Evaluating partial cutting in broadleaved temperate forest under strong experimental control: Short-term effects on herbaceous plants

Page 1

www.elsevier.com/locate/foreco

Forest Ecology and Management 214 (2005) 124–141

Evaluating partial cutting in broadleaved temperate forest

under strong experimental control: Short-term effects

on herbaceous plants

Frank Gotmark a,*, Heidi Paltto b, Bjorn Norden b, Elin Gotmark c

a Department of Zoology, Goteborg University, Box 463, SE-405 30 Goteborg, Swedenb Botanical Institute, Goteborg University, Box 461, SE-405 30 Goteborg, Sweden

c Department of Mathematics, Goteborg University and Chalmers University of Technology, SE-412 96 Goteborg, Sweden

Received 13 July 2004; received in revised form 19 November 2004; accepted 31 March 2005

Abstract

Partial harvesting of forest for biofuel and other products may be less harmful to biodiversity than clear-cutting, and may even

be beneficial for some species or groups of organisms such as herbs. There are, however, few well-controlled experiments

evaluating positive and negative effects, such as species losses directly after harvest. In closed canopy mixed oak forest in

Sweden, about 25% of the tree basal area and 50–90% of the understory was removed (mainly spruce, birch, aspen, lime, rowan

and hazel). In each of six forests, we studied herbs in an experimental (cutting) plot and a control plot (undisturbed) before, and

in the first summer, after the harvest (conducted in winter). Losses of species were similar in experimental and control plots (15–

16%). The harvest increased species richness by 4–31% (mean 18%); also species diversity (H0) increased. Several ruderals

increased in experimental plots, but most changes occurred in grassland and forest species; partial cutting led to complex, partly

unpredictable early changes in the herb community. A review of early effects of partial cutting (eight experiments) indicated that

it increases herb species richness in stands of broadleaves, but apparently not in conifer stands; there was no evidence that partial

cutting increases species losses. Thus, with respect to early changes after harvest, we found no negative effects of partial cutting

on herbs. We suggest, however, that some proportion of closed-canopy mixed oak forest should not be harvested, to protect rare,

potentially sensitive herbs, and to create stand diversity.

# 2005 Elsevier B.V. All rights reserved.

Keywords: Forest succession; Oak Quercus; Disturbance; Forest management; Biodiversity

* Corresponding author. Tel.: +46 31 7733650;

fax: +46 31 416729.

E-mail address: [email protected] (F. Gotmark).

0378-1127/$ – see front matter # 2005 Elsevier B.V. All rights reserved

doi:10.1016/j.foreco.2005.03.052

1. Introduction

In forestry, management alternatives combining

wood production and conservation of biodiversity are

increasingly considered and tested. Partial cutting of

forest or ‘green tree retention’ where a substantial

.

Page 2

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 125

proportion of the high trees remains after cutting, has

been examined as a potentially sustainable form of

management (Metzger and Schultz, 1984; Reader,

1987; Reader and Bricker, 1992a; Halpern and Spies,

1995; Brunet et al., 1996; Franklin and Kohm, 1997;

Hunter, 1999; Sullivan et al., 2001; Nyland, 2002;

Schumann et al., 2003).

Restoration and protection of broadleaved forest

often take place in areas earlier used for agriculture. In

parts of Europe, oak woodland meadows and pastures

were once common (Rackham, 1980; Vera, 2000;

Hansson, 2001) but many of them were abandoned

during the twentieth century, leading to succession and

invasion of other tree species, and finally to closed

canopy forest. In Sweden, the forestry boards classify

many stands of this type as woodland key habitats

(NBF, 1999a; Gustafsson, 2000; Gotmark and Thorell,

2003), stating that partial cutting may be one

management alternative, in particular for saving old

oaks (Quercus robur and Q. petraea) that may die in

competition with other tree species (NBF, 1999b,

2001). Therefore, wood production might be com-

bined with biodiversity conservation. To evaluate this

possibility, in 2000 we started a long-term partial

cutting experiment in Sweden, studying herbs and

other groups of organisms (see Norden et al., 2004a,b;

Økland et al., 2005).

Harvesting of trees, especially clear-cutting,

changes the species composition of herbs in temperate

forests (Metzger and Schultz, 1984; Schoonmaker and

McKee, 1988; Kirby, 1990; Hannertz and Hanell, 1993;

Bergstedt and Milberg, 2001). Several mid to long term

studies of cutting concluded that most forest herbs are

relatively resistant to changes in tree density, and over

time recover (and even benefit) from disturbance

(McComb and Noble, 1982; Metzger and Schultz,

1984; Halpern and Spies, 1995; Brunet et al., 1996,

1997; Ruben et al., 1999; Bergstedt and Milberg, 2001;

but see Meier et al., 1995). Forest ecosystems are

subjected to dramatic natural disturbances (Lorimer,

2001) and many forest herbs, though not all, may be

adapted to disturbance. Partial harvesting implies that

trees remain after cutting, potentially providing wildlife

habitat (e.g., shelter for forest herbs).

In this study, we focus on the early effects of partial

cutting on herbs. In broadleaved stands with high

conservation values, it is important to examine losses

of species due to cutting (Reader, 1987; Reader and

Bricker, 1992a; Meier et al., 1995). Also, species

richness and diversity might increase immediately

after cutting, and early effects may be crucial in

directing future vegetation change, a possibility which

has rarely been addressed. In experimental work, one

common approach is to compare cut stands and

undisturbed stands after cutting. In the present project,

we established cutting plots that were studied before

and after the harvest, and control plots studied before

and after harvest. Unfortunately, this experimental

design (referred to below as strong control) is rare,

which motivates more work to confirm conclusions

reported in earlier studies. In forest, strong experi-

mental control is essential, because the herb flora is

highly dynamic and subject to temporal changes,

caused by annual differences in precipitation, tem-

perature, and other factors (Brunet and Tyler, 2000;

Tyler, 2001; Tyler et al., 2002). Therefore, a herb

community may change because of cutting, or because

of other factors between years. Below, we illustrate the

benefits of our approach.

We addressed the following questions: (1) Does

partial cutting (harvest) increase losses of species that

initially occurred in the stands? (2) Does species

richness change immediately after partial cutting and

if so, in which direction and to what extent? (3) Which

species are favoured and disfavoured by cutting, and

are the observed responses related to habitat prefer-

ences of the species? In addition, we review published

experimental studies of the early effects of partial

harvesting on herbs in temperate forest (in Section 4).

2. Materials and methods

2.1. Study area, site and plot characteristics

We studied six sites in southern Sweden (Fig. 1)

located in the boreonemoral zone, i.e., between the

boreal forest in northern Europe and the temperate

(nemoral) forest in the middle parts of Europe (Ahti

et al., 1968; Esseen et al., 1997; Nilsson, 1997). About

55% of Sweden is covered by forest (Gustafsson and

Ahlen, 1996). Natural (primary) broadleaved forest

hardly exists in southern Sweden, but semi-natural

stands form about 2–4% of the productive (producing

>1 m3/ha per year) forest. Pasture woodlands with

grazing domestic animals were relatively common

Page 3

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141126

Fig. 1. Study area in southern Sweden, with the six forest sites where cutting experiments were conducted. The sites are located in the

boreonemoral vegetation zone (mixed hardwood zone), south of the boreal forest.

until the early part of the twentieth century, but later

many of them were gradually abandoned. These

relatively open woods often contained oaks, but during

secondary succession other trees invaded (Norway

spruce Picea abies or broadleaved trees). We studied

stands with relatively old oaks (about 80–200 years),

located 5–230 m above sea level. The mean monthly

precipitation (July) decreases from about 80 mm at the

western site (Fig. 1) to about 55 mm at the eastern

coastal sites (www.smhi.se). The mean temperature in

July varies from about 14 8C in the west to about 17 8Cin the east.

The study sites were nature reserves (Rya asar,

Lindo) or woodland key habitats (Norra Vi, Ulvsdal,

Ytterhult, Farbo), obtained through authorities and

forest owners. We selected stands at the sites with

almost closed canopy, mesic moraine soil, and

relatively level and usually a bit stony surface. At

the time of the survey, usually at least 70% of the

ground was covered with herbs. Species recorded at

five or six sites were the ones that dominated (see

Appendix A). In each stand, we delimited two plots

(each 1 ha); one experimental and one control plot.

The plots were 100 m � 100 m, except at Rya (two

plots of 83.5 m � 120 m) and at Ytterhult, where the

experimental plot (83.5 m � 120 m) was combined

with five smaller control plots (sum 1 ha), due to

patchiness of the oak forest. The mean distance

between experimental and control plot was 95 m

(range 40–250 m, n = 6 sites). Experimental and

control plots were selected to be as similar as possible

with respect to forest habitat. After plots were

established, we selected experimental (cutting) plot

randomly from each pair of plots.

Canopy closure was measured in each plot from

eight photographs taken with a digital camera (28 mm

lens) from ground level towards the sky, near transects

where herbs were studied (see below). We converted

colour pixels to binary black-and-white pixels using

the program NIH Image, and calculated the mean

proportion of sky visible for each plot (Table 1). Tree

basal area was measured for stems >5 cm diameter

(dbh, at 1.3 m). Oaks made up most of the basal area

(Table 1) and were on average larger than the other

trees, so in terms of stems they were in minority.

Norway spruce made up >90% of the conifer basal

area, and other broadleaved trees (Table 1) were

mainly (in this order) birches Betula pubescens and B.

pendula, aspen Populus tremula, lime Tilia cordata,

rowan Sorbus intermedia, and hazel Corylus avellana

(a large bush). To characterize soil conditions, we took

eight samples from the topsoil (0–5 cm depth, litter

removed, each sample 150 cm3) in each plot along

transects (see below), pooling the eight samples to

one, and analysing pH (H2O), total-C (%), and total-N

(%). All samples were treated in the same way during

collection, preparation and analyses (samples were

frozen for six months before analyses).

Page 4

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 127

Table 1

The characteristics of experimental (Exp) and control (Con) plots at the six study sites in Sweden (Fig. 1) before experimental cutting (harvest)

Study site Light conditions

(% sky visible

from ground)

Basal area at 1.3 m

(m2 per ha, for

trees �5 cm dbh)

Oaksa (% of

basal area)

Conifersa (% of

basal area)

Other broadleaveda

(% of basal area)

Soil pH (H2O) Soil C/N

Rya

Exp 18 27.8 78 4 18 4.56 15.7

Con 18 30.0 54 9 37 5.01 14.9

Norra Vi

Exp 11 34.3 67 11 22 5.47 15.5

Con 14 25.1 57 21 22 5.67 14.3

Ulvsdal

Exp 19 26.6 51 7 42 5.65 15.9

Con 15 24.7 37 32 31 5.91 14.3

Ytterhult

Exp 23 23.8 53 31 16 5.20 15.5

Con 12 32.5 67 4 29 5.92 12.0

Farbo

Exp 18 31.3 61 34 5 5.61 16.0

Con 11 31.0 44 26 30 5.66 15.1

Lindo

Exp 10 25.7 41 4 55 5.48 17.8

Con 11 17.4 42 21 37 5.06 15.0

Mean (S.D.)

Exp 16.5 (5.0) 28.2 (3.9) 58.5 (13.0) 15.2 (13.7) 26.3 (18.5) 5.33 (0.41) 16.1 (0.9)

Con 13.5 (2.7) 26.8 (5.6) 50.2 (11.1) 18.8 (10.5) 31.0 (5.6) 5.54 (0.41) 14.3 (1.1)a Oaks were Quercus robur (mainly) and Q. petraea; conifers were Picea abies (mainly) and Pinus sylvestris; other broadleaved consisted

mainly of (in this order) Betula pubescens/pendula, Populus tremula, Tilia cordata, and Sorbus intermedia.

The experimental (n = 6) and control plots (n = 6)

did not differ with respect to plot characteristics (mean

values in Table 1; 0.31 > P > 0.10, for test see below),

except in C/N ratio (P = 0.01). However, the mean

difference in C/N ratio between plot types was small

(11%; Table 1), and pH is probably the most important

soil factor for the flora (Diekmann, 1994).

2.2. Experimental design and procedures

Our overall objective is to examine biodiversity of

closed (presently undisturbed) stands and partially

harvested stands, without prejudice with respect to

‘best’ option for management. One common goal for

stands containing large oaks (NBF, 1999b) is to favour

these trees. Therefore, careful cutting (thinning) was

conducted around them, providing space and light.

There were few oak saplings, so an additional goal was

to favour oak recruitment by cutting conifers (almost

all) and other broadleaved trees of intermediate size

(also some oaks of intermediate size). Old (large)

individuals of other broadleaved trees were retained.

Although more trees were cut near large oaks, the cut

trees were distributed fairly evenly across plots.

Understory trees (0.5–5 cm dbh) were not measured or

marked; about 50–90% of them were cut and

harvested, a higher proportion if there were many

stems (this wood fraction is increasingly used for

biofuel). Tops and branches of larger trees were

usually left in the plots, to create some deadwood. All

trees (>5 cm dbh) that were to be cut (harvested) were

marked in the summer 2002 by one of us (F.G.).

Protocols and detailed instructions were then sent

landowners and forestry entrepreneurs, who cut and

harvested the six plots October 2002–March 2003.

Cutting was done manually, except at Farbo where

also a harvester was used for spruce cutting. At all

sites, cut stems were taken from the plots by

Page 5

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141128

Fig. 2. Study plot and sampling design (two plots per site, one

experimental and one control plot) with transects (2), transect

sections (8) and squares (8). Squares were not used at two sites,

Norra Vi and Ulvsdal.

forwarders, relatively heavy machines that created

tracks in the ground where the vegetation was

disturbed, and soil often exposed. The length of these

tracks was about 300 m per experimental plot (not

measured).

A relatively small proportion of the basal area

(stems �5 cm dbh) was harvested. In forest of this

type, landowners often cut selectively few trees at a

time (cf. Kittredge et al., 2003), for fuel or other

purposes. Also, our stands had high conservation

values and small harvest was in line with precau-

tionary principles. The proportion of basal area

(�5 cm dbh) harvested was on average 24%

(Table 2); as 50–90% of the thin stems (<5 cm

dbh) also were removed, the true value is about 25–

30%. Based on photos (same positions before-after

harvest), the proportion of visible sky more than

doubled after harvest (Table 2). The increase in light

was relatively patchy, and higher, e.g., around old oaks

or where spruces were cut. Harvesting of the

undergrowth also contributed to increased light levels

for herbs.

2.3. Sampling and analyses

Before cutting, field work was conducted 14–22

May (spring flora) and 16–31 July (summer flora),

either 2001 (three sites) or 2002 (three sites). After

cutting, we conducted field work 14–19 July 2003 (6

sites, regarding May, see below). Conducting surveys

in August would increase the risk that some herbs

wither and are overlooked (pers. obs.). At each site,

both plots were surveyed on the same day. In each plot,

we recorded species along two 100 m transects,

Table 2

Changes in experimental plots as a result of tree harvest in the winter of

Study site % of basal area removed (stems �5 cm dbh)

Rya 15.6

Norra Vi 21.6

Ulvsdal 21.6

Ytterhult 31.6

Farbo 37.4

Lindo 15.2

Mean (S.D.) 23.8 (8.9)a 2001 and 2003 refers to photographic measurements from before (July

level.

usually separated by 20 m and stretching from edge to

edge of the plot (Fig. 2). The two transects were

located to cover central, representative parts of the

plots. Along transects, we stretched a measuring tape

between permanent poles, recording all species within

a metre of the tape on one side (each transect area was

thus 100 m2). Each transect was divided into four 25 m

long sections (25 m2), which were sampling units in

most analyses based on plots (eight sections per plot;

Fig. 2). We walked very slowly along each transect

section, recording all encountered species (presence

only). Difficult specimens were collected and later

identified by specialists. We did not identify (record)

seedlings smaller than about 4–5 mm, which some-

times grew on disturbed ground; therefore, species

2002/2003

% of sky visible through canopya

2001 2003 Increase (%)

18 38 111

11 29 163

19 33 74

23 46 100

18 36 100

10 29 190

16.5 (5.0) 35.2 (6.4) 123 (44)

–August, 2001) and after (July–August 2003) harvest, from ground

Page 6

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 129

richness in experimental plots is probably slightly

underestimated. Larger seedlings that only could be

identified to genus were included in the analyses

(denoted ‘sp’.).

To study the response to cutting especially in

common species, we used a frame covering 1 m2

(1 m � 1 m). We randomly selected a square (1 m2) in

each 25 m-section (Fig. 2), placing the frame over it,

next to the measuring tape. The frame was subdivided

into 100 quadrats (10 cm � 10 cm) by thin strings and

carefully placed over the vegetation. Presence of each

species (visible above-ground parts) was counted and

a value from 1 to 100 obtained (tests indicated that a

frame with 25 subdivisions, 20 cm � 20 cm, would be

much less likely to detect differences in species

frequency). In total, we had eight permanently marked

squares per plot, studied before and after cutting at

four sites (no squares at Ulvsdal and Norra Vi, Fig. 1).

The metal markers for squares were hidden (under-

ground) before cutting, otherwise vegetation in the

squares might not be treated as vegetation elsewhere in

the plot during cutting (also, metal markers may

destroy tires of forwarders).

We sampled herbs in spring (May) before cutting,

but nearly all species could be examined in July and

we did not repeat spring sampling in 2003. Four spring

herbs that wither in early summer (May–June) were

excluded: Lathraea squamaria (recorded at one of six

sites 2001/2001), Gagea lutea (two of six sites),

Corydalis intermedia (two of six sites), and Ranun-

Table 3

Species richness (number of recorded species in transect sections (n)) in ex

species richness measured in number of species and in %

Study site Experimental plot

Mean (S.D.), no. of species Mean (S.D.), chang

Before

harvest

n After

harvest

n In species In %

Rya asar 12.8 (3.3) 10 14.0 (2.6) 10 +1.2 (1.6) +12 (1

Norra Vi 21.1 (5.3) 8 24.0 (5.0) 8 +2.9 (4.0) +17 (2

Ulvsdal 27.0 (2.7) 8 27.9 (2.2) 8 +0.9 (1.9) +3.6 (

Ytterhult 23.8 (3.5) 6 26.3 (1.5) 6 +5.8a (7.9) +25 (3

Farbo 23.0 (4.0) 8 29.4 (3.7) 8 +6.4 (5.0) +31 (2

Lindo 15.6 (4.8) 8 18.6 (5.7) 8 +3.0 (3.3) +21 (2

All sites (n = 6) 20.6 (5.3) 23.4 (5.9) +3.4 (2.3) +18.3a Change measured for each section (after minus before), therefore not

Variation in number of sections sampled was due to differences in shape an

the other sites).

culus ficaria (two of six sites). From square (1 m2)

samples (but not transect sections) we also excluded

Anemone nemorosa, a very abundant spring-flowering

herb that withers in June–July. In total, 158 species

(taxa) are included in the analyses (see Appendix A).

We classified these species by (major) habitat type in

which they occurred in the landscape (Appendix A)

using own field experience, Krok and Almquist (1984)

and Mossberg (1992). In the analyses below, we used

only three groups: (1) forest species, (2) grassland

species (including species growing mostly in open

habitat, see Appendix A), and (3) ruderals (on highly

disturbed ground). Species occurring mostly in (small)

forest openings and at forest edges were classified as

forest species, as natural forests commonly have open

spaces. Our classification of the species is partly

subjective, but indicates changes in species composi-

tion. The nomenclature follows Karlsson (1998,

updated at www.nrm.se).

In statistical analyses of changes in species

richness, for each site we examined the effect of

cutting by calculating change in the number of

recorded species in each transect section (number of

species after cutting minus number of species before

cutting). Repeated measurements in sections would

not be independent, but here we use change as test

variable, comparing changes in experimental and

control plot. For each site, the null hypothesis of no

difference in change in species number in sections in

the experimental plot (n = 8 at most sites, Table 3) and

perimental and control plots before and after harvest, and changes in

Control plot

ea Mean (S.D.), no. of species Mean (S.D.), changea

Before

harvest

n After

harvest

n In species In %

7) 17.0 (5.0) 10 17.5 (5.0) 10 +0.5 (1.5) +3.9 (8)

2) 21.1 (3.2) 8 21.1 (3.8) 8 0 (3.3) +3.9 (16)

6.9) 27.9 (5.7) 8 26.0 (4.6) 8 �1.9 (2.8) �5.6 (10)

0) 23.2 (8.5) 5 23.8 (10) 5 +0.6 (6.2) +4.3 (25)

6) 18.5 (2.0) 8 18.4 (3.0) 8 �0.1 (1.4) �1.2 (8)

6) 13.2 (3.5) 8 13.2 (3.5) 8 0 (1.8) +2.1 (13)

(9.7) 20.2 (5.1) 20.0 (4.6) �0.2 (0.9) +1.2 (3.9)

always congruent with difference in no. of species (before–after).

d set-up of plots; at Rya asar, each section was about 21 m2 (25 m2 at

Page 7

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141130

in the control plot (n = 8 at most sites) was tested by

permutation test (difference between two means,

15 000 resamplings; Resampling Stats and Inc., 2000).

Such tests are as powerful as parametric tests and have

few requirements regarding data, solving the problems

of skewed distribution and variance. In some analyses,

we use squares (1 m2) or sites (i.e., plots, n = 6 + 6) as

sample units in tests. Unless otherwise stated, P-

values are from resampling tests, two-tailed. For

analyses of species diversity (H0, Shannon index) and

changes in this index, the same test was used. Species

frequencies in transect sections and squares were used

to calculate H0 ¼ �P

ð pi � ln piÞ, where p is relative

frequency of species i. To illustrate changes in species

communities due to cutting, we used correspondence

analysis (CANOCO 4.5; ter Braak and Smilauer,

2002). The data consisted of 158 species as rows and

24 surveys as columns (six sites, two plots, visited

twice) and frequencies expressed as % transect

sections where species was recorded in each plot.

The output showed ‘‘arch’’ and ‘‘edge effects’’, so CA

was replaced with DCA (detrended correspondence

analysis; ter Braak and Smilauer, 2002). We used the

default options in the program.

Of the 158 species, only 16 occurred at all sites, and

as many as 88 occurred at one or two sites only

(Appendix A). This, in combination with high

patchiness of forest herbs in samples, meant that

changes in frequency of individual species could not

be analysed statistically. However, our experimental

design implies that estimates presented for individual

species are valuable (see below). We are not aware of

community-wide estimates of population changes due

to cutting for herbs, based on strong experimental

control. To analyse change in species frequencies, we

first took into account potential recording error. As

observers probably did not detect all species in

transect sections, increase/decrease of a species by one

section (1/8; 12.5%) was disregarded; change by at

least two transect sections was set up as criterion. For

each site, we quantified species changes due to cutting

by comparing experimental and control plot. For

instance, if a species increased by 75% in the

experimental plot, and by 25% in the control, it was

listed as increasing by 50% due to cutting (75 minus

25). Our experimental design also identified (say)

positive effects of cutting for species that did not

increase in the experimental plot, but decreased by

25% or more in the control between seasons (for other

reasons). Moreover, species that decreased in the

experimental plot, but decreased more (�25% units)

in the control were identified as favoured by partial

cutting. For species that colonized or were lost in

experimental plot, and did not occur in control plot,

the experimental plot was the sole basis for inclusion/

listing in Table 4. Our analysis of colonizing and lost

species (see below) included also species with change

in a single transect section.

We used squares to analyse responses in 31

common species (at 4 sites), comparing changes in

experimental and control plot in a similar way. To be

reported in this analysis of change, a species or

population had to occur in both experimental and

control plot, and in sufficient frequencies [in general,

more than a sum of 10 (dm2) in each plot, before-

after]. This meant that most species (17 of 31) are

reported for one site only (10 for two, 4 for three sites).

For each plot and species/population, frequency

values from the eight squares were added together

for a single number (for before and after cutting). The

response of a species/population was classified as (1)

none or slight, (2) favoured by cutting, or (3)

disfavoured by cutting. This was partly a subjective

classification, taking into account potential error and

accuracy in recording of different species in the frame.

3. Results

3.1. Loss of species, changes in species richness,

and species diversity (H0)

We considered a species to be lost when it was not

re-observed in any of the eight sections in a plot after

cutting, and used plots as units for this statistical

analysis. There was no indication that a higher

proportion of species was lost from experimental

plots (mean 16.0%, S.D. 5.7%, n = 6) than from

control plots (mean 15.0%, S.D. 2.2%; n = 6,

P = 0.34). Measured in number of species, the mean

values for experimental and control plots were 7.3 and

5.8 species, respectively (P = 0.14). For the four sites

with squares, we pooled the eight squares for each

plot, examining species losses after cutting in the same

way. Species losses in squares did not differ between

experimental plots (mean 14.8%, S.D. 7.6%, n = 4)

Page 8

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 131

Table 4

Partial cutting in experimental plots: species with different habitat preferences that were favoured/disfavoured; effect (change in proportion of

transect sections, or general response for common species in squares) established by matched comparison of experimental plot-control plot as

described in Section 2 (see also footnote below)

Main habitat and response Transect sectionsa Squaresa

Site (6 sites;

F, L, N, R, U, Y)

Change (mean if �2 sites,

and range)

Site and changes

(0 = none or slight; +; or �)

(4 sites; F, L, R, Y)

Ruderal

Favoured

Convolvulus arvensis Y +67%

Senecio sylvaticus F, L, N, U, Y +52% (33–75)

Poa annua F +50%

Cirsium sp. F +38%

Rubus idaeus F, N, Y +32% (25–47)

Epilobium montanum U +25%

Galeopsis bifida F +25%

Juncus bufonius F +25%

Mean change (n = 8 species) +39% (S.D. �16)

Grassland

Favoured

Stellaria graminea Y +67%

Agrostis capillaries (�) F, L, U +54% (38–62)

Polygonatum odoratum Y +50%

Succisa pratensis N +50%

Dactylis glomerata (�) L +38%

Solidaga virgaurea (�) N +38%

Ajuga pyramidalis F, L, Y +36% (25–50) Y: 0

Saxifraga granulata Y +33%

Potentilla erecta (�) Y +33%

Veronica officinalis F, L +32% (25–38)

Anthriscus sylvestris Y +30%

Festuca rubra Y +30%

Anthoxantum odoratum L, N, U +29% (25–38) Y: 0

Geum urbanum/rivale F, L, U +29% (25–38)

Carex pallescens N +25%

Fragaria vesca (�) U +25% F: 0, Y: +

Geranium robertianum (�) F, N +25% (25–25) Y: +

Hieracium sect. Vulgata (�) F +25%

Lathyrus pratensis F, N +25% (25–25)

Scorzonera humilis N +25%

Trifolium repens F +25%

Veronica chamaedrys L +25% F: 0, L: �, Y: 0

Campanula persicifolia Y: +

Mean change (n = 22) +34% (S.D. �11)

Disfavoured

Dactylis glomerata (+) Y �60%

Laserpitium latifolium Y �40%

Festuca ovina L �38%

Deschampsia cespitosa U �38%

Geranium robertianum (+) Y �37%

Hylotelephium telephium Y �37%

Allium oleraceum Y �33%

Luzula multiflora Y �33%

Page 9

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141132

Table 4 (Continued )

Main habitat and response Transect sectionsa Squaresa

Site (6 sites;

F, L, N, R, U, Y)

Change (mean if �2 sites,

and range)

Site and changes

(0 = none or slight; +; or �)

(4 sites; F, L, R, Y)

Deschampsia flexuosa U, Y �29% (25–33) R: 0, Y: 0

Fragaria vesca (+) Y �27%

Galium uliginosum N �25%

Agrostis capillaries (+) N �25%

Hieracium sect. Vulgata (+) N �25%

Poa compressa L �25%

Potentilla erecta (+) U �25%

Solidaga virgaurea (+) F, U �25% (25–25)

Vicia sepium F, U �25% (25–25) F: �, Y: 0

Mean change (n = 17) �32% (S.D. �9)

Forest

Favoured

Lathyrus linifolius F +62% Y: �Moehringia trinervia F +62% Y: 0

Trifolium medium F +62%

Pulmonaria obscura (�) F +50%

Melampyrum sylvaticum (�) U, Y +46% (30–62) Y: +

M. pratense N +38%

Melica nutans (�) U +38%

Vicia sylvatica F, U +38% (38–38)

Viola riviniana (�) N +38% F: 0, L: +, Y: 0

Luzula pilosa (�) F, N, Y +30% (25–38) F: 0, R: +, Y: 0

Milium effusum (�) U, Y +29% (25–33) F: 0

Dryopteris filix-mas Y +27%

Carex digitata F, N +25% (25–25) F: 0

Hieracium sect. Hieracium (�) U +25% F: +, Y: �Mycelis muralis (�) F, N +25% (25–25)

Poa nemoralis (�) L +25%

Rubus saxatilis (�) U +25% Y: 0

Melica uniflora F: 0, Y: +

Trientalis europaea R: +

Stellaria holostea L: +

Mean change (n = 17) +38% (S.D. �14)

Disfavoured

Poa nemoralis (+) F �62% F: 0, Y: �Cardamine bulbifera F �50%

Pulmonaria obscura (+) Y �43%

Elymus caninus Y �40%

Laserpitium latifolium Y �40%

Melampyrum sylvaticum (+) F �38%

Melica nutans (+) F �38%

Viola riviniana (+) F, Y �36% (25–47)

Primula veris F, Y �34% (30–38)

Convallaria majalis R �30% R: �, Y: 0

Hieracium sect. Hieracium (+) R �30%

Anemone hepatica N �25% F:�, Y: 0

Calamagrostis arundinacea F �25%

Dryopteris carthusiana L �25%

Equisetum sylvaticum F �25%

Lathyrus vernus N �25% F: 0, Y: 0

Page 10

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 133

Table 4 (Continued )

Main habitat and response Transect sectionsa Squaresa

Site (6 sites;

F, L, N, R, U, Y)

Change (mean if �2 sites,

and range)

Site and changes

(0 = none or slight; +; or �)

(4 sites; F, L, R, Y)

Luzula pilosa (+) U �25%

Milium effusum (+) N �25%

Mycelis muralis (+) U �25%

Paris quadrifolia N �25%

Rubus saxatilis (+) F �25%

Mean change (n = 21) �33% (S.D. �10)

Mercurialis perennis L: �Oxalis acetocella F: +, L: �, R: �Pteridium aquilinum R: �Vaccinium myrtillus F: 0, R: 0, Y: �Gymnocarpium dryopteris R: 0

Maianthemum bifolium R: 0

Melampyrum pratense Y: 0a + and � in habitat group refers to species that were favoured and disfavoured (in different sites/plots); they are therefore listed twice. For

squares (pooled), only 31 common species were analysed (relatively abundant in both experimental and control sites). Sites: F = Farbo;

L = Lindo; N = Norra Vi; R = Rya asar; U = Ulvsdal; Y = Ytterhult. Numbers of transect sections in each plot: F = 8; L = 8; N = 8; R = 10;

U = 8; Y = 5 (control) and 6 (experimental).

and control plots (mean 21.4%, S.D. 8.8%; n = 4,

P = 0.14).

At all six sites, the number of recorded species in the

experimental plot increased after the partial harvest

(Table 3); the increase in species number ranged from

4% (Ulvsdal) to 31% (Farbo). Species richness in the

six control plots did not change much (slight decrease in

two, slight increase in two). Tests based on changes in

transect sections in the plots revealed significant

increases in species richness at three sites; Ulvsdal

(P = 0.024), Farbo (P = 0.001) and Lindo (P = 0.026).

We found tendencies in the same direction for Rya asar

(P = 0.14) and Norra Vi (P = 0.076), and for Ytterhult

(P = 0.077) where a smaller sample size might explain

lack of significance (see Table 3). For Ulvsdal, with

lowest increase in species number after harvest

(Table 3), the significant effect was due to a decrease

in species number in the control. Thus, in a single forest,

a cutting experiment lacking strong control may not

detect changes in species richness.

Using plots instead of sections as sample units, and

testing mean plot (mean transect section) change in

species number (Table 3), experimental (n = 6) versus

control plots (n = 6), we found a highly significant

difference (P = 0.001) indicating that partial cutting

increases species richness of herbs in this forest type

already in the first season after harvest.

For transect data pooled to plot level, species

diversity (H0) in experimental and control plots did not

differ before cutting (P = 0.32, n = 12). After cutting

in experimental plots, in contrast to controls, diversity

(H0) increased, and the change in H0 (difference after-

before) differed significantly between plot types

(Fig. 3, P = 0.012, n = 12). For squares in experi-

mental plots, mean H0 increased slightly (1.48 before,

and 1.59 after cutting, n = 4); for control plots, the

means were similar (1.42 before, and 1.43 after

cutting, n = 4). Given non-significant tendency

(P = 0.18) and smaller sample size (n = 8), we

analysed changes at the site level with squares as

sample units (n = 6–8 per plot). For three of the four

sites, change in H0 did not differ between experimental

and control squares (0.25 < P < 0.46); Ytterhult

approached significance (P = 0.051) due to a drop

in H0 in control squares. Thus, cutting generally

increased H0 for transects, but not for squares.

3.2. Gains and losses of species

For experimental plots, we listed species that were

not observed before cutting and number of plots/sites

(1–6) they colonized. We recorded in total 69

colonizing species in 2003; 20 (29%) were ruderals,

13 (19%) were forest species, and 36 (52%) grassland

Page 11

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141134

Fig. 3. Species diversity (H0, Shannon index) in experimental and control plots at the six study sites before and after harvest (controls

undisturbed), based on transect data pooled for plots.

species. Sixteen species colonized two or more

experimental plots but only one colonized all six

plots, the wind dispersed Senecio sylvaticus, common

on clear-cuts in the region. Six of the 16 species were

ruderals, two were forest species, and eight grassland

species. Among 53 species colonizing one experi-

mental plot only, 14 (26%) were ruderals, 11 (21%)

were forest species, and 28 (53%) grassland species.

As many as 35 of the 69 species were not recorded in

any of the six experimental plots before cutting.

For control plots, we recorded in total 27

colonizing species in 2003; one (4%) was ruderal,

seven (26%) were forest species, and as many as 19

(70%) grassland species. Eleven species had not

earlier been recorded in any control plot, indicating

natural turnover in non-harvested forest of this type.

Based on changes in control plots, an estimate of the

number of species colonizing transect sections in

experimental plots due harvest disturbance is 42 (69

minus 27), or 29% of all 147 taxa recorded there.

Eighteen of the 27 species colonizing controls also

colonized one or more experimental plots.

With respect to lost species (not resighted), we

found no differences between experimental and

control plots. In experimental plots, we recorded 34

species that were lost from two (6 species) or one plot

(28). However, 12 of these were also recorded as

colonizing one or more experimental plots and

therefore were not lost. Three additional species were

lost also from control plots; these 15 (12 + 3) species

were excluded, as they were not lost due to cutting. Of

the remaining 19 species, one (5%) was ruderal, 11

(58%) were grassland species and seven (37%) forest

species. In control plots, 30 species were lost from two

(3 species) or one plot (27). Four of them also

colonized a control plot, and were omitted. Of the

remaining 26 species, one was ruderal (4%), 16 (61%)

were grassland species, and nine (35%) forest species.

Less common forest species in the study area, of

higher interest for conservation work, were lost (i.e.,

were not re-observed) from experimental plots (e.g.,

Viola mirabilis, Carex vaginata and C. divulsa) as well

as from control plots (e.g., Pyrola minor, Actea

spicata, and Bromopsis benekenii/ramosa).

3.3. Increasing (favoured) and decreasing

(disfavoured) species and populations

This analysis revealed a complex response in the

herbs to partial cutting (in the first season). In transect

sections, eight species of ruderals were favoured,

while in the grassland species group, we found 22

favoured and 17 disfavoured species (Table 4). In the

forest species group, 17 species were favoured and 21

Page 12

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 135

disfavoured (Table 4). Moreover, 17 species were both

favoured and disfavoured, in different plots; 13 of

them belonged to the forest species (see plus and

minus sign after Latin name in Table 4). When these

changes were compared with those in squares for the

31 common species (Table 4) we found further

variation and discrepancies, possibly related to scale

(squares covered a small area).

Since species richness increased in experimental

plots, and most of the ground was covered by herbs

before cutting, we expected reduced frequencies of the

31 common species in experimental but not in control

plots (resulting in space for new/increasing species).

We analysed changes at the population level (i.e.,

when a species occurred in two experimental plots, it

had two populations), measuring frequency as number

of quadrats (dm2 as unit) in the squares, and pooling

these frequencies for each plot. For 50 species

populations of the 31 common species, the mean

frequency per population before cutting was 92

quadrats for experimental and 69 for control plots.

After cutting, populations in experimental plots

decreased in frequency by on average 10 quadrats

(S.D. 38, n = 50), while those in controls increased by

Fig. 4. Ordination (DCA) of species composition, and change in

species composition, in experimental (exp) and control (con) plots,

before and after partial cutting. Study sites: RYA = Rya asar,

NOR = Norra Vi, ULV = Ulvsdal, FAR = Farbo, YTT = Ytterhult,

LIN = Lindo. Temporal change in community composition for each

plot (from before, to after partial cutting) is indicated by arrows. The

length of the first (x) axis in the DCA is 2.55, and the second (y) axis

1.85; eigenvalues for the first axis is 0.34 and the second 0.18 (total

inertia 1.78). The first (x) axis explained 19.1% of the variation in the

species data, the second (y) axis explained 10.4%.

on average four quadrats (S.D. 39, n = 50), a

significant difference, though weak (P = 0.05).

3.4. Ordination

The ordination of species communities and of

changes in the plots (Fig. 4) showed that the more

isolated sites in the west (Rya asar) and south (Lindo)

had positions that deviated from the other four sites in

northeast, which apparently was a geographical effect

(cf. Fig. 1). For two sites with marked changes in

species richness in the experimental plot (Farbo,

Lindo), the ordination also indicated larger change in

experimental than control plot. Also, the ordination

detected (verified) larger change in control than

experimental plot at Ulvsdal, an effect that produced

significant change in species richness there (see

above). Moreover, Rya asar showed least change in

both species richness and species composition (in

ordination). The direction of changes in the plots

varied; at least three directions were represented in

experimental plots (Fig. 4), suggesting complexity in

floristic change. For five of the six sites, the direction

of change differed between the experimental and the

control plot, suggesting that partial cutting influenced

species composition.

4. Discussion

Partial cutting in the six stands did not lead to

higher losses of herb species in experimental than in

control plots, and we found no evidence that species of

interest for conservation were lost at a higher rate in

experimental than in control plots. These results are

similar to, and confirm those reported by Reader

(1987) in one of the few earlier studies using strong

experimental control. Our results also suggest

between-season species turnover in undisturbed mixed

oak forest, and that the herb flora is highly dynamic in

the short term (cf. Brunet and Tyler, 2000; Tyler, 2001;

Tyler et al., 2002). However, species that were lost

might in most cases survive locally (for instance,

above-ground parts of some plant species may have

been reduced in 2003). Partial cutting increased

species richness of herbs in the first summer.

Colonizing and favoured species were of several

types; as expected, new ruderals were recorded, but

Page 13

F.

Gotm

ark

eta

l./Fo

restE

colo

gy

an

dM

an

ag

emen

t2

14

(20

05

)1

24

–1

41

13

6

Table 5

Experimental studies of the immediate effects of partial harvesting on herbs in temperate forest; effects analysed by means of control plots 1–1.5 years after cutting/harvest (including

2nd summer if cutting done late in spring/early summer)

Forest and stand Location # Sites Treatment

or % cut

Month of cutting Type of

controlaSpecies

richnessa

Species initially

present lostaSpecies

diversity

indexa

Changes species/

species groupsa

Other

aspectsa

Mainly coniferous (mesic) forest

(1) Douglas fir

and pine

Canada 5 Patch cut

(30–40%)

February–March,

June–July

Space NS Discussed NS Discussed Herb structural

diversity lowest

in controls

(2) Norway spruce Sweden 1 Shelterwood

(80%)

May Time �14–22%

(NU)

None No change

(NU)

�, minor

(3) Pine and oak USA 1 25% December Space NS Not studied NS �; less than in

study (5)

Physiographic

effects

Mainly broadleaved (mesic) forest

(4) Maple mainly USA 1 20–58% ? Space

(+time)

Increase

(NU)

? Increase

(NU)

Discussed

(5) Mixed

bottomland

USA 1 61% February Space NS, but total

richness

increased

Not studied NS Marked changes Physiographic

effects

(6) Oak/hazel UK 1 ‘‘Coppice and

group fell’’

January–March Space +60% (NU) Not studied Not studied +, light demanding,

+, nitrogen demand,

+, ruderals (NU)

Vernal species

unaffected (NU)

(7) Oak mainly Canada 2 33%, 66% November–April Time +

space

Not studied NS Not studied �, forest herbs,

see (9)

(8) Oak mainly Sweden 6 25% November–April Time +

space

+18% (SS) NS +19% (SS) +, ruderals, �,

many species

Heterogeneity

promoteda Control is undistrubed forest; control ‘in time’ = plots studied before treatment; ‘in space’ = controls after treatment; SS = statistically significant; NS = no statistically significant

difference(s), NU = not used statistical tests; + : increase/higher for experimental plot; –: decrease/lower for experimental plot. References: (1) Sullivan et al. (2001) (also data for

seed-tree, and clear-cut), (2) Hannertz and Hanell (1993) (also data for clear-cut), (3) McComb and Noble (1982), (4) Metzger and Schultz (1984) (also data for clear-cut), (5) McComb

and Noble (1982), (6) Kirby (1990) (also data for clearcut), (7) Reader (1987), Reader and Bricker (1992a), (8) present study, (9) Reader (1988), Reader and Bricker (1992b) See also

Jalonen and Vanha-Majamaa (2001), and Pykala (2004).

Page 14

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 137

most favoured species belonged to the grassland and

forest species groups. For wind-dispersed ruderals

(e.g., Senecio sylvaticus), seeds from the previous

summer brought into plots may explain colonization,

but for most species, activation of the soil seed bank or

vegetative expansion probably explains why species

colonized or were favoured. In addition, seeds may

have been brought into plots by forestry workers or

forwarders. Mayer et al. (2004) also studied plant

colonization of forest soil one year after disturbance

(clipped vegetation) and found that activation of the

seed bank (especially Rubus idaeus) was of major

importance, whereas ‘‘seed rain’’ was unimportant.

Before our survey in 2003, colleagues remarked

that little would happen in experimental plots, or that

there was no need for survey, in the first year. The

responses of species studied under strong experi-

mental control indicate surprisingly quick, and

complex changes in the herb community in the first

season after partial cutting (see also Reader, 1988,

and Reader and Bricker, 1992b). For instance, among

the forest herbs at least 13 species were favoured and

disfavoured at the same time, depending on site or

plot. We suggest that partial cutting increases spatial

heterogeneity in the herb community to a larger

extent than would be expected after clear-cutting or

under continued secondary succession in closed

canopy stands (cf. Reader, 1988, p. 807). Ground

disturbance due to tree harvesting apparently

diminishes some species populations and favours

expansion in others; however, since trees and

undisturbed ground remain, there are also refuges

for forest species from which they may spread. Given

differences between sites in initial conditions, in soil

seed bank and in other factors, our heterogeneity

scenario implies that it may be difficult to predict the

immediate response of herb communities to partial

harvest, at least for multiple sites at landscape or

regional level. Long-term studies of several sites

under strong experimental control are needed to

further investigate our proposed scenario (most

earlier studies concern single forest sites, see also

below).

With respect to species richness and diversity (H0),the increase was strong at Ytterhult and Farbo, two

stands that initially contained relatively high propor-

tion of conifers. The herb cover was usually sparse

under or near conifers, and herbs in adjacent patches

might have expanded where conifers were cut, and/or

species from the seed bank might have colonized. At

Ulvsdal, the site with the weakest (although sig-

nificant) increase in species richness and in H0 in the

experimental plot, the cover of forest grasses was

higher than at the other sites. Some grasses might

withstand ground disturbance well, and might

dominate after partial harvest. At Ulvsdal, both

Deschampsia species tended to decrease after harvest,

while Milium tended to increase (Table 4), but the

experimental plot had relatively high grass cover

before cutting (mainly Calamagrostis, Melica nutans,

Milium). More work is needed to evaluate the role of

grasses after partial cutting (relative to clear-cutting,

and closed-canopy secondary succession).

Using databases, we reviewed studies where partial

cutting in temperate forest was planned for research

and controls were available either in time (experi-

mental area surveyed before cutting) or space (after

cutting; cutting and control area), or both (strong

control; present study and Reader, 1987). The review

was limited to early effects following partial cutting/

harvest (first summer after, or within about 1.5 year).

We found eight studies with some form of experi-

mental control, three in forests dominated by conifers,

and five in forests dominated by broadleaved trees

(Table 5). Although stand types and study methods

were far from identical, the three studies of coniferous

stands reported reduced, or unchanged species

richness, while four of five studies of broadleaved

stands reported increased species richness (Table 5).

Therefore, early responses to partial cutting may differ

between these forest types, for instance due to

differences in seed bank size or sensitivity of species.

Competition with woody plants seems unlikely as

explanation, as most conifers die after cutting, in

contrast to broadleaved trees. Species composition

changed more or less in all studies (Table 5). Of the

three studies testing for loss of species, none found

increased losses due to partial cutting. In Canada

(study 7, Table 5), losses were slightly lower than in

the present study, and there was a tendency for higher

species losses when more trees were cut (33% versus

66%), but this tendency was not significant. Four

studies found no change in species diversity index, one

possibly an effect (Simpson’s index, study 4), while

we found increased H0 due to partial cutting in our

transects. H0 may be sensitive to either scale or type of

Page 15

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141138

frequency measure, as we found no change in H0 for

squares.

In conclusion, with respect to early effects on the

herb flora after partial cutting, we found that (1) the

rate of species losses did not increase, in this and two

other studies with good experimental control, (2)

species richness increased, and this may mostly be the

case for broadleaved stands, but apparently not for

stands dominated by conifers, (3) some ruderals

increased after partial harvesting, but more changes

occurred in species of grassland and forest habitats.

Partial harvesting is suggested to lead to high spatial

and compositional heterogeneity in the herb commu-

nity at early successional stages.

5. Implications for forest management and

research

We focused on partial or ‘selection’ cutting

(Nyland, 2002) which is common in broadleaved

forest in areas with non-industrial private forest

owners (Kittredge et al., 2003) and generally may

become more common in the future. With respect to

short-term effects, one could argue that they are

ephemeral, but at the landscape and regional level

many stands are cut each year. Knowledge of effects of

cutting at each successional stage is needed, and early

changes may be crucial in directing succession, which

should to be investigated in the future.

For herbs, semi-experimental studies in European

broadleaved stands suggest that, although not all forest

herbs benefit from cutting, the herb flora is resilient and

harvested stands often seem to have higher species

richness than uncut stands (e.g., Brunet et al., 1996,

1997; Graae and Heskjaer, 1997; Tybirk and Strand-

berg, 1999). Our partial harvest was designed to

consider biodiversity values, such that old oaks and

other valuable trees and bushes were not cut. Therefore,

immediate effects on the woody vegetation was of little

concern. Ruderals are not desirable in the forest stands

studied, as they are common elsewhere, but so far they

are a minor component. In many areas, non-native or

non-forest species (Reader, 1994) are undesirable, or a

threat to biodiversity. In Europe (Pysek et al., 2003) and

Sweden, many herbs of grassland and semi-open

habitat arrived more than 100 years ago, and presently

seem to pose no or little threat to other species.

Beside herbs, we suggest that partial cutting in

forestry and research should consider (1) changes in

species associated with herbs (many pollinators,

herbivores, and parasites such as fungi and wasps)

(2) changes in other taxa of interest (e.g., Norden et al.,

2004a,b; Økland et al., 2005), and (3) the surrounding

landscape and its composition of stand types. Partial

harvesting implies reduction of deadwood in the future

stand, compared to uncut stands. Thus, one component

of biodiversity, dead wood with its associated fungi

and invertebrates, would be reduced (Norden et al.,

2004a,b). We recommend that some proportion of

broadleaved oak stands should not be cut, to safeguard

rare and potentially sensitive herbs (that are difficult to

survey) and to create stand diversity at the landscape

level (Meier et al., 1995; Scheller and Mladenoff,

2002).

Acknowledgements

We thank the Swedish Research Council (VR), the

Swedish Energy Agency, and Goteborg University for

financial support. The manuscript was written while

F.G. was Visiting Scientist at University of Wisconsin,

Department of Forest Ecology and Management at

Madison. Thanks to Craig Lorimer, and Raymond

Gurie and others for hospitality in Madison. The

Swedish Foundation for International Cooperation in

Research provided financial support for the visit in

Madison. For permission to conduct research at the

Swedish sites, and for kindly conducting the experi-

mental harvest, we thank the County Administration,

Kalmar (Lindo), Sveaskog AB (Farbo), Anders

Heidesjo (Ytterhult), Holmen Skog AB (Ulvsdal),

the diocese of Linkoping (Norra Vi), and the forest

sector of the municipality of Boras (Rya asar). Jorg

Brunet, Ralph Harmer, Johan Ehrlen and anonymous

reviewers kindly commented on the manuscript. A.

Agebjorn, Y. Folkesson, J. Forsberg, L. Helmersson, I.

Johansson, K. Jungbark, A. Karlsson, A. Malmsten, O.

Sandberg and E. Rube provided field assistance.

Appendix A

All species recorded at the six study sites

throughout the study (2001–2003); classification of

Page 16

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 139

the species’ habitat type in the surrounding land-

scape in southern Sweden; and the number of sites

where they occurred. A few species identified only to

genera (‘‘sp.’’) at some sites are also listed. In

analyses, we used the first (=major) of two habitats

given below, and the following three groups: forest,

grassland (including open habitat), and ruderal

species.

Species (taxa)

Major habitat No. of sites

Actaea spicata

Forest 1

Aegopodium podagraria

Open/forest 1

Agrostis capillaris

Grassland 6

Agrostis vinealis

Open 1

Ajuga pyramidalis

Open/forest 5

Alchemilla sp.

Open/grassland 2

Alliaria petiolata

Ruderal 1

Allium oleraceum

Open 2

Allium scorodoprasum

Forest/open 1

Anemone hepatica

Forest 6

Anemone nemoralis

Forest 6

Anthoxantum odoratum

Grassland 5

Anthriscus sylvestris

Grassland 4

Asplenium trichomanes

Open/forest 1

Athyrium filix-femina

Forest 3

Bromopsis benekenii/ramosa

Forest 1

Calamagrostis arundinacea

Forest 5

Caltha palustris

Open/forest 1

Campanula persicifolia

Grassland 4

Cardamine bulbifera

Forest 4

Cardamine hirsuta

Open 1

Carex digitata

Forest 1

Carex divulsa

Forest 1

Carex montana

Forest/open 2

Carex pallescens

Grassland 5

Carex pilulifera

Forest 3

Carex vaginata

Forest 2

Cerastium fontanum

Grassland/forest 1

Chenopodium polyspermum

Ruderal 1

Cirsium sp.

Ruderal/grassland 1

Convallaria majalis

Forest 6

Convolvulus arvensis

Open/ruderal 1

Cornus suecica

Open/forest 1

Dactylis glomerata

Grassland 5

Dactylorhiza sambucina

Grassland 1

Deschampsia cespitosa

Grassland 4

Deschampsia flexuosa

Open/forest 6

Dryopteris carthusiana

Forest 3

Dryopteris cristata

Forest 2

Dryopteris expansa

Forest 1

Dryopteris filix-mas

Forest 6

Elymus caninus

Forest 2

Epilobium montanum

Ruderal 1

Epilobium sp.

Ruderal 3

Equisetum sylvaticum

Forest 1

Festuca ovina

Open/grassland 3

Festuca rubra

Grassland 3

Filipendula ulmaria

Open/grassland 1

Fragaria vesca

Open/forest 3

Galeopsis bifida

Ruderal 3

Galeopsis sp.

Ruderal 3

Galeopsis tetrahit

Ruderal 2

Galium album

Open 1

Galium aparine

Forest/open 3

Galium odoratum

Forest 2

Galium palustre

Open 1

Galium uliginosum

Grassland 2

Geranium robertianum

Grassland 3

Geranium sylvaticum

Open/forest 3

Geum rivale

Open/forest 2

Geum urbanum

Open/forest 3

Geum urbanum/rivale

Open/forest 4

Glechoma hederacea

Open/forest 1

Gnaphalium sylvaticum

Forest/open 1

Gymnocarpium dryopteris

Forest 4

Helictotrichon pubescens

Grassland 1

Hieracium sect. Hieracium

Forest/open 6

Hieracium sect. Vulgata

Open 3

Hieracium umbellatum

Open 1

Holcus lanatus

Grassland 1

Hylotelephium telephium

Open 1

Hypericum maculatum

Grassland/open 2

Hypericum perforatum

Grassland 3

Juncus articulatus

Ruderal 1

Juncus bufonius

Ruderal 1

Knautsia arvensis

Grassland 1

Lapsana communis

Open 3

Laserpitium latifolium

Forest/open 1

Lathyrus linifolius

Forest/open 6

Lathyrus niger

Forest 2

Lathyrus pratensis

Grassland 4

Lathyrus vernus

Forest 4

Lotus corniculatus

Grassland 1

Luzula multiflora

Grassland 3

Luzula pilosa

Forest 6

Lysimachia vulgaris

Open 1

Maianthemum bifolium

Forest 5

Melampyrum pratense

Forest/grassland 6

Melampyrum sp.

Forest 2

Melampyrum sylvaticum

Forest 4

Melica nutans

Forest 6

Melica uniflora

Forest 3

Mercurialis perennis

Forest 1

Milium effusum

Forest 4

Moehringia trinervia

Forest 5

Monotropa hypopitys

Forest 1

Mycelis muralis

Forest 3

Myosotis arvensis

Ruderal 1

Origanum vulgare

Open 1

Oxalis acetocella

Forest 6

Paris quadrifolia

Forest 3

Phegopteris connectilis

Forest 1

Page 17

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141140

Phleum pratense

Grassland 1

Pilosella officinarium

Open 1

Plantago major

Ruderal 1

Platanthera bifolia

Grassland/forest 1

Platanthera sp.

Grassland/forest 1

Poa annua

Ruderal 2

Poa compressa

Open 2

Poa nemoralis

Forest 6

Poa pratensis

Grassland 2

Poa trivialis

Open 1

Polygonatum multiflorum

Forest 1

Polygonatum odoratum

Open/forest 3

Polygonatum verticillattum

Forest 1

Polypodium vulgare

Open/forest 4

Potentilla erecta

Open/forest 5

Primula veris

Grassland/forest 4

Pteridium aquilinum

Open/forest 3

Pulmonaria obscura

Forest 3

Pyrola minor

Forest 1

Ranunculus acris

Grassland 4

Ranunculus auricomus

Forest 4

Ranunculus flammula

Open/forest 1

Ranunculus repens

Ruderal 1

Rubus idaeus

Open/ruderal 5

Rubus saxatilis

Forest 4

Rumex acetosa

Open/grassland 1

Sanicula europaea

Forest 1

Saxifraga granulata

Open 1

Scorzonera humilis

Grassland 2

Senecio sylvaticus

Ruderal 6

Solidaga virgaurea

Open/forest 4

Sonchus arvensis

Ruderal 1

Sonchus sp.

Ruderal 2

Spergula morisonii

Ruderal/open 1

Stellaria graminea

Grassland 3

Stellaria holostea

Forest/open 1

Stellaria media

Ruderal 2

Succisa pratensis

Grassland 3

Taraxacum sect. Ruderalia

Ruderal 2

Trientalis europaea

Forest 2

Trifolium medium

Forest/open 3

Trifolium repens

Grassland 2

Tripleurospermum perforatum

Ruderal 1

Trollius europaeus

Grassland 1

Urtica dioca

Ruderal/grassland 1

Urtica urens

Ruderal 1

Vaccinium myrtillus

Forest 6

Vaccinium vitis-idaea

Forest 2

Veronica chamaedrys

Grassland/forest 5

Veronica officinalis

Open/forest 5

Vicia sepium

Grassland/forest 5

Vicia sylvatica

Forest 2

Vincetoxicum hirundinaria

Open 1

Viola mirabilis

Forest 1

Viola palustris

Open/forest 1

Viola riviniana

Forest 6

References

Ahti, T., Hamet-Ahti, L., Jalas, J., 1968. Vegetation zones and their

sections in northwestern Europe. Annales Botanici Fennici 5,

169–211.

Bergstedt, J., Milberg, P., 2001. The impact of logging intensity on

field-layer vegetation in Swedish boreal forests. For. Ecol.

Manage. 154, 105–115.

Brunet, J., Falkengren-Grerup, U., Tyler, G., 1996. Herb layer

vegetation of south Swedish beech and oak forests — effects

of management and soil acidity during one decade. For. Ecol.

Manage. 88, 259–272.

Brunet, J., Falkengren-Grerup, U., Ruhling, A., Tyler, G., 1997.

Regional differences in floristic change in south Swedish oak

forests as related to soil chemistry and land use. J. Veg. Sci. 8,

329–336.

Brunet, J., Tyler, G., 2000. Interannual variability in abundance of

field layer species in a south Swedish deciduous wood. Flora

195, 97–103.

Diekmann, M., 1994. Deciduous forest vegetation in Boreo-nemoral

Scandinavia. Acta Phytogeographica suecica 80, 1–112.

Esseen, P.-A., Ehnstrom, B., Ericsson, L., Sjoberg, K., 1997. Boreal

forests. Ecol. Bull. 46, 16–47.

Franklin, J.F., Kohm, K.A., 1997. Creating a Forestry for the 21st

Century: The Science of Ecosystem Management. Island Press,

Washington, DC.

Gotmark, F., Thorell, M., 2003. Size of nature reserves: densities of

large trees and dead wood indicate high value of small con-

servation forests in southern Sweden. Biodiv. Conserv. 12,

1271–1285.

Gustafsson, L., Ahlen, I. (Eds.), 1996. National atlas of Sweden.

Geography of Plants and Animals. Almqvist & Wiksell, Stock-

holm.

Gustafsson, L., 2000. Red-listed species and indicators: vascular

plants in woodland key habitats and surrounding production

forest in Sweden. Biol. Conserv. 92, 35–43.

Graae, B.J., Heskjaer, V.S., 1997. A comparison of understorey

vegetation between untouched and managed deciduous forest in

Denmark. For. Ecol. Manage. 96, 111–123.

Halpern, C.B., Spies, T.A., 1995. Plant species diversity in natural

and managed forests of the Pacific Northwest. Ecol. Appl. 5,

913–934.

Hannertz, M., Hanell, B., 1993. Changes in the vascular plant

vegetation after different cutting regimes on a productive peat-

land site in central Sweden. Scand. J. For. Res. 8, 193–203.

Hansson, L., 2001. Traditional management of forests: plant and

bird community responses to alternative restoration of oak-hazel

woodland in Sweden. Biodiv. Conserv. 10, 1865–1873.

Hunter, M.L. (Ed.), 1999. Maintaining Biodiversity in Forest

Ecosystems. Cambridge University Press, Cambridge, UK.

Jalonen, J., Vanha-Majamaa, I., 2001. Immediate effects of four

different felling methods on mature boreal spruce forest under-

storey vegetation in southern Finland. For. Ecol. Manage. 146,

25–34.

Karlsson, T., 1998. The vascular plants of Sweden. Svensk Botanisk

Tidskrift 91, 241–560.

Page 18

F. Gotmark et al. / Forest Ecology and Management 214 (2005) 124–141 141

Kirby, K.J., 1990. Changes in the ground flora of a broadleaved

wood within a clear fell, group fells and a coppiced block.

Forestry 63, 241–249.

Kittredge Jr., D.B., Finley, A.O., Foster, D.R., 2003. Timber harvest-

ing as ongoing disturbance in a landscape of diverse ownership.

For. Ecol. Manage. 180, 425–442.

Krok, T., Almquist, S., 1984. Svensk Flora. Esselte, Uppsala

Lorimer, C.G., 2001. Historical and ecological roles of disturbance

in eastern North American forests: 9000 years of change. Wildl.

Soc. Bull. 29, 425–439.

Mayer, P., Abs, C., Fischer, A., 2004. Colonization by vascular

plants after soil disturbance in the Bavarian forest — key factors

and relevance for forest dynamics. For. Ecol. Manage. 188, 279–

289.

McComb, W.C., Noble, R.E., 1982. Response of understory vegeta-

tion to improvement cutting and physiographic site in two mid-

south forest stands. Castanea 47, 60–77.

Meier, A.J., Bratton, S.P., Duffy, D.C., 1995. Possible ecological

mechanisms for loss of vernal-herb diversity in logged eastern

deciduous forests. Ecol. Appl. 5, 935–946.

Metzger, F., Schultz, J., 1984. Understorey response to 50 years of

management of a northern hardwood forest in upper Michigan.

Am. Midl. Nat. 112, 209–223.

Mossberg, B., 1992. Den nordiska floran. Wahlstrom and Widstrand,

Brepols, Belgium.

NBF (National Board of Forestry), 1999a. Nyckelbiotopsinventer-

ingen 1993–1998, Slutrapport. Meddelande 1-1999. NBF, Jon-

koping (in Swedish).

NBF (National Board of Forestry), 1999b. Vard och skotsel: Nyck-

elbiotoper och andra vardefulla biotoper. NBF, Jonkoping (in

Swedish).

NBF (National Board of Forestry), 2001. Skogsbransle, hot eller

mojlighet? Vagledning till miljovanligt skogsbransleuttag.

Skogsstyrelsens forlag, Kristianstad (in Swedish).

Nilsson, S.G., 1997. Forests in the temperate-boreal transition:

natural and man-made features. Ecol. Bull. 46, 61–70.

Norden, B., Ryberg, M., Gotmark, F., Olausson, B., 2004a. Relative

importance of coarse and fine woody debris for the diversity of

wood-inhabiting fungi in temperate broadleaf forests. Biol.

Conserv. 117, 1–10.

Norden, B., Gotmark, F., Tonnberg, M., Ryberg, M., 2004b. Dead

wood in semi-natural temperate broadleaf woodland: contribu-

tion of coarse and fine dead wood, attached dead wood and

stumps. For. Ecol. Manage. 194, 235–248.

Nyland, R.D., 2002. Silviculture: Concepts and Applications.

McGraw-Hill, New York.

Økland, B., Gotmark, F., Norden, B., Franc, N., Kurina, O., Polevoi,

A., 2005. Regional diversity of mycetophilids (Diptera: Sciar-

oidea) in Scandinavian temperate forests. Biol. Conserv. 121, 9–

20.

Pykala, J., 2004. Immediate increase in plant species richness after

clear-cutting of boreal herb-rich forests. Appl. Vegetat. Sci. 7,

29–34.

Pysek, P., Sadlo, J., Mandak, B., Jarosik, V., 2003. Czech alien flora

and the historical pattern of its formation: what came first to

Central Europe? Oecologia 135, 122–130.

Rackham, O., 1980. Ancient Woodland: Its History, Vegetation and

Uses in England. Edward Arnold, London.

Reader, R.J., 1987. Loss of species from deciduous forest under-

storey immediately following selective tree harvesting. Biol.

Conserv. 42, 231–244.

Reader, R.J., 1988. Using the guild concept in the assessment of tree

harvesting effects on understory herbs: a cautionary note.

Environ. Manage. 12, 803–808.

Reader, R.J., 1994. Barriers to establishment of invading, non-forest

plants in deciduous forest reserves. Environ. Conserv. 21,

62–66.

Reader, R.J., Bricker, B.D., 1992a. Value of selectively cut decid-

uous forest for understory herb conservation: an experimental

assessment. For. Ecol. Manage. 51, 317–327.

Reader, R.J., Bricker, B.D., 1992b. Response of five deciduous

forest herbs to partial canopy removal and patch size. Am. Midl.

Nat. 127, 149–157.

Resampling Stats, Inc., 2000. Users’s guide. Arlington, Virginia.

Ruben, J.A., Bolger, D.T., Peart, D.R., Ayres, M.P., 1999. Unders-

tory herb assemblages 25 and 60 years after clearcutting of a

northern hardwood forest, USA. Biol. Conserv. 90, 203–215.

Scheller, R., Mladenoff, D., 2002. Understory species patterns and

diversity in old-growth and managed northern hardwood forests.

Ecol. Appl. 12, 1329–1343.

Schoonmaker, P., McKee, A., 1988. Species composition and

diversity during secondary succession of coniferous forests in

the Western Cascade mountains of Oregon. For. Sci. 34, 960–

979.

Schumann, M.E., White, A.S., Witham, J.W., 2003. The effects of

harvest-created gaps on plant species diversity, composition, and

abundance in a Maine oak-pine forest. For. Ecol. Manage. 176,

543–561.

Sullivan, T.P., Sullivan, D.S., Lindgren, P.M.F., 2001. Influence of

variable retention harvests on forest ecosystems I. Diversity of

stand structure. J. Appl. Ecol. 38, 1221–1233.

ter Braak, C.J.F., Smilauer, P., 2002. CANOCO Reference manual

and CanoDraw for Windows, User’s guide: Software for Cano-

nical Community Ordination (version 4.5). Microcomputer

Power, Ithaca, NY.

Tybirk, K., Strandberg, B., 1999. Oak forest development as a result

of historical land-use patterns and present nitrogene deposition.

For. Ecol. Manage. 114, 97–106.

Tyler, G., 2001. Relationships between climate and flowering of

eight herbs in a Swedish deciduous forest. Ann. Bot. 87, 623–

630.

Tyler, T., Brunet, J., Tyler, G., 2002. Sensitivity of the woodland

herb Anemone hepatica to changing environmental conditions.

J. Veg. Sci. 13, 207–216.

Vera, F.W.M., 2000. Grazing Ecology and Forest History. CABI

Publishing, Oxon, UK.