African savannas are the last stronghold of diverse large-mammal communities, and a major focus of savanna ecology is to understand how these animals affect the relative abundance of trees and grasses. However, savannas support diverse plant life-forms, and human-induced changes in large-herbivore assemblages—declining wildlife populations and their displacement by livestock—may cause unexpected shifts in plant community composition. We investigated how herbivory affects the prevalence of lianas (woody vines) and their impact on trees in an East African savanna. Although scarce (<2% of tree canopy area) and defended by toxic latex, the dominant liana, Cynanchum viminale (Apocynaceae), was eaten by 15 wild large-herbivore species and was consumed in bulk by native browsers during experimental cafeteria trials. In contrast, domesticated ungulates rarely ate lianas. When we experimentally excluded all large herbivores for periods of 8 to 17 years (simulating extirpation), liana abundance increased dramatically, with up to 75% of trees infested. Piecewise exclusion of different-sized herbivores revealed functional complementarity among size classes in suppressing lianas. Liana infestation reduced tree growth and reproduction, but herbivores quickly cleared lianas from trees after the removal of 18-year-old exclosure fences (simulating rewilding). A simple model of liana contagion showed that, without herbivores, the long-term equilibrium could be either endemic (liana-tree coexistence) or an all-liana alternative stable state. We conclude that ongoing declines of wild large-herbivore populations will disrupt the structure and functioning of many African savannas in ways that have received little attention and that may not be mitigated by replacing wildlife with livestock.

Detailed methods are provided in the Main Text and Supplementary Information of the associated article (Coverdale et al. 2021, Proc. Natl. Acad. Sci. U.S.A.; https://doi.org/10.1073/pnas.2101676118). Questions concerning datasets or methodology should be directed to Dr. Tyler Coverdale (tc684@cornell.edu) and/or Dr. Robert Pringle (rpringle@princeton.edu).

The experimental designs and baseline conditions of the UHURU (Ungulate Herbivory Under Rainfall Uncertainty) and GLADE (Glade Legacies and Defaunation Experiment) studies at Mpala Research Centre, Kenya, are described in detail by Goheen et al. 2013 (PLOS ONE; https://doi.org/10.1371/journal.pone.0055192) and 2018 (Ann. N.Y. Acad. Sci.; https://doi.org/10.1111/nyas.13848). The four UHURU treatments are coded in the included datasets as "Total" (corresponding to "-all" in the associated article), "Meso" (corresponding to "-meso" in the article), "Mega" (corresponding to "-mega" in the article), and "Control" (corresponding to "+all" in the article). These deviations from the naming conventions in the article are intended to reduce issues arising from the inclusion of plus and minus symbols in treatment names during analysis. Unless otherwise specified, all UHURU data are from the southern set of replicates (3 blocks, 12 plots, as described in Goheen et al. 2013); some files also contain data from the central and northern set of replicates (comprising an additional 6 blocks and 24 plots). Raw data from the first five years of the UHURU study have previously been published (Kartzinel et al. 2014, Ecology; https://doi.org/10.1890/13-1023R.1) and overlap with some of datasets presented here; a follow-up data paper presenting the first 12 years of data from UHURU has been submitted for publication (J.M. Alston et al., “Ecological consequences of large-herbivore exclusion in an African savanna: 12 years of data from the UHURU experiment”) and likewise overlaps with some of the datasets presented here (e.g., canopy-intercept surveys of Cynanchum prevalence and data underlying our estimates of tree biomass). Information regarding the most recent and complete version of any UHURU data is available from the senior corresponding author (Dr. Robert Pringle; rpringle@princeton.edu). 

1) Cafeteria trial

Removal of transplanted lianas by herbivores in a cafeteria-style feeding trial. This file contains data underlying Figure 1F. The feeding trial is also pictured in Figure 1H and Movies S1-S4. Columns J–N give the number of herbivory events by each of 5 common browser species on each transplanted liana, as recorded with camera traps. Three transplanted lianas (on recipient trees #5, #12, #14, corresponding to the labels on the x-axis of Fig. 1F) were not monitored with camera traps (thus, all herbivore counts are reported as "NA").

Feeding_Trial.csv

2) Desiccation trial

Desiccation of 5 transplanted lianas over 10 days. These data were used to infer how much of the weight lost during cafeteria-style feeding trials (Fig. 1F) was attributable to herbivory vs. desiccation.

Desiccation_Trial.csv

3) UHURU liana prevalence surveys

Prevalence and cover of Cynanchum viminale as a function of herbivore exclosure treatments in UHURU in 2017. Missing data coded as "NA". Tree species key: AE = Acacia etbaica; AM = Acacia mellifera; AB = Acacia brevispica; AD = Acacia drepanolobium; AN = Acacia nilotica; AS = Acacia seyal; AX = Acacia xanthophloea; BR = Balanites rotundifolia; BOS = Boscia angustifolia; Other = Grewia spp.,  Lycium europaeum, Croton dichogamus, Euclea divinorum, and other unknown species. This file contains data underlying Figure 2A–D. Data on liana prevalence in GLADE (rightmost bars in Fig. 2C and 2D) are in the file GLADE_Liana_Prevalence_And_Herbivore_Reintroduction_Survey.csv; data on tree density and biomass in UHURU (gray bars in Fig. 2B) are in the file UHURU_Tree_Biomass_and_Density.csv.

UHURU_Liana_Survey.csv

4) UHURU transect surveys

Effect of herbivore exclosure treatments on juvenile liana densities beneath and outside tree canopies in UHURU in 2016. Each transect was 50m x 4m (200 m²) and aligned with a permanent grid of stakes in the center of each plot. This file contains data underlying Figure 2E and 2F.

UHURU_Transect_Survey.csv

5) UHURU tree survey

Effects of herbivore exclosure treatments on tree density and aboveground biomass in UHURU; data encompass the years 2016–2018 (see article text). "Level" corresponds to the location of replicate UHURU blocks at Mpala Research Centre (3 blocks and 12 plots at each of South, Central, and North locations; see Goheen et al. 2013); only data from the "South" level were used in the associated article. Estimated biomass of trees in each size class (<1m, 1-2m, 2-3m, 3-4m, >4m) was calculated by multiplying the number of trees in each class by the estimated biomass of the average tree in that class (see article text for details). Total biomass (kg/ha) was estimated by summing the biomass of trees in all size classes >1m tall. Trees with negative biomass values according to the equation used for estimation (i.e., those with height < 1m) were excluded from analyses in the manuscript but are included in the dataset. Estimated tree density (#/ha) is similarly provided for each plot; total density was estimated by summing the densities of all trees >1m tall. This file contains data underlying Figure 2B.

UHURU_Tree_Biomass_and_Density.csv

6) GLADE liana prevalence and effects of herbivore reintroduction

Prevalence of lianas in GLADE (percent cover and number of lianas per tree) and effects of exclosure fence removal on liana cover. Each row corresponds to a tagged tree that was surveyed in both 2016 and 2017. Replicates from GLADE (where fences were removed prior to the resurvey in 2017) are coded as "Fence Removal" in the "Treatment" column; replicates from the UHURU experiment, used as controls for assessing the effect of herbivore reintroduction, are coded as "Total" (i.e., –all exclosure). Liana heights was not measured in UHURU. Tree species key: AE = Acacia etbaica; AM = Acacia mellifera; AB = Acacia brevispica; AD = Acacia drepanolobium; AN = Acacia nilotica; AS = Acacia seyal; BR = Balanites rotundifolia; BOS = Boscia angustifolia; GX = Grewia spp.; LE = Lycium europaeum; CD = Croton dichogamus; ED = Euclea divinorum. This file contains data underlying Figure 2C and 2D (liana prevalence in GLADE in 2016) and Figure 3 (effects of fence removal).

GLADE_Liana_Prevalence_And_Herbivore_Reintroduction_Survey.csv

7) Liana growth surveys

Effects of herbivore exclosure treatment on liana growth rate in UHURU from August 2016 to January 2017. Data presented are plant-level averages of multiple branch measurements from 117 individually tagged lianas. This file contains data reported only in the article text.

Liana_Growth_Survey.csv

8) Liana-removal experiment

Effects of experimental liana removal on tree growth in UHURU from June 2016 (“Initial”) to June 2017 (“Final”). Tree species key: AE = Acacia etbaica; AD = Acacia drepanolobium. "Triplet" denotes matched sets of trees (i.e., same species and similar height growing in close proximity within the same plot). This file contains data underlying Figure 4A.

Liana_Removal_Experiment.csv

9) Tree reproductive survey

Effects of liana infestation severity on reproductive output of 24 Acacia etbaica trees in UHURU in January 2017. The corresponding analysis presented in the paper is based on log-transformed reproductive data after averaging the replicate measurements for individual trees. This file contains data underlying Figure 4B.

Acacia_Reproductive_Survey.csv

10) UHURU canopy-intercept survey

Effect of herbivore exclosure treatment on liana abundance and mean height in UHURU in November 2018. "Level" corresponds to the location of replicate UHURU blocks (South, Central, North; see Goheen et al. 2013). “Total # of vegetation hits” includes all plant species measured at all heights from the ground to the top of the canopy; “# of Cynanchum hits” is the number of those hits corresponding to the lianas C. viminale and C. gerrardii; “Mean height (cm)” is the average height of all Cynanchum spp. hits in each plot. Eight plots where Cynanchum was not detected (i.e., 0 hits) are excluded from this file; these include the Mega (–mega) plot of Central Block 3 along with 7 of the 9 Control (+all) plots (all except those in South Blocks 1 and 3). This file contains data underlying Figure S1A and S1B.

UHURU_Canopy_Intercept_Survey.csv

11) UHURU Cynanchum understory survey

Effect of herbivore exclosure treatment on liana abundance in the understory (< 0.5 m height) in 20 surveys in UHURU from 2008–2019. The lianas Cynanchum viminale and Cynanchum gerrardii were lumped for the analysis presented in the article and were not distinguished in surveys prior to Survey 13 (2014). "Level" corresponds to the location of replicate UHURU blocks (South, Central, North; see Goheen et al. 2013). Surveys were conducted by placing a 10-pin frame inside a 0.25 m² quadrat at each of 49 evenly spaced locations in each plot (Goheen et al. 2013); surveys with missing data are denoted by values <49 in the "#Quadrats with data" column. "Total # of hits" includes all vegetation hits (i.e., the total number of contacts between plants and pins) along with bare-ground hits (maximum 1 possible for each pin). This file contains data underlying Figure S1C and S1D.

UHURU_Cynanchum_Understory_Survey.csv