CTFS-ForestGEO ARTHROPOD INITIATIVE 2008-2015: Monitoring barcodes? Long-term monitoring of tropical arthropods

Yves Basset, Smithsonian Tropical Research Institute, [email protected] Currently: 61 sites in 24 countries 6 million of trees monitored, representing 10,000 species 10 science initiatives: Arthropod monitoring: 9 sites CTFS-ForestGEO ARTHROPOD INITIATIVE Aims: to monitor key insect assemblages over the long-term at CTFS sites and to study insect-plant interactions across the CTFS network

At each CTFS site, 3 phases: - baseline survey to identify common species - monitoring (modeled on baseline survey) - interaction studies (different set of protocols)

Focus on a priority set of assemblages chosen for their • ecological relevance • taxonomic tractability • ease of sampling Monitoring Interactions only Priority assemblages

Litter ants: key organisms in tropical forests and often key predators [Formicidae]

Selected moths and butterflies: caterpillars leaf-chewers, adults often pollinators [Rhopalocera, Geometridae, Arctiinae & Pyraloidea]

Bees: important pollinators of many tropical trees [Apidae Euglossini]

Termites: important decomposers in tropical forests [Isoptera]

Tephritid fruit-flies: seed (fruit) predators [Tephritidae]

Seed predators: important influence on fruit/seed survival (whole guild) [Varia] Full suite of 15 taxa studied at BCI, Panama Methods: baseline survey & monitoring

Litter ants: extraction from litter with Winkler

Bees: attraction to chemical baits, (only Neotropical sites)

Tephritid fruit-flies: baited McPhail traps (not in the Neotropics)

Moths and other taxa: light traps

Termites: light traps & hand search in quadrats

Butterflies: walking transects INTERACTION STUDIES

• KHC (2010): Effects of litter composition on ants

• BCI (2010), KHC (2013), WAN (2013):

Insect seed predation: quantitative food webs TAXONOMY

• Local reference collections often managed

by “parataxonomists”

• DNA barcoding

• Collaborating experts (in-country or abroad)

But taxonomic impediment high in the tropics! Example: BARRO COLORADO ISLAND PANAMA (BCI) Progress in DNA barcoding since 2009

STATUS 2008: baseline survey 2009-2015: on-going monitoring (7th year) Collections: 36,388 pinned specimens; 1,864 spp. 68% of spp. with Barcode Index Numbers (BINs)

Four full-time research assistants, based at STRI MAIN USE OF DNA BARCODING

• Used to refine delineation of insect morphospecies

and identifications of common, abundant species,

likely to be tractable in the long-term

• Currently > 10,000 sequences:

Panama (7,414), Thailand (1,053), Papua New Guinea (1,567)

• Some of these areas have extensive barcode librairies

(PNG 25,000 barcodes for Lepidoptera), others not (Thailand):

approach for barcoding needs to be different for each site INTEREST OF DNA BARCODING

• Refinement of reference collections

BCI: Arctiinae, Geometridae, etc: best collections in Panama

• Baseline and ecological surveys (Guanacaste: Janzen et al.)

Of interest when monitoring protocols are similar to the most efficient

collecting methods for focal taxa. For example: Saturniidae collected at light on BCI:

40 (a)

30

20

10

Mean cumulative no. of species of no. cumulative Mean

0 0 200 400 600 800 Individuals

Species accumulation curve OK Documentation of local fauna Representative tree INTEREST OF DNA BARCODING

DNA barcoding allows:

• Matching sexes in case of sexual dimorphism

• Identification of caterpillars, pupae and reared parasitoids

• Matching casts of social insects (sometimes across different protocols)

Cornitotermes walkeri BOLD:AAP9751

Alate, Soldier, light trap termite transect INTEREST OF DNA BARCODING

DNA barcoding allows monitoring functional groups important to the forest ecosystem but not well known taxonomically

Flatidae (sap-sucking bugs): 10% of BOLD sequences from BCI

300 FLATIDAE, BCI 250 raw data

200 2009 2015

150

100 No. of individuals No. of

50

0 2 3 4 5 8 9 10 11 14 15 16 17 20 21 22 23 26 27 28 29 32 33 34 35 38 39 40 41 Light trap survey RESULTS: SELECTED EXAMPLES

• Yearly results: annual indices (butterflies & ants)

• Population dynamics (saturniid moths and butterflies)

• Yearly changes in assemblages (several taxa)

Immediate significance vs. Interpretation of long chronosequences Annual indices, BCI, Year 2011

50 All butterflies: 197 ± 9.8, 5.0% 7.1 precision, % 45 40 60,000 insects collected: 35 10.7 17,000 focal individuals (910 spp.) 30

25 19.2 14.3

20 For 56 spp. we can estimate annual indices 15 with good precision 17.2 10 17.9 15.7 5 56 spp. = 6% of total spp. but 0 Pierella luna Itaballia Detritivora Itaballia Cithaerias Mesosemia Phanus Mean Mean ind. and per site s.e. censued (annual index) 55% of total abundance of focal taxa pandosia hermodora demophile pireta hesperina marshalli

0.9 All ants: 1.00 Annual indices: 0.8 Non-social insects: mean per site (n=10) 0.7 Precision = s.e./mean 0.6

0.5 (< 20% very good, economic entomology)

0.4 0.3 Social insects: occurrence in samples, Occurence and 95% c.l. Occurence 0.2 transects or quadrats 0.1 Precision = 95% c.l. on occurrence data, 0 assuming a binomial distribution

3 1

rugiceps

Pheidole

ruidum

Pheidole

Pyramica gundlachi

multispina

minutus

Ectatomma

Hypoponera

Wasmannia

distinguenda

auropunctata

Cyphomyrmex

Solenopsis sp. Solenopsis sp.

Pyramica zeteki Annual indices

Immediate significance:

• Common spp. can be monitored with relative precision, even in tropical rainforests

• Few long-term monitoring programs in the tropics (butterflies: 10-11 years: Leidner et al. 2010, Grøtan et al. 2012)

• Indices for social insects need to be reported differently than for non-social insects (refinements needed for social insects; geometric mean for non-social insects)

• BCI: all common spp. (with annual indices) have BINs Population dynamics, BCI, 2009-2014, saturniid moths

Rhescyntis 2011 Hylesia sp. 2YB 2014 20 2 20 4.5 All Saturniidae 2011 hippodamia (LT) Time-series models 200 18 18 n.s. (ET) 2013 4 180 All saturniids 2012 16 (TRIM): 16 3.5 160 1.5 14 14 2009 3 (1) No time effects 140 2013 12 12 2010 2.5 120 2009 2010 (2) Linear trend 10 1 10 2011 100 2 8 2009 2012 2013 8 (3) Effect for each time 80 2012 1.5 6 2010 6 60 2014 1 point 0.5 4 4 40 2014 2 0.5 2 (separate parameters 20 0 0 0 0 0 0 10 20 30 40 50 60 70 for each time-point) 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70

Periphoba 2011 Automeris 2011 25 5 30 Automeris 2011 sp. 1YB fieldi 20 3 zuganaDHJ03 (ET) 4.5 n.s. 18 (ET) 25 20 4 2.5 16 3.5 20 14 2009 2 15 3 12 2010 2012 2.5 15 10 1.5 10 2 2010 8 2012 2009 10 2009 1 1.5 2012 2013 2014 6 2010 2013 2014 2013 5 1 5 4 2014 0.5 0.5 2 0 0 0 0 0 0 10 20 30 40 50 60 70 collected per survey per collected 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Pseudodirphia 2011 18 7 Automeris eupanamensis Pseudoautomeris 10 1.5 30 2011 6.5 salmonea vanschaycki 16 (ET) 6 n.s. 9 (NT) 14 5.5 25 8 5 2009 2010 2011 2012 2013 2014 12

7 4.5index Annual 20 1 6 10 4 2012 3.5 15 5 8 2014 3 4 2.5 6 10 0.5 2009 2010 2 2010 3 4 2012 2013 1.5 2 2013 5 1 2 2009 2014 1 0.5 0 0 0 0 0 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Automeris 2011 Oxytenis 14 6 Oxytenis 7 2013 6.5 belti 8 1.5 sp. 1YB naemia 6 (ET) 5.5 (ET) 12 (NT) 6 5 7 5.5 4.5 5 10 6 2009 2010 2011 2012 2013 2014 5 Number of individuals individuals of Number 4 2010 4.5 2014 1 4 Model 8 3.5 5 4 3.5 3 2014 4 3 6 2010 2012 2.5 3 2.5 2 3 2012 4 2009 0.5 2 2 1.5 2013 2 1.5 Raw data 1 2009 2011 2 1 1 1 0.5 0.5 0 0 0 0 0 0 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 Months since January 2009 Population dynamics, BCI, 2009-2014, butterflies

Exclusive feeders on: 180 3 Herbs (49 spp.): each time point 2012 160 Trees (35 spp.): model n.s. Liana (32 spp.): model n.s. 2011 2013 2.5 140

120 2 2010 2014 100 1.5 80 2009

60 1 indice Annual

Number of individuals of Number 40 0.5 20

0 0 0 10 20 30 40 50 60 70 Months since January 2009

Different feeding guilds, moderate increase for herb feeders since 2009 (p<0.01) Population dynamics

Immediate significance:

• We can detect significant (short-term) trends

• Nearly a quarter of species show significant changes with time (different groups tested)

• Population dynamics may depend on the ecology of species

• Species identity is crucial, therefore BINs are essential Changes in assemblages, BCI, 2009-2014

Butterflies 0.8 2014 Geometridae 1 2009 283 species 2013 254 species 0.6 0.8 Transects Light traps 0.6 0.4 2010 Matrices Spp. x Years 2011 0.4 0.2 NDMS 0.2

0 0 Different protocols, different taxa -2 -1.5 -1 -0.5 0 0.5 1 1.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 -0.2 2009 -0.2 2013 & 2014 6 out of 16 assemblages 2010 -0.4 -0.4

2012 -0.6 (different guilds)

-0.6 2011 -0.8 2012 Stress= 0.00 Stress= 0.00046 -0.8 -1

Euglossine bees 1 Flatidae 0.8 Mantel tests, p-values: 24 species 0.8 2009 & 2011 2010 27 species 0.6 Baits 0.6 Light traps 0.4 2013 Geometridae Bees Flatidae Ants Termites 0.4 0.2 2013 & 2014 Butterflies 0.152 0.482 0.815 0.044 0.617 0.2 2014 0 -2 -1.5 -1 -0.5 0 0.5 Geometridae1 *** 0.240 0.593 0.432 0.043 -0.2 0 2012 -1.5 -1 -0.5 0 0.5 1 1.5 2 Bees *** 0.094 0.987 0.221 -0.4 -0.2 2009 Flatidae *** 0.989 0.210 -0.6 2011 -0.4 -0.8 Ants *** 0.849 -0.6 -1 -0.8 2012 -1.2 2010 Stress= 0.035 -1 Stress= 0.00040 -1.4

2012 0.8 Termites 1.5 16 species 0.6 2014 Quadrats 2012 2011 1 0.4 Trajectories are mostly independent

2011 0.2 0.5 0 2010 -1.5 -1 -0.5 0 0.5 1 1.5 -0.2 2010 Directional changes are few 0 -0.4 2013 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

-0.6 2013 -0.5

-0.8 2014 -1 Formicidae 2009 -1 2009 -1.2 124 species Stress= 0.00049 Stress= 0.074 -1.4 Winkler -1.5 Changes in assemblages, BCI, 2009-2014

Geometridae 1 2009 Flatidae 0.8 254 species 0.8 27 species 2009 & 2011 Light traps 0.6 Light traps 0.6 0.4 2010 Matrices Spp. x Years 2011 0.4 0.2 2013 & 2014 NDMS 0.2 0 -2 -1.5 -1 -0.5 0 0.5 1 -0.2 0 2012 Same protocol (light trap), different taxa -2 -1.5 -1 -0.5 0 0.5 1 1.5 -0.4 -0.2 2013 & 2014 6 out of 12 taxa (different guilds) -0.6 -0.4

-0.8 -0.6 -1 -0.8 2012 -1.2 2010 Stress= 0.00046 -1 Stress= 0.00040 -1.4

Arctiinae 0.8 Passalidae 1.5 Mantel tests, p-values: 168 species 11 species 2014 Light traps 0.6 2014 2013 Light traps Flatidae Arctiinae Passalidae Reduviidae Termites 1 0.4 Geometridae 0.593 0.834 0.320 0.135 0.151 2011 Flatidae *** 0.225 0.981 0.635 0.875 0.2 0.5 2013 Arctiinae *** 0.665 0.624 0.189 0 -1.5 -1 -0.5 0 0.5 1 1.5 2 Passalidae *** 0.320 0.306

-0.2 0 Reduviidae *** 0.103 -1 -0.5 0 0.5 1 1.5 2012 2011 -0.4 2012 -0.5 2009 2009 -0.6

2010 2010 Stress= 0.00030 -0.8 Stress= 0.0146 -1

Reduviidae 0.8 Termites (alates) 0.4 Trajectories are independent 44 species 2009 30 species 2010, 2011, 2013 & 2014 0.6 Light traps Light traps 0.2

0.4 0 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 Directional changes are few -0.2 0.2 2010 & 2014 2013 2012 -0.4 0 -2 -1.5 -1 -0.5 0 0.5 1 1.5 -0.6

-0.2

-0.8

-0.4 2012 2011 2009 -1

Stress= 0.00050 -0.6 Stress= 0.00049 -1.2 Changes in assemblages, BCI, 2009-2014 All species: Common species: Species with BINs:

0.6 0.8 Reduviidae 0.8 Reduviidae Reduviidae 2009 All species 2009 2012 Common species Species with BINs 0.6 n = 44 0.6 0.4 n = 16 n = 23

0.4 0.2 0.4 2011 2010 & 2014 2014 2013 0.2 0 0.2 2010 & 2014 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2013 2013 0 -0.2 0 -2 -1.5 -1 -0.5 0 0.5 1 1.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5

-0.2 -0.4 -0.2 2011

2012 -0.4 -0.6 -0.4 2012 2011

2009 2010 Stress= 0.00050 Stress= 0.00597 Stress= 0.00050 -0.6 -0.8 -0.6

0.8 0.8 0.8 Crambidae Crambidae 2012 Crambidae 2012 2012 0.6 All species 0.6 Common species 0.6 Species with BINs 2009 n = 302 n = 109 n = 215 0.4 2009 0.4 2009 0.4 2013 2013 2013 0.2 0.2 0.2 2014 2014 2014 0 0 0 -1.5 -1 -0.5 0 0.5 1 1.5 2 -1.5 -1 -0.5 0 0.5 1 1.5 2 -1.5 -1 -0.5 0 0.5 1 1.5 2

-0.2 -0.2 -0.2 2010 2010 2010 -0.4 -0.4 -0.4

-0.6 -0.6 -0.6

-0.8 -0.8 -0.8

-1 -1 -1 2011 2011 2011 Stress= 0.00159 Stress= 0.00085 Stress= 0.00159 -1.2 -1.2 -1.2 Mantel tests, p-values: Matrices Spp. x Years Trajectories similar, Common BINs NDMS especially for species-rich Reduviidae, all 0.0012 0.0017 Species-poor and -rich assemblages Reduviidae, common *** 0.0013 assemblages Common BINs Crambidae, all 0.0017 0.0016 Crambidae, common *** 0.0014 Changes in assemblages

Immediate significance:

• Trajectories appear largely independent, even when considering a particular protocol: need to monitor an array of taxa

• Directional changes appear also to be few

• Species-rich and poorly known assemblages: their yearly changes may be estimated with common species, especially if they have BINs Challenge: publish or perish • Challenge: initial wait for quality data may be long (long chronosequences)

• Remedy: (a) comparison of insect data with other sites (b) faunistical surveys (if reasonably complete) (c) mine barcoding data; data release (d) mine historical data Example: mining historical data The butterflies of Barro Colorado Island, Panama: local extinction since the 1930’s

Compilation of butterfly records on BCI 1923-2013: 601 species Old period: 1923-1943, 267 species (Huntington, 1932) Recent period: 1993-2013, 373 species Barro Colorado Island history:

• ca 1880: 45% old growth forest, rest shifting agriculture

• ca 1910: became a 1542 ha island after damming the Chagres river, to create the Panama canal

• ca 1920-1923: 96.7% forested, rest clearing and small agricultural areas

• today: 100% forested

Main vegetational changes: fragmentation/disparition of herbs/crops Actualized list with BINs BINs for species collected only during the old period mined elsewhere from BCI Historical knowledge of BCI butterflies

New molecular methods, cryptic species

Comparison of old vs recent data: sampling effort old << recent: 23 abundant species were not recorded during the recent period We cannot discount misidentification of species but comparison of BINs reduces (but do not avoid) this possibility Phylogenetic tree for BCI butterflies, obtained with barcodes and constrained to match inter-familial relationships

Extinct species (in red or orange) do not cluster within the tree: phylogenetic signal weak Host growth form and wing size were significant predictors of probability of extinction, independently of phylogenetic relations

(f) Size 100% 100% (c) Host growth form

75% 75% >60mm 51-60mm Shrub or tree 41-50mm 50% 31-40mm 50% Liana 21-30mm Herb 11-20mm 0-10mm 25% 25% Extinct spp. feed Extinct spp. smaller

on herbs 0% (poor dispersers) 0% 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 Cryptic Common Settler Extinct Cryptic Common Settler Extinct The 23 missing species (6% of recent records) may be extinct locally Most likely candidates for extinction were small hesperiids feeding on herbs. Poor dispersers may find difficult to locate declining herbs

During the same period many more (50-60, 21%) bird species were lost from BCI small reserves still effective for invertebrate conservation

Locally extinct

BOLD:AAE3730 BOLD:AAB0089 The future of monitoring: meta-barcoding Example: Global Malaise Program (Biodiversity Institute of Ontario)

Abundance data better than presence-absence data for monitoring

Calibrations and test of the methods (with respect to seasonality) in species-rich tropical rainforests, where arthropod BINs are already available: ForestGEO sites...

BCI, Panama Conclusions

• Nascent mini-network focused on monitoring arthropods in tropical rainforests

• With adequate protocols (including DNA barcoding) common species of tropical insects can be precisely monitored in the long-term

• Because most insects have short generation times in the tropics, it may be possible to develop efficient warning systems that can yield results within 5-10 years (equivalent to ca. 40-80 insect generations) Invitation

• To join the mini-network (site; development of databasing and statistical tools)

• To collaborate to the analysis of monitoring and barcoding data (currently 0.6 M specimens, 6,000 species & 10,000 sequences)

Contact: [email protected] Acknowledgments

Biodiversity Institute of Ontario, Genome Canada, Smithsonian Institution, US NSF, ForestGEO-CTFS, Smithsonian Tropical Research Institute, Czech GACR, SENACYT, Universidad de Panama

S. Davies, S. Miller, H. Barrios, S. Bunyavejchewin, K. Somboon, W. Sackchoowong, P. Kongnoo, V. Novotny, G. Weiblen, C. Dahl, P. Hebert, A. Borisenko, R. Eastwood, D. Lohman, L. Sam, M. Leponce, D. Donoso, D. Roubik, S. Gripenberg, O. Lewis, S.J. Wright, T. Bonebrake, A. Nakamura