This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Assessing grassland restoration success: relative roles of seed additions and native ungulate activities
LEANNE M. MARTIN and BRIAN J. WILSEY
Iowa State University, Department of Ecology, Evolution and Organismal Biology, Ames, IA 50011, USA
Summary
1.
Grassland restorations often lack rare forb and grass species that are found in intactgrasslands. The possible reasons for low diversity include seed limitation, micrositelimitation and a combination of both. Native ungulates may create microsites forseedling establishment in tallgrass prairie restorations by grazing dominant species orthrough trampling activities, but this has never been tested in developing prairies.
2.
We experimentally tested for seed and microsite limitation in the largest tallgrassprairie restoration in the USA by adding rare forb and grass seeds in two trials insideand outside native ungulate exclosures. We measured seedling emergence because thisstage is crucial in recruiting species into a community. We also measured light, waterand standing crop biomass to test whether resource availability could help to explainseedling emergence rates.
3.
Ungulates increased light availability for each sampling time and also increasedabove-ground net primary productivity (ANPP) during summer.
4.
Seedling emergence of rare prairie forbs and grasses was consistently greater whenwe added seeds.
5.
Seedling emergence was conditionally greater with a combination of seed additionsand grazing, but grazing alone was unable to increase emergence.
6.
When ungulates increased seedling enhancement, the mechanism was partiallyassociated with increased water and light availability.
7.
Exotic and cosmopolitan weed seedling emergence was not affected by grazing.
8.
Synthesis and applications.
These results suggest that tallgrass prairie restorationsare primarily seed limited and that grazing alone may not be able to increase seedlingemergence of rare species without the addition of seeds. Therefore, adding seeds tograssland restorations may increase seedling emergence of rare species, and mimickingeffects of grazing may increase emergence when seeds are added.
Key-words
:
Bos bison
,
Cervus elaphus
, diversity, grazing, Iowa, net primary productivity,seedling emergence, tallgrass prairie
Journal of Applied Ecology
(2006)doi: 10.1111/j.1365-2664.2006.01211.x
Introduction
Ecosystem restoration is becoming a more commonway to increase native species habitat. Typically, resto-rations are attempted by adding seeds from nearbyremnants to a previously converted ecosystem (Sluis2002; Polley, Derner & Wilsey 2005). Seedlings ofmultiple species are expected to emerge, survive and
establish reproducing populations, and then popula-tions are expected to assemble into a communitysimilar to the original system. The seedling emergencestage is important in this process because it funnelsindividuals into the system. Contrary to expectations,restored ecosystems often have lower plant speciesrichness and diversity than their unaltered counterparts(Galatowitsch & van der Valk 1996; Martin, Moloney& Wilsey 2005; Polley, Derner & Wilsey 2005) andspecies richness has been observed to decline over timeinstead of increase as expected (Sluis 2002).
Typically, low species diversity is attributed to either(i) seed limitation or (ii) seedling microsite limitation
Correspondence: Leanne M. Martin, University ofNebraska at Omaha, Department of Biology, 6001 DodgeSt, Omaha, NE 68182, USA (fax 402 554 3532; [email protected]).
. 2004). The seed limitationhypothesis suggests that plant community richness anddiversity are limited by the species pool (Gough, Grace& Taylor 1994). Seed additions have increased speciesrichness and diversity of some native plant commun-ities and agriculturally improved grasslands (Pywell
et al
. 2002; Smith
et al
. 2002; but see Wilsey & Polley2003). If restorations are seed limited, then addingseeds of a large number of species should increasediversity and recruitment of rare species even in sys-tems with high dominance. Alternatively, the micrositehypothesis suggests that one or a few strongly dominantspecies suppress seedlings (Howe 2000; Sluis 2002;Camill
et al
. 2004). In this scenario, seeds or prop-agules are not limiting but seedlings fail to establishreproducing populations. Dominant grass patches areusually larger in restorations than in intact grassland(Derner
et al
. 2004). Dominance by C
4
grasses, whichcan occur as soon as 3 years after establishment, can beespecially high in the nutrient-rich environments thatcharacterize most restorations (Baer
et al
. 2002; Baer
et al
. 2004; Camill
et al
. 2004). Large grass canopiesand abundant litter can reduce light and water avail-ability, which are crucial to seedling survival (Fahnestock& Knapp 1993; Haugland & Froud-Williams 1999;Xiong & Nilsson 1999). Common management practices,such as frequent spring burning and grazing exclusion,could exacerbate this problem (Collins
et al
. 1998; Howe2000). Thus, anything that reduces grass dominanceshould alleviate competition with rare species andshould increase seedling establishment and diversity(Foster & Gross 1997).
According to the intermediate disturbance andgrazing optimization hypotheses, intermediate levelsof grazing should produce the highest levels of speciesdiversity, but also the highest levels of net primaryproductivity (NPP) in an ecosystem (Grime 1973;Connell 1978). The question of whether native ungulategrazing can increase seedling emergence and diversityis becoming more relevant because grazers such asbison
Bos bison
L. and wapiti
Cervus elaphus
L. areincreasingly being reintroduced (Knapp
et al
. 1999;Larkin
et al
. 2004). Management strategies, such asmowing, aimed at decreasing the biomass of dominantspecies have shown increased seedling survival insome experimental and pasture plantings (Burke &Grime 1996; Hutchings & Booth 1996; Lawson, Ford &Mitchley 2004). Moderate grazing by native ungulates,a common grassland disturbance, could have positiveimpacts on plant species diversity in intact grasslandsby reducing dominant grasses and increasing lightavailability (Hartnett, Hickman & Walter 1996; Collins
et al
. 1998; Knapp
et al
. 1999). Moderate grazingwould therefore be expected to produce non-lineareffects on diversity in restorations, with higher levels ofdiversity at intermediate grazing intensities (Smith
et al
. 2000). However, this may be restricted to intactsystems, where there is a propagule source available for
recruitment into the community (Hartnett, Hickman &Walter 1996; Collins
et al
. 1998). Furthermore, becauseintermediate levels of grazing can also produce thehighest levels of NPP (McNaughton 1979; Dyer,Turner & Seastedt 1993), grazing may lead to increasedresource uptake by plants. Productivity is alreadyhigh in grassland restorations, often higher than incomparable remnants (Baer
et al
. 2002; Camill
et al
.2004; Martin, Moloney & Wilsey 2005). If inter-mediate grazing increases production of dominant speciesabove and beyond what is already high, then interme-diate grazing in restorations, unlike in intact grasslands,may actually lessen positive effects on seedling emergenceand diversity. Thus increased productivity in restorationsbecause of moderate grazing might nullify potentiallypositive effects of grazing on microsite availability.
Finally, seed limitation and low seedling emergencebecause of grass dominance may interact to limit diver-sity in grassland restorations. A combination of addingseeds and increasing microsite availability may benecessary to favour seedling emergence (Burke &Grime 1996; Turnbull, Crawley & Rees 2000; Foster &Dickson 2004).
Our objectives were to determine: (i) if native ungu-lates increase availability of resources crucial to seed-ling emergence and (ii) whether seed additions, nativeungulate grazing or a combination of both enhancenative prairie seedling emergence while having little orno effect on non-native and cosmopolitan weeds in tall-grass prairie restorations. Our focus was on seedlingemergence, a key stage in the establishment of grass-land plants. Whether seedlings can establish viablepopulations is a longer term question that will not beconsidered here.
Materials and methods
The objective of the Neal Smith National WildlifeRefuge (NS) prairie project is to restore a large tallgrassprairie ecosystem using locally collected seeds combinedwith prescribed fire and grazing by native ungulates.The restoration is located on the Walnut Creek water-shed in Jasper County, Iowa, USA (41
°
33
′
N, 93
°
17
′
W).The refuge currently spans 2104 ha, approximately1200 ha of which have been seeded with tallgrassprairie species, beginning in 1992 and continuing to thepresent day. Grazing ungulates (
B. bison
and
C. elaphus
)were introduced to a 303-ha enclosure in 1996 and1998, respectively, which is where our study took place.Approximately 35
B. bison
and 15
C. elaphus
occupiedthe area during our study. Land use prior to prairieseeding included corn
Zea mays
L. and soybean
Glycinemax
(L.) Merr rotations and a few scattered pastures.There were 20 different plantings in this area (mean ofapproximately 14 ha each) and each planting was seededwith separate bulk seed mixes collected from local prairieremnants. Management practices after planting included
3
Native ungulates and seed additions in restorations
yearly spring burning during the early years followedby 2-year burn rotations, which is a common practicefor beginning restorations (Packard & Mutel 1997;Copeland, Sluis & Howe 2002). Mowing was donewhen necessary to control weedy and invasive species(P. Drobney, personal communication). Our plots werenot burned or mowed in 2003 or 2004, the years ofsampling.
Historically, precipitation at the site has a unimodaldistribution and peaks in May and/or June, with anaverage of approximately 880 mm year
−
1
. Weather in 2003was much warmer and drier than during 2004. BetweenMay and August, the peak growing months, tempera-tures and monthly precipitation averaged 21·7
°
C and13·3 mm in 2003 and 19·7
°
C and 143·8 mm in 2004.To standardize our sampling, we randomly selected
eight plantings within the enclosure that were seededbetween 1994 and 1996 on formerly cropped areas.Four plantings north of a dirt road included 6·7 kg ha
−
1
of
Elymus canadensis
L. in the seed mixture as a puta-tive cover crop, whereas four plantings south of theroad did not (for effects of the cover crop see Martin,Moloney & Wilsey 2005).
A randomized complete block split-plot design withunequal replication was used, with grazing or exclosuresapplied to main plots and seed addition treatments(described below) applied to subplots. Two 6
×
8-mgrazed plots were established 5 m away on either side ofa permanent 6
×
8-m permanent exclosure in June 2003in each of the eight plantings (blocks). Two grazedplots were sampled per planting because of increasedheterogeneity with grazing (Knapp
et al
. 1999). Byrequest of the refuge staff, exclosures were kept out ofview of visitors where possible and this precludedcompletely random locations.
Biomass and above-ground net primary productivity(ANPP) (general indicators of resource uptake andcompetition intensity) were estimated to comparegrazed and exclosed plots (Baer
et al
. 2004). Above-ground biomass was clipped to 2 cm in a 40
×
100-cmquadrat randomly placed in each exclosure (ng) andgrazed plot (gr), and surface litter was collected in Juneand August 2003 and in March, June and August 2004.Biomass was sorted into live and dead components,and live material was sorted by species, dried for 48 hat 65
°
C, and weighed. Subsequently, estimates weremade for the following biomass components: propor-tion of exotic biomass (exotic/total), proportion oftotal grass biomass (grass/total) and C
4
grass biomass(C
4
/total), and combined litter and standing dead bio-mass. Each of these different components of biomasscould suppress seedling emergence and species richnessin grassland restorations (Howe 2000; Camill
et al
.
2004). Plant species were designated as native or exoticbased on Eilers & Roosa (1994).
We estimated grazing intensity (GI) and used poly-nomial regressions to determine if GI was quadratic-ally related to response variables (McNaughton 1979,1985). Simply comparing grazed and ungrazed plotscan be misleading in cases where grazing is non-linearlyrelated to response variables (Grime 1973; Connell1978; McNaughton 1979). Above-ground NPP andGI were estimated for grazed plots (
n
=
16) using themoveable exclosure approach (McNaughton 1985;McNaughton, Milchunas & Frank 1996) during threeperiods: June–August 2003 and March–June andJune–August 2004. One 3
×
4-m temporary exclosurewas established at each site in March 2004 and wasmoved in June 2004 to measure consumption and GI(McNaughton 1985). Above-ground biomass from thecentre of each temporary exclosure was collected usingthe same quadrat size as explained above. Biomass fromthe permanent exclosure was used to estimate consump-tion during June–August 2003 (i.e. the first growingseason). Consumption (C) was estimated as (ng – g)/time, where ng was biomass inside and g was biomassoutside temporary exclosures at the end of the period,and time was the number of days exclosures were inplace (McNaughton 1985; Wilsey
et al
. 2002). Above-ground NPP (g m
−
2
day
−
1
) was calculated as a positivebiomass increment + consumption for each time period.Grazing intensity was calculated as GI
=
C/NPP(McNaughton 1985; Wilsey
et al
. 2002), with GI set to0 if consumption estimates were negative.
Environmental variables were measured to determinewhether grazing was creating microsites favourable forseedling emergence
.
Soil moisture and percentage lightat the soil surface were measured monthly from July toSeptember 2003 and from May to October 2004 (soilmoisture was not measured in July–August 2003 becauseof equipment failure) in each plot using a MoisturePoint® MP-917 Time Domain Reflectometry system(30 cm rods; Environmental Sensors, Victoria, Canada)and a 1-m Decagon® AccuPar Ceptometer (DecagonDevices Inc., Pullman, WA). Sampling points wererandomly located, and two measures of incident lightwere taken during each sampling time.
Species diversity was calculated from biomass todetermine if grazing exclusion affected diversity.Diversity was calculated at the quadrat scale for eachgrazed and exclosed plot. Diversity was quantifiedwith Simpson’s diversity (1/
D
), where and
p
i
=
relative biomass of each species
i
, and was thendecomposed into species richness (
S
) and evenness(1/
D
/
S
) to determine if each component of diversitydiffered (Buzas & Hayek 1996; Smith & Wilson 1996;Martin, Moloney & Wilsey 2005).
Two seed additions of rare native prairie forbs andgrasses were made to separate, randomly located 1-m
subplots within each plot using two different methods.These were compared to one control subplot (no seedaddition) within each main plot. Therefore, threesubplots were located in each main plot (exclosure orgrazed plot), with a total of nine subplots in each block.The first seed addition treatment consisted of addingseeds of 10 species collected by hand from local rem-nants in June 2003. The second treatment consisted ofadding seeds of 25 species from a local seed company(Allendan Seed Co., Winterset, IA, USA) in April 2004in a second set of subplots. More species were used in thesecond trial than the first because seeds were more readilyavailable from the seed company, and we wanted tomimic the high number of species found at theneighbourhood scale in remnants (Martin, Moloney &Wilsey 2005). Seeds were added with equal relativeabundances at a rate of 19 700 seeds m
−
2
for both trials(1970 and 788 per species for addition experiments 1and 2, respectively). Seed numbers were based onnumber of seeds in a typical seed rain rate found byRabinowitz & Rapp (1980) in a Missouri tallgrassprairie. Seed viability was not tested with seeds collectedfrom remnants, but all but one species readily germi-nated in greenhouse pots grown for seedling referencesamples. Mean seed viability for seeds obtained fromthe seed company was 81% (range 49–96%). Seedswere hand-scattered in each 1-m
2
subplot and existingvegetation and litter were shaken to aid seeds inreaching the soil surface. Species added in the firstexperiment were
Bouteloua curtipendula
(Michx.)Torrey,
Sporoblus asper
(Michx.) Kunth,
Solidagospeciosa
Nutt.,
Pycnanthemum virginianum
(L.) Dur. &Jackson,
Dalea purpurea
Vent.,
Chamaecrista fasciculata
(Michx.) Greene,
Amorpha canescens
Pursh,
Lespedezacapitata
Michx.,
Monarda fistulosa
L. and
Eryngiumyuccifolium
(Michx.) (Eilers & Roosa 1994). Speciesadded in the second experiment included all thoseadded in the first experiment, plus
Potentilla arguta
Pursh,
Silphium laciniatum
L.,
Echinacea pallida
Nutt.,
Ratibida pinnata
(Vent.) Barnh.,
Artemesia ludoviciana
Nutt.,
Liatris pycnostachya
(Michx.),
Verbena stricta
Vent.,
Helianthus rigidus
(Cass.) Desf.,
Gentianaandrewsii
Griseb.,
Tradescantia bracteata
Small,
Violapedatifida
G. Don,
Anemone cylindrica
Gray,
Phloxpilosa
L.,
Schizachyrium scoparium
(Michx.) Nash and
Solidago rigida
L. (Eilers & Roosa 1994).All forb seedlings, including species added from the
mix as well as volunteers, were identified to species andcounted in a randomly placed 20
×
50-cm quadratwithin each subplot, to estimate seedling emergence.Volunteers were included because some added specieswere already in the seed bank and therefore could notbe distinguished from experimentally sown seedlings.Exotic and cosmopolitan weed seedlings were countedto test concerns about disturbance from native ungu-lates increasing weeds in grasslands (Smith & Knapp1999). Grass seedlings were only counted if the specieswas added. Seedlings were counted if they were upto 7·5 cm tall or up to any height if they were annuals,
and were counted once per month during the grow-ing season, beginning the month after seeds wereadded.
Randomized block split-plot
s were used, withplanting as a random block term. All grazing effects weretested with the main plot error term (planting
×
grazed),and seed and seed–grazed interactions were tested withthe subplot error term. Repeated-measures
was used to compare grazed (
n
=
16) and exclosed plots(
n
=
8) for existing vegetation and resource variables,with time 0 data (measurements taken before exclos-ures were constructed) as a covariate (except for NPP,for which time 0 data could not be calculated). Wedropped the covariate from each model if it was notsignificant (
P
> 0·05). Variables were logarithmicallytransformed (biomass, standing dead and litter), square-root transformed (proportion of exotic biomass) orarcsin square-root transformed (proportion of C
4 andgrass) to improve normality when necessary. Allanalyses were done with in SAS (Littell,Stroup & Freund 2002).
The first and second seed additions were analysedseparately because they had different numbers of spe-cies, addition dates and weather conditions. A seedlingenhancement effect, ln[(added seedlings + 1)/(controlseedlings + 1)], was calculated to quantify the numberof seedlings that did not emerge from the existing seedbank but emerged from added seeds. This derived vari-able eliminated non-normality in data as a result ofhaving many zeroes in control subplots. To test if seedadditions increased seedling numbers above those ofcontrols, seedling enhancement effects were tested against0 with a t-test. Grazing effects on seedling enhancementover time were analysed with repeated-measures
of corresponding data. Exotic seedling and seedlingdiversity variables were analysed with repeated-measures(means for 2003 and 2004) for the first seed addi-tion, and with regular for the second addition.Data were averaged across months because raw datahad too many zeroes to analyse each sampling time.Seedling Simpson’s diversity (1/D), species richness (S )and evenness (1/D/S ) were calculated in each subplotto determine if seed additions or grazing improvedseedling diversity.
We used polynomial regression to test for quadraticand linear relationships between mean GI (grazed plotsonly) and response variables. Mean GI was calculatedby averaging GI across time because of non-normallydistributed data. We used path analysis to test for directand indirect associations of grazing on seedlingenhancement. A direct pathway was tested of GI onbiomass, biomass on light and water availability, andlight and water on seedling enhancement. An indirectpath was tested of GI on NPP, and NPP on seedlingenhancement. This indirect pathway could be signifi-cant if increased NPP had additional effects on seedlings
5Native ungulates and seed additions in restorations
because of non-light and water effects, such as increasednutrient uptake of vegetation.
Results
Only two response variables differed significantlybetween grazed and exclosed plots and another wasmarginally significant (Tables 1 and 2). Above-groundNPP m−2 was 1·2, 1·1 and 8·0 times as large in grazedplots depending on time period, and the difference wasonly significant for June–August 2004 (Table 1 andFig. 1a). Light availability at the soil surface was 1·7times as great in grazed plots, and this was fairlyconsistent across time periods (Table 1 and Fig. 1b).Combined standing dead and litter biomass m−2 wasmarginally significantly lower in grazed plots (Tables 1and 2).
No response variables were quadratically related tomean GI (grazing intensity) (F1,15 between 0·03 and3·07, P between 0·10 and 0·88). Mean GI was highest inJune–August 2004, when 68% of NPP was consumed(range 0–100%, SE 6), followed by 49% during June–August 2003 (range 0–100%, SE 8). It was muchlower during spring 2004 (mean 14%, range 0–37%,SE 3). Biomass m−2 was negatively related to mean GI
Fig. 1. Grazing (n = 16) or exclosure from grazing (n = 8)differences for (a) ANPP (P < 0·01) and (b) percentage lightavailability at soil surface (P < 0·01). Vertical bars are ± 1 SE. T
(F1,15 = 6·02, P = 0·03, r = −0·56, slope = −3·49; Fig. 2a),as expected. Species evenness at the 0·4 m−2 scalewas positively related to mean GI (F1,15 = 5·61, P = 0·03,r = 0·54, slope = 0·89; Fig. 2b).
-
Native species seedling emergence
Seed additions increased the number of native speciesseedlings 0·1 m−2 in both seed addition experiments(Table 3 and Fig. 3). Adding seeds increased seedlingnumbers by 2·5 times in 2003 (t = 1·97, P = 0·07) and2·0 times in 2004 (t = 3·22, P < 0·01) in the first experi-ment (Fig. 3a). In the second experiment, the seedlingenhancement ratio as a result of adding seeds was3·8 in May, 5·2 in June, 5·5 in July, 6·6 in August and17·9 in October (t = 6·94, 6·48, 3·60, 4·43 and 2·14,respectively; P-values < 0·01 in May–July, P = 0·05 inOctober; Fig. 3a). Overall, seedling numbers decreasedsignificantly between June and October (Table 4 andFig. 3a).
Grazing alone, without seed additions, did not increasenumber of seedlings 0·1 m−2 for either seed additionexperiment (Fig. 3a) but grazing conditionally affectedseedling enhancement. The seedling enhancementeffect did not differ between grazed and exclosed plotsin the first experiment but was on average 1·4 times aslarge in grazed than exclosed plots in the second(Table 4 and Fig. 3b). We did not find a significant
Fig. 2. Relationships between (a) biomass and (b) evennessand grazing intensity (n = 16).
Tab
le 2
.M
ean
(SE
) of
res
pons
e va
riab
les
mea
sure
d to
tes
t ef
fect
s of
nat
ive
ungu
late
gra
zing
in a
tal
lgra
ss p
rair
ie r
esto
rati
on
Gra
zed
(n =
16)
Exc
lose
d (n
= 8
)
Aug
ust
2003
June
200
4A
ugus
t 20
04A
ugus
t 20
03Ju
ne 2
004
Aug
ust
2004
Bio
mas
s (m
−2)
5·87
(0·
14)
5·82
(0·
14)
5·93
(0·
14)
6·13
(0·
17)
5·97
(0·
17)
6·12
(0·
17)
Lit
ter
and
stan
ding
dea
d (m
−2)
6·28
(0·
14)
6·09
(0·
14)
6·39
(0·
14)
6·36
(0·
14)
6·58
(0·
14)
6·58
(0·
14)
Pro
port
ion
exot
ic b
iom
ass
(0·4
m−2
)0·
17 (
0·06
)0·
33 (
0·06
)0·
32 (
0·06
)0·
17 (
0·07
)0·
31 (
0·07
)0·
22 (
0·07
)P
ropo
rtio
n ex
otic
S (
0·4
m−2
)0·
35 (
0·03
)0·
37 (
0·03
)0·
36 (
0·03
)0·
30 (
0·04
)0·
31 (
0·04
)0·
36 (
0·04
)P
ropo
rtio
n gr
ass
(0·4
m−2
)1·
13 (
0·06
)1·
01 (
0·06
)1·
17 (
0·06
)1·
22 (
0·08
)1·
19 (
0·08
)1·
08 (
0·08
)P
ropo
rtio
n C
4 gra
ss (
0·4
m−2
)0·
95 (
0·08
)0·
83 (
0·08
)0·
96 (
0·08
)1·
13 (
0·11
)0·
99 (
0·11
)1·
01 (
0·11
)A
lpha
S (
0·4
m−2
)10
·53
(0·8
2)12
·63
(0·8
2)10
·86
(0·8
2)9·
44 (
1·10
)9·
86 (
1·10
)11
·02
(1·1
0)A
lpha
E (
0·4
m−2
)0·
25 (
0·02
)0·
24 (
0·02
)0·
26 (
0·02
)0·
23 (
0·03
)0·
25 (
0·03
)0·
24 (
0·03
)A
lpha
1/D
(0·
4 m
−2)
2·55
(0·
33)
2·89
(0·
33)
2·63
(0·
33)
2·27
(0·
44)
2·46
(0·
44)
2·76
(0·
44)
Gra
zed
(n =
16)
Exc
lose
d (n
= 8
)
Sept
embe
r 20
03M
ay 2
004
June
200
4Ju
ly 2
004
Sept
embe
r 20
04O
ctob
er 2
004
Sept
embe
r 20
03M
ay 2
004
June
200
4Ju
ly 2
004
Sept
embe
r 20
04O
ctob
er 2
004
Per
cent
age
wat
er38
·92
(1·0
4)38
·24
(1·0
4)37
·23
(1·0
4)34
·84
(1·0
4)35
·10
(1·0
4)26
·66
(1·0
4)39
·98
(1·4
0)38
·41
(1·4
0)35
·75
(1·4
0)33
·82
(1·4
0)34
·69
(1·4
0)28
·23
(1·4
0)
7Native ungulates and seed additions in restorations
relationship with GI in either experiment (first experi-ment, linear effects F1,13 = 0·01, P = 0·93, quadraticeffects F1,13 = 1·68, P = 0·22; second experiment, lineareffects F1,13 = 0·63, P = 0·44, quadratic effects F1,13 =2·14, P = 0·17).
Path analysis from both experiments indicated thatbiomass m−2 was negatively related to light and soilwater availability, and that water was positively relatedto seedling enhancement more regularly and stronglythan light (Fig. 4 and Table 5). The indirect pathwayindicated that GI was positively related to NPP in thefirst (significant for 2004 only) and second experiments,but NPP never significantly explained seedling enhance-ment beyond effects of light and water (Fig. 4 and Table 5).
Exotic species seedling emergence
Exotic and cosmopolitan weed seedling emergencewere not clearly affected by treatments in either experi-ment (Table 4). Exotics such as Taraxacum officinale
Fig. 3. The (a) number of seedlings 0·1 m−2 in grazed andexclosed plots when seeds were added or were not and (b)effects of grazing on seedling enhancement for two experi-mental seed addition trials in a tallgrass prairie restoration(n = 16 for grazed; n = 8 for ungrazed). The first seed additionexperiment is presented before the break in the x axis and thesecond experiment is after the break. Vertical bars are ± 1 SE.
Tab
le 4
.
r
esul
ts (
F-v
alue
s) fo
r se
edlin
g en
hanc
emen
t (i
ncre
ase
in s
eedl
ing
num
bers
wit
h se
ed a
ddit
ions
) an
d se
edlin
g nu
mbe
rs b
etw
een
graz
ed p
lots
and
plo
ts e
xclo
sed
from
gra
zing
in a
tal
lgra
ss p
rair
iere
stor
atio
n. S
igni
fican
ce is
indi
cate
d by
†P
bet
wee
n 0·
06 a
nd 0
·1, *
P b
etw
een
0·02
and
0·0
5 an
d **
P ≤
0·0
1
Sour
ce
Fir
st e
xper
imen
tSe
cond
exp
erim
ent
Fir
st e
xper
imen
t se
edlin
g nu
mbe
rsSe
cond
exp
erim
ent
seed
ling
num
bers
d.f.
Seed
ling
enha
ncem
ent
d.f.
Seed
ling
enha
ncem
ent
Num
/Den
d.f.
Div
ersi
tyR
ichn
ess
Eve
nnes
sE
xoti
csD
iver
sity
Ric
hnes
sE
venn
ess
Exo
tics
Gra
zed
10·
021
6·3*
1/15
0·3
0·2
0·7
0·1
0·0
1·0
1·1
2·1
Seed
add
itio
n1/
224·
0†5·
1*0·
02·
060
·2**
54·0
**6·
5*0·
2G
raze
d ×
seed
add
itio
n1/
220·
70·
20·
40·
20·
31·
81·
00·
4E
rror
1515
Tim
e1
0·5
47·
3**
1/44
10·7
**9·
3**
1·0
7·8*
*T
ime
× gr
azed
10·
64
0·5
1/44
0·1
0·0
0·2
0·2
Tim
e ×
graz
ed ×
seed
add
itio
n2/
441·
21·
50·
90·
4E
rror
(ti
me)
2288
9Native ungulates and seed additions in restorations
and Daucus carota were among the most abundantspecies in both experiments, and cosmopolitan weedsAster pilosus and Ambrosia artemisiifolia were alsoabundant in the first experiment (Table 3). The meannumber of exotic seedlings did not significantly differbetween grazed and exclosed plots in either experiment[mean (SE) number of exotics 0·1 m−2, n = 48; firstexperiment 2003, grazed 1·51 (0·28), exclosed, 1·72(0·31); 2004, grazed 2·03 (0·28), exclosed 1·91 (0·31);second experiment, grazed 2·19 (0·29), exclosed 1·92(0·31); Table 4]. Grazing effects did not interact withseed additions in either experiment (Table 4). The
mean number of exotics was not related to mean GIfor either experiment (first, linear, exotics F1,13 = 0·07,P = 0·79; quadratic, exotics F1,13 = 2·08, P = 0·17;second, linear, F1,13 = 0·59, P = 0·45, quadratic,F1,13 = 0·23, P = 0·64).
Seed additions increased seedling diversity and richnessin both experiments and slightly decreased seedlingevenness in the second experiment. Mean diversity inthe first experiment was 1·2 times as great with seedadditions in 2003 and 2004, but this difference was onlymarginally significant [mean (SE) 1/D 0·1 m−2, n = 48;2003, 2·76 (0·22), control 2·24 (0·22); 2004, 2·10 (0·22),control 1·82 (0·22); Table 4]. Mean diversity in thesecond experiment was 1·9 times as great in the seedaddition subplots than in controls [mean (SE) 1/D0·1 m−2, n = 48; addition 3·37 (0·20), control 1·82 (0·20);Table 4]. Mean richness in the first experiment was 1·3and 1·2 times as great in seed addition than control sub-plots in 2003 and 2004, respectively, and was 2·3 timesas great in addition subplots in the second experiment[mean (SE) S 0·1 m−2, n = 48; first experiment, 2003,addition 3·81 (0·32), control 2·88 (0·32); 2004, addition2·96 (0·32), control 2·53 (0·32); second experiment,addition 5·82 (0·37), control 2·52 (0·37); Table 4]. Meanevenness did not differ in the first experiment but wasslightly lower in seed addition subplots in the second[mean (SE) E 0·1 m−2, n = 48; first 2003, addition 0·75(0·03), control 0·77 (0·03); 2004, addition 0·74 (0·03),control 0·72 (0·03); second, addition 0·65 (0·03), con-trol 0·72 (0·03); Table 4].
Seedling species diversity did not differ betweengrazed and exclosed plots for either experiment [mean(SE) for grazed and exclosed, respectively, n = 48; firstexperiment, diversity 2003 2·47 (0·19), 2·53 (0·25), 2004
Fig. 4. Path analysis diagram that tests direct effects ofgrazing intensity on seedling enhancement through biomass,light and water, and indirect effects of grazing intensity onseedling enhancement through NPP effects in a tallgrassprairie restoration. Ten and 25 rare prairie species were addedin plots inside and outside grazing exclosures in two separateexperiments.
Table 5. Path analysis results to determine direct effects of grazing intensity on biomass, biomass on light and water, and light andwater on seedling enhancement, and indirect effects of grazing intensity on NPP and NPP on seedling enhancement for the firstand second seed addition experiment. Ten (first) or 25 (second) rare prairie species were added to plots. Numbers representstandardized regression coefficients. Significance is indicated by *P between 0·02 and 0·05 and **P ≤ 0·01
Variable
First experiment Second experiment
Second half 2003 First half 2004 Second half 2004 First half 2004 Second half 2004
1·88 (0·19), 2·04 (0·25), richness 2003 3·28 (0·29), 3·41(0·36), 2004 2·68 (0·29), 2·81 (0·36), evenness 2003 0·75(0·03), 0·77 (0·04), 2004 0·71 (0·03), 0·75 (0·04); secondexperiment diversity 2·58 (0·19), 2·61 (0·22), richness4·39 (0·33), 3·95 (0·42), evenness 0·66 (0·02), 0·71(0·04); all at the 0·1 m2 scale; Table 4]. Diversityenhancement declined with GI in the first experiment(linear effect F1,13 = 5·05, P = 0·04; quadratic effectF1,13 = 2·29, P = 0·15; data not shown) and was quad-ratically related to GI in the second experiment (F1,13 =6·02, P = 0·03; quadratic equation y = 2·4 ± 11·3x +15·5x2) with an outlier included, and unrelated withan outlier excluded (linear effect F1,13 = 3·22, P = 0·10,r = 0·45, slope = 1·83; quadratic effect F1,13 = 0·53,P = 0·48; data not shown). The evenness enhancementeffect was negatively related to GI in the first experi-ment (F1,13 = 4·8, P = 0·05, r = 0·53, slope = −0·53) only.
Discussion
Previously, we found that conventional prairie restora-tion at the study site was able to restore common nativespecies but not species diversity of nearby prairieremnants (Martin, Moloney & Wilsey 2005). Here wetested hypotheses regarding why diversity was lower.Our results suggest that seedling emergence in lowdiversity restorations is seed limited but that nativeungulates can sometimes increase emergence as well.Seedling enhancement increased with water and lightavailability, which suggests that, when grazing is enhanc-ing emergence, the mechanism may be associated withgrazers having a direct effect on water and light. Acombination of seed additions and grazing led to thehighest amount of seedling emergence, but this resultwas conditional, i.e. it was found only in the secondtrial and year. However, grazing alone did not increaseseedling emergence in either trial.
Biomass and NPP, general indicators of resourceuptake in grasslands, are usually affected by grazing(Semmartin & Oesterheld 1996). We found that totalbiomass declined with grazing intensity, and water andlight availability were higher when biomass was lower,which suggests that grazing decreased biomass enoughto increase resource availability. However, ANPP wasalso higher with grazing, suggesting that defoliated plantswere readily recovering from defoliation and utilizingavailable resources (Knapp et al. 1999; Wilsey et al.2002). Nevertheless, increased levels of ANPP didnot appear to have effects above and beyond thosecorrelated with water and light availability in reducingseedling enhancement.
Seedling emergence was limited by seed availability inthese tallgrass prairie restorations, which suggests thatrestorations are similar to many old fields in their lackof propagule availability (Tilman 1997; Zobel et al.2000; Pywell et al. 2002). Many tallgrass prairie restora-
tions are initiated by harvesting seeds from remnantsin the autumn (Polley, Derner & Wilsey 2005), when C4
grass seed is most abundant relative to other species.Evidence from this study suggests that seedlingemergence of rare forbs is very low nearly 10 years afterinitial seeding, and that adding seeds of rare forb andgrass species that are typically lacking in restorationscan increase seedling emergence, the first step inrecruiting species into the community.
Our finding that seed additions and grazing com-bined could increase seedling emergence suggests thatgrazing mammals might increase seedling recruitmentin some situations. Turnbull, Crawley & Rees (2000)found that, overall, a combination of adding seeds andinducing disturbance to reduce dominant vegetationwas most important for recruitment. Rhinanthus minor,a parasitic plant, also increased seedling recruitmentof added species by reducing competitive effects ofdominant vegetation (Pywell et al. 2004). However, theeffect of grazing on seedling emergence with seed addi-tions in the restoration was conditional. This condi-tionality may have been the result of very differentweather between years, but our design could not deter-mine this definitively. Although grazing combined withseed additions conditionally improved seedling emer-gence, we did not find that emergence was quadraticallyrelated to grazing intensity, as predicted by the inter-mediate disturbance hypothesis. Knapp et al. (1999)proposed that target grazing intensity in intact tallgrassprairies should be about 25% of annual above-groundprimary production, based on historic grazing intensi-ties. We observed grazing intensities in the restorationthat were sometimes double that estimate. In our study,it appears that grazing had an increasingly beneficial(i.e. linear) effect on seedling emergence when seedswere added.
Increases in exotic or cosmopolitan weed species maynegatively impact diversity and are a primary concernin grazed grasslands (Smith & Knapp 1999; Hulme &Bremner 2006). Grazing, which is utilized in manage-ment of both intact grasslands and restorations(Collins et al. 1998; Knapp et al. 1999), could increaseweeds by the same mechanisms that increase nativeplant recruitment (Smith & Knapp 1999). In contrast,high dominance may enhance seedling emergence insome grasslands (Wilsey & Polley 2002; Smith et al.2004). We found no relationships with grazing onexotics, suggesting that grazing may not be importantto exotic recruitment in these restored grasslandcommunities.
Diversity components responded differently to seedadditions and grazing. With grazing, we found nochanges in species richness in the vegetation after1.5 years of grazing, but evenness increased linearlywith grazing intensity. An increase in evenness can be
11Native ungulates and seed additions in restorations
associated with a decrease in dominance from ungulateactivities (Hartnett, Hickman & Walter 1996). How-ever, grazers may not be able to recruit new species andenhance richness if rare plant species are not availableeither in the seed bank or as vegetative propagules,which appears to be the case in this restoration. Seedadditions, on the other hand, tended to increase rich-ness but had a smaller negative effect on evenness.
It is important to point out that we only tested forseed limitation with the emerging seedling community.Although seedling emergence is a crucial step towardsrecruitment into a community, seedlings must establishviable populations before they will influence diversityin the longer term. Wilsey & Polley (2003) found that,even when seedling emergence was high, plant diversitywas unchanged because of low seedling survival in aTexas grassland. It is too early to estimate establish-ment success, and longer term monitoring is needed totest the hypothesis that seed additions and grazing willincrease diversity of vegetation in the long term.
Seedling numbers of rare species increased greatly whenseeds were added, suggesting that these restorationswere severely seed limited. Grazers increased water andlight availability by decreasing above-ground biomass,and water increased seedling enhancement more regu-larly and strongly than did light. Grazing increasedANPP, but ANPP did not explain seedling emergenceabove and beyond water and light effects. Undercertain conditions, seedling emergence was positivelyinfluenced by grazing when seeds were added. However,grazing by itself did not increase seedling emergence,probably because seeds of rare species were notavailable to emerge. These results suggest that it may beadvantageous to mimic positive effects of native un-gulates in restorations when seeds or propagules of rarespecies are available to emerge. If propagule availabilityis low, positive effects of grazing alone will not increaseseedling emergence. Improving seedling emergence inrestorations may therefore require adding seeds evenafter dominant plants are established, while simultane-ously mimicking grazing effects to increase light andwater availability. In contrast to management strategiesin intact grasslands (Collins et al. 1998), utilizinggrazing without propagules may not improve seedlingemergence in restorations. However, longer term moni-toring is necessary to determine if seed additions andgrazing promote long-term plant diversity.
Acknowledgements
We thank Pauline Drobney at Neal Smith NWR,Andrea Blong, Brennan Dolan, Dan Haug, and DavidLosure for help in the field, and Tim Dickson andtwo anonymous referees for comments on an earlierversion of this manuscript.
References
Baer, S.G., Blair, J.M., Collins, S.L. & Knapp, A.K. (2004)Plant community responses to resource availability andheterogeneity during restoration. Oecologia, 139, 617–629.
Baer, S.G., Kitchen, D.J., Blair, J.M. & Rice, C.W. (2002)Changes in ecosystem structure and function along achronosequence of restored grasslands. Ecological Applica-tions, 12, 1688–1701.
Burke, M.J.W. & Grime, J.P. (1996) An experimental study ofplant community invasibility. Ecology, 77, 776–790.
Eilers, L.J. & Roosa, D.M. (1994) The Vascular Plants of Iowa.University of Iowa Press, Iowa City, IA.
Eriksson, O. & Ehrlen, J. (1992) Seed and microsite limitationof recruitment in plant populations. Oecologia, 91, 360–364.
Fahnestock, J.T. & Knapp, A.K. (1993) Water relations andgrowth of tallgrass prairie forbs in response to selectivegrass herbivory by bison. International Journal of PlantSciences, 154, 432–440.
Foster, B.L. & Dickson, T.L. (2004) Grassland diversity andproductivity: the interplay of resource availability andpropagule pools. Ecology, 85, 1541–1547.
Foster, B.L. & Gross, K.L. (1997) Partitioning the effects ofplant biomass and litter on Andropogon gerardi in old-fieldvegetation. Ecology, 78, 2091–2104.
Foster, B.L., Dickson, T.L., Murphy, C.A., Karel, I.S. &Smith, V.H. (2004) Propagule pools mediate communityassembly and diversity–ecosystem regulation along agrassland productivity gradient. Journal of Ecology, 92,435–449.
Galatowitsch, S.M. & van der Valk, A.G. (1996) The vegeta-tion of restored and natural prairie wetlands. EcologicalApplications, 6, 102–112.
Gough, L., Grace, J.B. & Taylor, K.L. (1994) The relationshipbetween species richness and community biomass: theimportance of environmental variables. Oikos, 70, 271–279.
Grime, J.P. (1973) Competitive exclusion in herbaceousvegetation. Nature, 242, 344–347.
Hartnett, D.C., Hickman, K.R. & Walter, L.E.F. (1996)Effects of bison grazing, fire, and topography on floristicdiversity in tallgrass prairie. Journal of Range Management,49, 413–420.
Haugland, E. & Froud-Williams, R.J. (1999) Improvinggrasslands: the influence of soil moisture and nitrogenfertilization on the establishment of seedlings. Journal ofApplied Ecology, 36, 263–270.
species recruitment and species richness as soil resourcesrise. Journal of Ecology, 92, 339–347.
Howe, H.F. (2000) Grass response to seasonal burns in exper-imental plantings. Journal of Range Management, 53, 437–441.
Hulme, P.E. & Bremner, E.T. (2006) Assessing the impact ofImpatiens glandulifera on riparian habitats: partitioningdiversity components following species removal. Journal ofApplied Ecology, 43, 43–50.
Hutchings, M.J. & Booth, K.D. (1996) Studies on the feasi-bility of re-creating chalk grassland vegetation on ex-arableland. II. Germination and early survivorship of seedlingsunder different management regimes. Journal of AppliedEcology, 33, 1182–1190.
Knapp, A.K., Blair, J.M., Briggs, J.M., Collins, S.L.,Hartnett, D.C., Johnson, L.C. & Towne, E.G. (1999) Thekeystone role of bison in North American tallgrass prairie.Bioscience, 48, 39–50.
Lawson, C.S., Ford, M.A. & Mitchley, J. (2004) The influenceof seed addition and cutting regime on the success of grass-land restoration on former arable land. Applied VegetationScience, 7, 259–266.
Littell, R.C., Stroup, W.W. & Freund, R.J. (2002) SAS forLinear Models, 4th edn. SAS Publishing, Cary, NC.
McNaughton, S.J. (1979) Grazing as an optimization process:grass–ungulate relationships in the Serengeti. AmericanNaturalist, 113, 691–703.
McNaughton, S.J. (1985) Ecology of a grazing ecosystem: theSerengeti. Ecological Monographs, 55, 259–294.
McNaughton, S.J., Milchunas, D.G. & Frank, D.A. (1996)How can net primary productivity be measured in grazingecosystems? Ecology, 77, 974–977.
Martin, L.M., Moloney, K.A. & Wilsey, B.J. (2005) Anassessment of grassland restoration success using speciesdiversity components. Journal of Applied Ecology, 42, 327–336.
Packard, S. & Mutel, C.F. (1997) The Tallgrass RestorationHandbook for Prairies, Savannas, and Woodlands. IslandPress, Washington, DC.
Polley, H.W., Derner, J.D. & Wilsey, B.J. (2005) Patterns ofplant species diversity in remnant and restored tallgrassprairies. Restoration Ecology, 13, 480–487.
Pywell, R.F., Bullock, J.M., Hopkins, A., Walker, K.J.,Sparks, T.H., Burke, M.J.W. & Peel, S. (2002) Restorationof species-rich grassland on arable land: assessing thelimiting processes using a multi-site experiment. Journal ofApplied Ecology, 39, 294–309.
Gregory, S.J. & Stevenson, M.J. (2004) Facilitatinggrassland diversification using the hemiparasitic plantRhinanthus minor. Journal of Applied Ecology, 41, 880–887.
Rabinowitz, D. & Rapp, J.K. (1980) Seed rain in NorthAmerican tall grass prairie. Journal of Applied Ecology, 17,193–802.
Semmartin, M. & Oesterheld, M. (1996) Effect of grazingpattern on primary productivity. Oikos, 75, 431–436.
Sluis, W.J. (2002) Patterns of species richness and compositionin recreated grassland. Restoration Ecology, 10, 677–684.
Smith, B. & Wilson, J.B. (1996) A consumer’s guide to even-ness indices. Oikos, 76, 70–82.
Smith, M.D. & Knapp, A.K. (1999) Exotic plant species in aC4-dominated grassland: invasibility, disturbance, andcommunity structure. Oecologia, 120, 605–612.
Smith, M.D., Wilcox, J.C., Kelly, T. & Knapp, A.K. (2004)Dominance not richness determines invasibility of tallgrassprairie. Oikos, 106, 253–262.
Smith, R.S., Shiel, R.S., Millward, D. & Corkhill, P. (2000)The interactive effects of management on the productivityand plant community structure of an upland meadow: an8-year field trial. Journal of Applied Ecology, 37, 1029–1043.
Smith, R.S., Shiel, R.S., Millward, D., Corkhill, P. &Sanderson, R.A. (2002) Soil seed banks and the effectsof meadow management vegetation change in a 10-yearmeadow field trial. Journal of Applied Ecology, 39, 279–293.
Tilman, D. (1997) Community invasibility, recruitment limita-tion, and grassland biodiversity. Ecology, 78, 81–92.
Turnbull, L.A., Crawley, M.J. & Rees, M. (2000) Are plantpopulations seed-limited? A review of seed sowing experi-ments. Oikos, 88, 225–238.
Wilsey, B.J. & Polley, H.W. (2003) Effects of seed additionsand grazing history on diversity and productivity of sub-humid grasslands. Ecology, 84, 920–931.
Xiong, S. & Nilsson, C. (1999) The effects of plant litter onvegetation: a meta-analysis. Journal of Ecology, 87, 984–994.
Zobel, M., Otsus, M., Liira, J., Moora, M. & Möls, T. (2000)Is small-scale species richness limited by seed availability ormicrosite availability? Ecology, 81, 3274–3282.
Received 31 October 2005; final copy received 28 May 2006Editor: Phil Hulme