone. The genetic marker we targeted in our eDNA analysis is a small fragment of the 12S rRNA gene in the hypervariable V5 region (~110–120 base pairs), which has been shown to be efficient in reconstructing entire vertebrate communities in a range of habitats. We therefore downloaded 12S rRNA sequences and associated taxonomic information from GenBank for expected vertebrate species in the vicinity of the Acarai-Corentyne Corridor using the CRABS (version 1.0.9) software and workflow. A total of 101 12S sequences, representing 64 fish species, were also generated in-house and added to our database. Sequences were trimmed to the primer regions, and all sequences with unique taxonomic annotations were retained, even if the sequences were duplicates (dereplication). Our local reference database contained a total of 1258 sequences: 325 sequences of 194 fish species, 228 sequences of 84 amphibians, 93 sequences of 56 reptiles, 296 sequences of 190 birds, and 316 sequences of 167 mammals. All reference sequences were quality-checked for nomenclatural errors in GenBank by plotting neighbor-joining sequence trees for each family. The neighbor-joining trees were constructed using the ‘muscle’ and ‘ape’ packages in R v4.4.0.  Fieldwork eDNA samples were collected over a 21-day period across the three camps. At each camp, we sampled between 14 and 20 sites, generating 30–34 filters per camp including one blank per camp for a total of 97 eDNA filters. Our sampling coverage represented all aquatic habitats accessible by boat or by trail. Sampled water bodies encompassed two different river basins (the Essequibo and the Corentyne) and included the main river channels, tributaries, and streams, as we aimed to characterize species variation among the different habitat types. We collected 1–4 water samples at each site, depending on the width of the waterbody and available microhabitat types. Water was pumped through self-preserving polyethersulfone (PES) membrane filters of 0.8-µM pore size and 47-mm diameter (Smith-Root Inc.) using either an eDNA Citizen Scientist Sampler vacuum pump (Smith-Root Inc.) or a peristaltic pump powered by a drill. Water was passed through each filter until the filter clogged. For each sample taken, we recorded total water volume filtered, microhabitat descriptions, as well as water temperature, pH, and dissolved oxygen, which were measured with a multiparameter water sonde (YSI). Due to the nature of trace eDNA, contamination—or the accidental introduction of non-target DNA—can be easily introduced by improper handling. We controlled for contamination between sites by frequently changing gloves, and contamination between camps by sterilizing filter equipment with a 10% bleach solution at the beginning of each camp. One field blank per campsite was generated by filtering 1 L of distilled water with the sampling gear, in order to measure contamination introduced during handling and control for potential false positive results. Collected filters were dried and stored in dark bags to prevent DNA degradation by UV light in the field. All filters were stored at ambient temperature in the dark until our return to Georgetown, at which point we placed the filters in –20°C freezers at the University of Guyana.  Laboratory work eDNA metabarcoding was carried out in two stages: 1) minION subset: A subset of 18 field samples (six per camp), representing a variety of habitat types, was processed at the Centre for Study of Biological Diversity at the University of Guyana using portable lab equipment to generate a preliminary list of species (Fig. 42) on 25–27 November 2024. 2) Full dataset: All 97 field samples, including three blanks, were then processed at the Field Museum from January to March 2025 using the Museum’s Biomolecular Clean Lab and the Pritzker Laboratory for Molecular Systematics and Evolution. Here we report results from both sequencing runs. DNA extractions MinION subset: 18 filters were cut into two pieces with sterilized scissors. One piece was then extracted at the University of Guyana with a Qiagen extraction kit (DNeasy Blood and Tissue kit) with some minor modifications. The other piece was preserved in a CTAB buffer and transported to the Field Museum for more thorough processing using the Illumina pipeline. To avoid contamination, extractions and PCR setup were conducted in a lab space free from animal tissues and PCR products. Full dataset: All remaining eDNA samples were extracted at the Field Museum in the Biomolecular Clean Lab to avoid contamination using phenol-chloroform extraction methods as described in Picq et al. (2024). Amplification of vertebrate DNA For both the MinION and full datasets, amplification was carried out in the V5 region of the 12S ribosomal RNA region with a general vertebrate primer set that amplifies a ~100bp fragment (Forward: 5'‐ACTGGGATTAGATACCCC‐3', Reverse: 5'‐TAGAACAGGCTCCTCTAG‐3') developed by Riaz et al. (2011). Primers were modified to include an overhang on the 5'-end to allow for indexing and the addition of the appropriate sequencing adapters (ONT or Illumina) during the two-step PCR-based library preparation process. A 25μL first step PCR reaction was used with the following recipe: 14.95μL sterile water, 2.5μL 10× Amplitaq Gold buffer, 0.6μL 10mM dNTPs, 2μL 25mM MgCl2, 0.8μL 10μM of each 12S-V5 primer, 0.25μL 5U/μL Amplitaq Gold DNA Polymerase (Applied Biosystems), 0.1μl 50 mgl/ml Bovine Serum Albumin, and 3μL of DNA. Thermocycling conditions were as follows: initial denaturation at 95°C for 10 minutes followed by 40 cycles of denaturation at 95°C for 10 seconds, annealing at 54°C for 30 seconds, extension at 72°C for 30 seconds, and f inal extension at 72°C for 5 minutes. For the minION subset, all field samples were analyzed with three PCR replicates, while for the full dataset, 12 PCR replicates were performed, to better capture the diversity of DNA in each f ilter. Corresponding field negatives and lab negative controls were included to quantify the non-target DNA that may have been introduced by handling. Exogenous positive control communities of vertebrate species were included to quantify the rate of barcode-jumping, so that sequences could be assigned to localities with better confidence. All replicates were pooled and amplification was confirmed by gel electrophoresis. PCR cleanup was conducted with ExoSAP-IT Express for the minION subset and with Serapure magnetic beads at a 1.8X bead-toPCR ratio for the full dataset. PCRs were then quantified with a Qubit Fluorometer.   Library Preparation For the MinION subset, eDNA library preparation was performed with the Oxford Nanopore Technologies Ligation Sequencing Kit (LSK-114) and PCR Barcoding Expansion 1-96 (EXP-PBC096). Sample concentrations were adjusted to contain 5 ng of DNA and unique barcode sequences were attached to each sample, negative, and positive. Samples were normalized to a standard concentration and pooled equimolarly. Bead cleanups were conducted with a ratio of 1.4X with Ampure beads. Following equimolar pooling, all library preparation was conducted according to the Oxford Nanopore Technologies Ligation protocol for end prep, adapter ligation, and loading. All samples were loaded on a MinION FlowCell (R10.4.1), and the MinION portable sequencer was run for a total of 10 hours. For the full dataset, sample concentrations were also adjusted to contain 5 ng of DNA and unique barcode sequences were attached to each sample, negative, and positive. All samples were again cleaned with Serapure at a ratio of 1.8X. Following equimolar pooling, all samples were loaded on an Illumina NovaSeq platform at the Northwestern University sequencing core.   Bioinformatic analyses For the MinION subset, sequencing was carried out on a MacBook Pro M1 8-Core CPU and 32 GB of unified memory on macOS Sonoma Version 14.5. Basecalling and subsequent bioinformatic analyses were performed at Michigan State University’s High-Performance Computing Center. Our bioinformatics workflow is based on a pipeline developed by the Map of Life Rapid Assessments XPRIZE team. We used the ‘high accuracy’ (hac) model run via dorado software to perform basecalling during sequencing, resulting in FASTQ f iles of demultiplexed sequences. NanoFilt v2.8.0 was used to remove reads with an average quality score <10. We used seqkit v2.8.0 to remove raw reads outside of the expected length range (i.e., amplicon length + primer length + adapter length). Length filtering at this step avoids the need for chimera removal. Next, we trimmed off the primers from each read using Cutadapt v4.8, accounting for every combination of forward and reverse primer sequences. We set the Cutadapt error rate to 0.2 and combined the results from all Cutadapt passes, removing duplicated reads. We then used porechop v0.2.4 to remove any ONT adapters and barcodes from reads in which Cutadapt did not find both a forward and reverse primer sequence (i.e., reads with only one primer sequence removed via Cutadapt). Next, we clustered reads using the cluster_fast function from VSEARCH v2.27.1 with a global clustering threshold (id) of 0.95 to generate MOTUs (Molecular Operational Taxonomic Units). To guard against PCR and sequencing errors, we discarded any cluster with ≤5 reads. To assign taxonomy to our environmental sequences, we performed BLAST (Basic Local Alignment Search Tool; the blastn function in BLAST+ v.2.15.0) searches on the consensus sequences from each cluster against our local reference sequence database. For any consensus sequences without a match of 95% or greater to the local reference database, we used blastn to perform a second search against the broader NCBI nucleotide sequence database for all eukaryotes (nt_euk), downloaded on 30 October 2024. We processed the blastn output using R, including removing sequences with an alignment length (query coverage) <90 base pairs. We used a threshold of ≥98% match to assign sequences to species and ≥95% to assign a genus-level identification. Within a threshold level, some sequences matched to more than one taxa; in these cases, we assigned the taxa of the most recent common ancestor (e.g., genus or family level). We removed common contaminant species, including humans, domesticated animals, and food items (Gallus, Felis, Canis, Rattus, Mus, Equus, Ovis, Sus, Bos, Capra, Thunnus). In metabarcoding studies, barcode tag jumping can cause species detected in one sample to erroneously appear in another sample. To guard against this, we included an exogenous positive control containing a mock DNA community of species (seven fish, one reptile, one amphibian, and one bird) that do not occur in the study area. We determined the maximum present in a negative control). If the resulting number of reads was equal to or less than zero, we removed this cluster. Bioinformatic processing of the full Illumina dataset was conducted using QIIME2 v2024.10. Briefly, raw demultiplexed reads were trimmed using Cutadapt v-4.2 (Martin et al. 2011), with a maximum error rate of 0.25. Trimmed reads were then denoised to remove putative PCR and sequencing errors using DADA2 v1.26.0 and the ‘pseudo’ pooling setting, with the additional parameters ‘p-trunc-len-f 95’ and ‘p-trunc-len-r 95,’ resulting in 3553 amplicon sequence variants (ASVs). As in Picq et al. (2024), taxonomic assignment was carried out via a hybrid approach, first prioritizing exact matches using the global-aligner VSEARCH and then retaining Naive Bayes classifications with sufficient taxonomic information. The Bayesian classifier was trained using the RESCRIPt command ‘qiime rescript evaluate-fit-classifier’, producing the QIIME object required for kmer-based classification. For VSEARCH, 100% identity across at least 94% query coverage was required with ‘qiime feature-classifier classify-consensus-vsearch’. Only ASVs with at least family-level taxonomic classifications were retained from this initial alignment-based classification (n = 240). All other remaining ASVs were then classified with the Bayesian classifier, using ‘qiime feature-classifier classify-sklearn’ with default parameters. Similarly, only ASVs with family-level taxonomic classifications were retained (n = 3185). Lastly, nucleotide BLAST was used to query all other ASVs that were not assigned taxonomy by previous classifiers to account for other possible vertebrate species not present in the taxonomic database, requiring matches ≥98% percent identity for at least 95% query coverage. A total of 128 ASVs with a minimum family-level taxonomy were retained. Sequences that had either no taxonomic assignment, or no assignment at least at the phylum level, were discarded. Similarly, sequences assigned to humans, domesticated, or farm animals were considered as common contaminants and removed. To control for tag jumping, we included two exogenous positive controls containing two mock DNA communities of species that do not occur in the study area. We determined the maximum within-sample proportion of any ASV assigned to a positive-control species (excluding the positive control and negative controls). We then moved ASVs equal to or less than this proportion (0.0009) to the ‘unclassified’ sample category. For each sample, we subtracted the maximum number of reads found in negative samples for each ASV (if that ASV was present in a negative control). If the resulting number of reads was equal to or less than zero, we removed this ASV from the dataset. For both datasets, all taxonomic identifications were reviewed by taxonomic experts within our team to control for potential false positives. Rarefaction curves were generated using the R statistical software package ‘iNEXT’. For more information on the sampling protocol, including maps and figures, see the Rapid Inventory 32 report (Picq, Sophie, et al., in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 151 - 159.)	Picq, Sophie, et al., in Pitman, Nigel C., et al. RI-32 Guyana: Acarai-Corentyne Corridor. 151 - 159.	FALSE	FALSE		FALSE		TRUE	FMNH	TRUE	DNA from water pumped through self-preserving polyethersulfone (PES) membrane filters of 0.8-µM pore size and 47-mm diameter	Sophie Picq | Mariel Vandegrift | Gyanpriya Maharaj | Annalise Bayney | Ronnell Lewis | Samantha Garcia	TRUE			
