Selection of studies identified in the systematic map

Most of the evidence on which this systematic review will be based is included in
the recently completed systematic map of biodiversity impacts of active management
in forest set-asides 19]. Of the 812 studies in the map, 149 reported on how plants or invertebrates were
affected by manipulation of grazing or browsing. Nearly all of these dealt with grazing/browsing
by wild or domesticated ungulates—the few that focused on other herbivores will be
excluded from this review.

The systematic map was based on searches using 13 publication databases, 2 search
engines, 24 specialist websites and 10 literature reviews. The majority of searches
were performed in May–August 2014. In March 2015, a search update was made using Web
of Science and Google Scholar.

Search update

A comprehensive search for additional potentially relevant studies will be made in
the bibliographies of existing reviews of forest grazing/browsing.

Furthermore, in order to identify recently published literature, we will perform an
additional search update, using the following subset of search terms applied for the
systematic map:

Subject:: Forest*, woodland*, “wood* pasture*”, “wood* meadow*”.

Forest type:: Boreal, boreonemoral, hemiboreal, nemoral, temperate, conifer*, deciduous,
broadlea*, “mixed forest”, spruce, “Scots pine”, birch, aspen, beech, “Quercus robur”, Swed*.

Intervention:: Graz*, brows*, fenc*, exclos*.

Outcomes:: *Diversity, species AND (richness OR focal OR target OR keystone OR umbrella
OR red-list* OR threatened OR endangered OR rare), “species density”, “number of species”,
indicator*, abundance, habitat*.

The terms within each category (‘subject’, ‘forest type’, ‘intervention’ and ‘outcomes’)
will be combined using the Boolean operator ‘OR’. The four categories will then be
combined using the Boolean operator ‘AND’. An asterisk (*) is a ‘wildcard’ that represents
any group of characters, including no character.

Searches for peer-reviewed or grey literature published in 2014 or later will be made
in Web of Science and Google Scholar. In the latter case, the first 200 hits (based
on relevance) will be examined for appropriate data. No language or document type
restrictions will be applied.

Article screening and study inclusion criteria

Articles will be evaluated for inclusion at three successive levels. First, they will
be assessed by title. Next, each article found to be potentially relevant on the basis
of title will be judged for inclusion on the basis of abstract. Finally, each article
found to be potentially relevant on the basis of abstract will be judged for inclusion
based on the full text. At all stages of this screening process, the reviewer will
tend towards inclusion in cases of uncertainty. Articles identified by one reviewer
as potentially useful based on full text will be assessed by a second reviewer; and
care will be taken to ensure reviewers do not assess studies authored by themselves.
Final decisions on whether to include doubtful cases will be taken by the review team
as a whole.

A list of studies rejected on the basis of full-text assessment will be provided in
an appendix together with the reasons for exclusion.

In order to be included, each study must pass each of the following criteria (a subset
of those used for the systematic map):

Relevant subjects Forests in the boreal or temperate vegetation zones.

Any habitat with a tree layer is regarded as forest, which means that studies of e.g.
wooded meadows and urban woodlands may be included.

As an approximation of the boreal and temperate vegetation zones we will use the cold
Köppen-Geiger climate zones (the D zones) and some of the temperate ones (Cfb, Cfc
and Csb), as defined by Peel et al. 27]. The other temperate Köppen-Geiger climate zones are often referred to as subtropical
and are therefore considered to fall outside the scope of this review.

Nevertheless, forest stands dominated by ponderosa pine (Pinus ponderosa) will be considered as relevant even if located outside the climate zones mentioned
above. These forests constitute a well-studied North American habitat type that shares
several characteristics with the pine forests in boreal and temperate regions.

Relevant types of intervention Manipulation of ungulate grazing/browsing pressure, e.g. by fencing or by introduction
or culling of ungulates.

Studies of areas where the grazing/browsing pressure varies for reasons other than
direct manipulation (e.g. because of natural differences in accessibility or food
availability) will not be included.

Relevant type of comparator Non-intervention or alternative levels of intervention.

Both temporal and spatial comparisons of how grazing/browsing manipulations affect
biodiversity are considered to be relevant. This means that we will include both before/after
(BA) studies, i.e. comparisons of the same site prior to and following an intervention,
and control/impact (CI) studies, i.e. comparisons of treated and untreated sites (or
sites that had been subject to different kinds of treatment). Studies combining these
types of comparison, i.e. those with a before/after/control/impact (BACI) design,
will also be included.

Relevant types of outcome Abundance, diversity or composition of plants and/or invertebrates (including exotic
species); tree regeneration (seedlings and saplings); performance (e.g. growth, reproduction)
of focal/target species (individual plant or invertebrate species that the intervention
was intended to benefit or control).

Relevant type of study Primary field studies.

Based on this criterion, we will exclude e.g. simulation studies, review papers and
policy discussions.

Language Full text written in English, French, German, Danish, Norwegian, Swedish, Finnish,
Estonian or Russian.

Study quality assessment

Studies that have passed the relevance criteria described above will be subject to
critical appraisal. Based on assessments of their validity, they will be categorised
as having high, medium or low susceptibility to bias.

Studies will be excluded from the review due to high susceptibility to bias (low quality)
if any of the following factors apply:

No true replication.

Methodological description insufficient.

Intervention and comparator sites not well-matched.

Severely confounding factors present.

Outcomes difficult to interpret.

Intervention data difficult to interpret.

Confounding factors may include interventions performed in addition to manipulation
of the grazing pressure. Historically, wood-pastures were often used for multiple
purposes that combined grazing with, for example, mowing, acorn collecting, litter
raking and field crop cultivation 15]. However, present-day reserve management typically requires separate consideration
of each intervention; therefore, studies of such combined activities (even if historically
relevant) will be excluded due to severely confounding factors unless the main effect
of grazing can be distinguished.

Studies that are not excluded due to low quality will be considered to have medium
susceptibility to bias (medium quality) if any of the following factors apply:

Location of study plots potentially biased.

BA study (not CI or BACI).

No data on variance or sample sizes.

No quantitative data on grazing/browsing pressure.

If none of the above factors apply, the study will be considered to have low susceptibility
to bias (high quality).

Detailed reasoning concerning critical appraisal will be recorded in a transparent
manner. In general, the quality of a study will be assessed by one reviewer, but again,
care will be taken to ensure reviewers do not assess studies authored by themselves.
The final ruling on doubtful cases will be made by the review team as a whole. Full
justification for our decisions on susceptibility to bias will be provided for each
study in an appendix to the final report. The significance of quality issues and other
limitations of the available data will be considered in our discussion of review findings.
In particular, we will discuss how our results may have been affected by the exclusion
of studies that are relevant in principle but have high susceptibility to bias, e.g.
due to confounding factors or lack of replication.

Data extraction strategy

Outcome means, measures of variation and uncertainty (standard deviation, standard
error, confidence intervals) and sample sizes will be extracted from tables and graphs,
using image analysis software when necessary. The selection of outcomes to be extracted
from an article will be made by one reviewer. A second reviewer will reassess this
selection and perform the actual data extraction, and the records will be double-checked
by a third reviewer. Data on interventions and other potential effect modifiers will
also be extracted from the included articles.

It may in some cases be useful to ask authors of relevant articles to supply data
in digital format. This will primarily be done where useful data have been published
in graphs from which they are difficult to extract accurately enough, or when it is
known or assumed that considerable amounts of relevant but unpublished data may be
available in addition to the published results. If raw data are provided, summary
statistics will be calculated by us. Extracted data records will be made available
as an additional file.

Potential effect modifiers and reasons for heterogeneity

To the extent that data are available, the following potential effect modifiers will
be considered and recorded for all studies included in this review:

Geographical coordinates.

Altitude.

Climate (and climate change).

Mean age of forest stand.

Dominant tree species.

Forest density (e.g. basal area or overstorey canopy cover).

Primary target vegetation for grazers/browsers.

Type of intervention (exclosures, enclosures, culling etc.)

Duration and seasonality of intervention.

Size of exclosures/enclosures and/or study sites.

Size of sampling plots.

Ungulate grazer/browser species subject to manipulation.

Grazing/browsing pressure (e.g. no. of animals per km
2
).

Other interventions at study sites (harvesting, thinning, understorey removal, mowing,
burning etc.)

Landscape aspects (such as degree of isolation).

History of land use, grazing and protection.

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 will
be performed. In these cases effect sizes will be standardised and weighted appropriately.
Judging from the availability of data, meta-analyses are likely to focus on the species
richness, abundance and performance of tree seedlings, saplings and field-layer vegetation,
and on the species richness and abundance of shrubs and major groups of invertebrates.

Where meta-analysis is possible, it will take the form of random-effects models. Meta-regressions
or subgroup analyses of categories of studies will also be performed where sufficient
studies report common sources of heterogeneity. Special efforts will be made to analyse
how effects depend on the density of grazing/browsing ungulates. Publication-bias
and sensitivity analysis will be carried out where possible. Overall management effects
will be presented visually in plots of mean effect sizes and variance.