Submitted 24 July 2018Accepted 14 June 2019Published 13 August 2019

Corresponding authorDavid Anthony Kirkdavidaquilaecologycom

Academic editorAlastair Culham

Additional Information andDeclarations can be found onpage 19

DOI 107717peerj7296

Copyright2019 Kirk et al

Distributed underCreative Commons CC-BY 40

OPEN ACCESS

Grazing effects on woody and herbaceousplant biodiversity on a limestonemountain in northern TunisiaDavid Anthony Kirk14 Katherine Heacutebert2 and Frank Barrie Goldsmith3

1Department of Geography University College London University of London London United Kingdom2Deacutepartement de biologie Faculteacute des Sciences Universiteacute de Sherbrooke Sherbrooke Queacutebec Canada3Department of Biology University College London University of London London United Kingdom4Current affiliation Aquila Conservation amp Environment Consulting Ottawa Ontario Canada

ABSTRACTMediterranean maquis vegetation is highly biodiverse but widespread grazing posesa challenge for management and conservation We sampled woody and herbaceousplants separately on a limestonemountainwith strongmesic-xeric gradients inTunisiarsquosParc National de LrsquoIchkeul assessed grazing pressure (on a scale of 1ndash3) and askedwhether grazing had a significant effect on plant compositional abundance beforeand after controlling for environmental covariates Sites on the more mesic lakesideface of the mountain were most compositionally unique and forbs contributed mostto the herbaceous beta-diversity on the mountain We used variance partitioningto separate the collective and individual effects of the abiotic environment grazinghuman activity and space on herbaceous and woody beta-diversity However theindividual effect of grazing on overall plant community composition was confoundedwith space due to the spatially autocorrelated grazing pressure on the mountainImportantly we found that herbaceous and woody communities responded differentlyto increasing levels of grazing intensity herbaceous beta-diversity was highest betweensites with no grazing pressure while woody beta-diversity peaked under light grazingHerbaceous community composition was sensitive to any intensity of grazing pressureand biotic homogenization occured under moderate-to-high grazing pressure On theother hand woody community composition remained relatively similar under no tolight grazing pressure but differed under moderate-to-heavy grazing Using a one-waypermutational analysis of variance analysis we showed that grazing had a significanteffect when controlling for abiotic and spatial covariates Our findings offer insightinto the effects of grazing on maquis vegetation at Jebel Ichkeul acting as a microcosmof similar conservation and management issues elsewhere in the Mediterranean Wesuggest that a combination ofmonitoring and carefully controlled grazingmay enhanceplant diversity and maintain the regionrsquos biodiverse maquis vegetation potentiallymaintaining a key climate refugium for vulnerable endemic species Importantly ourstudy provides a useful baseline of the plant assemblages at Jebel Ichkeul with which tocompare future vegetation changes

Subjects Biodiversity Conservation Biology Ecology Plant Science Environmental ImpactsKeywords Beta diversity Mediterranean maquis vegetation Conservation management NorthAfrica Goat grazing Biotic homogenization

How to cite this article Kirk DA Heacutebert K Goldsmith FB 2019 Grazing effects on woody and herbaceous plant biodiversity on a lime-stone mountain in northern Tunisia PeerJ 7e7296 httpdoiorg107717peerj7296

INTRODUCTIONHumans and their livestock have influenced Mediterranean shrublands for millennia(Naveh 1990) This long history of widespread human activity has played an integraland pervasive role in shaping the regionrsquos disturbed yet remarkably biodiverse landscape(Papanastasis 1998 Rundel 1998 Mazzoleni et al 2004 Vogiatzakis Mannion amp Griffiths2006 Blondel et al 2010) Evaluating the impacts of humansrsquo coevolution with theMediterranean shrublands is thus an ongoing challenge for the management andconservation of this regionrsquos biodiversity (Falcucci Maiorano amp Boitani 2007 Rundelet al 2016)

Historically low intensity grazing by domestic livestock and fires set by shepherdsmaintained the floristically rich and heterogeneous landscapes of the Mediterraneanmaquis which were previously shaped by more abundant native herbivores (LeHouerou 1981Papanastasis 1998Papanastasis Kyriakakis amp Kazakis 2002) The resultingshrubland communities are of high ecological value due to their rich diversity of annualsand geophytes from the Orchidaceae Iridaceae and Liliaceae (Pons 1981 Queacutezel 1981)Under certain conditions grazing can also effectively restore plant diversity after shrubencroachment in dry grasslands (Elias amp Tischew 2016 Elias Houmllzel amp Tischew 2018 seealso Toumlroumlk et al 2016 Toumlroumlk et al 2018) and in heathlands (Rupprecht Gilhaus amp Houmllzel2016) Without such disturbance many shrublands develop into forests leading to a lossof plant diversity and a build-up of organic matter rendering them susceptible to wildfires(Rackham ampMoody 1996 Perevolotsky amp Seligman 1998Henkin 2011) Although grazingcan maintain endemic species and regulate ecosystems (Gomez-Campo 1985) overgrazingand uncontrolled wood-cutting can also reduce plant diversity and cause erosion ordesertification (Hill et al 1998 Papanastasis 1998 Papanastasis Kyriakakis amp Kazakis2002) Evaluating the effects of different intensities of grazing disturbance on biodiversityis therefore fundamental to inform management strategies in maquis ecosystems

The effects of grazing on plant biodiversity are equivocal being subject to varioustemporal spatial and ecological contingencies (Olff amp Ritchie 1998 Olsvig-Whittakeret al 2006) On islands intensive overgrazing by goats has devastated vegetation andthreatened the persistence of many island endemics (Campbell amp Donlan 2005) HowevermanyMediterranean plant species have coevolved with herbivores and developed strategiesto resist grazing including spininess chemical repulsion prostrate growth and an ability togrow on remote rocky cliffs that are inaccessible to livestock (Papanastasis 1998) Grazinghas therefore applied a continuous selective pressure on plant species over extensive timescales likely influencing plant community composition and diversity across the entireregion (Rundel et al 2016)

Characterized by extremely high floristic richness theMediterranean region holds about10 of global plant species diversity (25000 species) on only 2 of the worldrsquos terrestrialsurface area and half of these plant species are endemic to the region (Meacutedail amp Queacutezel1997 Radford Catullo amp De Montmollin 2011) The stability of climate across timescalesincluding provision of Pleistocene refugia is thought to have prevented speciesrsquo extinctionsand facilitated speciation leading to this remarkably high biodiversity in theMediterranean

Kirk et al (2019) PeerJ DOI 107717peerj7296 226

Basin (Nogueacutes-Bravo et al 2008 Nogueacutes-Bravo Loacutepez-Moreno amp Vicente-Serrano 2012)The resulting characteristic shrub formations occur only in five areas California (UnitedStates) Chile South Africa Australia and the Mediterranean Basin in southern Europeand North Africa (Tomaselli 1977 Di Castri 1981 Cowling et al 1996)

Given the long history of human activity in the Mediterranean basin implementingsuitable management strategies to maintain and protect these biodiversity hotspots is amajor conservation dilemma (Blondel et al 2010) Primary lsquolsquonaturalrsquorsquo vegetation remainsin only 47 of the Mediterranean basin (Geri Amici amp Rocchini 2010) and many ofthese remaining wildlands are restricted to mountainous or coastal regions with steepslopes that preclude cultivation Some protected areas were implemented to conserveMediterranean shrublands but do not cover enough land to effectively buffer themagainst the erosion of vegetation diversity (Wilson et al 2007) With forecasted increasesin the frequency and magnitude of droughts and desertification due to climate changeareas of high plant endemism in the Mediterranean will likely lose their biodiversity andfunctional heterogeneity (De la Riva et al 2016Gauquelin et al 2016) Thus the challengeof integrating multiple uses of Mediterranean landscapes to effectively protect the regionrsquosremarkable diversity is fundamental to the conservation of Mediterranean flora

Jebel Ichkeul a limestone mountain within Le Parc National de LrsquoIchkeul in the northof the Republic of Tunisia represents a microcosm of the conservation issues faced byMediterranean vegetation including grazing In a recent assessment of the threats toImportant Plant Areas (IPAs) in the southern and eastern Mediterranean Radford Catulloamp De Montmollin (2011) identified overgrazing as the main threat to IPAs in Tunisia(see also Underwood et al 2009) Assessing the influence of grazing pressure and humanactivity on plant community composition at Jebel Ichkeul can therefore provide insightinto how these threats might be affecting other Mediterranean shrublands and to informtheir protection and management

In this paper we investigate the influence of grazing pressure on woody and herbaceousplant communities on Jebel Ichkeul a limestone mountain with strong mesic-xericgradients More specifically we ask (1) how grazing the abiotic environment and spacecollectively and separately shape herbaceous and woody plant community compositionon Jebel Ichkeul with a special focus on the influence of grazing (2) how the intensity ofhuman activity (usingO europaea size and density as a proxy) influences plant communitycomposition on Jebel Ichkeul in concert with and independently of these ecological driversand (3) how different grazing intensity levels influence plant beta-diversity on Jebel Ichkeuland which plant taxonomic and functional groups are associated with each level

We predicted that grazing as well as the abiotic environment and spatial relationshipsbetween sites would collectively and individually explain variation in plant communitycomposition across sites (Osem Perevolotsky amp Kigel 2007) Second we expected thathuman activities like woodcutting would apply a selective pressure on communitycomposition explaining additional plant beta-diversity across sites (Agra amp Nersquoeman2009) Third we predicted that grazing intensity should influence beta-diversity wherebeta-diversity should be highest under light browsing and should be lowest under bothgrazing extremes (ie unbrowsed and moderate-to-heavily intensity Laacutezaro et al 2016)

Kirk et al (2019) PeerJ DOI 107717peerj7296 326

Figure 1 Map of study sites on Jebel Ichkeul within Le Parc National de LrsquoIchkeul and location withinTunisia (inset) Points are coloured according to grazing class where yellow unbrowsed orange lightbrowsing and red moderate-to-high browsing

Full-size DOI 107717peerj7296fig-1

METHODSStudy areaSet in the Mateur plain Jebel Ichkeul is 25 km southwest of Bizerte in northeastern Tunisiaand 15 km south of the Mediterranean Sea (3710primeN 0940primeE Fig 1) It is surroundedalong its northern flanks by Garaet el Ichkeul an internationally important wetland boththe Jebel and lake are included within the National Park and World Heritage Site (IUCN2003) South of the southern perimeter of the Jebel there is intensive arable farmingpasture and orchards (UNEP-WCMC amp IUCN 2017) Formerly an island within LacIchkeul the dolomitic massif covers an area of 1363 ha (136 km2) 690 ha of this wasgazetted in the original declaration of the National Park in 1977 (Hollis 1977Hollis 1986)Jebel Ichkeul was listed as an Important Plant Area (IPA) in 2000 (Radford Catullo amp DeMontmollin 2011) and is nationally important (Peterken amp Radford 1971) Of the 29 plantfamilies known to occur in the Mediterranean region (Queacutezel 1981) 24 occur on the Jebelincluding the rare Tunisian endemic Teucrium schoenenbergeri (Fay 1980)

Jebel Ichkeul was listed in the European Red List of Habitats as one of the mosttypical examples of Olea europaea var sylvestris with Ceratonia siliqua and Pistacia lentiscusalong with southern Andalusia Menorca Sardinia Sicily Calabria and Crete (EuropeanEnvironment Agency 2017) At the time of our study the Jebelrsquos forests and maquisprovided grazing for livestock especially in autumn and winter (Hollis 1977) At othertimes of year the wetlands andmarshes of the National Park were a source of forage for over2500 + livestock (Anonymous 1998 unpublished data) Combined with uncontrolledwoodcutting for firewood on the southern slopes grazing pressure caused loss of vegetationcover and reduced species diversity leading to increased soil erosion and the spread ofinvasive plants (Fay 1980)

Kirk et al (2019) PeerJ DOI 107717peerj7296 426

The vegetation of Jebel Ichkeul is extremely heterogeneous due to its varied geology(dolomite calc-schist and marble) dissected relief high altitude (maximum 512 masl)the adjacent Lac Ichkeul and spatially-variable anthropogenic activity (Daoud-BouattourGammar Ghrabi amp Limam Ben Saad 2007) The northern slopes facing Lac Ichkeul aregenerallymesic with relatively continuous vegetation cover By contrast the southern slopesof the mountain feature xeric plant communities often in various states of degradationThese effects were most apparent close to gourbi village settlements and the Hammams(hot springs) located at the foot of the mountain Illegal quarries since closed also minedlimestone on the southern slopes of the Jebel and contributed to the xeric conditions atthese sites and potentially dust pollution (Kirk 1983) A road running along the southernflanks of themountain links the villages and quarries and is served by some secondary roadsfrom the Mateur Plain (Fig 1) On the southern slopes of the mountain calc-schist parentmaterial results in sites with high soil pH while the northern slopes are mostly dolomiticJebel Ichkeul occurs in the Mediterranean bioclimatic zone with a summer drought (Daget1977) Mean monthly temperatures range from 113 C in January (winter minimum 0 C)to a mean of 252 C in July (summer maximum 40 C) The average annual rainfall is 625mm with only 4 per cent of this falling in summer (Hollis 1977)

Plant community surveys and sampling designThe first author conducted all sampling between 18 June and 7 August 1983 To describeplant species distribution and abundance we located 78 quadrats on the Jebel stratifiedusing the data and vegetation maps in Fay (1980) for guidance Except for inaccessiblecrags and steep cliffs we sampled all vegetated areas Although the general areas of surveysites were not chosen randomly they were widely dispersed and spaced over the Jebel andsampled all community types in Fay (1980) At each site we threw a quadrat randomlyto locate the central stake We used a nested quadrat design with dimensions of 2 times 2m 5 times 5 m 707 times 707 m 10 times 10 m and 14 times 14 m (Bunce 1982) We estimatedcover-abundance for herbaceous and woody species within the 2 times 2 m quadrat andthe 10 times 10 m quadrat respectively using a modified Braun Blanquet scale (eg lt11ndash5 6ndash10 11ndash25 26ndash50 51ndash75 gt75) In the remaining quadrats we recordedpresence-absence of species but here present only data from the 2 times 2 m and 10 times 10 mquadrats On average three sites (nested quadrats) were surveyed during 8ndash10 h per day

Identifying plant species and development of functional groupsPlant names and occurrences in the national park were verified by a plant ecologist (ADaoud Bouattour University of Tunis pers comm 2012) and final taxonomy is basedon a catalogue of plants of Tunisia (Le Flocrsquoh Boulos amp Vela 2010) as well as a flora of thenational park compiled since our study (Daoud-Bouattour Gammar Ghrabi amp Limam BenSaad 2007) and the Euro-Med Plantbase (2017)

Because we surveyed woody and herbaceous speciesrsquo abundance at different spatialscales we considered herbaceous and woody plant communities separately in all analysesWe further classified species into taxonomic and functional groups predicted to responddifferently to grazing intensity Poaceae (Graminoids) Legumes Geophytes Forbs (annual

Kirk et al (2019) PeerJ DOI 107717peerj7296 526

and perennial) shrub legumes and shrubstrees (similar to those in Fernaacutendez-Lugo etal 2013) The lsquolsquoForbsrsquorsquo functional group contains ferns cactuses club mosses and someunidentified mosses We did not differentiate between annual and perennial species as inFernaacutendez-Lugo et al (2013) because our study took place over a single season

Assessing grazing intensityAt each of the quadrat locations within the 10 times 10 m square we ranked browsing bylivestock on the principal woody plant species We recognized that the effect of grazingcould vary between woody and herbaceous species that certain species might be preferredand that these preferences may differ by livestock type (Toacuteth et al 2018) However we wereunable to quantify grazing on herbaceous species and therefore assumed that browsing onwoody plants (probably almost entirely by goats) was indicative of overall grazing intensityon both woody and herbaceous species

We categorized grazing intensity into three levels (1) unbrowsed (ie dense forestsforests with gaps or high altitude arborescent matorral) (2) lightly browsed (ie densemiddle or low matorral with trees showing signs of browsing on shrubs) and (3) moderateto heavy browsing (ie bushes clipped to a hedge-like shape or severely clipped to shortstunted bushes) following Tomaselli (1977) We also recorded the activity of herbivorousmammals qualitatively by signs such as hair on tree trunks droppings and feeding signsfrom wild boars (Sus scrofa) such as uprooted tubers and soil disturbance but these werenot quantified Of the 78 sampled sites 71 (n= 55) were classified as unbrowsed (level1) 128 (n= 10) as lightly browsed (level 2) and 167 (n= 13) as moderately to heavilybrowsed (level 3) Moderately to heavily grazed sites (level 3) were generally of low altitudewith more rock outcropping and smaller-diameter O europaea trees See Fig 2 for anassessment of percent occurrence and relative abundance of woody species in each grazingclass

Measuring abiotic environmental variablesTo disentangle the effect of grazing from environmental covariates on plant speciescompositional turnover (Areacutevalo et al 2011a Areacutevalo et al 2011b) we Areacutevalo abioticvariables at the centre of each quadrat (Table 1 see below) We recorded altitude with abarometric field altimeter Because slope aspect plays an important role in the distributionof woody vegetation in the Mediterranean (Sternberg amp Shoshany 2001) we also measuredslope using an inclinometer (Abney LevelmdashEugene Dietzgen Chicago USA) and recordedcompass aspect as a proxy for solar insolation Parent material characteristics play a strongrole in shaping Mediterranean shrub communities (Molina-Venegas et al 2016) so werecorded the percentage of rock outcropping or rock cover at quadrats

Olea europaea as a proxy for anthropogenic activityWoody vegetation shades other plant species and thus modulates the composition andabundance of herbaceous plant species (Agra amp Nersquoeman 2009 Segoli et al 2012) Woodyvegetation especially O europaea is also a key indicator and surrogate for the intensityof anthropogenic activities including grazing and woodcutting Therefore we measuredO europaea stems (gt4 cm) in the 10 times 10 m quadrat at breast height (dbh c 15 m) and

Kirk et al (2019) PeerJ DOI 107717peerj7296 626

Figure 2 (A) Percent occurrence and (B) relative abundance of woody plant species on Jebel Ichkeulwithin each of three grazing classes Bars are coloured according to grazing class where yellowunbrowsed orange light browsing and red moderate-to-high browsing

Full-size DOI 107717peerj7296fig-2

at 30 cm above ground level for 69 sites (data were not available for 9 sites) We presentresults using stem size at 30 cm height because most trees were too small to obtain a breastheight measurement in shallow soils or heavily grazed sites We then calculatedO europaeadensities in 0ndash5 6ndash10 11ndash15 16ndash20 and gt20 cm circumference size classes Where treesbifurcated all stems were measured and a calculation used to provide a single diametercomparable with a tree of the same surface area (Supplemental Information 1)

Spatial autocorrelation between sitesTo assess positive spatial autocorrelation in plant community composition across themountain we computed the local Moranrsquos I for each site based on Hellinger distancesand tested for significance using 999 random permutations We also made the assumptionthat certain unmeasured environmental covariates were spatially autocorrelated Forexample prevailing westerly winds carry moisture from the lake and may create a humiditygradient from northerly mesic vegetation to xeric southern vegetation at Jebel Ichkeul(Fay 1980) Explicitly incorporating spatial relationships between sites can help accountfor this spatial gradient in humidity and soil moisture which can have a strong influence onMediterranean maquis vegetation (Sardans amp Pentildeuelas 2013) To account for the positivespatial autocorrelation between sites in subsequent analyses we computed positive distance-based Moranrsquos eigenvector maps (dbMEMs Dray Legendre amp Peres-Neto 2006 Legendreamp Legendre 2012 Legendre amp Gauthier 2014) based on site coordinates determined from

Kirk et al (2019) PeerJ DOI 107717peerj7296 726

Table 1 Biophysical variables measured at Jebel Ichkeul

Variable Meanplusmn SE Range N

Altitude masl 1430plusmn 127 6ndash404 78Aspect 20ndash360 78Slope 230plusmn 11 43-645 78Rock ( cover) 369plusmn 32 0ndash90 78pH 736plusmn 003 67ndash77 50Density Olea A (0ndash5 cm)30 cm 148plusmn 045 0ndash25 69DBH 203plusmn 063 0ndash32 69Density Olea B (6ndash10 cm)30 cm 203plusmn 042 0ndash15 69DBH 181plusmn 038 0ndash15 69Density Olea C (11ndash15 cm)30 cm 141plusmn 024 0ndash8 69DBH 120plusmn 021 0ndash7 69Density Olea D (16ndash21 cm)30 cm 074plusmn 018 0ndash8 69DBH 039plusmn 010 0ndash3 69Density Olea E (21 cm+)30 cm 041plusmn 011 0ndash4 69DBH 020plusmn 008 0ndash4 69

Google Earth Local Moranrsquos I and the positive dbMEMs were calculated using the Rpackage adespatial (Dray et al 2018)

STATISTICAL ANALYSISBeta-diversity of herbaceous and woody communitiesWe assessed the compositional differences between sites in herbaceous and woodycommunities as the Hellinger distances between sites for each community type (Andersonet al 2011) To determine which sites harboured more ecologically unique communitieswe quantified each sitersquos local contribution to beta-diversity (LCBD) for each communitytype (Legendre amp De Caacuteceres 2013) To identify the plant groups that most strongly drivebeta-diversity between sites we also assessed each speciesrsquo contribution to beta-diversity(SCBD) summed for each functional group (Poaceae legumes forbs geophytes Legendreamp De Caacuteceres 2013) We tested the significance of LCBD values using permutation testswith 999 iterations

Disentangling the ecological and anthropogenic drivers of plantbeta-diversityWe first determined the combined and separate influences of ecological drivers on plantcommunity structure We used distance-based redundancy analyses (db-RDA) to quantifythe combined influence of the abiotic environment (altitude slope aspect and rock-crop cover) grazing intensity and spatial autocorrelation (dbMEMs) on the Hellinger

Kirk et al (2019) PeerJ DOI 107717peerj7296 826

dissimilarities between sites in the herbaceous and the woody plant communities Modelsignificance tests were based on 999 permutations We then used variation partitioningto disentangle the individual influences of the abiotic environment grazing and spatialautocorrelation on the herbaceous and woody plant beta-diversity (Borcard Legendre ampDrapeau 1992 Peres-Neto et al 2006) For the two plant community types we partitionedHellinger dissimilarities into fractions of variation explained by [a] abiotic environmentalvariables (altitude slope aspect and rock outcrop cover) [b] grazing classes [c] spatialautocorrelation (dbMEMs) We then tested the significance of each individual fractionusing permutation tests with 999 iterations

To determine the influence of anthropogenic activity on plant community compositionwe repeated the db-RDA and variation partitioning analyses including only sites withO europaea sizes and densities For the twoplant community types we therefore partitionedHellinger dissimilarities into fractions of variation explained by [a] abiotic environmentalvariables (altitude slope aspect and rock outcrop cover) [b] grazing classes [c] humanactivity (O europaea sizes and densities) and [d] spatial autocorrelation (db-MEMs)

All db-RDA and variation partitioning analyses were conducted using the R packagevegan (Oksanen et al 2019)

Influence of grazing intensity on plant beta-diversityTo determine how grazing intensity influences plant beta-diversity we tested for significantdifferences in the variance of Hellinger dissimilarities between the three levels of grazingintensity We computed a test of the multivariate homogeneity of group dispersionsbetween grazing classes and assessed significance of between-group differences using 999permutations (Anderson 2006) Smaller multivariate site distances to the grazing-specificcentroids are interpreted as biotic homogenization whereas larger distances are interpretedas biotic differentiation The multivariate homogeneity of group dispersions test wasperformed using the R package vegan Oksanen et al (2019)

To further tease out the influence of grazing intensity from the effects of abiotic spatialand anthropogenic covariates we performed one-way permutational multivariate analysisof variance (PERMANOVA) models with grazing as a factor and controlling for covariatesWe selected abiotic and spatial variables through sequential AICc comparison of db-RDAmodels We first compared db-RDA models including all abiotic covariates and grazingand retained abiotic variables from the best model We repeated this with spatial variables(dbMEMs) A PERMANOVA was then conducted to evaluate grazing intensityrsquos effectson beta-diversity with the retained abiotic and spatial variables as covariates Becausegrazing intensity is spatially structured we forced the order of entry in the PERMANOVAby entering grazing intensity first in one set of models and then entering it last in a secondset of models PERMANOVAs were conducted first for all 78 sites then separately forthe subset of 69 sites for which anthropogenic variables (ie O europaea densities by sizeclass) were available We assessed explanatory power using Type I sums of squares in thePERMANOVAs We performed the db-RDAs (best model subsets) and permutationalmultivariate analysis of variance using PERMANOVA + (Anderson Gorley amp Clarke2008)

Kirk et al (2019) PeerJ DOI 107717peerj7296 926

Figure 3 Local contributions to beta diversity (LCBD) of sites grouped by grazing class for (A) herba-ceous and (B) woody plant communities on Jebel Ichkeul

Full-size DOI 107717peerj7296fig-3

RESULTSSpatial autocorrelation between sitesMoranrsquos I tests demonstrated positive spatial autocorrelation in plant communitycomposition within grazing classes particularly in the woody communities (Fig S1)Herbaceous community composition showed significant positive spatial autocorrelationin 21 of 55 (382) of unbrowsed sites five of 10 (50) of lightly-browsed sites and11 of 13 (846) of moderately-to-heavily browsed sites Most strikingly the sites withthe highest positive spatial autocorrelation for herbaceous species were on the southernheavily grazed slopes of the Jebel (Fig S1) Woody community composition showed highlevels of positive spatial autocorrelation across all grazing classes 47 out of 55 (855)in unbrowsed sites nine out of 10 sites (90) in lightly-browsed sites and 11 out of 13(846) in moderately-to-heavily browsed sites The influence of grazing and of otherecological and anthropogenic processes therefore needed to be disentangled from thepervasive effect of spatial autocorrelation between sites in both community types

Beta-diversity of herbaceous and woody communitiesSites with the highest local contribution to beta-diversity (LCBD) (ie morecompositionally unique sites) were generally located on the lakeside face of the mountaincontributing up to 204 (site 44) of total herbaceous beta-diversity and up to 247(site 51) of the overall woody beta-diversity on the Jebel (Fig S2) LCBD also differedsignificantly among grazing classes for herbaceous plant species (F2 = 431 df = 2P = 002) but not for woody species (F2 = 035 df = 2 P = 07 Fig 3)Forbs contributed most to herbaceous beta-diversity (SCBD) followed by Poaceae

Legumes and Geophytes (Fig 4) Woody beta-diversity was largely attributed to woodyspecies other than legumes (Fig 4) Legumes almost exclusively contributed to beta-diversity in unbrowsed and lightly browsed sites It is also notable that of the 35 woodyspecies only 12 occurred in heavily grazed sites (Fig 2)

Kirk et al (2019) PeerJ DOI 107717peerj7296 1026

Figure 4 Species contributions to beta diversity (SCBD) of each functional group in sites of each graz-ing class for (A) herbaceous and (B) woody plant communities

Full-size DOI 107717peerj7296fig-4

Disentangling the ecological and anthropogenic drivers of plantbeta-diversityThe importance of ecological drivers on plant beta-diversity differed between herbaceousand woody communities In herbaceous communities the db-RDA showed the abioticenvironment grazing intensity and spatial variables collectively explained relatively little(Radj = 0047 p = 0001) of the variation in beta-diversity between the 78 sites Variationpartitioning indicated that abiotic variables ([a]) significantly explained a small portion ofthis variation (Radj = 0015 p = 0007) while grazing intensity ([b]) and spatial variables([c]) showed no significant effects (Fig 5A) The individual effect of grazing intensity ([b])could not be partitioned from the effects of abiotic and spatial variables (Radj =minus0003 p= 08) Here a negative Radj indicated that grazing intensity only explained the variationin beta-diversity when considered with the abiotic and spatial covariates (Legendre ampLegendre 2012) This suggests that grazing intensity is highly correlated with the abioticand spatial variables making it difficult to isolate its influence on herbaceous communitycomposition

In the woody community (Fig 5B) the db-RDA demonstrated that 125 (Radj = 0125p = 0001) of the variation in beta-diversity could be attributed to the combined effectof the abiotic environment grazing intensity and spatial variables As in the herbaceouscommunity the individual effect of grazing intensity ([b]) could not be partitioned from

Kirk et al (2019) PeerJ DOI 107717peerj7296 1126

Figure 5 Variation partitioning of herbaceous (A C) and woody (B D) beta-diversity In (A) and (B)beta-diversity explained by abiotic grazing and spatial variables across 78 sites In (C) and (D) beta-diversity explained by abiotic grazing O europaea sizes and densities (as proxy for human activity) andspatial variables across 69 sites

Full-size DOI 107717peerj7296fig-5

the effects of the abiotic environment and spatial variables (Radj =minus0005 p= 09) Spacealone ([c]) significantly explained the greatest portion of the variation in woody beta-diversity across sites (Radj = 0044 p = 0001) while the individual effect of the abioticenvironment ([a]) did not significantly explain plant beta-diversity after partitioning(Radj = 0011 p = 011)

To investigate the influence of anthropogenic activity on plant beta-diversity werepeated these analyses on 69 sites with the inclusion of O europaea densities as a proxyfor human activities such as woodcutting Including O europaea density covariates slightlyincreased the explanatory power of both db-RDAmodels the combined effect of the abioticenvironment grazing intensity O europaea densities and spatial variables explained 76(Radj = 0076 p= 0001) of herbaceous beta-diversity and 164 (Radj = 0164 p= 0001)of woody beta-diversity

Variation partitioning further discerned the individual effects of the ecological andanthropogenic covariates In the herbaceous community space alone ([d]) still explained

Kirk et al (2019) PeerJ DOI 107717peerj7296 1226

Figure 6 Test for homogeneity of multivariate dispersions between grazing classes for (A) herbaceousand (B) woody species

Full-size DOI 107717peerj7296fig-6

most variation (Radj = 0019 p = 003) followed closely by the abiotic environment([a]) (Radj = 0021 p = 003) (Fig 5C) The individual effects of grazing intensity ([b])and O europaea densities ([c]) were both unable to significantly explain variation inherbaceous beta-diversity ([b] Radj = 0001 p= 04 [c] Radj = 0012 p= 009) althoughthe effect of O europaea densities was significant when considered with all other covariates(Radj = 0015 p = 001) Importantly in the woody community the individual effects ofgrazing intensity ([b]) and O europaea densities ([c]) were successfully partitioned fromthe effects of spatial variables and the abiotic environment respectively explaining 22 (p= 0001) and 21 (p= 003) of woody beta-diversity (Fig 5D) However spatial variablesstill explained the largest fraction of woody beta-diversity (Radj = 0024 p = 001)

Influence of grazing intensity on plant beta-diversityThe test for the homogeneity of multivariate dispersions indicated that beta-diversitydiffered significantly among grazing classes in the herbaceous communities (p lt0001)and was marginally significantly different among grazing classes in the woody communities(p= 0065 Fig 6) Herbaceous communities were more differentiated among unbrowsedsites and most homogenized in moderately-to-heavily grazed sites (Fig 6) Although thepattern was less pronounced in the woody community moderately-to-heavily grazed sitesshowed more homogenization than sites with less grazing pressure (Fig 6)

Pairwise tests demonstrated significant differences in herbaceous beta-diversity betweenunbrowsed and lightly browsed sites (p lt 0001) and between unbrowsed and moderately-to-heavily browsed sites (p lt 0001) However herbaceous communities did not appearto differ significantly under light browsing and moderate-to-heavy browsing (p= 098)Woody beta-diversity differed significantly between unbrowsed and moderately-to-heavilybrowsed sites (p= 022) However beta-diversity did not differ significantly between lightlybrowsed and unbrowsed sites (p = 055) or between lightly browsed and moderately-to-heavy browsed sites (p = 016)

Kirk et al (2019) PeerJ DOI 107717peerj7296 1326

Table 2 Permutational multivariate analysis of variance (Hellinger distance) testing herbaceous and woody species composition among graz-ing classes (crossed design grazing as fixed effect) at 78 sites controlling for abiotic and spatial covariates (a) Herbaceous (b) Woody

Source Df SS MS Pseudo F P (perm) unique permutations

(a) Herbaceous speciesGrazing first

Grazing 2 26875 13437 1995 00003 9830MEM1 1 15919 15919 2363 00013 9880Residuals 74 49845 06736Total 77 5412

Grazing lastMEM1 1 15664 15664 2326 00006 9858Grazing 2 2713 13565 2014 00002 9839Residuals 74 49845 06735Total 77 54124

(b) Woody speciesGrazing first

Grazing 2 31109 15555 5399 00001 9897MEM1 1 10529 10529 3655 00004 9926MEM2 1 1626 1626 5644 00001 9936MEM3 1 067419 067419 2340 00127 9926MEM4 1 15297 15297 5309 00001 9924Residuals 71 21456 028811Total 77 28449

Grazing lastMEM1 1 11044 11044 3833 00004 9918MEM2 1 1915 1915 6647 00001 9928MEM3 1 16641 16641 5776 00001 9921MEM4 1 24528 24528 8514 00001 9927Grazing 2 085743 042871 1488 00791 9902Residuals 71 20456 028811Total 77 28449

To determine how the abiotic and spatial covariates influenced these differences weperformed a one-way permutational multivariate analysis of variance with grazing levelsas a factor and controlling for covariates derived from the lsquobestrsquo subset db-RDA models(Tables 2 and 3) For both herbaceous and woody communities grazing intensity hada significant effect (p lt 0001) for herbaceous and woody species for 69 sites and forherbaceous species for 78 sites regardless of the order of entry in the model Howevergrazing had a marginally significant effect on beta-diversity in woody communities for 78sites when it was entered last confirming that was spatially structured

DISCUSSIONOur results confirmed that the abiotic environment grazing and spatial relationshipsbetween sites collectively shaped plant communities at Jebel Ichkeul although their

Kirk et al (2019) PeerJ DOI 107717peerj7296 1426

Table 3 Permutational multivariate ANOVA (Hellinger distance) testing herbaceous and woody species composition among grazing classes(crossed design grazing as fixed effect) at 69 sites withO europaea densities controlling for abiotic and spatial covariates (a) Herbaceous (b)Woody

Source Df SS MS Pseudo F P (perm) unique permutations

(a) Herbaceous speciesGrazing first

Grazing 2 21619 1081 1575 00055 9833MEM1 1 16556 16556 2413 00004 9868Residuals 65 44603 06862Total 68 48421

Grazing lastMEM1 1 16396 16396 2389 00003 9871Grazing 2 2178 1089 1587 00066 9827Residuals 65 44603 06862Total 68 48421

(b) Woody speciesGrazing first

Grazing 2 32019 16009 5296 00001 9904MEM1 1 10699 10699 3539 00004 9920MEM2 1 10382 10382 3435 00007 9926MEM3 1 12522 12522 4143 00002 9920MEM12 1 09975 09975 3300 00007 9923Residuals 62 18741 030227Total 68 263

Grazing lastMEM1 1 11956 11956 3955 00001 9924MEM2 1 13177 13177 4359 00001 9918MEM3 1 22144 22144 7326 00001 9914MEM12 1 14074 14074 4656 00001 9934Grazing 2 14246 071229 2357 00010 9898Residuals 62 18741 030227Total 68 263

influence was strongest in woody communities However the individual effect of grazingcould not be disentangled from the strong effect of space on overall plant beta-diversityin both the herbaceous and woody communities potentially reflecting the spatiallyautocorrelated nature of the grazing regime Grazing pressure and human activityincorporated using O europaea sizes and densities as a proxy explained additionalbeta-diversity only in the woody community types and had no individual effects onthe herbaceous community Human activity thus appears to exert a slight selective pressurein the woody community overall though not in its herbaceous counterpart

Although the overall effect of grazing onbeta-diversitywas highly confoundedwith spacegrazing intensities drove significant differences in beta-diversity in both the herbaceousand woody communities Moderate-to-heavy grazing appears to homogenize herbaceouscommunities and to a lesser extent woody communities Herbaceous communities in

Kirk et al (2019) PeerJ DOI 107717peerj7296 1526

grazed sites regardless of intensity differed from ungrazed sites On the other hand woodybeta-diversity differed only between unbrowsed and moderately-to-heavily browsed sitessuggesting that woody communities were able to withstand light levels of browsingGrazingrsquos main effect thus appeared to lie in shaping communities according to speciesrsquosensitivity to grazing especially at high intensity levels although its overall effect oncommunity composition was small

It is perhaps unsurprising that grazing had little influence on plant communitycomposition in Jebel Ichkeul given that grazingrsquos effect on community compositionand species diversity is scale-dependent and can be challenging to isolate (Brinkmann et al2009Areacutevalo et al 2011aAreacutevalo et al 2011b) As space almost universally explainedmostof the variation in plant beta-diversity some unmeasured spatially-structured variablesmight be confounding the effect of grazing It is also important to note that this studywas initially designed to sample floristically diverse communities and thus intentionallyneglected the denuded southern border of the mountain rather than ensuring an evensampling of different grazing regimes Had we employed a grid sampling design stratifiedby grazing the overall effects of grazing on plant community composition might have beenclearer

Despite these sampling issues we found that different levels of grazing intensity structurecompositionally dissimilar communities Our results also suggested that herbaceous andwoody communities responded differently to grazing intensity Herbaceous communitiesappeared to be sensitive to both light and moderate to heavy grazing and becamehomogenized under higher grazing pressure On the other hand woody communitiesonly differed from their unbrowsed state in response to moderate to heavy grazingsuggesting that woody communities were robust to intermediate grazing pressure (GabayPerevolotsky amp Shachak 2006 Miguel-Ayanz Garciacutea-Calvo amp Fernaacutendez-Olalla 2010)This is consistent with a similar study in the Canary lslands where woody shrubs declinedstrongly from abandoned to heavily grazed areas (Fernaacutendez-Lugo et al 2013) Assumingthat grazing intensity on woody species is representative of grazing pressure on herbaceousspecies our findings suggest that management strategies should vary for herbaceous andwoody species under grazing regimes However we caution that our interpretations couldbe confounded by differences in species diversity (144 herbaceous species 35 woodyspecies) and by spatial autocorrelation in grazing pressure and beta-diversity within thetwo community types

Although the signal of grazing intensityrsquos effect on plant community composition wasweak this is likely due to limitations imposed by sampling design categorical grazingpressure measurements and the assumption of equal grazing on herbaceous and woodycommunities irrespective of livestock preferences In fact plant species on Jebel Ichkeuldiffer in their responses to grazing where some species are sensitive to increased grazingpressure some are maintained by moderate grazing pressure and some are able to persistunder even heavy grazing pressure This range of responses resulted in a heterogeneouscover of community types along the mountainrsquos grazing gradient For example somewoody browse species (eg Ceratonia siliqua Coronilla valentina Erica arborea Ericamultiflora and Rhamnus lycioidesmdash(Le Houerou 1981) are sensitive to grazing and were

Kirk et al (2019) PeerJ DOI 107717peerj7296 1626

negatively correlated with the grazing gradient Instead the increased humidity on thenorth side of the lake and deep valley soils allowed the development of closed canopyforest dominated by O europaea Ceratonia siliqua and Phillyrea angustifolia Althoughthese communities are species poor they are of high ecological value because they arerepresentative of the Mediterranean basinrsquos lsquointactrsquo vegetation presenting an importanteducation and scientific opportunity for researchers and visitors to the national parkElsewhere in the Mediterranean the Olea-Ceratonion formation is restricted to zonesbelow 300 m (Tomaselli 1977)

Low matorral vegetation comprised of species with the ability to persist under heavygrazing pressure was scattered on denuded xeric slopes along the mountainrsquos heavilygrazed southeastern perimeter near gourbi villages on the southern face and the westernend of the mountain These highly grazed areas harboured herbaceous indicators ofovergrazing including weedy species (Asteraceae) like Atractylis cancellata Carthamuslanatus and Scolymus hispanicus and were lacking many sensitive woody browse species(eg Ceratonia siliqua Coronilla valentina Erica arborea and Rhamnus lycioides) EvenPistacia lentiscus which is normally resistant to grazing (Le Houerou 1981) was sparsein these areas These sensitive woody species only occur on the southern slopes of themountain when sheltered from grazing at higher elevations below the ridge crest outsidethe range of goat herds This overgrazing creates inhospitable conditions for establishmentof other protective vegetation on the southern slopes of the mountain increasing insolationand aridity and potentially exacerbating this problem

While overgrazing erodes plant diversity moderate grazing can maintain openingsbetween woody vegetation patches in some areas of Jebel Ichkeul removing competingplant biomass providing nutrients from livestock manure and ultimately increasing plantspecies richness For example cattle (45ndash60) and numerous wild boar that foraged in theJebel Ichkeul created the open conditions which suit Orchidaceae and Liliaceae (Fay 1980)as well as various socio-economically important pasture grass species (Noy-Meir amp Oron2001) Most of the orchid species in the Ampelodesma stand at Saida Lalia Hadan whichcontains the largest O europaea in the Park are also maintained by intermediate levels ofgrazing (Fay 1980) Nevertheless Gramineae which include some valuable pasture speciesappeared to be excluded from heavily grazed sites in Jebel Ichkeul The level of grazingneeded to maintain the richness of different plant groups should thus be investigated infurther research

Importantly more knowledge of the herbivore community might have shed morelight on the effects of grazing on plant community structure An estimated 2500 grazinganimals browsed on the mountain in 1980 (Fay 1980 Anonymous 1998 unpublisheddata) primarily comprised of goats in addition to sheep and cattle (Hollis 1977) Livestocktype affects grazing impacts on vegetation structure and composition (Toacuteth et al 2018)with goats generally being the most destructive species (Tomaselli 1981) Although thecomposition and abundance of the herbivore community were not assessed in this studytheir effect on the outcomes of grazing merit further investigation in future research

Translating these findings into conservation and management recommendationsrequires an acknowledgement of three major changes in human occupation and abiotic

Kirk et al (2019) PeerJ DOI 107717peerj7296 1726

conditions at Jebel Ichkeul since sampling occurred in 1983 First around 1000 peoplelived in the gourbi villages along the southern flanks of the Jebel from 1983 to 2004 butthis number dropped to 400 by 2008 declining until these villages were entirely evictedAlthough grazing pressure has not been completely eradicated the change in livestockabundance and distribution is likely substantial Second the limestone quarries whichwere contributing to xeric conditions through dust deposition have been closed down(UNEP-WCMC amp IUCN 2017) Third Tunisia has undergone climate change since atleast the 1970s (Paeth et al 2009) and faces a predicted 20 decrease in precipitationand 1 and 3 C increase in temperature by 2050 (Tramblay El Adlouni amp Servat 2013Dakhlaoui et al 2017) These changing climatic conditions in addition to reduced dustdeposition and altered grazing regimes have probably substantially altered the relationshipsinvestigated in this study While the dams installed at all of the major rivers flowing intoLac Ichkeul (UNEP-WCMC amp IUCN 2017) could have created more xeric conditionsespecially on the northern slopes of the Jebel this effect was likely temporary and minimaldue to substantial rainfall between 2003 and 2006 which restored former water and salinitylevels

CONCLUSIONS AND MANAGEMENT RECOMMENDATIONSMaintaining the heterogeneity and diversity of vegetation types and structure at JebelIchkeul is an important management goal in order to maintain the regionrsquos high ecologicalvalue (see Gabay Perevolotsky amp Shachak 2006 Miguel-Ayanz Garciacutea-Calvo amp Fernaacutendez-Olalla 2010) A key step towards this goal is the consistent monitoring of the spatialdistribution and structure of vegetation particularly with regards to rare herbaceousspecies and woody species Importantly our study provides a useful baseline of the plantassemblages at Jebel Ichkeul withwhich to compare future vegetation changes Re-surveyingplant assemblages at our survey sites could reveal how changes in grazing intensityclimate and human occupation have affected plant community composition and diversityover several decades Going forward vegetation structure could be monitored using acombination of remote sensing and on-the-ground metrics of patch size patch densityor edge density (Glasser et al 2013) However ground truthing of species compositiongap dimensions and height of vegetation would be essential given that grazing pressure isfocused below the canopy (Glasser et al 2013) though LiDAR imagery may circumventthis problem

Recovering the maquis vegetation particularly on the southern base of the mountainwould protect the watershed by reducing soil erosion and consequent sedimentation ofadjacent marshland and should therefore be a high priority goal Restricting grazing andwoodcutting on the Jebel would directly facilitate this recovery According to Fay (1980)the spatially heterogeneous grazing pressure is the Jebel Ichkeul parkrsquos main conservationand management problem Managed grazing would be preferential to address this issue(Areacutevalo et al 2012 Laacutezaro et al 2016) rather than complete eradication of livestockwhich can have unintended consequences such as biological invasions or successioninto less desirable plant communities (Mata et al 2014 Areacutevalo et al 2012) As such

Kirk et al (2019) PeerJ DOI 107717peerj7296 1826

implementing strategies to control grazing on the southern slopes of the mountaincombined with wood-cutting restrictions could enable vegetation to re-establish on thedegraded southern slopes

Effective management of the park should balance biodiversity goals with multiple usesincluding traditional pastoralism (Verduacute Crespo amp Galante 2000 Perevolotsky 2005)to maintain valuable ecosystem services provided by the Jebel within Ichkeul nationalpark (Daly-Hassen 2017) A systematic conservation planning tool could prioritize areasbased on maximizing spatiotemporal variation in plant assemblages (Levin et al 2013)effectively maintaining vegetation diversity across the landscape delineating zones forcontrolled grazing and setting biodiversity targets for specific vegetation types

Given the predicted decreases in precipitation and increases in aridity (Tramblay ElAdlouni amp Servat 2013 Dakhlaoui et al 2017) in North Africa it is vital to consider therole of the Jebelrsquos complex topography and its proximity to Lake Ichkeul in creatingmicroclimate refugia (McLaughlin et al 2017) which can locally buffer climate extremesfor many endemic plant species (Harrison amp Noss 2017) The parkrsquos potential function as aclimate refugium solidifies the need to effectively manage grazing and other anthropogenicactivities in order to maintain plant diversity and ecosystem functioning in this vulnerableregion

ACKNOWLEDGEMENTSThis research was conducted as part of an MSc Thesis in Conservation at UniversityCollege London by the first author Special thanks to JM Fay for pioneering this project(a plant inventory of Jebel Ichkeul completed in 1980) and his devotion to protectedarea conservation in Africa AC Stevenson of UCL and R Vickery of the British Museum(Natural History) helped with plant species identifications We would also like to thankJM Fay B Green the late GE Hollis JD Skinner A Warren and JB Wood for logisticalsupport or advice BT Collins helped with the equations for olive diameter measurementsWe are especially indebted to MJ Anderson for statistical advice L Olson for help withpreliminary variance partititioning C Fauvelle and C Widdifield for help with GIS andA Daoud-Bouattour and M Ouali for checking plant nomenclature We thank A Daoud-Bouattour RG Gavilan PW Rundel and AC Stevenson for general comments on earlierdrafts of this manuscript Two anonymous referees provided extremely helpful commentsduring formal review of the manuscript

ADDITIONAL INFORMATION AND DECLARATIONS

FundingThis work was supported by the Natural Environment Research Council (UK) Fundingfor the manuscript for DAK was provided by Ivan and Thelma Kirk KH was supportedby an FQRNT doctoral research scholarship The funders had no role in study design datacollection and analysis decision to publish or preparation of the manuscript

Kirk et al (2019) PeerJ DOI 107717peerj7296 1926

Grant DisclosuresThe following grant information was disclosed by the authorsNatural Environment Research Council (UK)FQRNT doctoral research scholarship

Competing InterestsThe authors declare there are no competing interestsHoweverDavidAnthonyKirk is ChiefExecutive Officer of and owns the ecological research consultancy Aquila Conservationamp Environment Consulting There is no financial relationship of any kind between thesubmitted research and the company lsquoAquila Conservation amp Environment Consultingrsquoor any other organization The original fieldwork and preparation of the original thesiswere funded by the Natural Environment Research Council (UK) and done at UniversityCollege London in 1983

Author Contributionsbull David Anthony Kirk conceived and designed the experiments performed theexperiments analyzed the data contributed reagentsmaterialsanalysis tools preparedfigures andor tables authored or reviewed drafts of the paper approved the final draftbull Katherine Heacutebert analyzed the data contributed reagentsmaterialsanalysis toolsprepared figures andor tables authored or reviewed drafts of the paper approved thefinal draftbull Frank Barrie Goldsmith conceived and designed the experiments approved the finaldraft helped with field design supervised project commented on manuscript

Data AvailabilityThe following information was supplied regarding data availability

Kirk DA Heacutebert K Goldsmith F B (2019) Plant biodiversity data andenvironmental and spatial data from Jebel Ichkeul a limestone mountain in northernTunisia (1983) NERC Environmental Information Data Centre httpsdoiorg1052854e36cdfa-0281-423f-99a8-e7c331b2e0d1

Supplemental InformationSupplemental information for this article can be found online at httpdxdoiorg107717peerj7296supplemental-information

REFERENCESAgra H Nersquoeman G 2009Woody species as landscape modulators their effect on

the herbaceous plants in a Mediterranean maquis Plant Ecology 205165ndash177DOI 101007s11258-009-9606-3

AndersonMJ 2006 Distance-based tests for homogeneity of multivariate dispersionsBiometrics 62245ndash253 DOI 101111j1541-0420200500440x

AndersonM Gorley RN Clarke KR 2008 PERMANOVA+ for PRIMER guide tosoftware and statistical methods Plymouth Primer-E

Kirk et al (2019) PeerJ DOI 107717peerj7296 2026

AndersonMJ Crist TO Chase JM VellendM Inouye BD Freestone AL Sanders NJCornell HV Comita LS Davies KF Harrison SP Kraft NJB Stegen JC SwensonNG 2011 Navigating the multiple meanings of szlig diversity a roadmap for thepracticing ecologist Ecology Letters 1419ndash28 DOI 101111j1461-0248201001552x

Areacutevalo JR De Nascimento L Fernaacutendez-Lugo S Mata J Bermejo L 2011a Grazingeffects on species composition in different vegetation types (La Palma CanaryIslands) Acta Oecologica 37230ndash238 DOI 101016jactao201102006

Areacutevalo JR De Nascimento L Fernandez-Lugo S Saro I Camacho A Mata J BermejoL 2011b Effects of abandoning long-term goat grazing on species compositionand species richness of pastures at La Gomera Canary Islands Spanish Journal ofAgricultural Research 9113ndash123 DOI 105424sjar20110901-076-10

Areacutevalo JR Fernaacutendez-Lugo S De Nascimento L Bermejo LA Naranjo A 2012Grazing management and impact in the Canary islands rethinking sustainable useIn Areacutevalo JR ed Grazing ecology vegetation and soil impact Hauppauge NovaScience Publishers Inc

Blondel J Aronson J Bodiou J-Y Boeuf G 2010 The mediterranean region biologicaldiversity in space and time Oxford Oxford University Press

Borcard D Legendre P Drapeau P 1992 Partialling out the spatial component ofecological variation Ecology 731045ndash1055 DOI 1023071940179

Brinkmann K Patzelt A Dickhoefer U Schlecht E Buerkert A 2009 Vegetation pat-terns and diversity along an altitudinal and a grazing gradient in the Jabal al Akhdarmountain range of northern Oman Journal of Arid Environments 731035ndash1045DOI 101016jjaridenv200905002

Bunce RGH 1982 A field key for classifying British woodland vegetation Part 1 Cam-bridge Institute of Terrestrial Ecology 103 pp

Campbell K Donlan CJ 2005 Feral goat eradications on islands Conservation Biology191362ndash1374 DOI 101111j1523-1739200500228x

Cowling RM Rundel PW Lamont BB ArroyoMK ArianoutsouM 1996 Plant diver-sity in mediterranean-climate regions Trends in Ecology and Evolution 11362ndash366DOI 1010160169-5347(96)10044-6

Daget P 1977 Le bioclimat mediterraneen analyse des formes climatiques par le systemedrsquoemberger Vegetation 3487ndash103 DOI 101007BF00054477

Dakhlaoui H Ruelland D Tramblay Y Bargaoui Z 2017 Evaluating the robustnessof conceptual rainfall-runoff models under climate variability in northern TunisiaJournal of Hydrology 550201ndash217 DOI 101016jjhydrol201704032

Daly-Hassen H 2017 Valeur eacuteconomique des services eacutecosysteacutemiques du Parc National delrsquoIchkeul Tunisie Gland Switzerland Malaga IUCN

Daoud-Bouattour A Gammar Ghrabi Z Limam Ben Saad S 2007Guide illustreacute desPlantes du Parc National de LrsquoIchkeul Ariana Eco-Ressources International

De la Riva EG Lloret F Peacuterez-Ramos IM Marantildeoacuten T Saura-Mas S Diacuteaz-Delgado RVillar R 2016 The importance of functional diversity in the stability of Mediter-ranean shrubland communities after the impact of extreme climatic events Journalof Plant Ecology 10rtw027 DOI 101093jpertw027

Kirk et al (2019) PeerJ DOI 107717peerj7296 2126

Di Castri F 1981 Mediterranean-type shrublands of the world In Di Castri F GoodallDWSR Specht RL eds Ecosystems of the World II Mediterranean Type ShrublandsLondon Elsevier 1ndash52

Dray S Legendre P Peres-Neto PR 2006 Spatial modelling a comprehensive frame-work for principal coordinate analysis of neighbour matrices (PCNM) EcologicalModelling 196483ndash493

Dray S Bauman D Blanchet G Borcard D Clappe S Guenard G Jombart T LarocqueG Legendre P Madi NWagner HH 2018Multivariate multiscale statisticalanalysis package lsquolsquoAdespatialrsquorsquo R Foundation for Statistical Computing Availableat https cranr-projectorgwebpackagesadespatial indexhtml

Elias D Houmllzel N Tischew S 2018 Goat paddock grazing improves the con-servation status of shrub-encroached dry grasslands Tuexenia 38215ndash233DOI 1014471201838017

Elias D Tischew S 2016 Goat pasturingmdasha biological solution to counteract shrubencroachment on abandoned dry grasslands in Central Europe AgricultureEcosystems and Environment 23498ndash106 DOI 101016jagee201602023

Euro-Med Plantbase 2017 The information resource for Euro-Mediterranean plantdiversity Available at httpwwwemplantbaseorghomehtml (accessed on 01January 2017)

European Environment Agency 2017 G24 Olea europaea-Ceratonia siliqua woodlandEuropean red list of habitatsmdashForests habitat group Available at https forumeioneteuropeaeu european-red-list-habitats library terrestrial-habitats g-forests g24-olea-europea-ceratonia-siliqua-woodland G24 Olea-europea-Ceratonia-siliqua-woodland

Falcucci A Maiorano L Boitani L 2007 Changes in land-useland-cover patternsin Italy and their implications for biodiversity conservation Landscape Ecology22617ndash631 DOI 101007s10980-006-9056-4

FayM 1980 Flora of the National Park of Ichkeul Tunisia Unpublished ReportLondon UK University College London 199 pp

Fernaacutendez-Lugo S Areacutevalo JR De Nascimento L Mata J Bermejo LA 2013 Long-termvegetation responses to different goat grazing regimes in semi-natural ecosystemsa case study in Tenerife (Canary Islands) Applied Vegetation Science 1674ndash83DOI 101111j1654-109X201201211x

Gabay O Perevolotsky A ShachakM 2006 Landscape mosaic for enhancing biodiver-sity on what scale and how to maintain it Options 4945ndash49

Gauquelin T Michon G Joffre R Duponnois R Geacutenin D Fady B Dagher-KharratMB Derridj A Slimani S Badri W Alifriqui M Auclair L Simenel R AderghalM Baudoin E Galiana A Prin Y Sanguin H Fernandez C Baldy V 2016Mediterranean forests biocultural heritage and climate change A social-ecologicalperspective Regional Environmental Change DOI 101007s10113-016-0994-3

Geri F Amici V Rocchini D 2010Human activity impact on the heterogeneity of aMediterranean landscape Applied Geography 30370ndash379DOI 101016japgeog200910006

Kirk et al (2019) PeerJ DOI 107717peerj7296 2226

Glasser T Hadar L Navon Y Perevolotsky A 2013 Innovative monitoring of goatgrazing effects on landscape structural properties DOI 10131402126808003

Gomez-Campo C 1985 Plant conservation in the mediterranean Dordrecht W JunkHarrison S Noss R 2017 Endemism hotspots are linked to stable climatic refugia

Annals of Botany 119207ndash214 DOI 101093aobmcw248Henkin Z 2011 Cattle grazing and vegetation management for multiple use of

Mediterranean shrubland in Israel Israel Journal of Ecology amp Evolution 5743ndash51DOI 101560IJEE571-243

Hill J Hostert P Tsiourlis G Kasapidis P Udelhoven T 1998Monitoring 20 years ofintense grazing impact on the Greek island of Crete with earth observation satellitesJournal of Arid Environments 39165ndash178 DOI 101006jare19980392

Hollis GE 1977 A proposed management plan for the internationally important ParcNational de LrsquoIchkeul Tunisia Ecology and Conservation Unit University CollegeLondon Report Series 10 London UK

Hollis GE 1986 The modelling and management of the internationally importantwetland at Garaet El Ichkeul Tunisia IWRB Special Publication No 4 InternationalWaterfowl Research Bureau Slimbridge England

Kirk DA 1983 The plant associations of Djebel Ichkeul northern Tunisia in relation tosite factors and conservation implications MSc thesis University College LondonEngland UK

Laacutezaro A Tscheulin T Devalez J Nakas G Stefanaki A Hanlidou E PetanidouT 2016Moderation is best effects of grazing intensity on plant-flower visitornetworks in Mediterranean communities Ecological Applications 26796ndash807DOI 105061dryadp3c75

Le Flocrsquoh E Boulos L Vela E 2010 Catalogue synonymique commenteacute de la Floretunisie Tunis Reacutepublique Tunisienne Ministeacutere de lrsquoEnvironnement et dudeacuteveloppement durable Banque Nationale de Geacutenes

Le Houerou HN 1981 Impact of man and his animals on Mediterranean vegetation InDi Castri F Goodall DW Specht RL eds Ecosystems of the World II MediterraneanType Shrublands London Elsevier 479ndash522

Legendre P De Caacuteceres M 2013 Beta diversity as the variance of communitydata dissimilarity coefficients and partitioning Ecology Letters 16951ndash963DOI 101111ele12141

Legendre P Gauthier O 2014 Statistical methods for temporal and spacendashtime analysisof community composition data Proceedings of the Royal Society B Biological Sciences281(1778)20132728 DOI 101098rspb20132728

Legendre P Legendre LFJ 2012Numerical ecology 3rd English edition AmsterdamElsevier Science BV xvi + 990

Levin NWatson JEM Joseph LN GranthamHS Hadar L Apel N PerevolotskyA DeMalach N PossinghamHP Kark S 2013 A framework for systematicconservation planning and management of Mediterranean landscapes BiologicalConservation 158371ndash383 DOI 101016jbiocon201208032

Kirk et al (2019) PeerJ DOI 107717peerj7296 2326

Mata J De Nascimento L Fernaacutendez-Lugo S Areacutevalo JR Viera JJ Camacho A BermejoLA 2014 The inefficient planning of goat grazing causes and consequences ThePalmera breed case (Canary Islands) Small Ruminant Research 121125ndash130DOI 101016jsmallrumres201403010

Mazzoleni S Di Pasquale G MulliganM Di Martino FRM 2004 Recent dynamics of theMediterranean vegetation and landscape West Sussex Wiley

McLaughlin BC Ackerly DD Klos PZ Natali J Dawson TE Thompson SE 2017Hydrologic refugia plants and climate change Global Change Biology 232941ndash2961DOI 101111gcb13629

Meacutedail F Queacutezel P 1997Hot-spots analysis for conservation of plant biodiversityin the Mediterranean Basin Annals of the Missouri Botanical Garden 84112ndash127DOI 1023072399957

Miguel-Ayanz AS Garciacutea-Calvo RP Fernaacutendez-Olalla M 2010Wild ungulates vsextensive livestock Looking back to face the future Options Mediteacuterraneennes9227ndash34

Molina-Venegas R Aparicio A Lavergne S Arroyo J 2016How soil and elevationshape local plant biodiversity in a Mediterranean hotspot Biodiversity and Conser-vation 251133ndash1149 DOI 101007s10531-016-1113-y

Naveh Z 1990 Fire in the Mediterraneanmdasha landscape ecological perspective InGoldammer JFJMJ ed Fire in ecosystems dynamics The Hague SPB AcademicPublishing

Nogueacutes-Bravo D ArauacutejoMB Lasanta T Loacutepez-Moreno JI 2008 Climate change inMediterranean Mountains during the 21st Century AMBIO A Journal of the HumanEnvironment 37280ndash285 DOI 1015790044-7447(2008)37[280CCIMMD]20CO2

Nogueacutes-Bravo D Loacutepez-Moreno JI Vicente-Serrano SM 2012 Climate change and itsimpact In Vogiatzakis IN edMediterranean mountain environments HobokenJohn Wiley amp Sons Ltd 185ndash200 DOI 1010029781119941156ch9

Noy-Meir I Oron T 2001 Effects of grazing on geophytes in Mediterranean vegetationJournal of Vegetation Science 12749ndash760 DOI 1023073236862

Oksanen J Guillaume Blanchet F Friendly M Kindt R Legendre P McGlinn DMinchin PR OrsquoHara RB Simpson GL Solymos P HenryM Stevens H SzoecsE Wagner H 2019 vegan community ecology package R package version 25-4Available at httpsCRANR-projectorgpackage=vegan

Olff H Ritchie ME 1998 Effects of herbivores on grassland plant diversity Trends inEcology and Evolution 13261ndash265 DOI 101016S0169-5347(98)01364-0

Olsvig-Whittaker L Frankenberg E Perevolotsky A Ungar ED 2006 Grazing over-grazing and conservation changing concepts and practices in the Negev rangelandsScience et changements planeacutetaires Seacutecheresse 17195ndash199

Osem Y Perevolotsky A Kigel J 2007 Interactive effects of grazing and shrubs onthe annual plant community in semi-arid Mediterranean shrublands Journal ofVegetation Science 18869ndash+ DOI 101111j1654-11032007tb02603x

Kirk et al (2019) PeerJ DOI 107717peerj7296 2426

Paeth H Born K Girmes R Podzun R Jacob D 2009 Regional climate change intropical and Northern Africa due to greenhouse forcing and land use changesJournal of Climate 22114ndash132 DOI 1011752008JCLI23901

Papanastasis VP 1998 Livestock grazing in Mediterranean ecosystems an historicaland policy perspective In Papanastasis VP Peters D eds Ecological basis of livestockgrazing in mediterranean ecosystems 5ndash9 Proceedings of the International Workshopheld in Thessaloniki (Greece) on October (1997) 23ndash25

Papanastasis VP Kyriakakis S Kazakis G 2002 Plant diversity in relation to overgraz-ing and burning in mountain mediterranean ecosystems Journal of MediterraneanEcology 353ndash63

Peres-Neto PR Legendre P Dray S Borcard D 2006 Variation partitioning of speciesdata matrices estimation and comparison of fractions Ecology 872614ndash2625DOI 1018900012-9658(2006)87[2614VPOSDM]20CO2

Perevolotsky A 2005 Livestock grazing and biodiversity conservation in Mediterraneanenvironments the Israeli experience Sustainable Grazing Nutritional Utilization andQuality of Sheep and Goat Products 5651ndash56

Perevolotsky A Seligman NG 1998 Role of grazing in mediterranean rangelandecosystems BioScience 481007ndash1017 DOI 1023071313457

Peterken GL Radford GF 1971 Report on field-trials in Tunisia In IBPCT ProgressReport 36ndash94

Pons A 1981 The history of Mediterranean shrublands In Di Castri F Goodall DWSpecht RL edsMediterranean-type shrublands Amsterdam Elsevier 131ndash138

Queacutezel P 1981 Floristic composition and phytosociological structure of sclerophyllousmatorral around the Mediterranean In Di Castri F Goodall DW Specht RLeds Ecosystems of the World II Mediterranean Type Shrublands London Elsevier107ndash121

RackhamOMoody J 1996 The making of the cretan landscape Manchester ManchesterUniversity Press

Radford E Catullo G DeMontmollin B 2011 Important plant areas of the south and eastmediterranean region priority sites for conservation IUCN Gland Switzerland andMalaga Spain Gland Switzerland and Malaga Spain IUCN VIII + 108 pp

Rundel PW 1998 Landscape disturbance and biodiversity in Mediterranean-type ecosys-tems New York Springer 3ndash22

Rundel PW ArroyoMTK Cowling RM Keeley JE Lamont BB Vargas P 2016Mediterranean biomes evolution of their vegetation floras and climate AnnualReview of Ecology Evolution and Systematics 47383ndash407DOI 101146annurev-ecolsys-121415-032330

Rupprecht D Gilhaus K Houmllzel N 2016 Effects of year-round grazing on the vegetationof nutrient-poor grass- and heathlandsmdashevidence from a large-scale surveyAgriculture Ecosystems and Environment 23416ndash22 DOI 101016jagee201602015

Sardans J Pentildeuelas J 2013 Plant-soil interactions in Mediterranean forest and shrub-lands impacts of climatic change Plant and Soil 3651ndash33DOI 101007s11104-013-1591-6

Kirk et al (2019) PeerJ DOI 107717peerj7296 2526

Segoli M Ungar ED Giladi I Arnon A ShachakM 2012 Untangling the positive andnegative effects of shrubs on herbaceous vegetation in drylands Landscape Ecology27899ndash910 DOI 101007s10980-012-9736-1

SternbergM ShoshanyM 2001 Influence of slope aspect on Mediterranean woodyformations comparison of a semiarid and an arid site in Israel Ecological Research16335ndash345 DOI 101046j1440-1703200100393

Tomaselli R 1977 The degradation of the Mediterranean maquis Ambio 6356ndash362Tomaselli R 1981 Main physionomic types and geographic distribution of shrub

systems related to Mediterranean climates In Di Castri F Goodall DW Specht RLeds Ecosystems of the World II Mediterranean Type Shrublands London Elsevier95ndash106

Toumlroumlk P Penksza K Toacuteth E Kelemen A Sonkoly J Toacutethmeacutereacutesz B 2018 Vegetationtype and grazing intensity jointly shape grazing effects on grassland biodiversityEcology and Evolution 810326ndash10335 DOI 101002ece34508

Toumlroumlk P Valkoacute O Deaacutek B Kelemen A Toacuteth E Toacutethmeacutereacutesz B 2016Managing forspecies composition or diversity Pastoral and free grazing systems in alkali steppesAgriculture Ecosystems and Environment 23423ndash30DOI 101016jagee201601010

Toacuteth E Deaacutek B Valkoacute O Kelemen A Migleacutecz T Toacutethmeacutereacutesz B Toumlroumlk P 2018Livestock type is more crucial than grazing intensity traditional cattle and sheepgrazing in short-grass steppes Land Degradation and Development 29231ndash239DOI 101002ldr2514

Tramblay Y El Adlouni S Servat E 2013 Trends and variability in extreme precipi-tation indices over maghreb countries Natural Hazards and Earth System Sciences133235ndash3248 DOI 105194nhess-13-3235-2013

Underwood EC Viers JH Klausmeyer KR Cox RL ShawMR 2009 Threats andbiodiversity in the mediterranean biome Diversity and Distributions 15188ndash197DOI 101111j1472-4642200800518x

UNEP-WCMC IUCN 2017 Ichkeul National Park Datasheets for natural and mixedworld heritage sites Available at httpworld-heritage-datasheetsunep-wcmcorg

Verduacute JR CrespoMB Galante E 2000 Conservation strategy of a nature reserve inMediterranean ecossystems the effect of protection from grazing on biodiversityBiodiversity amp Conservation 91707ndash1721 DOI 101023A1026506725251

Vogiatzakis IN Mannion AM Griffiths GH 2006Mediterranean ecosystemsproblems and tools for conservation Progress in Physical Geography 30175ndash200DOI 1011910309133306pp472ra

Wilson KA Underwood EC Morrison SA Klausmeyer KR MurdochWW Reyers BWardell-Johnson G Marquet PA Rundel PWMcBride MF Pressey RL BodeM Hoekstra JM Andelman S Looker M Rondinini C Kareiva P ShawMRPossinghamHP 2007 Conserving biodiversity efficiently what to do where andwhen PLOS Biology 5(9)e223 DOI 101371journalpbio0050223

Kirk et al (2019) PeerJ DOI 107717peerj7296 2626