Follow the forest: Slow resilience of West African rainforest frog assemblages after selective logging

Highlights

• Amphibian communities were assessed at identical sites more than 15 years apart.

• Amphibian composition in previously logged forests gradually recovered following forest regeneration.

• Even 45 years post-logging, frog assemblages still differ between old and re-growth forests.

• Recovery speed of frog assemblages depends on the speed of forest structure recovery.

Abstract

Assessing resilience of ecological communities, the ability of species assemblages to return to a previous state following disturbance, is an essential task when designing biodiversity friendly concepts for silvicultural management regimes. However, ecological communities may demonstrate resilience only after long time periods, and long-term data, collected on permanent study sites within standardized sampling frameworks, are scarce. We assessed the resilience of amphibian assemblages to disturbance via selective logging within the largest remaining rainforest in Ivory Coast, the Taï National Park. We analyzed extensive amphibian and habitat data obtained during two major assessment periods > 15 years apart (period P1 = in 2000, period P2 = in 2016–2017) and spanning 45 years post-logging recovery (period P1 = 30 years post-logging). We revisited identical sites and used identical, standardized methodology in both study periods. Forest structure in formerly selectively logged sites slowly recovered towards old growth forest structure over the course of 45 years, with most visible changes occurring between P1 and P2. While species richness and diversity in both old growth and secondary forests remained largely unchanged across the study periods, frog assemblage composition in formerly disturbed sites followed the forest recovery process. Yet, their composition was still distinctly different from old growth forest assemblages. The different levels of resilience over time indicate that the slowest organisms in our system (forest trees) set the pace for the recovery of associated taxa (frogs). Effective post-harvesting restoration schemes need to incorporate these delayed response times. Thus, significantly longer felling cycles than those commonly applied in regular harvesting schemes are needed if the goal is to conserve and restore original forest diversity in logged forest ecosystems.

Introduction

Tropical forests are under unprecedented anthropogenic pressure from agricultural conversion, development of road networks, and direct exploitation due to logging, hunting and the collection of non-timber forest products (Fitzherbert et al., 2008, Geist and Lambin, 2002, Gibbs et al., 2010, Kleinschroth et al., 2016). This results in human-modified landscapes (anthromes sensu Ellis et al., 2010) consisting of an increasing proportion of highly degraded, fragmented or fully cleared forest areas (Lugo and Helmer, 2004, Silvestrini et al., 2012, FAO, 2016, Rozendaal et al., 2019). However, successional recovery following timber extraction or agricultural land abandonment may result in the formation of secondary forests (Aide et al., 2000, Peña-Claros, 2003, Lugo and Helmer, 2004, Silvestrini et al., 2012, FAO, 2016). These forests can play important roles in the provision of ecosystem services and biodiversity conservation (Edwards et al., 2014, Edwards et al., 2017, Chazdon et al., 2016, Rocha et al., 2018).

The study of ecological succession has a long history, particularly in plant ecology (Bazzaz, 1979). More recently it has attracted increased research interest in the context of resilience theory (Gunderson, 2000, Hanna et al., 2019). Resilience, defined as the time required for a disturbed system to recover and return to its pre-disturbance state (sensu Pimm, 1984), has been the focus of a large body of literature, particularly in forest ecology. For example, several studies demonstrated that return and recovery times for species abundance, species richness and biomass in logged forests vary between 40 and 60 years until old growth/mature forest states are reached (Aide et al., 2000, Bonnell et al., 2011, Dent et al., 2013, Rozendaal et al., 2019, Hu et al., 2020). In contrast, plant species composition may take centuries to return to a pristine/primary state (Finegan, 1996, Aide et al., 2000, Bonnell et al., 2011, Xu et al., 2015, Rozendaal et al., 2019). Despite their key roles in tropical forest ecosystems (Muscarella and Fleming, 2007, Avila-Cabadilla et al., 2012), vertebrate communities have only rarely been studied with respect to their recovery potential and resilience following disturbance (Dunn, 2004, Hernández-Ordóñez et al., 2015, Edwards et al., 2017, Sayer et al., 2017, Hilje et al., 2020). Moreover, the few available studies investigating the resilience of vertebrates mainly used a chronosequence approach; i.e. sites that are assumed to have had the same history and setup with respect to both abiotic and biotic components, but differing in time since disturbance (sensu Johnson and Miyanishi, 2008). The use of chronosequences has been criticized mainly on grounds of the underlying assumption of stability, i.e. biotic and abiotic constancy (Johnson and Miyanishi, 2008). Thus, using space for time substitutuons in community ecology may lead to erroneous conclusions, particularly if spatial variation data is exclusively used to study ecological processes in communites. A combination of time series and spatial variation data has therfore been recommended when addressing processes in community ecology (Damgaard, 2019). Unfortunatly, these kind of data, although essential for the study of resilience and recovery processes in floral and faunal communities following anthropogenic disturbance, are scarce.

In this study we use amphibian communities as a vertebrate response model since amphibians react particularly sensitive to environmental change (Stuart et al., 2004). Various studies demonstrated that rainforest amphibians respond negatively to even comparatively low levels of disturbance (Ernst and Rödel, 2005, Ernst et al., 2006, Jongsma et al., 2014, Caballero-Gini et al., 2019, Anunciação et al., 2021) making them an ideal model to investigate the effects of forest disturbance (Hopkins, 2007, Hölting et al., 2016, Brüning et al., 2018). Usually, this is achieved by comparing sites with different impact intensities and/or varying age (space for time approach), thereby assuming that the initial faunal composition was identical in all sites (Ernst et al., 2006, Adum et al., 2013, Jongsma et al., 2014, Konopik et al., 2015). However, several studies have shown that amphibian species composition in tropical forests may vary tremendously over comparatively small geographic scales (Doan and Arriaga, 2002, Ernst and Rödel, 2005, Burivalova et al., 2014). Here we use spatially explicit time-series, spanning a maximum period of 45 years after initial disturbance, to analyze structural changes in a selectively logged West African rainforest and the associated faunal responses.

The combined data sets that we analyzed consist of extensive amphibian assemblage data from Taï National Park (TNP), western Ivory Coast, acquired in standardized assessments in the early 2000s, and reassessed on identical sites using the same standardized methodology, 15 years later (Rödel and Ernst, 2004, Ernst and Rödel, 2005). In their 2005 and 2008 studies, Ernst and Rödel have shown that, 30 years after selective logging, frog assemblages in selectively logged forest in TNP differed significantly from those in unlogged forest. In the present study, we aimed to investigate whether the amphibian assemblages show signs of recovery after an additional 15 years in which no logging occurred.

We therefore compared structural habitat and amphibian richness, compositional and abundance data from two assessment periods: an initial period (P1: 30 years after selective logging) and a re-assessment period (P2: 45 years after selective logging). We hypothesize that: (1) forest structure observed during P1 in previously logged sites gradually recovers toward an unlogged state, whereas forest structure in unlogged sites remains largely unchanged; (2) amphibian species richness, composition and abundance in previously logged forest recovers to unlogged forest states in P2; (3) forest specialists, absent from previously logged sites in P1, recolonize late successional areas in P2; and (4) that recovery of amphibian assemblages follows forest recovery.