Lianas – woody climbers – are great competitors, affecting tree survival and growth (Tobin et al. Reference TOBIN, WRIGHT, MANGAN and SCHNITZER2012) as well as forest structure (Schnitzer & Bongers Reference SCHNITZER and BONGERS2002). The abundance of lianas is reported to have been increasing in the Amazon forest over the last 20 y (Laurance et al. Reference LAURANCE, ANDRADE, MAGRACH, CAMARGO, VALSKO, CAMPBELL, FEARNSIDE, EDWARDS, LOVEJOY and LAURANCE2014), which may have a global impact due to lower carbon uptake capacities in tropical forests with high liana densities (Schnitzer et al. Reference SCHNITZER, VAN DER HEIJDEN, MASCARO and CARSON2014).

In the Amazon forest, a limited number of liana (Burnham Reference BURNHAM2002) and tree species (Macía & Svenning Reference MACÍA and SVENNING2005, Pitman et al. Reference PITMAN, TERBORGH, SILMAN, NÚÑEZ, NEILL, CERÓN, PALACIOS and AULESTIA2001, ter Steege et al. Reference TER STEEGE, PITMAN, SABATIER, BARALOTO, SALOMÃO, GUEVARA, PHILLIPS, CASTILHO, MAGNUSSON and MOLINO2013) dominate the plant communities. Most of the dominant tree species are widely distributed, but dominant only in a few places (ter Steege et al. Reference TER STEEGE, PITMAN, SABATIER, BARALOTO, SALOMÃO, GUEVARA, PHILLIPS, CASTILHO, MAGNUSSON and MOLINO2013). The dominance of few widely distributed species characterizes the oligarchy hypothesis (Pitman et al. Reference PITMAN, TERBORGH, SILMAN, NÚÑEZ, NEILL, CERÓN, PALACIOS and AULESTIA2001). High local diversity of plants is often related to local habitat variability, with most species occurring in some parts of the forest. Many species are associated with a particular kind of topography, which is often associated with particular edaphic conditions, forest structure and dynamics (Baldeck et al. Reference BALDECK, HARMS, YAVITT, JOHN, TURNER, VALENCIA, NAVARRETE, DAVIES, CHUYONG and KENFACK2013, John et al. Reference JOHN, DALLING, HARMS, YAVITT, STALLARD, MIRABELLO, HUBBELL, VALENCIA, NAVARRETE, VALLEJO and FOSTER2007, Queenborough et al. Reference QUEENBOROUGH, BURSLEM, GARWOOD and VALENCIA2007, Svenning Reference SVENNING1999). Most of the dominant tree species in the Amazon have, in contrast, been characterized as generalists (Duque et al. Reference DUQUE, CAVELIER and POSADA2003, Pitman et al. Reference PITMAN, TERBORGH, SILMAN, NÚÑEZ, NEILL, CERÓN, PALACIOS and AULESTIA2001), occurring on a range of soil types and topographic positions (Pitman et al. Reference PITMAN, TERBORGH, SILMAN, NÚÑEZ, NEILL, CERÓN, PALACIOS and AULESTIA2001). Not only can the distribution of trees be related to environmental factors, but also their vital rates. In the Amazon, tree mortality can be related to soil factors (Phillips et al. Reference PHILLIPS, BAKER, ARROYO, HIGUCHI, KILLEEN, LAURANCE, LEWIS, LLOYD, MALHI and MONTEAGUDO2004) and water availability (Phillips et al. Reference PHILLIPS, ARAGÃO, LEWIS, FISHER, LLOYD, LÓPEZ-GONZÁLEZ, MALHI, MONTEAGUDO, PEACOCK and QUESADA2009). Although the density of lianas has increased in the Amazon forest in the last 30 y (Phillips et al. Reference PHILLIPS, MARTÍNEZ, ARROYO, BAKER, KILLEEN, LEWIS, MALHI, MENDOZA, NEILL, VARGAS, ALEXIADES, CERÓN, FIORE, ERWIN, JARDIM, PALACIOS, SALDIAS and VINCETI2002), there is no work demonstrating how the vital rates of dominant liana species relate to environmental heterogeneity.

Our aim was to investigate how sensitive the vital rates of a dominant Amazonian liana species are to environmental variation using a long-term study. We tested the influence of environmental variation (maximum canopy height, understorey density, floristic composition, soil components and topographic steepness) on the relative growth rate (RGR) and mortality of the liana Machaerium cuspidatum Kuhlm. & Hoehne (Fabaceae) over an 11-y period at Yasuní National Park, Ecuador. As palms and their relation with the environment are well known in the Amazon forest (Kristiansen et al. Reference KRISTIANSEN, SVENNING, EISERHARDT, PEDERSEN, BRIX, MUNCH KRISTIANSEN, KNADEL, GRÁNDEZ and BALSLEV2012, Svenning Reference SVENNING1999, Reference SVENNING2002; Svenning et al. Reference SVENNING, HARLEV, SØRENSEN and BALSLEV2009), we used the palm species composition (Vormisto et al. Reference VORMISTO, PHILLIPS, RUOKOLAINEN, TUOMISTO and VÁSQUEZ2000) as a proxy for general floristic variation and related environmental drivers.

The study site in the Yasuní National Park in Amazonian Ecuador (00°40′S, 76°23′W) is located in an old-growth tropical moist forest with little annual variation in precipitation and temperature (Nabe-Nielsen Reference NABE-NIELSEN2001) and with relatively small natural gaps (Svenning Reference SVENNING1999). Floodplain, terra firme and swamp are the three local major habitats (Svenning Reference SVENNING1999). The forest has a high diversity and low density of lianas (Nabe-Nielsen Reference NABE-NIELSEN2001). The soil is acidic and silt is prominent (Tuomisto et al. Reference TUOMISTO, POULSEN, RUOKOLAINEN, MORAN, QUINTANA, CELI and CAÑAS2003). The average yearly precipitation is 2255 mm, and the average temperature is 24.7ºC (data from 1995 to 2009 – Estación Científica Yasuní, http://www.yasuni.ec/).

Machaerium cuspidatum is a shade-tolerant liana (Nabe-Nielsen Reference NABE-NIELSEN2004) that is abundant and widely distributed in the Amazonian rain forest of Colombia, Peru, Bolivia, Brazil and Ecuador (Lozano & Klitgaard Reference LOZANO and KLITGAARD2006). It is one of the most abundant liana species in the Yasuní National Park, eastern Ecuador (Burnham Reference BURNHAM2002, Burnham & Romero-Saltos Reference BURNHAM, ROMERO-SALTOS, Schnitzer, Bongers, Burnham and Putz2015, Nabe-Nielsen Reference NABE-NIELSEN2002). The individuals of M. cuspidatum with diameter ≥ 1 cm in Yasuní National Park are more abundant in low-canopy and dense-understorey areas (Nabe-Nielsen & Hall Reference NABE-NIELSEN and HALL2002). The population growth rate is most influenced by survival of large individuals and relatively insensitive to recruitment due to low survival rate of seedlings (Nabe-Nielsen Reference NABE-NIELSEN2004).

All individuals of M. cuspidatum in seven 20 × 250-m transects in a terra firme area were censused in 1998 and re-censused in 2009 (1378 individuals in total). Three transects were located in terra firme areas close to the Tiputini river and four transects were far from the river, c. 8 km further west. In both years, the diameter of each individual was measured with a calliper at the first point above the roots where the stem was regular, which was painted to assure the next measurement was taken in the same place. Mortality was recorded in the last census.

The transects were divided into 5 × 5-m subplots; in each subplot we measured: (1) maximum canopy height, classified into six categories: 0–2 m, 2–5 m, 5–10 m, 10–20 m, 20–40 m and ≥ 40 m; (2) topographic steepness, divided in four classes: depression, flat (< 10º), slope (≥ 10º and < 30º) and steep slope (≥ 30º); and (3) the understorey vegetation density, divided into dense vegetation and sparse vegetation. Canopy height was measured in both censuses, and we used the mean value in the data analyses. The understorey density did not change between years (χ² = 0.420, df = 1, P = 0.516); we used data from 2009 in the analyses. Three soil samples were collected in each transect for chemical analyses, which were made following van Reeuwijk (Reference VAN REEUWIJK2002) in the laboratories of the Department of Bioscience at Aarhus University. All palm individuals were identified in a sub-transect of 5 × 250-m within each 20 × 250-m transect. Soil and palm data were sampled in 2012. From 1995 to 2012 the palm community remained stable (Balslev, unpubl. data).

For data analysis, the M. cuspidatum population was divided into three classes: seedling-sized plants, non-climbers and climbers. We used this classification because the transition to becoming a climber is a key phase in the life history of lianas and is known to improve their fitness (Gianoli Reference GIANOLI2015). We refer to individuals ≤ 30 cm long as seedling-sized. Non-climbers have length > 30 cm and can be upright or lie on the ground; climbers are > 30 cm long and attached to a host plant. The soil data were summarized as the first axis of a Principal Component Analysis (PCA) with the soil components calcium, potassium, magnesium and sodium. We used Beals smoothing function on palm data (De Caceres & Legendre Reference DE CACERES and LEGENDRE2008) to obtain a probability of occurrence of palms in the 5 × 5-m subplot where there were no palms. Then, we summed the calculated value for every five subplots, repeated the summed value in these plots, and did Hellinger transformation (Legendre & Gallagher Reference LEGENDRE and GALLAGHER2001). We summed the values to avoid many subplots with zeros. We used the second axis of a PCA with the transformed palm species distribution to characterize variation in the palm species composition that were associated with environmental variation. The first axis only captured spatial variations in the species composition. In the areas of the 20 × 250-m transect where there were no palm data available, we used the PCA second axis score of the adjacent 5 × 5-m subplot. The influence of environmental factors on RGR of each class was tested by Akaike Information Criterion (AIC) model selection with linear regression and linear mixed model (LMM). RGR was logit transformed for seedling-sized plants. Mortality was treated as a binary variable, and the influence of environmental factors on it was tested by AIC model selection with generalized linear model (GLM) and generalized mixed model (GLMM). Mean maximum canopy height, steepness, soil, palm species composition and understorey density were treated as fixed effects in all models. Transect number, region (close or far from the river), and transect within regions were treated as random effects. We used backward stepwise for model selection, as recommended by Zuur et al. (Reference ZUUR, IENO, WALKER, SAVELIEV and SMITH2009). For Beals smoothing function and Hellinger transformation we used the vegan package (http://CRAN.R-project.org/package=vegan) for R (R Development Core Team). The mixed effect models were made in lme4 package (http://www.inside-r.org/packages/lme4/versions/1-0-4) for R (R Development Core Team).

Over 11 y, 421 of 448 seedling-sized plants died, 539 of 732 non-climbers died, and 107 of 198 climbing individuals died. The mean relative growth rate of the surviving seedling-sized plants (± 1SE) was 0.164 ± 0.062 mm mm⁻¹ y⁻¹, of non-climbers it was 0.096 ± 0.021 mm mm⁻¹ y⁻¹ and for climbers it was 0.094 ± 0.018 mm mm⁻¹ y⁻¹. The models selected for RGR based on AIC did not include any random factor, and only the model for climber growth was significant (F = 3.98, r² = 0.04, P = 0.049). The only variable retained in this model, understorey density, had a weak positive effect on the RGR (estimates = −0.010, t test = −1.99, P = 0.05). For mortality, only the model selected for seedling-sized plants was significant. It included mean maximum canopy height and slope, but no random effects. The mortality of seedling-sized plants was higher (58.4%) in areas with intermediate (10º–30º) steepness than elsewhere (estimates = −0.690, t test = −2.04, P = 0.042). The models selected for non-climbers and climbers included transect as a random effect, but none of the variables was significant in these models.

Many species of liana in tropical forests are generally described as light demanding (Castellanos Reference CASTELLANOS, Putz and Mooney1991, Schnitzer & Bongers Reference SCHNITZER and BONGERS2002). Over the short term (1998–2000), M. cuspidatum seedlings (individuals ≤ 30 cm high) have been found to grow more in places with higher light availability (Nabe-Nielsen Reference NABE-NIELSEN2002). However, in the long term (1998–2009) the vital rates of the species were uncorrelated with a range of environmental variables that are all related to variation in light availability (maximum canopy height, steepness and understorey density). Although the mortality of seedling-sized plants was higher in areas with intermediate slope and the RGR of climbing individuals was higher in dense understorey, the effect was weak for the RGR of climbing individuals, and the vital rates were not related to any other type of environmental variation. The most parsimonious models did not suggest that the vital rates varied among transects with different environmental conditions either, e.g. between bottomland and ridgetop habitats.

The weak relationship between the vital rates of M. cuspidatum and the studied environmental factors indicate that the species can be regarded as an ecological generalist. Whereas lianas are generally characterized as species that respond to high-light environments and availability of suitable supports by growing fast (Putz Reference PUTZ1984), this is not the case for M. cuspidatum. Two studies from the Ecuadorian Amazon have shown that the three most dominant tree species are all relatively indifferent to environmental heterogeneity (Pitman et al. Reference PITMAN, TERBORGH, SILMAN, NÚÑEZ, NEILL, CERÓN, PALACIOS and AULESTIA2001). The present study suggests that ecological generalism may play a similar role in explaining the prevalence of the most abundant liana species.