Pollen productivity estimates from old-growth forest strongly differ from those obtained in cultural landscapes: Evidence from the Białowieża National Park,Poland

Abstract

Pollen productivity estimates of individual plant taxa are necessary when determining changes of vegetation cover during the Holocene. To date, studies describing this parameter in lowland temperate Europe have been carried out in cultural landscapes showing low forest cover and dominated by human activities. However, these may be of limited use when applied to reconstruct past land cover, for instance, from pre-agricultural landscapes. The aim of this paper is to ascertain whether pollen productivity from the closed-canopy old-growth forest in the Białowieża National Park, Poland, where human impact has been minimal for nearly a century, is different from that calculated in much more open landscapes. We ask: how much does forest antiquity and structure influence the amount of pollen released from particular taxa? We implemented maximum likelihood estimation of relative pollen productivity for seven tree species and for Poaceae using 18 modern pollen assemblages and distance-weighted plant abundances. Our results demonstrate that the ratio of pollen productivity between high producers (Pinus sylvestris and Quercus robur) and low producers (Poaceae, Corylus avellana) is on an average six times greater in Białowieża than across other European cultural landscapes. Pollen from forest Poaceae and C. avellana is six times more under-represented in old-growth forest than hitherto estimated from cultural landscapes. This finding reinforces the idea that pollen productivity can vary in response to changes in the prevailing environmental settings and we present for the first time a quantification of this variability, likely induced by differences in light availability.

Keywords

canopy structure ERV-models landscape openness moss polsters old-growth forest relative pollen productivity

Introduction

Advances in quantitative reconstruction of vegetation cover using fossil pollen over the last decade have made it possible to estimate land cover throughout the Holocene for the most of Europe (Gaillard et al., 2008, 2010; Marquer et al., 2014; Mazier et al., 2012; Nielsen et al., 2012), an important parameter for climate modelling in relation to climate change (Gaillard et al., 2010; Strandberg et al., 2014; Trondman et al., 2015). Landscape openness is also critical to studies focusing on the impact of past civilisations on their environment, with respect to their cultural, economic and demographic dynamics (Nielsen et al., 2012).

These advances are due to the conceptual and technical progress in landscape reconstruction algorithms that translate pollen assemblages from small and large sedimentary basins into the estimation of per cent vegetation cover or biomass (Sugita, 2007a, 2007b). Sugita’s methodology used in these models solves a major difficulty in the analysis of fossil pollen, that is, the confounding influence of spatial distribution of vegetation onto pollen proportion in sediments (Davis, 2000). One crucial parameter within Sugita’s algorithms is the estimation of relative pollen productivity that is specific to different plant species (Broström et al., 2008; Hellman et al., 2008). Relative pollen productivity is the ratio between the pollen productivity of two species (Sugita et al., 1999).

To date, relative pollen productivity estimates (PPEs) have been established for all the major tree species throughout cultural landscapes of temperate Europe where agriculture and urbanisation directly modify between 40% and 90% of the plant cover or from boreal and subalpine environments where climate is significantly colder and where lower pollen productivity is predictable (Table 1). Other studies estimate pollen productivity for various herbs but no trees (e.g. Broström et al., 2004; Bunting and Hjelle, 2010; Hjelle, 1998).

Table 1.

Vegetation cover and climatic characteristics of pollen productivity studies using the same methodology that used in this paper.

Reference	Country	Region	% Forest	% Grassland and arable land	% Others	Forest	Sample media	AMT (°C)	TS (°C)	TAP (mm)	PS (mm)
This paper	Poland	Białowieża National Park (Miścicki, 2012)	97	3		Temperate	Moss	7.0	8.1	579	31
		Forest of Białowieża (Mikusińska et al., 2013)	87	10	3	Temperate
Nielsen (2004)	Denmark	Denmark	10	65	25	Temperate	Lake	7.7	6.0	611	22
Theuerkauf et al. (2012)	Germany	States of Mecklenburg – Vorpommern and Brandenburg	22	59	19	Temperate	Lake	8.1	6.7	579	22
Soepboer et al. (2007)	Switzerland	Swiss Plateau	25	50	25	Temperate	Lake	8.7	6.4	1103	25
Matthias et al. (2012)	Germany	State of Brandenburg	37	45	18	Temperate	Lake	9.0	6.8	567	22
Sugita et al. (1999)	Sweden	1. Southern Skane	20–40	60–80		Temperate	Moss	7.3	6.7	494	24
		2. Northern Skane and Smaland	50–60	40–50		Temperate
Poska et al. (2011)	Estonia	South Estonia	51	31	18	Temperate/boreal	Lake	5.0	8.7	605	33
Mazier et al. (2008)	Switzerland	Parc Jurassien Vaudois	60	40		Montane	Moss	4.5	6.1	1622	18
Abraham and Kozáková (2012)	Czech Rep.	Central Bohemia	Forest fragments within agricultural landscape			Temperate	Moss	7.9	6.9	533	40
Bunting et al. (2005)	England	Norfolk	Forest fragments within agricultural landscape			Temperate	Moss	10.3	4.8	618	15
Twiddle et al. (2012)	Scotland	Cairngorm National Park	Forest fragments within agricultural landscape			Temperate/boreal	Moss	7.2	4.3	785	16
Von Stedingk et al. (2008)	Sweden	Skalstugan, Scandes Mountains	<30		>70	Boreal	Moss	1.6	7.0	962	21
Räsänen et al. (2007)	Finland	Northern Finland	Large pine forests, with treeless mires and mountain tops			Boreal	Moss	−0.8	8.8	438	42

AMT: annual mean temperature; TS: temperature seasonality (SD of all months); TAP: total annual precipitation; PS: precipitation seasonality (SD of all months). Climatic data are from Hijmans et al. (2005).

However, what is still unknown is whether results from cultural landscapes can be utilised to interpret fossil pollen assemblages from landscapes that have undergone different disturbance regimes such as old-growth forests or ancient landscapes with vegetation cover relatively untouched by agriculture. For instance, the composition and structure of vegetation before the advent of agriculture in Europe, as reconstructed using fossil pollen, have been at the heart of a polarised debate in recent decades following Vera’s (2000) hypothesis. Are pre-agricultural landscapes best described as high forest (i.e. continuous forest cover with a closed canopy wherever the relief and hydrology permit tree growth) or wood pasture (i.e. patchy forest cover within a mosaic of regeneration stages driven by large herbivores, such as bison, aurochs, horses and other natives)? This question is of importance to guide nature conservation and forest management, because ancient landscapes reconstructed with fossils are a source of inspiration and can serve as a baseline for restoration. Many palaeoecologists have argued for high forest on the basis of the fossil pollen record (e.g. Birks, 2005; Bradshaw et al., 2003; Mitchell, 2005) and the fossil beetle assemblages (Buckland, 2005; Smith et al., 2010). However, these studies and others (e.g. Hall, 2008; Nielsen et al., 2012; Svenning, 2002) also highlight the importance of natural forces, such as wildfires, wind throws, hydrology and soil fertility, along with grazing, to create the dynamic necessary to open forest gaps and promote tree recruitment and regeneration of more light-demanding species. In this respect, validation of these competing hypotheses with more accurate quantitative reconstructions of past vegetation is still urgently needed.

The overarching aim of our study is to estimate new pollen productivity data from an old-growth forest that had minimum human impact over time and that can serve as an important contrast to the cultural landscapes previously used to calculate PPE. We chose the forest located in the Białowieża National Park (BNP) in Poland, because this is often regarded as a recent analogue for primeval lowland woodland where large herbivores are among the most important disturbance factors (Bobiec, 2002; Faliński, 1986). The BNP is covered by extensive old-growth forest not exploited by forestry or other human activities since nearly 100 years and is nested at the heart of the largest fragment of temperate woodland in lowland Europe – the Białowieża Forest. Our key objectives are as follows:

To establish relative PPEs for the main trees and Poaceae in the old-growth forest of Białowieża;

To compare PPE from the Białowieża Forest to those obtained in cultural landscapes across Europe;

To discuss any differences in the light of the disturbance regimes and environmental settings that prevail in old-growth forests and cultural landscapes.

Study area

The Białowieża Forest is a flagship ecosystem in European nature conservation (Bobiec, 2002; Peterken, 1996). It covers about 1450 km² (52°29′–52°57′N; 23°31′–24°21′E) straddling the border between Poland and Belarus (Figure 1). Białowieża Forest is the largest expanse of lowland temperate closed-canopy forest in Europe and it is unique for the complete assemblage of native trees, large herbivores and carnivores, for its size and for its antiquity (Faliński, 1986).

Figure 1.

Location of Białowieża Forest in Europe and of the Białowieża National Park within the Białowieża Forest (compiled according to different sources). See Figure 2 for the sampling locations and dominant tree stands within the BNP.

The whole area of Białowieża Forest has international protection as a UNESCO Biosphere Reserve and World Heritage site, a Ramsar site, and a Natura 2000 site as well as having protected status at the national and local levels. Special protection applies to 105 km² within the Polish BNP and to 157 km² within the Belarus Belovezhskaya Pushcha National Park. Strict protection, where no human intervention is allowed and access is limited to research or restrictive tourism, started as early as 1921 for 47.5 km² of old-growth forest within today’s BNP. Outside this area and outside the numerous other smaller nature reserves present across Białowieża Forest, both hunting and forestry are current practice, thereby directly influencing the populations of large herbivores and trees (Jędrzejewski et al., 2006; Kuijper et al., 2010a)

The Białowieża Forest has been under protection from agricultural development from the Middle Ages when it became a game reserve for Lithuanian and Polish rulers (Samojlik, 2005). As a result, it has remained remarkably untouched by human activities in contrast with the rest of Europe (Faliński, 1986; Latałowa et al., 2015). No large-scale agriculture or industry has ever been prominent. However, small-scale activities such as subsistence farming and various forest crafts are all reported from historical archives. Historical variation of these disturbance regimes explains much of the forest diversity today (Bobiec, 2012; Niklasson et al., 2010; Pigott, 1975). In spite of these activities, the area has experienced exceptional stability with regard to land cover for the last 200 years (Mikusińska et al., 2013). The main natural disturbance factors that have been shaping the forest composition and structure are as follows: large herbivore selective browsing (Kuijper et al., 2010a; Smit et al., 2012), pathogen outbreaks such as the European spruce bark beetle (Ips typographus) (Bobiec et al., 2011; Miścicki, 2012) and wind throws (Faliński, 1978). Large herbivores and carnivores in this forest include bison (Bison bonasus), moose (Alces alces), red deer (Cervus elaphus), roe deer (Capreolus capreolus), wild boar (Sus scofa), wolves (Canis lupus) and lynx (Lynx lynx) (Kuijper et al., 2010a).

In terms of vegetation, this large forest complex is located at the transition between the boreal and the nemoral biogeographic zones. This is reflected by the co-occurrence of sub-oceanic, central-European and boreal forest communities, that is, with deciduous trees and the evergreen spruce (Picea abies). The boreal elements of the flora and vegetation structure are promoted by a relatively short growing season (179 days) and long duration of snow cover (92 days) (Boczoń, 2006; Faliński, 1986). For the period 1950–2003, the average annual precipitation was 627.5 mm and the temperatures were as follows: 6.8°C (annual), –4.3°C (January) and 17.7°C (July) (Boczoń, 2006). However, in the last decade, precipitation has systematically decreased, while temperatures have systematically increased, resulting in extension of growth seasons up to around 220 days long (Malzahn et al., 2009).

The Białowieża Forest is dominated by mesophile oak–linden–hornbeam forest community (Tilio-Carpinetum), having a multi-layered structure with P. abies in the top-most layer (up to 50 m high), oak (Quercus robur), linden (Tilia cordata) and Norway maple (Acer platanoides) below it, and hornbeam (Carpinus betulus) in the lowest tree-layer closing space between other trees. Hazel (Corylus avellana) occurs in the shrub layer, together with juvenile forms of other trees. Ground water level and soil fertility determine the particular dominance and composition of the herb layer of this very biodiverse community (Faliński, 1986; Sokołowski, 1993). Other communities include mixed conifer forests with P. abies or pine (Pinus sylvestris) as dominant species, and lower abundances of Q. robur and birches (Betula spp.). These conifer forests appear to be in a process of transformation, manifested by a reduction of Picea abies and Pinus sylvestris and an increase in mesophile taxa, mainly Carpinus betulus and T. cordata (Bobiec, 2012; Faliński, 1988). Also, small stands of pine forest (Vaccinio vitis idaeae–Pinetum) are restricted to the nutrient-poor substrate found in areas of inland sand-dune systems.

Over 20% of the BNP is covered by forest growing on waterlogged soils, with a variety of fertility levels and hydrological regimes (Czerepko, 2008). Ash-alder forest (Circaeo-Alnetum) and ash-elm forest (Ficario-Ulmetum) occur along streams and in river valleys. These tree stands are dominated by alder (Alnus glutinosa), elm (Ulmus glabra), ash (Fraxinus excelsior) and Picea abies, while Corylus avellana grows in the shrub layer. In local depressions with stagnant water, alder carr develops with A. glutinosa as dominant species and lower abundances of P. abies and B. pubescens. Raised bogs and transitional peatlands support marshy forests dominated by either Pinus sylvestris or Betula spp. or a mixture of both, and have a lower abundance of Picea abies. Transitional bogs have the typical character of those found in the boreal zone, with spruce forests of the type Sphagno girgensohnii-Piceetum (Czerepko, 2008; Faliński, 1986). This is characterised by weak growth forms of P. abies, B. pubescens in lower abundances, a limited dwarf shrub and herb layers, and a very rich moss layer. We refer the reader to the phytosociology literature for more information on the forest communities present in the Białowieża Forest and the BNP (e.g. Faliński, 1986; Sokołowski, 1993).

Methods

Pollen

All the field work was carried out within the BNP (52°45′07″Ν, 23°52′44″E) in August 2011. Moss polsters were collected from the forest floor, at 18 locations well spread in space and showing diverse old forest stands (Figure 2; ESM Table S1, available online). Totally, 12 sampling sites were located in oak–linden–hornbeam forest and six in different types of coniferous forests. We sampled three or four moss sub-samples including both green and brown parts (Räsänen et al., 2004) within a 1 m² quadrat using a ring of 21.2 cm². The sub-samples were mixed in a plastic bag and kept in the dark at 4°C. The moss polsters were treated in the laboratory according to Hicks et al.’s (1999) protocol: rinsed in distilled water and sieved through 200 µm mesh, then boiled in 10% KOH for 5 min, followed by Erdman’s acetolysis, staining and mounting in glycerine. Over 1000 pollen grains per sample were counted and identified with specific keys (Beug, 2004; Moore et al., 1991; Punt et al., 1976–2003), and the reference collection of the Laboratory of Palaeoecology and Archaeobotany, University of Gdańsk.

Figure 2.

Location of the pollen sampling sites in the Białowieża National Park and dominant tree stands distribution; numbers of sites as in ESM Table S1 (available online) (map by Kwiatkowski and Gajko, 2009).

Vegetation survey

The vegetation survey was conducted at three spatial scales and followed standard practice for the estimation of pollen productivity based on moss polsters (Broström et al., 2008), in particular Bunting et al. (2013) for the first 10 m and Mazier et al. (2008) from 10 to 100 m. It was conducted around each 1 m² quadrat sampled for pollen. The percentage cover of ground flora was estimated visually using four transects of four 1 m² quadrats to 10 m away from the sampling point (distances from 0.5, 1.5, 3.5 and 7.5 m). Damage from large herbivores was recorded as present or absent for each species in each quadrat. All plant identifications followed Flora Europaea (Tutin et al., 1964–1980) and Flora Vegetativa (Eggenberg and Möhl, 2008) for diagnostic vegetative characters. The canopy cover was also recorded using four transects along a distance of 100 m. For each transect, the canopy composition and percentage cover were recorded within four 5-m-radius relevés (distance 15, 30, 50 and 90 m), while canopy directly above the sampling quadrat was recorded in a 10-m radius relevé. Only mature trees producing pollen were recorded during the canopy survey. The vegetation data within 1000 m radius from the sampling point were prepared on the basis of existing floristic and phytosociological studies (Sokołowski, 1993, 2004). For each forest type, we compiled a list of tree and Poaceae species and their average per cent cover within the community. The maps of forest habitat types and tree stands were drawn using the Numerical Map of BNP 2001 and a geographic information system (GIS; Quantum GIS ver. 2.0). In a GIS environment (ArcGIS 10.1), we compiled the survey and forest map data into a single vegetation map and split it into adjoining concentric rings at regular distances from the sampling point (0.5, 1.5, 3, 6, 10, 23.7, 41.2, 72.8, 100 m and then every 50 m from 100 to 1000). Thus, we obtained the surface cover of all tree and herb species (in m² or in per cent) within each concentric ring.

Data handling and modelling

To calculate PPE of main trees at our research sites, we undertook a whole modelling approach similar to that of Prentice and Sugita (Parsons and Prentice, 1981; Prentice and Parsons, 1983; Sugita, 1993; Sugita et al., 1999). This approach calculates PPE by maximum likelihood using pollen counts as a response variable and vegetation data as explanatory variable. We implemented three types of distance-weighting transformation on the vegetation data (Prentice, 1985; Prentice and Parsons, 1983; Sugita, 1993; Sugita et al., 1999) by down-weighting plant species surface cover in the concentric rings, proportionally with distances from the sampling points. This method simulates pollen dispersal such that pollen accumulation is influenced by the distance of a sampling point from the pollen source. The three methods are inverse distance (1/d), inverse square distance (1/d²) and pollen-type specific (pt-s) distance-weighting (Prentice, 1985). Pollen-type specific distance-weighting is based on the travelling of small particles in turbulent air above the canopy and has been successfully applied in similar situations (Broström et al., 2008). To implement this type of distance-weighting, we used the pollen fall speed as in Sugita et al. (1999).

Three maximum likelihood algorithms were used to obtain relative PPE. These were implemented using vegetation data, distance-weighted in three ways as presented above, resulting in nine combinations of algorithm-distance-weighting. The algorithms are named as ERV 1, ERV 2 and ERV 3 (Prentice and Parsons, 1983; Sugita, 1993) and make different assumptions regarding the background pollen contribution across sites (summarised in Broström et al., 2008). All three algorithms make reasonable assumptions in many situations and there is no a priori reason to favour one over another, so all three algorithms are usually applied (Broström et al., 2008). The assumptions underlying each algorithm are reviewed in detail in Broström et al. (2008). The relevant source area of pollen (Sugita, 1994) was established by implementing each of the nine combinations of algorithm-distance-weighting at the same regular distances used for distance-weighting. We used the log-likelihood indicator to select the distance and combination of algorithm-distance-weighting that fitted best our pollen and vegetation data. The relevant source area of this best combination of parameters is reached when the log-likelihood indicator stops improving with distance (i.e. it plateaus off when plotted against distance).

The final selection of plant species was obtained by repeating the above analysis until all PPE values returned by the modelling were plausible (between 0.01 and 100) and had reasonable standard errors (<10% of the PPE). The distance-weighting and maximum likelihood calculations were implemented with the ‘ERV.Analysis.v1.3.0’ program (S. Sugita, unpublished).

To compare the PPE obtained in the BNP with those of other studies, we calculated three sets of PPE using Pinus sylvestris, Quercus robur and Poaceae as reference taxa, respectively (Table 2). In addition, we compiled from the published literature all relevant estimates obtained with similar methods and the most recent consensus values given by Mazier et al. (2012) (Table 3).

Table 2.

Relative pollen productivity estimates in Białowieża Forest.

	Relative to Poaceae				Relative to Pinus P. sylvestris				Relative to Quercus robur
	ERV 1	SE	ERV 3	SE	ERV 1	SE	ERV 3	SE	ERV 1	SE	ERV 3	SE
Poaceae	1.00	na	1.00	na	0.051	0.0048	0.039	0.0246	0.066	0.0066	0.049	0.0406
Alnus glutinosa	33.28	0.8581	15.95	0.6622	0.488	0.0143	0.680	0.0280	0.697	0.0230	0.873	0.0404
Betula	30.08	0.4084	13.94	0.2293	0.505	0.0072	0.604	0.0095	0.726	0.0104	0.758	0.0199
Carpinus betulus	10.43	0.1461	4.48	0.0301	0.162	0.0015	0.195	0.0020	0.234	0.0025	0.244	0.0066
Corylus avallana	2.78	0.1418	1.35	0.0512	0.045	0.0018	0.057	0.0021	0.066	0.0026	0.072	0.0032
Pinus sylvestris	51.38	0.7011	23.12	0.2388	1.000	na	1.000	na	1.457	0.0217	1.256	0.0323
Quercus robur	44.76	0.5937	18.47	0.1032	0.694	0.0077	0.808	0.0071	1.000	na	1.000	na
Tilia cordata	2.18	0.0596	0.98	0.0263	0.034	0.0008	0.043	0.0010	0.049	0.0011	0.053	0.0020

Table 3.

Ratio between high pollen producers and low pollen producers in this study and across Europe.

		Highest producers		Lowest producers		Range of ratio between high and low producers
	Location	Pinus	Quercus	Corylus	Poaceae	Range of ratio between high and low producers
This study – ERV 1	Poland	51.38	44.76	2.78	1	51–16
This study – ERV 3	Poland	23.12	18.47	1.35	1	23–14
Mazier et al. (2012)	Europe-wide	6.38	5.83	1.99	1	6–3
Räsänen et al. (2007)	Finland	8.4	–	–	1	8
Sugita et al. (1999)	S Sweden	5.66	7.53	1.4	1	8–4
Bunting et al. (2005)	England	–	5.83	1.51	1	6–4
Poska et al. (2011)	Estonia	5.07	7.39	–	1	7–5
Soepboer et al. (2007)	Switzerland	1.35	2.56	2.58	1	3–0.5
Abraham and Kozáková (2012)	Czech Rep.	6.17	–	–	1	6

Results

Pollen proved abundant enough for our analysis in all samples collected. All pollen spectra were characterised by a very high arboreal/non-arboreal proportion, between 93.8% and 98.3% (ESM Figure S1, available online). In most of the 12 samples from oak–linden–hornbeam forest, the pollen of Carpinus (5) and Quercus (3) absolutely dominated. However, in two other samples, Picea pollen and Pinus pollen were co-dominant with a lower percentage of deciduous tree pollen. In one other sample, Betula was the most common pollen type. The pollen spectra from oak–linden–hornbeam forest had several tree taxa well represented, reflecting the complex floristic composition of this vegetation type. In all moss polsters from coniferous forests (6), Pinus pollen was dominated; it reached 80–90% in four samples just under 40% in the two remaining ones.

Plant taxa abundance at each site is presented in ESM Table S2 (available online). Across all vegetation plots, damage from herbivory was systematically absent from herbaceous plants (including Poaceae) and was common on juvenile woody plants in the shrub layer of the forest’s undergrowth.

The taxa which displayed a positive relationship between pollen per cent and distance-weighted per cent cover were A. glutinosa, Betula spp. (including B. pendula and B. pubescens), C. betulus, Corylus avellana, Pinus sylvestris, Poaceae, Q. robur and T. cordata (ESM Figure S2 A and B, available online). Other pollen types were not suitable for modelling with the methodology used. Notably, the models including Picea abies, an important component in Białowieża Forest, returned unrealistic results for that taxon (PPEs = ~10⁻⁶) with very large standard errors. In addition, the inclusion of P. abies in models was significantly increasing the PPE standard errors for all other taxa. Sedges (Cyperaceae) did not show a good enough spread of data (pollen between 0% and 0.33% of pollen sum) and produced very large standard errors and unrealistic results when included in models.

Among distance-weighting methods, 1/d and pt-s returned nearly equally good results with all three algorithms (Figure 3a–c), while 1/d² systematically returned worse log-likelihoods. In such situations, it is appropriate to recommend 1/d, the simplest distance-weighting method.

Figure 3.

Model performance (log-likelihood values), in relation to distance, for the selection of distance-weighting method (a–c) and for selection of the reference taxa (d–f). Algorism ERV 1 (a and d), ERV 2 (b and e), ERV 3 (c and f); plant abundance distance-weighting ‘1/d’ is inverse distance, ‘1/d²’ is inverse squared distance, ‘p-t specific’ is pollen-type specific. Note that symbols overlap when model performance is similar.

When using Poaceae as a reference taxon, the ERV 1 algorithm did not perform as consistently as ERV 2 and ERV 3, as shown by the variability of the log-likelihood when plotted against the distance gradient (Figure 3d in comparison to Figure 3e and f). Because this behaviour is not observed when using Pinus or Quercus as reference taxa, it was concluded that in our study, Poaceae are less suitable for this purpose (Figure 3), but we provide PPEs relative to Poaceae, for comparison with similar published analyses.

The relevant source area is within a 400-m-radius distance from sampling points with ERV 1 and ERV 3. Using these algorithms, PPE for all species stabilise before 400 m along a distance gradient (ESM Figure S3, available online). In contrast, the PPEs obtained with the algorithm ERV 2 do not show any stable pattern along the distance gradient while log-likelihood values do not reach a horizontal asymptote as expected with this method (Sugita, 1994). We decided not to retain results of model ERV 2 on this basis although the PPEs were in the same range of values than those obtained with models ERV 1 and ERV 3.

The PPEs relative to Pinus, Poaceae and Quercus obtained with the best setting identified for our data, that is, algorithms ERV 1 and ERV 3, 1/d distance-weighting and wind speed of 3 m s⁻¹, are presented in Table 2. Results indicate that A. glutinosa, Betula, Q. robur and P. sylvestris are high pollen producers, Carpinus betulus is an intermediate pollen producer, and Poaceae, Corylus avellana and T. cordata are relatively very low producers. There is between 23- (ERV 3) and 50-fold (ERV 1) difference of PPE between the highest producers (Pinus and Quercus) and the lowest producers (Poaceae and Tilia).

Discussion

Relative PPE from Białowieża Forest

Comparing our results with PPE values from similar studies across Europe indicates that the difference between low and high producers is much higher at our research sites than in any other place in the rest of lowland Europe (Table 3). The ratio of PPE between the highest producers (Pinus and Quercus) and the lowest producers (Poaceae and Corylus) is on average six times greater in the BNP than across European cultural landscapes. This means that either in closed-canopy forest Poaceae and Corylus are six times more under-represented in the pollen rain than hitherto estimated from cultural landscapes, or Pinus and Quercus are six times more over-represented, or any mixture between these two extreme explanations. However, because it is based on ratios, the modelling method used here does not allow us to ascertain whether high producers in the BNP are higher producers in absolute terms, or whether low producers produce an absolute lower quantity of pollen. This main finding, however, remained true whether using Pinus, Quercus or Poaceae as reference taxa. There may be several reasons why this ratio is so high in our study, including different methodologies between studies, extrinsic abiotic factors such as climate influencing pollen productivity rates and intrinsic biotic factors such as the composition of the vegetation and the disturbance regimes that the forest has been undergoing. Each of these will be discussed in turn in the next sections.

Reference taxa

Our results highlight how the choice of reference taxa may influence the data interpretation. As underlined by Sugita et al. (1999) and Bunting et al. (2013), in theory any taxa can serve as the reference unit, if it represents a wide range of values in both pollen and vegetation data, and is expected as an intermediate pollen producer. For practical reasons, most studies in Europe included Poaceae or P. sylvestris and utilised them as reference taxa (summarised by Mazier et al., 2012). For this reason, in Figure 4, we display our data using these reference taxa and compare them with all other existing studies. However, Pinus and Poaceae probably represent two extremes in the range of absolute pollen productivity in the BNP, so, in our case, a better intermediate pollen producer would have been Carpinus. Unfortunately, only two other European studies have produced PPE values for this taxon (Soepboer et al., 2007; Sugita et al., 1999) drastically limiting potential comparison. Nevertheless, the very high ratio between high and low pollen producers is independent from the choice of a reference taxon.

Figure 4.

Comparison of the pollen productivity estimates obtained in this study with those from the rest of Europe. Full and empty squares are this study’s ERV 1 and ERV 3 results, the dot is the recommended productivity from Mazier et al. (2012), the triangles are the productivity estimates in all other studies from temperate Europe (see Table 1 for references). Note that sometimes triangles overlap and that y-axes have different scales. See text for the species names.

Media sampled and vegetation survey

Methodology has been highlighted by some authors as a potential source of result discrepancy between studies (e.g. Broström et al., 2008; Bunting et al., 2013; Bunting and Hjelle, 2010; Hellman et al., 2008; Hjelle and Sugita, 2012; Theuerkauf et al., 2012), in particular, first, the media sampled, second, the models for pollen dispersal and number of years of pollen accumulation found in mosses or lake sediments and, third, the way vegetation is surveyed.

Most studies show that different media are more or less effective in trapping certain pollen types; however, it is also emerging that variation between pollen assemblages collected from different media is not always significant (e.g. Broström et al., 2008; Giesecke and Fontana, 2008; Lisitsyna et al., 2012; Pardoe et al., 2010; Räsänen et al., 2004; Wilmshurst and McGlone, 2005). Some herb taxa show lower PPE values with lake sediment sampling than with moss polster sampling and this has been attributed to the poor dispersal of certain pollen types (Broström et al., 2008). Results from northern Europe have also shown the same pattern (Lisitsyna et al., 2012), and this may impact the estimation of pollen productivity values. In addition, occasional differences in PPE may arise as a result of strong over-representation of some taxa in the local littoral vegetation (e.g. Alnus spp., Filipendula) which, if under-estimated in the vegetation survey, may produce higher than expected PPE from lake sediments (Broström et al., 2008).

Another consideration regarding the media is that mosses may better retain some pollen types than others. Several studies indicate that bisaccate pollen, especially Pinus, is usually much better represented in moss polsters than in Tauber traps (Caramiello et al., 1991; Lisitsyna et al., 2012; Pardoe et al., 2010; Vermoere et al., 2000). This bias was also noted in the BNP where seven pairs combining moss samples and Tauber traps (averaged 2-year pollen data) showed that Pinus proportions in moss polsters were in average two times higher (Zimny, 2014). These differences are usually explained by the specific structures of the moss cushions which results in different effectiveness in trapping and then preserving pollen of various size and form (Joosten and De Klerk, 2007). Pollen deposition in mosses may also be biased by occasional extreme meteorological events (Sjögren et al., 2006), or pollen addition by insects (Bunting et al., 2013) which induce variation in the PPE of some taxa and not others.

Because of the annual variability in pollen deposition, PPE calculations should be made on samples containing several years of accumulation (Bunting et al., 2013). However, in most studies, the number of years in both moss polsters and sediment samples cannot be determined with great confidence (Hjelle and Sugita, 2012; Pardoe et al., 2010). In our study, we addressed the above issues by sampling mosses forming thick mats, including the lower brown parts, thus representing as many years as possible. Moreover, we sampled similar types of forest floor mosses (pleurocarps or Sphagnum) for all our samples so as to avoid large discrepancies in pollen rain accumulation time between samples (ESM Table S1, available online).

The deposition environment may also be a source of uncertainty. Although pt-s distance-weighting better takes into account the different flight abilities of different pollen types, the pt-s model currently in use (Prentice, 1985; Sugita et al., 1999) only reflects pollen dispersal from a ground level source and above the forest canopy (Jackson and Lyford, 1999; Theuerkauf et al., 2012). As a result, we preferred 1/d (inverse distance) distance-weighting for our results, a robust methods that do not make any assumption regarding the deposition environment. The forest in the BNP is unique for its complex structure and it remains unknown how this may have affected pollen deposition. More research regarding the penetration of pollen rain under the canopy will be required in order to better understand the nature of background pollen rain in our samples.

Methods used to survey the plant species abundances around pollen sampling sites may influence estimation of pollen productivity (Broström et al., 2008; Bunting et al., 2013), and this has been demonstrated for various herbs and Ericaceae species (Bunting and Hjelle, 2010). However, to avoid any discrepancies in this respect, we followed the methodology used in other similar studies (Mazier et al., 2008; Bunting et al., 2013) for, respectively, 10–100 m and 0–10 m away from the sampling point. In addition, the relevant source area of pollen in our research (400 m) was similar to other research based on moss polsters (Broström et al., 2004; Von Stedingk et al., 2008) giving less weight to the field survey data in the estimation of pollen productivity. Our estimates of tree cover slightly diverge from most studies of PPE in that we explicitly excluded immature trees. This follows the recommendation of Matthias et al. (2012) who found a significant difference in PPE when immature trees are included or excluded from calculations. However, this difference remains too small to explain the discrepancy between our results and those of other studies.

Finally, another potential source of uncertainty comes from the fact that some of our sites were relatively close to each other, that is, there could be a problem of spatial autocorrelation. Because of extremely difficult access in many parts of the BNP, in our study, sampling location was a compromise between safety, needs for adequate moss cushions and spatial distribution. It is not clear how much this factor could have impact on PPE calculations and how it may have influenced our relative source area of pollen (Twiddle et al., 2012).

We acknowledge that some of the discrepancies in PPE results from different studies can be attributed to the variety of vegetation survey methodology used. However, considering all the above methodological issues, it can be inferred that the much higher ratio between high and low pollen producers observed in the BNP cannot be explained on methodological grounds only.

The influence of extrinsic abiotic factors such as climate

Plant species experience limited fitness towards the extreme edge of their distribution range because of sub-optimal climatic conditions, and, as a result, produce pollen less abundantly (Barnekow et al., 2007; Sjögren et al., 2006). All the eight tree species dealt with here are more or less at the centre of their distribution range (Jalas and Suominen, 1973, 1976), and it can be expected that they experience optimal climatic conditions for growth and flowering. Potentially, this might result in higher pollen productivity of deciduous trees in Białowieża Forest than in northern Europe.

Climate may not only affect annual pollen production but also its year-to-year variation, including biological cycles responsible for frequency of years with high and low pollen production (Nielsen et al., 2010). Three of the taxa studied here (e.g. Betula, Alnus and Quercus) have a more or less distinct biennial or triennial alternating pattern in many aerobiological stations in Europe, however, in some others such regularities are not observed (Spieksma et al., 2003). In Białowieża Forest, Carpinus and Tilia also show fluctuation in intensity of flowering (Pawlaczyk, 2009). Annual airborne pollen counts in high pollen seasons may be more than 10 times higher than in low ones as evidenced for Betula (Latałowa et al., 2002) or Quercus (Grewling et al., 2014). Therefore, presence or absence of the alternating patterns and frequency of high and low pollen years might be of importance when considering PPE. This potential issue may have affected Picea abies in particular, a tree showing a large variation of pollen production from year to year. This may explain why our data for this taxon could not be modelled and require further investigation. Also, some pollen types may comprise different species in different climatic regions of Europe. For instance, the Quercus pollen recorded in our samples comes from the local native Q. robur. However, in other studies (Soepboer et al., 2007; Theuerkauf et al., 2012), both Q. robur and Q. petraea were included for the calculation of Quercus PPE. There is a similar situation with A. incana, a common tree in central Europe that produces pollen morphologically similar to that of A. glutinosa but that is scarce in the BNP. However, it is sometimes included in PPE of Alnus spp. (Poska et al., 2011). We are not aware of any direct evidence specifically demonstrating that these pairs of species (Q. robur–Q. petraea, A. glutinosa–A. incana, B. pendula–B. pubescens, T. platyphyllos–T. cordata) have similar or different pollen productivity, but it cannot be excluded that taxonomic and genetic diversity within the geographic range of a taxa might be among the important factors deciding on its different response to climate variability (Hjelle and Sugita, 2012). However, because the existing data from different parts of Europe show PPE values within a similar range, there is no reason to believe that the tree species assemblage present in the climatic condition of Białowieża Forest is sufficient to explain the unique characteristics of our results.

Age of trees

Matthias et al. (2012) suggest that in cultural landscapes where forests are managed, PPEs are likely to be lower than expected from a closed-canopy old-growth forest. They argue that trees reach sexual maturity only after 10–50 years depending on the species and hardly produce any pollen until then. As a result, young tree plantations would not produce pollen, yet this fraction of forest cover tends to be included in datasets for the calculation of PPE. The result would be an over-estimation of the tree species’ surface cover, resulting in an under-estimation of PPE values.

The old-growth forest of the BNP is also a dynamic ecosystem where young trees are an integral portion of the vegetation cover (Faliński, 1988; Sokołowski, 1993) and this can be extended to the general dynamics of the forest (Faliński, 1988). There is an ample body of evidence showing that in Białowieża Forest, tree recruitment is increasing (Miścicki, 2012), and being constantly shaped by natural disturbance factors, such as browsing (Bobiec et al., 2011; Kuijper et al., 2010a, 2010b), disease or wind throws (Bobiec, 2002). However, in our study, the impact of immature trees on PPE calculation was minimised due to their exclusion during the field inventory. Moreover, even if our estimation of mature/immature trees in the plant cover was not in line with previous studies (except Matthias et al., 2012), we can assume that proportion of young trees in the BNP is much lower than in cultural landscapes – tall Pinus, Quercus and Alnus over 100 years old (the highest pollen producers in this study) are very common in the forest. It is also important to stress that in the deciduous forest communities in the BNP P. sylvestris, a key species in our study, occurs in the form of veteran trees (more than 300-year old) reaching up to 45 m in height (Faliński, 1977), but the younger generation of this tree is almost absent (Bobiec, 2012). The very high Pinus PPE found at our study sites may be partly explained by the presence of such trees in the deciduous forest we sampled.

Canopy structure and light limitation

Our results may be explained by the multi-layered forest and highly diversified structure occurring in the BNP (Bobiec, 2012; Faliński, 1986). The highest pollen producers in our study (Pinus sylvestris, Q. robur and A. glutinosa) reach the upper forest layer (Faliński, 1977) where their large crowns are fully exposed to sunlight enabling good flowering conditions and pollen dispersal. In addition, these taxa have relatively low pollen fall speed (Sugita et al., 1999), which means that they may be over-represented in the pollen rain in relation to the surveyed vegetation (Theuerkauf et al., 2012). Betula, a light-demanding tree, is also among the highest pollen producers in our data. In the BNP, Betula is frequent in the forest gaps (B. pendula) and is common on peat bogs (B. pubescens) where it grows in full light and is exposed to wind – good conditions for high pollen production and dispersal. Again, we cannot exclude that its pollen is over-represented because pollen dispersal in those conditions is higher than expected.

In Białowieża Forest, Corylus avellana is restricted to very shaded forest undergrowth and never reaches the canopy (Sokołowski, 1993); therefore, light limitation is likely to reduce its reproductive fitness. This is in direct contrast with open or semi-open habitats promoting flowering, such as hedgerows and forest edges where it is generally found in cultural landscapes (e.g. Ellenberg, 1988). We suggest that this might well account for the much lower PPE of Corylus than hitherto assumed. The same may concern T. cordata. Although some of the T. cordata trees in Białowieża Forest reach the canopy level, many are located under the canopy, where they do not reach their full potential for pollen production (Bobiec, 2012; Keczyński, 2005). Carpinus is among the low pollen producers but its PPE is higher than those of Poaceae, Corylus and Tilia. When calculated in relation to Poaceae (ERV 3), the value (4.48) is very similar to that calculated for the Swiss Plateau (4.56) (Soepboer et al., 2007).

In the BNP, Poaceae flowering is also possibly reduced by the low light levels in the forest understorey in comparison to situations where agriculture prevails. It has been highlighted that Poaceae comprises numerous species and as a result, their PPEs from different areas may be different (Broström et al., 2008). In fact, many of the Poaceae taxa found in Białowieża Forest during our survey are specialists of forest environments (Poa nemoralis, Festuca altissima, Deschampsia cespitosa subsp. parviflora, Dactylis aschersoniana, Brachypodium sylvaticum and Melica nutans) that are expected to be lower pollen producers than grass species growing in full light conditions outside forest. In addition, other more ubiquitous Poaceae recorded during our survey (Phragmites australis, Molinia caerulea, Festuca ovina, Deschampsia flexuosa, Danthonia decumbens, Calamagrostis arundinacea) are expected to show reduced fitness as a result of the shading from the canopy.

Grazing and browsing by large herbivores

Large grazers directly impact on herbaceous plant flowering, either by physically suppressing flowering in a continuous way (Groenman-van Waateringe, 1993; Vera, 2000) or by enhancing the production of inflorescences and lengthening the flowering season, a phenomenon known as over-compensation (e.g. Massad, 2013). However, in Białowieża Forest, grazing is unlikely to explain the low Poaceae PPE for three reasons. First, we did not find any signs of grazing during our survey and Poaceae cover is low in the quadrats near our sampling points (mean = 3.1%, n = 306), that is, it never consists in a grazing lawn likely to sustain regular grazing. Second, ungulates in Białowieża Forest are primarily browsers and tend to graze and browse preferentially in forest gaps (Kuijper et al., 2009) not covered in our study. Third, European bison, the only large herbivore truly adapted to grazing within the local faunal assemblage, occurs in a low density of 0.45 individuals km⁻² (Kuijper et al., 2010a); therefore, its impact on grass flowering within the BNP may be neglected. Browsing has a pervasive effect on the forest composition in Białowieża, where herbivores literally select the canopy trees of tomorrow (Kuijper et al., 2010a). However, they can only reach young seedlings and saplings and in this respect, browsing herbivores do not impact pollen productivity in a direct way in the short term.

Nutrient cycling

Nitrogen availability is an important factor stimulating higher pollen productivity in some species. Long-term manipulation field experiments in the United Kingdom show that flowering can increase up to three times in woody plants with nitrogen addition (Phoenix et al., 2012). Similar results have been obtained for North American Quercus spp. (Callahan et al., 2007). The current nitrogen levels in the BNP area are increasing with a large amount of dead wood decomposition (Paluch, 2001) and the deposition of atmospheric nitrogen at the current rate of 11 kg ha⁻¹ year⁻¹ (Malzahn, 2009). In addition, the long-term transformation of some conifer forests into oak–linden–hornbeam (Bobiec, 2012) is likely to change nitrogen cycling at those locations. On the whole, it is not known whether the nitrogen availability levels currently experienced by trees in Białowieża Forest may differ from those in cultural landscapes but this is likely. However, this is a subject that deserves further consideration.

Also, relatively high density of ungulates in the BNP may be of importance, because large herbivores redistribute nitrogen in a more heterogeneous way (Augustine and Frank, 2001; Bump et al., 2009; Hobbs, 2006; Murray et al., 2013) and modify the cycling of nutrients (reviewed in Pastor et al., 2006). It is difficult to assess the impact of nutrient manipulation by large vertebrates on our results; however, this is another factor that probably needs further consideration.

Potential implications of the PPEs from Białowieża Forest for palaeoenvironmental reconstructions and conservation science

PPEs are the core element in quantitative reconstructions of past vegetation cover based on mechanistic models developed by Sugita (2007a, 2007b; Sugita et al., 2010); one underlying assumption of these models is that PPEs are constant in space and time.However, our results reinforce the idea that pollen productivity can vary in response to changes in the prevailing environmental settings and we present for the first time a quantification of this variability, likely induced by differences in tree cover and canopy structure. Our results indicate that it would be constructive to use, in parallel consensus PPEs calculated mostly from cultural landscapes (Mazier et al., 2012) and PPEs from an old-growth forest, when running the landscape reconstruction models. We suspect that our PPEs will result in interpreting higher proportions of open-land than hitherto estimated, for example, for the mid-Holocene landscapes in the temperate zone of lowland Europe (Nielsen et al., 2012; Trondman et al., 2015).

Tree cover during the Holocene is an important parameter in a number of studies on climate change modelling (Gaillard et al., 2010; Strandberg et al., 2014; Trondman et al., 2015). In fact, vegetation is an inherent part of climate systems influencing circulation of energy, water and greenhouse gases between land and the atmosphere. Therefore, climate change studies will benefit from a more informed choice of PPEs to quantify past changes in vegetation cover.

As there is a growing consensus about the importance of vegetation history for conservation science and ecosystem management (Froyd and Willis, 2008; Jeffers et al., 2015), palynological tools need continuing improvement. Applying our new set of PPE to pollen-based quantitative reconstructions of past vegetation will help to better understanding vegetation cover in temperate Europe prior to development of agriculture. Landscape openness and the drivers shaping vegetation structure for this period is the focus of a long-standing debate that has a direct impact on different nature conservation concepts (e.g. Bradshaw et al., 2003; Mitchell, 2005; Vera, 2000).

Conclusion

To date, pollen productivity has only been estimated in temperate Europe from cultural landscapes where human activities, such as agriculture, industrialisation and urbanisation are a preponderant source of disturbance. We found that pollen productivity in the closed-canopy old-growth forest of BNP (where disturbance by human activities is minimal) was different from that measured in cultural landscapes. In fact, the ratio of PPEs between high producers (Pinus sylvestris and Quercus robur) and low producers (Poaceae and Corylus avellana) is on an average six times greater in the BNP than across European cultural landscapes. We discuss several potential factors likely to explain our distinctive results, including the methodology used and the environmental settings in the BNP.

We conclude that our results cannot be explained on methodological grounds only. We followed standard methods for the vegetation survey and moss polsters, as recommended by previous studies of cultural landscapes. In addition, we excluded from our vegetation dataset trees not mature enough to produce pollen, a potential bias highlighted by Matthias et al. (2012).

From the environmental factors considered, we proposed that light availability is the most important. This is the direct result of the forest structure and age, a unique characteristic of the BNP forest. All low pollen producers in our data (Corylus, Poaceae and Tilia) grow in relatively shaded situations and do not receive sufficient light for their pollen production to be as abundant as in cultural landscapes. In addition, dense, multi-layered forest may limit the dispersal of their pollen, while high producers (Pinus sylvestris and Quercus robur) are tall trees that reach the canopy where they receive sufficient light for abundant pollen production and where pollen gets freely dispersed.

Besides light, we highlight that enhanced nitrogen availability induces a significant increase in pollen production in some species and that nitrogen cycling depends on atmospheric pollution levels, dead wood decomposition and spatial redistribution by large herbivores among other factors. We believe this is an aspect of pollen productivity that deserves more attention in the future.

The PPE for high pollen producers relative to Pinus and PPE for low pollen producers relative to Poaceae are comparable with estimates previously reported in other studies, but none of these reference taxa give consistent results for the whole range of taxa used in this study. This is an interesting finding which not only underlines specific characters of the data from an old-growth forest, but also indicates that more attention should be drawn to the role of a reference taxa used for relative PPE calculation.

Our results reinforce the idea that pollen productivity can vary depending on the prevailing environmental settings, ecological conditions and disturbance regimes. We present for the first time a quantification of this variability and suggest our PPE results to be used in parallel to consensus PPE from cultural landscapes (Mazier et al., 2012) when interpreting pollen assemblages potentially coming from closed-canopy forest. Applying the two PPE sets should help to test hypothesis, for instance, regarding the interpretation of pollen spectra from the mid-Holocene forest maximum at the heart of the ‘Vera debate’ (Mitchell, 2005) or different scenarios of human-induced changes in land cover used for climate modelling (Marquer et al., 2014; Trondman et al., 2015). These improved insights into vegetation reconstruction are necessary to better understand the drivers of change that occurred in the past and the consequences they had on land cover and climate.

Footnotes

Acknowledgements

We thank the Białowieża National Park for permission to carry out the field work and Georgina Southon and Grzegorz Zimny for assistance in the field work. We are grateful to Shinya Sugita for help with the application of his method and computer program, to Pim van der Knapp for advice regarding sampling method, to Wiesław Klimiuk for his GIS work and to the Mammal Research Institute (Polish Academy of Sciences) in Białowieża, and especially to Tomek Samojlik, for technical support during our stay in Białowieża. Finally, we would like to thank the two reviewers for their insightful comments and recommendations.

Funding

AGB was supported by a studentship from the UK Natural Environment Research Council; This study is a part of the project funded by the Ministry of Science and Higher Education in Poland (grant code N N 305 167839).

References

Abraham

Kozáková

(2012) Relative pollen productivity estimates in the modern agricultural landscape of Central Bohemia (Czech Republic). Review of Palaeobotany and Palynology 179: 1–12.

Augustine

Frank

(2001) Effects of migratory grazers on spatial heterogeneity of soil nitrogen properties in a grassland ecosystem. Ecology 82(11): 3149–3162.

Barnekow

Loader

Hicks

. (2007) Strong correlation between summer temperature and pollen accumulation rates for Pinus sylvestris, Picea abies and Betula spp. in a high-resolution record from northern Sweden. Journal of Quaternary Science 22(7): 653–658.

Beug

(2004) Leitfaden der Pollenbestimmung fur Mitteleuropa und angrenzende Gebiete. München: Verlag Friedrich Pfeil.

Birks

HJB

(2005) Mind the gap: How open were European primeval forests? Trends in Ecology & Evolution 20(4): 154–156.

Bobiec

(2002) Białowieża Primeval Forest. International Journal of Wilderness 8(3): 33.

Bobiec

(2012) Białowieża Primeval Forest as a remnant of culturally modified ancient forest. European Journal of Forest Research 131(5): 1269–1285.

Bobiec

Jaszcz

Wojtunik

(2011) Oak (Quercus robur L.) regeneration as a response to natural dynamics of stands in European hemiboreal zone. European Journal of Forest Research 130(5): 785–797.

Boczoń

(2006) Characteristics of thermal and pluvial conditions in the Białowieża Primeval Forest between 1950 and 2003. Leśne Prace Badawcze 1: 57–72.

10.

Bradshaw

RHW

Hannon

Lister

(2003) A long-term perspective on ungulate – Vegetation interactions. Forest Ecology and Management 181: 267–280.

11.

Broström

Sugita

Gaillard

M-J

(2004) Pollen productivity estimates for the reconstruction of past vegetation cover in the cultural landscape of southern Sweden. The Holocene 14(3): 368–381.

12.

Broström

Nielsen

Gaillard

M-J

. (2008) Pollen productivity estimates of key European plant taxa for quantitative reconstruction of past vegetation: A review. Vegetation History and Archaeobotany 17(5): 461–478.

13.

Buckland

(2005) Palaeoecological evidence for the Vera hypothesis. In: Hodder

(ed.) Large Herbivores in the Wildwood and Modern Naturalistic Grazing Systems (English nature research report 648). Peterborough: English Nature, pp. 62–116.

14.

Bump

Webster

Vucetich

. (2009) Ungulate carcasses perforate ecological filters and create hotspots in forest biogeochemical herbaceous layers allowing trees a competitive advantage. Ecosystems 12(6): 996–1007.

15.

Bunting

Hjelle

(2010) Effect of vegetation data collection strategies on estimates of relevant source area of pollen (RSAP) and relative pollen productivity estimates (relative PPE) for non-arboreal taxa. Vegetation History and Archaeobotany 19(4): 365–374.

16.

Bunting

Armitage

Binney

. (2005) Estimates of ‘relative pollen productivity’ and ‘relevant source area of pollen’ for major tree taxa in two Norfolk (UK) woodlands. The Holocene 15(3): 459–465.

17.

Bunting

Farrell

Broström

. (2013) Palynological perspectives on vegetation survey: A critical step for model-based reconstruction of Quaternary land cover. Quaternary Science Reviews 82: 41–55.

18.

Callahan

Del Fierro

Patterson

. (2007) Impacts of elevated nitrogen inputs on oak reproductive and seed ecology. Global Change Biology 14(2): 285–293.

19.

Caramiello

Siniscalco

Piervittori

(1991) The relationship between vegetation and pollen deposition in soil and in biological traps. Grana 30(2): 291–300.

20.

Czerepko

(2008) A long-term study of successional dynamics in the forest wetlands. Forest Ecology and Management 255: 630–642.

21.

Davis

(2000) Palynology after Y2K – Understanding the source area of pollen in sediments. Annual Review of Earth and Planetary Sciences 28: 1–18.

22.

Eggenberg

Möhl

(2008) Flora Vegetativa. Lausanne: Rossolis.

23.

Ellenberg

(1988) Vegetation Ecology of Central Europe. Cambridge: Cambridge University Press.

24.

Faliński

(1977) Research on vegetation and plant population dynamics conducted by Białowieża Geobotanical Station of the Warsaw University in the Białowieża Primeval Forest and in the environ (1952–1977). Phytocoenosis 6(1–2): 1–132.

25.

Faliński

(1978) Uprooted trees, their distribution and influence in the primeval forest biotope. Vegetatio 38(3): 175–183.

26.

Faliński

(1986) Vegetation Dynamics in Temperate Lowland Primeval Forests: Ecological Studies in Białowieża Forest. Dordrecht: Dr. W. Junk Publishers.

27.

Faliński

(1988) Succession, regeneration and fluctuation in the Białowieża Forest (NE Poland). Vegetatio 77: 115–128.

28.

Froyd

Willis

(2008) Emerging issues in biodiversity and conservation management: The need for palaeoecological perspective. Quaternary Science Reviews 27: 1723–1732.

29.

Gaillard

M-J

Sugita

Bunting

. (2008) The use of modelling and simulation approach in reconstructing past landscapes from fossil pollen data: A review and results from the POLLANDCAL network. Vegetation History and Archaeobotany 17(5): 419–443.

30.

Gaillard

M-J

Sugita

Mazier

. (2010) Holocene land-cover reconstructions for studies on land cover-climate feedbacks. Climate of the Past 6(4): 483–499.

31.

Giesecke

Fontana

(2008) Revisiting pollen accumulation rates from Swedish lake sediments. The Holocene 18(2): 293–305.

32.

Grewling

Jackowiak

Smith

(2014) Variations in Quercus sp. pollen seasons (1996–2011) in Poznań, Poland, in relation to meteorological parameters. Aerobiologia 30: 149–159.

33.

Groenman-van Waateringe

(1993) The effects of grazing on the pollen production of grasses. Vegetation History and Archaeobotany 2(3): 157–162.

34.

Hall

(2008) A comparative analysis of the habitat of the extinct aurochs and other prehistoric mammals in Britain. Ecography 31: 187–190.

35.

Hellman

Gaillard

M-J

Broström

. (2008) The REVEALS model, a new tool to estimate past regional plant abundance from pollen data in large lakes: Validation in southern Sweden. Journal of Quaternary Science 23(1): 21–42.

36.

Hicks

Tinsley

Pardoe

. (1999) European Pollen Monitoring Programme. Supplement to the Guidelines. Oulu: University of Oulu.

37.

Hijmans

Cameron

Parra

. (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25(15): 1965–1978.

38.

Hjelle

(1998) Herb pollen representation in surface moss samples from mown meadows and pastures in western Norway. Vegetation History and Archaeobotany 7: 79–96.

39.

Hjelle

Sugita

(2012) Estimating pollen productivity and relevant source area of pollen using lake sediments in Norway: How does lake size variation affect the estimates? The Holocene 22(3): 313–324.

40.

Hobbs

(2006) Large herbivores as sources of disturbance in ecosystems. In: Danell

Bergström

Duncan

. (eds) Large Herbivore Ecology, Ecosystem Dynamics and Conservation. Cambridge: Cambridge University Press, pp. 261–288.

41.

Jackson

Lyford

(1999) Pollen dispersal models in Quaternary plant ecology: Assumptions, parameters, and prescriptions. The Botanical Review 65(1): 39–75.

42.

Jalas

Suominen

(1973) Atlas Florae Europaeae: Distribution of Vascular Plants in Europe, 2: Gymnospermae (Pinaceae to Ephedraceae). Helsinki: The Committee for Mapping the Flora of Europe and Societas Biologica Fennica Vanamo.

43.

Jalas

Suominen

(1976) Atlas Florae Europaeae: Distribution of Vascular Plants in Europe, 3: Salicaceae to Balanophoraceae. Helsinki: The Committee for Mapping the Flora of Europe and Societas Biologica Fennica Vanamo.

44.

Jeffers

Nogué

Willis

(2015) The role of palaeoecological records in assessing ecosystem services. Quaternary Science Reviews 112: 17–32.

45.

Jędrzejewski

Spaedtke

Kamler

. (2006) Group size dynamics of red deer in Bialowieża Primeval Forest, Poland. Journal of Wildlife Management 70(4): 1054–1059.

46.

Joosten

De Klerk

(2007) In search of finiteness: The limits of fine-resolution palynology of Sphagnum peat. The Holocene 17(1): 1023–1031.

47.

Keczyński

(2005) Zmiany struktury drzewostanów wykształconych na glebach hydrogenicznych – na przykładzie wybranych powierzchni badawczych w Białowieskim Parku Narodowym. Leśne Prace Badawcze 4: 87–102.

48.

Kuijper

DPJ

Cromsigt

JPGM

Churski

. (2009) Do ungulates preferentially feed in forest gaps in European temperate forest? Forest Ecology and Management 258(7): 1528–1535.

49.

Kuijper

DPJ

Cromsigt

JPGM

Jędrzejewska

. (2010a) Bottom-up versus top-down control of tree regeneration in the Białowieża Primeval Forest, Poland. Journal of Ecology 98(4): 888–899.

50.

Kuijper

DPJ

Jędrzejewska

Brzeziecki

. (2010b) Fluctuating ungulate density shapes tree recruitment in natural stands of the Białowieża Primeval Forest, Poland. Journal of Vegetation Science 21(6): 1082–1098.

51.

Kwiatkowski

Gajko

(2009) Opracowanie i redakcja map BPN: Drzewostany. In: Okołów

Karaś

Bołbot

(eds) Białowieski Park Narodowy. Poznać – Zrozumieć – Zachować. Białowieża: Białowieski Park Narodowy, pp. 48–49.

52.

Latałowa

Miętus

Uruska

(2002) Seasonal variations in the atmospheric Betula pollen count in Gdańsk (Southern Baltic coast) in relation to meteorological parameters. Aerobiologia 18: 33–43.

53.

Latałowa

Zimny

Jędrzejewska

. (2015) Białowieża Primeval Forest – 2000-year interplay of environmental and cultural forces in Europe’s best preserved temperate woodland. In: Kirby

Watkins

(eds) Europe’s Changing Woods and Forests: From Wildwood to Cultural Landscapes. Wallingford: CABI, pp. 243–264.

54.

Lisitsyna

Hicks

Huusko

(2012) Do moss samples, pollen traps and modern lake sediments all collect pollen in the same way? A comparison from the forest limit area of northernmost Europe. Vegetation History and Archaeobotany 21(3): 187–199.

55.

Malzahn

(2009) Bio-monitoring of the forest environment of the Białowieża Primeval Forest. Ochrona Środowiska i Zasobów Naturalnych 40: 439–447.

56.

Malzahn

Kwiatkowski

Pierzgalski

(2009) II: Przyroda nieożywiona. In: Okołów

Karaś

Bołbot

(eds) Białowieski Park Narodowy: Poznać, Zrozumieć, Zachować. Białowieża: Białowieski Park Narodowy, pp. 18–36.

57.

Marquer

Gaillard

M-J

Sugita

. (2014) Holocene changes in vegetation composition in northern Europe: Why quantitative pollen-based vegetation reconstructions matter. Quaternary Science Reviews 90(15): 199–216.

58.

Massad

(2013) Ontogenetic differences of herbivory on woody and herbaceous plants: A meta-analysis demonstrating unique effects of herbivory on the young and the old, the slow and the fast. Oecologia 172(1): 1–10.

59.

Matthias

Nielsen

Giesecke

(2012) Evaluating the effect of flowering age and forest structure on pollen productivity estimates. Vegetation History and Archaeobotany 21(6): 471–484.

60.

Mazier

Broström

Gaillard

M-J

. (2008) Pollen productivity estimates and relevant source area of pollen for selected plant taxa in a pasture woodland landscape of the Jura Mountains (Switzerland). Vegetation History and Archaeobotany 17(5): 479–495.

61.

Mazier

Gaillard

M-J

Kuneš

. (2012) Testing the effect of site selection and parameter setting on REVEALS-model estimates of plant abundance using the Czech Quaternary Palynological Database. Review of Palaeobotany and Palynology 187: 38–49.

62.

Mikusińska

Zawadzka

Samojlik

. (2013) Quantifying landscape change during the last two centuries in Białowieża Primeval Forest. Applied Vegetation Science 16(2): 217–226.

63.

Miścicki

(2012) Structure and dynamics of temperate lowland natural forest in the Białowieża National Park, Poland. Forestry 85(4): 473–483.

64.

Mitchell

FJG

(2005) How open were European primeval forests? Hypothesis testing using palaeoecological data. Journal of Ecology 93: 168–177.

65.

Moore

Webb

Collinson

(1991) Pollen Analysis. 2nd Edition. Oxford: Blackwell Scientific Publications.

66.

Murray

Webster

Bump

(2013) Broadening the ecological context of ungulate-ecosystem interactions: The importance of space, seasonality, and nitrogen. Ecology 94(6): 1317–1326.

67.

Nielsen

(2004) Modelling pollen sedimentation in Danish lakes at c.ad 1800: An attempt to validate the POLLSCAPE model. Journal of Biogeography 31(10): 1693–1709.

68.

Nielsen

Giesecke

Theuerkauf

. (2012) Quantitative reconstructions of changes in regional openness in north-central Europe reveal new insights into old questions. Quaternary Science Reviews 47: 131–149.

69.

Nielsen

Møller

Giesecke

. (2010) The effect of climate conditions on inter-annual flowering variability monitored by pollen traps below the canopy in Draved Forest, Denmark. Vegetation History and Archaeobotany 19: 309–323.

70.

Niklasson

Zin

Zielonka

. (2010) A 350-year tree-ring fire record from Białowieża Primeval Forest, Poland: Implications for Central European lowland fire history. Journal of Ecology 98(6): 1319–1329.

71.

Paluch

(2001) Wpływ drewna martwego na zasobność gleby określoną za pomocą roślinności runa w Białowieskim Parku Narodowym. Parki Narodowe i Rezerwaty Przyrody 20(4): 15–25.

72.

Pardoe

Giesecke

van der Knaap

. (2010) Comparing pollen spectra from modified Tauber traps and moss samples: Examples from a selection of woodlands across Europe. Vegetation History and Archaeobotany 19(4): 271–283.

73.

Parsons

Prentice

(1981) Statistical approaches to R-values and the pollen-vegetation relationship. Review of Palaeobotany and Palynology 32: 127–152.

74.

Pastor

Cohen

Hobbs

(2006) The roles of large herbivores in ecosystem nutrient cycling. In: Danell

Bergström

Duncan

. (eds) Large Herbivore Ecology, Ecosystem Dynamics and Conservation. Cambridge: Cambridge University Press, pp. 289–325.

75.

Pawlaczyk

(2009) III: Zbiorowiska Leśne. In: Okołów

Karaś

Bołbot

(eds) Białowieski Park Narodowy: Poznać- Zrozumieć-Zachować. Białowieża: Białowieski Park Narodowy, pp. 38–58.

76.

Peterken

(1996) Natural Woodland: Ecology and Conservation in Northern Temperate Regions. Cambridge: Cambridge University Press.

77.

Phoenix

Emmett

Britton

. (2012) Impacts of atmospheric nitrogen deposition: Responses of multiple plant and soil parameters across contrasting ecosystems in long-term field experiments. Global Change Biology 18(4): 1197–1215.

78.

Pigott

(1975) Natural regeneration of Tilia cordata in relation to forest-structure in the forest of Białowieża, Poland. Philosophical Transactions of the Royal Society B: Biological Sciences 270(904): 151–179.

79.

Poska

Meltsov

Sugita

. (2011) Relative pollen productivity estimates of major anemophilous taxa and relevant source area of pollen in a cultural landscape of the hemi-boreal forest zone (Estonia). Review of Palaeobotany and Palynology 167(1–2): 30–39.

80.

Prentice

(1985) Pollen representation, source area, and basin size: Toward a unified theory of pollen analysis. Quaternary Research 23(1): 76–86.

81.

Prentice

Parsons

(1983) Maximum likelihood linear calibration of pollen spectra in terms of forest composition. Biometrics 39(4): 1051–1057.

82.

Punt

. (1976–2003) The Northwest European Pollen Flora, vols I–VIII. Amsterdam: Elsevier.

83.

Räsänen

Hicks

Odgaard

(2004) Pollen deposition in mosses and in a modified ‘Tauber trap’ from Hailuoto, Finland: What exactly do the mosses record? Review of Palaeobotany and Palynology 129(1–2): 103–116.

84.

Räsänen

Suutari

Nielsen

(2007) A step further towards quantitative reconstruction of past vegetation in Fennoscandian boreal forests: Pollen productivity estimates for six dominant taxa. Review of Palaeobotany and Palynology 146(1–4): 208–220.

85.

Samojlik

(2005) Conservation and Hunting: Białowieża Forest in the Time of Kings. Białowieża: Zakład Badania Ssaków PAN.

86.

Sjögren

van Leeuwen

JFN

van der Knaap

. (2006) The effect of climate variability on pollen productivity, AD 1975–2000, recorded in a Sphagnum peat hummock. The Holocene 16(2): 277–286.

87.

Smit

Kuijper

DPJ

Prentice

. (2012) Coarse woody debris facilitates oak recruitment in Białowieża Primeval Forest, Poland. Forest Ecology and Management 284: 133–141.

88.

Smith

Whitehouse

Bunting

. (2010) Can we characterise ‘openness’ in the Holocene palaeoenvironmental record? Modern analogue studies of insect faunas and pollen spectra from Dunham Massey deer park and Epping Forest, England. The Holocene 20(2): 215–229.

89.

Soepboer

Sugita

Lotter

. (2007) Pollen productivity estimates for quantitative reconstruction of vegetation cover on the Swiss Plateau. The Holocene 17(1): 65–77.

90.

Sokołowski

(1993) Phytosociological Characteristics of Forest Communities in the Białowieża National Park. Białowieża: Białowieża National Park, p. 190.

91.

Sokołowski

(2004) Lasy Puszczy Białowieskiej. Warszawa: Centrum Informacyjne Lasów Państwowych.

92.

Spieksma

FTM

Corden

Detandt

. (2003) Quantitative trends in annual totals of five common airborne pollen types (Betula, Quercus, Poaceae, Urtica, and Artemisia), at five pollen-monitoring stations in Western Europe. Aerobiologia 19: 171–184.

93.

Strandberg

Kjellström

Poska

. (2014) Regional climate model simulations for Europe at 6 k and 0.2 k yr BP: Sensitivity to changes in anthropogenic deforestation. Climate of the Past 10: 661–680.

94.

Sugita

(1993) A model of pollen source area for an entire lake surface. Quaternary Research 39(2): 239–244.

95.

Sugita

(1994) Pollen representation of vegetation in quaternary sediments: Theory and method in patchy vegetation. Journal of Ecology 82(4): 881–897.

96.

Sugita

(2007a) Theory of quantitative reconstruction of vegetation I: Pollen from large sites REVEALS regional vegetation composition. The Holocene 17(2): 229–241.

97.

Sugita

(2007b) Theory of quantitative reconstruction of vegetation II: All you need is LOVE. The Holocene 17(2): 243–257.

98.

Sugita

Gaillard

M-J

Broström

(1999) Landscape openness and pollen records: A simulation approach. The Holocene 9(4): 409–421.

99.

Sugita

Parshall

Calcote

. (2010) Testing the Landscape Reconstruction Algorithm for spatially explicit reconstruction of vegetation in northern Michigan and Wisconsin. Quaternary Research 74(2): 289–300.

100.

Svenning

(2002) A review of natural vegetation openness in north-western Europe. Biological Conservation 104: 133–148.

101.

Theuerkauf

Kuparinen

Joosten

(2012) Pollen productivity estimates strongly depend on assumed pollen dispersal. The Holocene 23(1): 14–24.

102.

Trondman

A-K

Gaillard

M-J

Mazier

. (2015) First pollen-based quantitative reconstructions of Holocene regional vegetation cover (plant functional types and land-cover types) in Europe suitable for climate modelling. Global Change Biology 21: 676–697.

103.

Tutin

Heywood

Burges

. (1964–1980) Flora Europaea. Cambridge: Cambridge University Press.

104.

Twiddle

Jones

Caseldine

. (2012) Pollen productivity estimates for a pine woodland in eastern Scotland: The influence of sampling design and vegetation patterning. Review of Palaeobotany and Palynology 174: 67–78.

105.

Vera

FWM

(2000) Grazing Ecology and Forest History. Wallingford: CABI.

106.

Vermoere

Vanhecke

Walelkens

. (2000) A comparison between modern pollen spectra of moss cushions and Cundill pollen traps. Grana 39: 146–158.

107.

Von Stedingk

Fyfe

Allard

(2008) Pollen productivity estimates from the forest – tundra ecotone in west-central Sweden: Implications for vegetation reconstruction at the limits of the boreal forest. The Holocene 18(2): 323–332.

108.

Wilmshurst

McGlone

(2005) Origin of pollen and spores in surface lake sediments: Comparison of modern palynomorph assemblages in moss cushions, surface soils and surface lake sediments. Review of Palaeobotany and Palynology 136(1–2): 1–15.

109.

Zimny

(2014) Późnoholoceńska historia roślinności Puszczy Białowieskiej. PhD Thesis, University of Gdańsk.