Background

While tropical silviculture still remains little-known or poorly acknowledged among decision-makers, there is also some strong antagonism even within the forestry scientific community, namely between forest ecologists and conservation ecologists. The former defend silvicultural interventions as tools to stimulate forest production whereas the latter accept only the most gentle interventions and promote biodiversity conservation through the creation of protected areas even within production forests. One source of confusion and conflict is that depending on how it is carried out, logging can represent either forest exploitation or a major silvicultural intervention designed to maintain forest productivity. From the outset, tropical silviculture faced the problem of reconciling timber production with long-term conservation of forest ecosystems, though the main objective in the early days was indeed to improve timber production [1]. The main challenge lies in finding thresholds of extraction intensity coupled with silvicultural treatments that are compatible with: (1) the maintenance of biodiversity and the main ecosystem services targeted for a given forest management unit, (2) the profitability for the actors involved.

Although it is essential to preserve undisturbed natural forests through the creation of protected areas, these areas alone will not be able to ensure the conservation of all species on a pan-tropical scale, for social, economic and political reasons. According to SEDAC (http://sedac.ciesin.columbia.edu/data/collection/wildareas-v2, cited by Hobbs et al. [2]), 83% of the land surface of the earth is impacted to some degree. Tropical forests are no exception, with (at most) only 1/2 of them being considered primary [3]. The conservation of tropical forests of tomorrow will mostly take place within human-modified (logged, domesticated) forests, but only if they are well managed. “Well managed" means that the implemented forest management must guarantee the perpetuation of timber resources, along with biodiversity and most of the other ecosystem services provided by forests. With advances in the agricultural sciences and increased demand for plantation crops and agricultural commodities (e.g., palm oil and beef), the opportunity costs of forest retention have also increased of late. Counterbalancing that trend is the emergence of markets for the payment of ecosystem services provided by forests which opens up economic incentives to support the implementation of silvicultural treatments.

Currently, about 400 million hectares of tropical moist forests worldwide are designated production forests, about a quarter of which are managed by rural communities and indigenous peoples [3]. There is now clear evidence that, with current logging practices based on a single diameter cutting limit and a rotation cycle of around 30 years, timber volumes will not recover within this rotation cycle [4]. The only way to keep an economically viable commercial volume is to expand the pool of species harvested and/or to reduce the minimum cutting diameter [4-6]. Today this challenge is further exacerbated because most tropical forests have already been logged, and because there is a greater diversity of stakeholders in the forestry sector (farmers, communities, urban society), which was long dominated by large logging companies exploiting forest concessions for timber production. In these forests, the volume of timber extracted over the first cutting cycle was mostly from large trees that grew over several centuries in the absence of strong anthropogenic pressures. When these forests are re-logged, that volume has not been recovered and, in many cases, the young trees expected to make up the next harvest are lacking in numbers or have not yet attained larger sizes as no silvicultural treatments were applied to favour them [7]. Figure 1 illustrates this reduced yield over time in contrast with sustained yield after silvicultural treatments.

Figure 1
figure 1

Expected impact of post-logging silviculture on both sustained timber volume and duration of rotation cycle.

Full size image

This depletion of commercial stocks of timber reduces the financial value of standing forest and puts the forests at risk of being further severely degraded and ultimately converted to other uses. This problem was highlighted in Africa and South-East Asia in the 1960s by foresters who originated from temperate countries and who undertook a large variety of silvicultural trials with the goal of enriching tropical forests with commercial species, by fostering their regeneration either naturally or artificially. Those trials involved contrasting technical operations, from full planting to liberation of naturally regenerated future crop trees by poison-girdling competitors. Those trials generated a considerable amount of data and knowledge that has mostly remained unavailable, published in technical reports and/or non-English language journals, and fully under-valued – or even forgotten – as logging has developed into a vast profitable 'mining' operation in most forests dedicated to production.

Objective of the review

A systematic review of the impact of past silvicultural practices, and not just logging, can inform us about the available opportunities for managing and restoring the economic and environmental values of natural tropical production forests. The main objective of this systematic review is therefore to assess past and current knowledge of the impact of silviculture on the conservation values of tropical forests (its natural capital, including biodiversity, and the full range of ecosystem services that it underpins) and to identify silvicultural approaches that are appropriate for the forests and forestry sectors of the Congo Basin, Amazon Basin, and tropical regions of Southeast Asia (Table 1 gives a full list of countries and regions included in the review). One important goal of this systematic review is to evaluate the compatibility of silviculture for enhanced timber production with the maintenance of other ecosystem services. In so doing, we want to address the following primary and secondary questions:

Table 1 List of terms used to build the search string and define the inclusion perimeter of the papers retrieved
Full size table

Primary review question

How do silvicultural interventions affect timber production and conservation values in natural tropical production forests?

Secondary review questions

  1. 1.

    How do different silvicultural treatments affect both production and conservation values?

  2. 2.

    Where silviculture for sustained timber harvests and maintenance/conservation of other ecosystem services and biodiversity has been practiced, how do differences in these treatments affect socio-economic outcomes?

Methods

This systematic review will undertake an extensive search of literature to assess the impacts of different silvicultural interventions on timber production and the conservation value of natural tropical forests, and to further determine whether these outcomes can both be achieved by the same interventions and how different silvicultural interventions for combined sustained timber harvests and maintenance/conservation of other ecosystem services and biodiversity affect socio-economic outcomes. Data will be extracted for meta-analysis of at least a sub-set of the review questions. Findings are expected to help inform policy and develop evidence-based practice guidelines on the value (or otherwise) of silvicultural practices in tropical natural forests.

Search strategy

Given that the scope of this systematic review is broad and that it can be difficult to identify relevant studies from titles and abstracts, the authors are aware of the necessity of balancing comprehensiveness with specificity in the searches. “Forest management”, for example, is in general a broader concept than the focus of this review, but “silviculture” alone is too narrow. A subgroup of authors tested various search strategies, using combinations of key words and phrases suggested in an initial two-day scoping meeting. These strategies were tested against a reference list of papers (see Additional file 1: Annex1) to refine the search and maintain a balance between, at the one end, extracting an unwieldy set of forestry references from the 1920s onwards which ‘might’ contain a table of useful data, and, at the other end, a narrow set of papers with known useful data, which is not representative of the overall body of evidence. Keywords and phrases for searches in the main bibliographic databases (indicated by an asterisk in Table 2) were derived from the following PICOs listed below. The terms and search logic are shown in Table 1. We recognise the importance of studies published in languages other than English. The search strategy will therefore be applied (and documented carefully for peer review) to the following languages: English, French, Spanish, Portuguese, German, and Dutch, subject to time and resources. Where time and resources permit, and recognizing the potentially high cost of translating relevant studies, we will also search in: Bahasa Indonesia, Japanese, and Thai.

Table 2 Databases and websites
Full size table

PICOs

Population

Closed canopy natural forest used for timber production in the tropics (= natural tropical production forests). (A full list of countries and regions in the tropics is included in Table 1).

Interventions

Any silvicultural activity, including pre-harvest, harvesting (logging), post-harvest treatments, construction of secondary roads and skid trails (A full list of silvicultural activities is included in Table 1).

Comparator

Temporal or spatial comparisons of silvicultural practices.

Outcomes

Increase of timber yield or timber volume recovery, or changes in ecological conditions of the forest or socio-economic outcomes (latter only if studies also reporting timber yield and ecological condition outcomes).

Other

In addition to systematic searching (as above), a call for grey literature will be issued to stakeholders, including through the IUFRO subgroup on tropical silviculture, the International Society of Tropical Forests, LinkedIn and Research Gate and the personal contacts of authors. References in relevant books will also be searched as time and resources permit.

Search terms & Boolean logic

Searches in CAB Abstracts, Web of Science and Scopus will be based on the logic A AND B AND C, where A-C are the sets of terms listed in Table 1. Within each set, alternative terms will be searched, using OR. CAB Abstracts, having a sub-database dedicated to forestry, will be searched using a modified strategy that makes use of CABICODES and descriptors from the CAB Thesaurus (Table 1 – section D). The search was devised in consultation with the Editor of Forestry Abstracts, and the former editor of Forest Products Abstracts (one of the review authors). In our scoping studies a simpler search performed better than the strategy used for Scopus and Web of Science. A much simpler search strategy, based on that devised for CAB Abstracts, but excluding the CABICODES, will be used to search resources that cannot accommodate complex search strings. These are NOT marked with an asterisk in Table 2.

Inclusion criteria

Studies will be included if they are from populations in the target countries or regions (Table 1), and include information on the silvicultural (and other) interventions and outcomes set out in the PICO list above. It is likely that a larger set of relevant Outcomes will be included once reviewers screen the full text of papers. Consequently, a complete list of silvicultural interventions and outcomes will be documented iteratively as these emerge. A subgroup of authors will select papers retrieved from the searches, following removal of duplicates. A preliminary inclusion/exclusion assessment of a randomly selected set of 100 articles will be conducted to test consistency between researchers’ screening judgment. Cohen’s Kappa metrics will be used to indicate a measure of consistency. A consistency co-efficient of 0.6 or greater is usually accepted as sufficient in the literature [8]. The first stage of inclusion/exclusion will entail screening for relevance of articles by title only. This stage will be followed by screening using titles and abstracts together; finally the assessment will be based on reading the full text. At each stage, studies will remain in the set to be included if they MAY contain relevant information about silvicultural interventions or outcomes. This will help reduce the possibility that we are overly-restrictive about relevant outcomes at the outset. Results of screening will be documented (including reasons for exclusion) and will be available as supplementary material to the systematic review paper.

Exclusion criteria

Papers will be excluded if they are exclusively about

  • mangroves or woodlands (e.g. miombo) or pure conifer forests

  • agroforestry systems

  • tree plantations following clear-cutting

Effect modifiers

The following will be recorded, where available, as potential effect modifiers and sources of heterogeneity:

  • Geographical location

  • Altitude

  • Climate

  • Soil type

  • Time after logging or any other silvicultural intervention

  • Dominant tree species

  • Type, duration and intensity of silvicultural activity and logging intensity

  • Time elapsed from silvicultural intervention to sampling

  • Land use and history of land use in the neighbouring landscape

  • Proximity to urban centres (which have different sizes, socio-economic status, etc.)

Critical appraisal of studies

Studies included in the review after full text screening will be critically appraised for study design, the robustness of their study design and the extent to which authors have attempted to limit biases.

Variables outlined in Table 3 will be used to construct a list from which the quality of individual studies can be assessed. Studies will be categorised as having high or low susceptibility to bias on the basis of presence/absence of replication, measures of variance and potential effect modifiers. Studies with high susceptibility to bias will not be excluded from the narrative review, but they will be subjected to sensitivity analysis following Brooks et al. [9]. We are aware of the proposal by Bilotta et al. [10] that environmental systematic reviews should adapt the Environmental- Risk of Bias Tool and we will check the feasibility of this approach for our set of studies. We know that our review will include a number of older studies and we are aware of advances in study design in recent decades, and also of the general problem of pseudo-replication in published studies [11], and presumably also in unpublished studies that may be included in our review, and we will amend our proposed critical appraisal method if (a) it is too difficult to apply by reviewers (requiring subjective judgment that is difficult to standardise across reviewers, despite repeated kappa analysis), and (b) the criteria are too ‘strict’ (resulting in a very small set of studies that are considered to be of sufficiently high quality to be included for quantitative analysis in the review).

Table 3 Criteria for critical appraisal and data extraction from studies selected after full text screening
Full size table

Data synthesis and presentation

A narrative synthesis of data from all studies included in the review will describe the quality of the results along with the study findings. Tables will be produced to summarise these results. Where studies report similar outcomes, meta-analysis may be possible. In these cases effect sizes will be standardised and weighted appropriately. Details of the quantitative analysis will emerge once full-text screening has yielded a set of studies that can be assessed for content and quality.

Subgroup analysis of categories of studies will also be performed where sufficient studies report common sources of heterogeneity. In addition to publication bias (see below), sensitivity analysis will be carried out where possible following Brooks et al. [9]. Silvicultural effects will be presented visually in plots of mean effect sizes and variance. If meta-analysis of effect sizes is possible, it will take the form of random-effects models, and meta-regression will be performed where effect modifiers cause significant heterogeneity between studies.

Publication bias

There is evidence that in many research areas papers are more likely to be published if they demonstrate clear, positive results (or strong negative effects), and that papers that show little or no effect are less likely to be published. To assess the possibility of such publication bias, we will compare studies in the unpublished ‘grey’ literature with studies in academic journals [12] to assess whether there is evidence of publication bias. If data allow, we will assess bias using funnel plots, which show effect sizes and standard error or sample sizes [13].

Endnote

aFor the CAB Abstracts database, searching is limited to Title, Abstract and keyword fields using the field limiters shown in Table 1. Web of Science and Scopus have a default search mode that interrogates only those fields, which will be used. This eliminates papers that contain any of our keywords in, for example, journal titles, institution addresses, etc.

KK110 is one of a suite of CABICODES, which are applied as additional indexing categories by abstractors compiling CAB Abstracts. It relates to Forest management & Silviculture. For a full list see http://www.cabi.org/Uploads/CABI/publishing/training-materials/additional-resources/cabi-codeguide.pdf.